Programming Tools User Guide

May 2014
 Copyright
Copyright © 2014 Lattice Semiconductor Corporation.

This document may not, in whole or part, be copied, photocopied, reproduced,

translated, or reduced to any electronic medium or machine-readable form without
prior written consent from Lattice Semiconductor Corporation.

Trademarks
Lattice Semiconductor Corporation, L Lattice Semiconductor Corporation (logo), L
(stylized), L (design), Lattice (design), LSC, CleanClock, Custom Mobile Device,
DiePlus, E2CMOS, ECP5, Extreme Performance, FlashBAK, FlexiClock, flexiFLASH,
flexiMAC, flexiPCS, FreedomChip, GAL, GDX, Generic Array Logic, HDL Explorer,
iCE Dice, iCE40, iCE65, iCEblink, iCEcable, iCEchip, iCEcube, iCEcube2, iCEman,
iCEprog, iCEsab, iCEsocket, IPexpress, ISP, ispATE, ispClock, ispDOWNLOAD,
ispGAL, ispGDS, ispGDX, ispGDX2, ispGDXV, ispGENERATOR, ispJTAG, ispLEVER,
ispLeverCORE, ispLSI, ispMACH, ispPAC, ispTRACY, ispTURBO, ispVIRTUAL
MACHINE, ispVM, ispXP, ispXPGA, ispXPLD, Lattice Diamond, LatticeCORE,
LatticeEC, LatticeECP, LatticeECP-DSP, LatticeECP2, LatticeECP2M, LatticeECP3,
LatticeECP4, LatticeMico, LatticeMico8, LatticeMico32, LatticeSC, LatticeSCM,
LatticeXP, LatticeXP2, MACH, MachXO, MachXO2, MachXO3, MACO, mobileFPGA,
ORCA, PAC, PAC-Designer, PAL, Performance Analyst, Platform Manager,
ProcessorPM, PURESPEED, Reveal, SensorExtender, SiliconBlue, Silicon Forest,
Speedlocked, Speed Locking, SuperBIG, SuperCOOL, SuperFAST, SuperWIDE,
sysCLOCK, sysCONFIG, sysDSP, sysHSI, sysI/O, sysMEM, The Simple Machine for
Complex Design, TraceID, TransFR, UltraMOS, and specific product designations are
either registered trademarks or trademarks of Lattice Semiconductor Corporation or its
subsidiaries in the United States and/or other countries. ISP, Bringing the Best
Together, and More of the Best are service marks of Lattice Semiconductor
Corporation.

Other product names used in this publication are for identification purposes only and
may be trademarks of their respective companies.

Disclaimers
NO WARRANTIES: THE INFORMATION PROVIDED IN THIS DOCUMENT IS “AS IS”
WITHOUT ANY EXPRESS OR IMPLIED WARRANTY OF ANY KIND INCLUDING
WARRANTIES OF ACCURACY, COMPLETENESS, MERCHANTABILITY,
NONINFRINGEMENT OF INTELLECTUAL PROPERTY, OR FITNESS FOR ANY
PARTICULAR PURPOSE. IN NO EVENT WILL LATTICE SEMICONDUCTOR
CORPORATION (LSC) OR ITS SUPPLIERS BE LIABLE FOR ANY DAMAGES
WHATSOEVER (WHETHER DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR
CONSEQUENTIAL, INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS OF
PROFITS, BUSINESS INTERRUPTION, OR LOSS OF INFORMATION) ARISING
OUT OF THE USE OF OR INABILITY TO USE THE INFORMATION PROVIDED IN
THIS DOCUMENT, EVEN IF LSC HAS BEEN ADVISED OF THE POSSIBILITY OF
SUCH DAMAGES. BECAUSE SOME JURISDICTIONS PROHIBIT THE EXCLUSION
OR LIMITATION OF CERTAIN LIABILITY, SOME OF THE ABOVE LIMITATIONS MAY
NOT APPLY TO YOU.

LSC may make changes to these materials, specifications, or information, or to the

products described herein, at any time without notice. LSC makes no commitment to
update this documentation. LSC reserves the right to discontinue any product or
service without notice and assumes no obligation to correct any errors contained
herein or to advise any user of this document of any correction if such be made. LSC
recommends its customers obtain the latest version of the relevant information to
establish, before ordering, that the information being relied upon is current.

ii Programming Tools User Guide

 Type Conventions Used in This Document
Convention Meaning or Use

Bold Items in the user interface that you select or click. Text that you type
into the user interface.

<Italic> Variables in commands, code syntax, and path names.

Ctrl+L Press the two keys at the same time.

Courier Code examples. Messages, reports, and prompts from the software.

... Omitted material in a line of code.

. Omitted lines in code and report examples.
.
.

[ ] Optional items in syntax descriptions. In bus specifications, the

brackets are required.
( ) Grouped items in syntax descriptions.
{ } Repeatable items in syntax descriptions.

| A choice between items in syntax descriptions.

Programming Tools User Guide iii

iv Programming Tools User Guide
 Contents

Chapter 1 Programming Tools Description 1

Programmer 1
Deployment Tool 1
Programming File Utility 2
Download Debugger 2
Model 300 2
Embedded Flow 2
Driver Installation 3

Chapter 2 Programmer Overview 5

Usage and flow 6
Programmer Design Flow 6
Programming Basics 6
In-System Programming 8
Programming Algorithm Basics 8
Programming Times 9
USERCODE 10
Programming Hardware 11
Programming Software 11
Embedded Programming 12
FPGA Configuration 12
Serial Peripheral Interface Flash 12

Chapter 3 Deployment Tool Overview 13

Deployment Tool Function Types 15
Output File Types 15
File Conversion Output File Types 15
Tester Output File Types 16
Embedded System Output File Types 17
External Memory Output File Types 18

Programming Tools User Guide v

CONTENTS

Chapter 4 Embedded Flow Overview 19

JTAG Full VME Embedded 30
VME File Format 30
JTAG Full VME Embedded Flow 33
JTAG Full VME Embedded System Memory 34
JTAG Full VME Embedded Basic Operation 34
VME Source Code 35
JTAG Full VME Embedded Programming Engine 36
RAM Size Requirement for VME 37
ROM Size Requirement for JTAG Full VME Embedded 39
JTAG Full VME Embedded Required User Changes 39
Program Memory Requirement 41
Program Memory Allocation 41
Sample Program Size 42
VME File Size 43
Using JTAG Full VME Embedded 51
Generating VME Files 51
Testing VME Files 52
Converting an SVF File to VME File 52
Choosing the File-Based or EPROM-Based Version 52
Customizing for the Target Platform 53
Advanced Issues 53
EPROM-based JTAG Full VME Embedded User Flow 53
Programming Engine Flow 55
VME Byte Codes 69
Unsupported SVF Syntax 72
JTAG Slim VME Embedded 73
JTAG Slim VME Embedded Source Code 74
Using the PC-based JTAG Slim VME Embedded 75
Using the 8051-based JTAG Slim VME Embedded 76
VME Algorithm Format 76
VME Data Format 79
VME Required User Changes 80
Program Memory Requirement 81
Program Memory Allocation 82
Sample Program Size 83
VME File Size 83
Generating JTAG Slim VME Embedded Files 84
JTAG Slim VME Embedded Source Code 85
8051 JTAG Slim VME Embedded User Flow 86
Programming Engine Flow 87
VME Algorithm and Format 98
Slave SPI Embedded 100
Requirements 100
Slave SPI Embedded Algorithm Format 101
Slave SPI Embedded Data Format 102
Generating Slave SPI Embedded Files 103
Modifications 103
Usage 109
Return Codes from Slave SPI Embedded 110
I2C Embedded 110
Masters and Slaves 111
MachXO2 or MachXO3L Slave I2C Programming 111

vi Programming Tools User Guide

 CONTENTS

Using the PC-based I2C Embedded Programming 112

Using the 8051-based I2C Programming 113
I2C Algorithm Format 114
I2C Data Format 115
I2C Embedded Programming Required User Changes 116
Generating I2C Files 117
sysCONFIG Embedded 120
sysCONFIG Embedded Flow 121
sysCONFIG Embedded Bitstream Format 122
sysCONFIG Embedded Bitstream Structure 123
sysCONFIG Embedded Basic Operation 124
sysCONFIG Embedded Source Code 125
sysCONFIG Embedded Engine 126
Sample Program Size 128
Generating a sysCONFIG Embedded Bitstream 129
sysCONFIG SPI Port AC Parameters 130
sysCONFIG Interface 133

Programming Tools User Guide vii

CONTENTS

viii Programming Tools User Guide

 Chapter 1

Programming Tools Description

This user guide is intended to provide users with basic information, and
references on where to find more detailed information, to assist in configuring
and programming Lattice devices using Diamond Programmer and related
tools including Deployment Tool, Programming File Utility, Download
Debugger, and Model 300 Programmer.

Programmer
Diamond Programmer is a system for programming devices. The software
supports both serial and concurrent (turbo) programming of Lattice devices
using PC and Linux environments. The tool also supports embedded
microprocessor programming. Refer to “Programmer Overview” on page 5.

Deployment Tool
Deployment Tool is a stand-alone tool available from the Diamond
Accessories. The Deployment Tool graphical user interface (GUI) is separate
from the Diamond design environment.
The Deployment Tool allows you to generate files for deployment for single
devices, a chain of devices, and can also convert data files to other formats
and use the data files it produces to generate other data file formats. Refer to
“Deployment Tool Overview” on page 13.

Programming Tools User Guide 1

PROGRAMMING TOOLS DESCRIPTION : Programming File Utility

Programming File Utility

Programming File Utility allows the viewing and editing of different format
programming files.The Programming File Utility is a stand-alone tool that
allows you to view and compare data files. When comparing two data files, the
software generates an output (.out) file with the differences highlighted in red.

Detailed information and procedures on how to use the Programming File

Utility are contained in the “Using Programming File Utility” section of the
Lattice Diamond online help or in the stand-alone Programming File Utility
online help.

Download Debugger
Download Debugger is a utility for debugging programming for debugging
Serial Vector Format (SVF) files, Standard Test And Programming Language
(STAPL) files, and Lattice Embedded (VME) files. Download Debugger allows
you to program a device, and edit, debug, and trace the process of SVF,
STAPL, and VME files. Download Debugger also allows you to create, edit, or
view a VME file in hexadecimal format.

Detailed information and procedures on how to use the Download Debugger

are contained in the “Debugging SVF, STAPL, and VME Files” section of the
Lattice Diamond online help or in the stand-alone Download Debugger online
help.

Model 300
The Model 300 Programmer is a simple engineering device programmer that
allows you to perform single-device programming directly from a PC or Linux
environment. The Model 300 Programmer software and hardware support all
JTAG devices produced by Lattice, with device Vcc of 1.8, 2.5, 3.3, and 5.0V.

Detailed information and procedures on how to use the Model 300 hardware
and software are contained in the “Using the Model 300 Programmer” section
in the Lattice Diamond online help or in the stand-alone Model 300 online
help.

Embedded Flow
Programming flow using a processor to read the contents of a stored
programming file and programming the FPGA. Lattice provides the option to
generate several different file formats for different embedded target options.
Refer to “Embedded Flow Overview” on page 19.

2 Programming Tools User Guide

 PROGRAMMING TOOLS DESCRIPTION : Driver Installation

Driver Installation
A utility is available for installing programming drivers after the Diamond or
Programmer software has been installed. Refer to the topic “Installing/
Uninstalling Parallel Port Driver and USB Driver on a PC” in the Lattice
Diamond online help or in the stand-alone Programmer online help.

Programming Tools User Guide 3

PROGRAMMING TOOLS DESCRIPTION : Driver Installation

4 Programming Tools User Guide

 Chapter 2

Programmer Overview

Diamond Programmer is integrated into the Lattice Diamond software, and is

also available as a standalone tool. Both versions come with online help that
provides essential information for using the tool to program Lattice devices.
The two versions are virtually identical, except for the following differences:

Integrated Mode When Programmer is used in integrated mode, it checks

that the bitstream (.bit) or JEDEC file (.jed) selected in the Diamond project
(.xcf) matches the file generated by the project. If the file does not match
when the file is originally selected, a warning dialog informs the user that the
file is not the one generated by the project. If an existing .xcf is opened, and
the file is not the correct one generated by the project, the file section for that
device is highlighted in red in the Diamond graphical user interface.

Standalone Mode The stand-alone Programmer supports mature Lattice

devices. A “Mature Device” license feature (LSC_PROGRAMMER_MATURE)
is required in your license file to enable these devices. Without the license,
Programmer will list the mature devices in the Device Selection and will be
able to Scan the devices, but will not be able to program mature devices. The
only operation available is Bypass. With the Mature Device license, full
support is available.

Diamond Integrated Programmer (Programmer integrated into the Lattice

Diamond software) will list the mature devices in the Device Selection and will
be able to Scan the devices, but will not be able to program mature devices,
even with the Mature Device license. The only operation available is Bypass.

Full support is only available when running Diamond Programmer as a stand-

alone tool.

Programming Tools User Guide 5

PROGRAMMER OVERVIEW : Usage and flow

Usage and flow

Programming is the process changing the state of a non-volatile
programmable element (such as embedded Flash and external SPI Flash
devices) by downloading data from JEDEC, bitstream, or hex files transmitted
to the download cable through the host computer’s serial communications
port; from an embedded system; or from a third party programmer.

Configuring is the process of changing the state of a volatile programmable

element (such as SRAM in the FPGA).

Before programming or configuring an FPGA, you need to create and verify

your design, and then generate data files. To download a data file into the
target device, use the Programmer tool which is integrated into the Diamond
software, and also available in a standalone version.

Programmer Design Flow

The design flow for creating and downloading a design chain is the same,
whether you select the devices and settings from the software or use an
existing chain configuration file.
1. Create a design and compile it to a JEDEC, ISC, hex, or bitstream data
file.
2. Using Programmer, create a new chain configuration or open an existing
one.
3. Add devices to the chain, and select the data file and operation for each.
4. Arrange the order of the devices in the chain and edit the options for each
device as needed.
5. Program the daisy-chained devices using the Program toolbar command.

Figure 1 on page 7 describes the Lattice programming process.

Programming Basics
To successfully program devices in-system, there are a few simple
requirements that must first be met. The first of these requirements is that the
devices on the board must be correctly connected into an 1149.1 scan chain.
This scan chain can be used for either programming or testing the board.

To program using Programmer a description of the scan chain must be

developed. This description, called a chain file, contains basic information
about all of the devices in the chain. For the Lattice devices, it includes the
device type, the operation to be performed, and the data file, if required by the
operation. Additional information in the chain file can include the state of the I/
O pins during programming, along with security requirements. If the chain
includes non-Lattice devices, the instruction register length is required for
these devices. The instruction register length can be found from the BSDL file
or the SVF file for the device.

6 Programming Tools User Guide

 PROGRAMMER OVERVIEW : Usage and flow

Figure 1: Programming Design Flow

Another requirement for successful programming is thoughtful board design.

The signals used in a scan chain (TCK, TMS, TDI, and TDO) rarely operate
as fast as the data path signals on the board. However, correct board layout
methodologies, such as buffering for large chains and termination resistors,
are required to ensure trouble-free operation. Some Lattice devices have
some additional pins (TRST, ispEN, PROGRAMN, INITN, DONE, SLEEPN,
and TOE) that can affect boundary scan programming and test if not taken
care of properly.

After all of these requirements have been met, it should be relatively

straightforward to program any number of devices on a board. You can

Programming Tools User Guide 7

PROGRAMMER OVERVIEW : Usage and flow

program the devices using a PC or Linux system and a board test system
connected by one of the following cables:
 A Lattice parallel port cable with the 8-pin AMP connector or 10-pin
JEDEC connector
 A Lattice USB port cable

In-System Programming
After you have compiled your design to a data file (JEDEC, hex, or bitstream)
and device programming is necessary, you must serially shift the fuse map (a
fuse map file is a design file that has the fuse data already pre-arranged in
exactly the same format as the physical layout of the fuse array of the device)
data into the device along with the appropriate addresses and commands.

Lattice non-volatile FPGA devices use embedded flash memory and require
only TTL-level programming signals. An integrated state machine controls the
sequence of programming operations, such as identifying the device, shifting
in the appropriate data and commands, and controlling internal signals to
program and erase the embedded Flash in the device. Programming consists
of serially shifting the logic implementation stored in a data file into the device
along with appropriate addresses and commands, programming the data into
the embedded Flash, and shifting the data from the logic array out for device
programming verification.
Most of Lattice’s devices use the IEEE 1149.1-1993 Boundary Scan Test
Access Port (TAP) as the primary interface for in-system programming.

Programming Algorithm Basics

Programming a device is similar to programming any piece of memory, such
as an EPROM or a Flash memory. The device can be thought of as an array
that is programmed one row at a time. The programming information is
provided to the software in the form of a data file that must be converted into
the row and column fuse map data. Before a non-volatile device can be
programmed, it first has to be erased. After the device has been erased, the
programming data can be loaded and the device programmed. After the
device has been programmed, it will be verified by reading out the data in the
device and comparing it to the original.

Figure 2 on page 9 shows the basic programming flow for the device. It does
not include the data file conversion into fuse map data, for it assumes that

8 Programming Tools User Guide

 PROGRAMMER OVERVIEW : Usage and flow

step has already been done. This programming flow will be the same,
regardless of the programming hardware used.

Figure 2: Basic Device Programming Flow

Note
If the device will not be programmed in-circuit (that is, by a cable or an embedded
processor), it is not necessary to preload or save the I/O states.

Programming Times
The time it takes to program a device can often determine where in the
manufacturing process a device, or group of devices, is programmed. A board
test system costing hundreds of thousands of dollars to purchase and as
much as one dollar per minute to operate can be an expensive alternative to
programming if programming times are too long. In many instances, it is more
cost-effective to buy a couple of PCs and program the devices using these
much cheaper systems.

The time it takes to completely program a device is based on the time it takes
to first erase the device, then to program each row in the device, and finally to

Programming Tools User Guide 9

PROGRAMMER OVERVIEW : Usage and flow

verify the device. The erase time for all devices is between 100 ms and 200
ms. A single row is programmed in 10 ms to 50 ms, depending on the device.
The verification process is the quickest of the required steps in the
programming sequence and mainly depends on the time required to shift the
verification data out of any given device.

The benefit of minimal programming times is much more obvious on board

test systems, because they are included as a part of the test program and are
running at the fastest speed possible. Additionally, there is no translation
needed to or from the data file, since this has already been done by
Programmer.

USERCODE
User-programmable identification can ease problems associated with
document control and device traceability. Every Lattice 1149.1-compliant
device contains a 32-bit register that is accessible through the optional IEEE
1149.1 USERCODE instruction. This user-programmable ID register is
basically a user’s notepad provided in Flash or SRAM cells on each device.

In the course of system development and production, the proliferation of PLD

architectures and patterns can be significant. To further complicate the
record-keeping process, design changes often occur, especially in the early
stages of product development. The task of maintaining which pattern goes
into what device for which socket becomes exceedingly difficult. Once a
manufacturing flow has been set, it becomes important to “label” each PLD
with pertinent manufacturing information, which is beneficial in the event of a
customer problem or return. A USERCODE register is incorporated into
devices to store such design and manufacturing data as the manufacturer’s
ID, programming date, programmer make, pattern code, checksum, ISC data
file CRC, PCB location, revision number, or product flow. This assists you with
the complex chore of record maintenance and product flow control. In
practice, the user-programmable USERCODE register can be used for any of
a number of ID functions.

Within 32 bits available for data storage, you may find it helpful to define
specific fields to make better use of the available storage. A field may use
only one bit (or all bits), and can store a wide variety of information. The
possibilities for these fields are endless, and their definition is completely up
to you.

Even with the device’s security feature enabled, the USERCODE register can
still be read. With a pattern code stored in the USERCODE register, you can
always identify which pattern has been used in a given device. As a second
safety feature, when a device is erased and re-programmed, the USERCODE
identification is automatically erased. This feature prevents any situation in
which an old USERCODE might be associated with a new pattern.

It is your responsibility to update the USERCODE when reprogramming. The

USERCODE information is not included in the fuse map checksum reading.

10 Programming Tools User Guide

 PROGRAMMER OVERVIEW : Usage and flow

Loading the USERCODE instruction makes the USERCODE available to be

shifted out in the Shift-DR state of the TAP controller. The USERCODE
register can be read while the device is in normal functional operation,
allowing the device to be scanned while operating.

Programming Hardware
All programming specifications, such as the programming cycle and data
retention, are guaranteed when programming devices over the commercial
temperature range (0 to 70 degrees C). It is critical that the programming and
bulk erase pulse width specifications are met by the programming platform to
ensure proper in-system programming. The details of device programming
are invisible to you if you use Lattice programming hardware and software.

Computer Hardware
Programming is most commonly performed on a PC or Linux system using
the parallel port cable or the USB port cable.

Parallel Port Cable

The cable uses the parallel port of a PC or Linux system for in-system
programming of all Lattice devices. Programmer generates programming
signals from the parallel port and passes them through the cable to the JTAG
port of the devices. With this cable and a connector on the printed circuit
board, no additional components are required to program a device. Refer to
the cable data sheet for more detailed specifications and ordering information.

Hardware design considerations for new boards include whether the

hardware designer will be using boundary scan test operations or low-voltage
(3.3 V–1.8 V) devices. In a system using 3.3 V devices, the cable version 2.0
should be used. This cable operates with either a 3.3 V or 5 V power source.
In a system using 1.8 V devices, cable version 3.0 must be used. This cable
operates with a power of 1.8 V to 5.0 V.

USB Port Cable

The USB port cable uses the USB port of a PC or Linux system for in-system
programming of all Lattice devices. Programmer generates programming
signals from the USB port and passes them through the USB port cable to the
JTAG port of the device.

Programming Software
Programmer supports programming of all Lattice devices in a serial daisy
chain programming configuration in a PC or Linux environment. The software
is built around a graphical user interface. Any required data files are selected
by browsing with a built-in file manager. The software supports both serial and

Programming Tools User Guide 11

PROGRAMMER OVERVIEW : Usage and flow

concurrent (turbo) programming of all Lattice devices. Any non-Lattice

devices that are compliant with IEEE 1149.1 can be bypassed after their
instruction register lengths are defined in the chain description. Any non-
Lattice devices that are compliant with IEEE 1532 can be programmed using
an IEEE 1532-compliant BSDL and ISC data file. Programmable devices from
other vendors can be programmed through the vendor supplied SVF file.

Embedded Programming
Programmer embedded source code is available for programming devices in
an embedded or customized environment. The programming source code is
written in ANSI-standard C language, which can be easily incorporated into
an embedded system or tester software to support programming of devices.
This code supports such common operations as Erase, Program, Verify, and
Secure. After completion of the logic design and creation of a JEDEC file,
Programmer can create the data files required for in-system programming on
customer-specific hardware: PCs, testers, or embedded systems.

FPGA Configuration
Programmer provides efficient and economical alternatives to large and
expensive PROMs that are normally used for configuring FPGA devices.

Because SRAM-based FPGA devices are volatile, they require

reconfiguration on power cycles. This means that external configuration data
must be held in a non-volatile device. On systems that require quick
configurations or that do not have processor resources readily available, a
dedicated serial PROM can be used. But such a PROM has to be large to
accommodate large FPGA devices or multiple devices.

A much easier solution is to use a low-cost, industry-standard flash memory

device combined with a LatticeECP/EC, LatticeECP2, LatticeECP3, and
LatticeSCM/SC device.

Serial Peripheral Interface Flash

Programmer, combined with a Lattice cable fly-wire, supports the
programming of Serial Peripheral Interface (SPI) flash devices.
Several Lattice FPGAs can be configured directly from a serial peripheral
interface (SPI) flash memory devices. Each of these devices contains
Lattice's Reveal IP Programmer, which uses the JTAG interface to program
the SPI flash device. Because of their bitstream compression capability, these
Lattice FPGAs allow the use of smaller-capacity SPI memory devices.

For an up-to-date list of Lattice devices that can be configured using SPI
flash, as well as a list of supported SPI flash devices, refer to the topic “Serial
Peripheral Interface (SPI) Flash Support” in the Lattice Diamond online help
or in the stand-alone Programmer online help.

12 Programming Tools User Guide

 Chapter 3

Deployment Tool Overview

Deployment Tool is a stand-alone tool available from the Diamond

Accessories and is also available with stand-alone Programmer. The
Deployment Tool graphical user interface (GUI) is separate from the Diamond
and stand-alone Programmer design environment.

The Deployment Tool allows you to generate files for deployment for single
devices, a chain of devices, and can also convert data files to other formats
and use the data files it produces to generate other data file formats. A four-
step wizard allows you to select deployment type, input file type, and output
file type.

A basic block diagram of the Deployment Tool flow is shown in Figure 3 on

page 14.

Programming Tools User Guide 13

DEPLOYMENT TOOL OVERVIEW :

Figure 3: Deployment Tool Flow

14 Programming Tools User Guide

 DEPLOYMENT TOOL OVERVIEW : Deployment Tool Function Types

Deployment Tool Function Types

There are four function types available in Deployment Tool:
 File Conversion
 Tester
 Embedded System
 External Memory

The function types are accessed from the Function Type dropdown menu on
the Deployment Tool Getting Started dialog box, as shown in Figure 4.

Figure 4: Deployment Tool Function Types

Output File Types

Each function type outputs different file types. This section describes all of the
file types that are output by the five function types.

File Conversion Output File Types

The File Conversion function outputs four different file types, as shown in
Figure 5 on page 16. The output types are defined as follows:

IEEE 1532 ISC Data File Converts JEDEC files to IEEE 1532 compliant
ISC (In System Configuration) data files, which are used in conjunction with
IEEE 1532 compliant BSDL files to program a device.

Application Specific BSDL File Converts a generic BSDL (Boundary Scan

Description Language) file to an Application Specific BSDL file, using the
signal names from the input file (JEDEC or ALT file). Also, for any I/Os that
support VREFs or LVDS pairs and are configured as VREFs or LVDS pairs,
the application-specific BSDL file changes to accurately reflect the behavior of
the VREF or LVDS pair. When generating the Application Specific BSDL file,
you have the option to convert bi-directional I/O's to inputs or outputs based

Programming Tools User Guide 15

DEPLOYMENT TOOL OVERVIEW : Output File Types

Figure 5: File Conversion Output File Types

on your design, or to keep all I/Os as bi-directional. The generic BSDL files
are available on the Lattice website.

JEDEC File Converts the following file types JEDEC, Binary Bitstream,
ASCII Bitstream, or IEEE 1532 ISC into a JEDEC file. The USERCODE,
USERCODE format, and set the Program Security Fuse for the JEDEC file.

Bitstream Takes a JEDEC, Binary Bitstream, or ASCII Bitstream file and

can convert it into the following output formats Binary Bitstream, ASCII
Bitstream, Intel Hex, Motorola Hex, and Extended Tektronix Hex. Users can
specify the Program Security Bit, Verify ID Code, Frequency, Compression,
CRC Calculation, USERCODE format, and USERCODE.

Refer to the Deployment Tool online help for information about specific device
support.

Tester Output File Types

The Tester function outputs five different file types, as shown in Figure 6.

Figure 6: Tester Output File Types

The output types are defined as follows:

16 Programming Tools User Guide

 DEPLOYMENT TOOL OVERVIEW : Output File Types

SVF - Single Device SVF Single Takes one of the following user data files
types JEDEC, ASCII Bitstream, Binary Bitstream, or IEEE 1532 ISC and then
select an operation to generate an SVF (Serial Vector Format) file. Depending
on the data file selected then a certain set of operation for the device are
available to be selected. The user is able to check several options which will
modify the SVF file.

SVF - JTAG Chain Takes an XCF file generated by Programmer and

generates an SVF file. There are several options available that modify the
SVF file including write header and comments, and set maximum data size
per row.

STAPL - Single Device Takes a JEDEC, ASCII Bitstream, Binary

Bitstream, or IEEE 1532 ISC and then depending on the input file type gives a
set of available operation that can be performed on the device. A STAPL
(Standard Test And Programming Language) file is generated using the data
file and operation.

STAPL - JTAG Chain Generates a STAPL file for testing using only an XCF
file generated by Programmer.

ATE Takes an XCF file and then the user is able to specify Tester Type,
whether or not to skip the verify step in erase program verify or to split into
separate files. An ATE (Automated Test Equipment) is a serial vector file
specific to a test equipment vendor.

Refer to the Deployment Tool online help for information about specific device
support.

Embedded System Output File Types

The Embedded System function outputs five different file types, as shown in
Figure 7.

Figure 7: Embedded System Output File Types

The output types are defined as follows:

Programming Tools User Guide 17

DEPLOYMENT TOOL OVERVIEW : Output File Types

JTAG Full VME Embedded Takes an XCF as an input file, then the user
can check options such as Compress VME, include Header along with
several other options. This operation generates a VME file which is a
compressed hexadecimal representation of an SVF files.

JTAG Slim VME Embedded VME is a compressed version of a VME file.

To generate a Slim VME file an XCF file must be specified, then specify
whether it is a Compressed VME file and whether or not to generate a HEX
file. This operation outputs an algorithm VME file and a data VME file.

Slave SPI Embedded This file type allows field upgrades via the slave SPI
port. This operation can be given an XCF, Binary Bitstream, and ASCII
Bitstream as an input file. If an Bitstream file is given then the operation for the
device must be specified along with whether or not to compress the
embedded file and whether or not to generate a HEX file. If an XCF file is
given there are no other operations or options the user needs to provide. This
operation will output an algorithm file (.sea) and a data file (.sed).

I2C Embedded I2C embedded files enable field upgrades via the I2C port. If
an XCF file is specified then the user is given the option to compress the
embedded files, generate a hex file, include comments, and if there should be
a fixed pulse width. If a Bitstream file is specified then the previous options are
available along with selecting the device operation and specifying the length
of the I2C Slave Address. Two files will be generated a data file (.ied) and an
algorithm file (.iea).

sysCONFIG Embedded Takes an XCF file as input and generates a CPU

(.cpu) file which can be used for field upgrades via the slave parallel or slave
serial modes.

Refer to the Deployment Tool online help for information about specific device
support.

Also, refer to “Embedded Flow Overview” on page 19.

External Memory Output File Types

The External Memory function outputs four different file types, as shown in
Figure 8 on page 19. The output types are defined as follows:

Hex Conversion converts a file JEDEC, Binary Bitstream, ASCII Bitstream,

Binary, or Hex to various Hexadecimal file formats which are used to
configure the external SPI Flash memory of a device. The output file formats
are Intel Hex, Motorola Hex, and Extended Tektronix Hex. The user is also
able to set the Program Security bit, Verify ID Code, Frequency, compression,
CRC Calculations and also the Starting Address.

Dual Boot Takes two JEDEC, Binary Bitstream or ASCII Bitstream files and
then creates a single hex file to configure primary and golden sectors of an
external SPI Flash. The output format can be Intel Hex, Motorola Hex, and
Extended Tektronix Hex. The device will usually boot form the primary sector
unless there is a problem then it will boot from the gold sector.

18 Programming Tools User Guide

 DEPLOYMENT TOOL OVERVIEW : Output File Types

Figure 8: External Memory Output File Types

Advanced This operation is for generating hex files which handles more
complicated operations such as Multiple Boot, and Quad I/O to configure
external memory. Users can set the output hex format, how big the SPI Flash
size is, whether or not to do a byte wide bit mirror, retain the bitstream header,
and Whether or not to optimize the memory space. Another option is to set
multiple user data file and where each of those data file's starting address
should be in memory.

sysCONFIG Daisy Chain. This is used when multiple devices are in a daisy
chain and configured from a single SPI flash or CPU. This operation will take
two Binary or ASCII bitstreams and convert them into a single hex file.

Refer to the Deployment Tool online help for information about specific device
support.

Programming Tools User Guide 19

DEPLOYMENT TOOL OVERVIEW : Output File Types

20 Programming Tools User Guide

 Chapter 4

Embedded Flow Overview

Lattice Embedded VME enables in-field upgrades of Lattice programmable

devices by suitable embedded processors, and consists of the following:

JTAG Full VME Embedded Enable field upgrades via the JTAG port.

JTAG Slim VME Embedded Featured s reduced foot print and is designed
for microcontrollers with limited resources, such as 8051 processors.

Slave SPI Embedded Enable field upgrades via the slave SPI port.

I2C Embedded Enable field upgrades via the I2C port.

sysCONFIG Embedded Enable field upgrades via the slave parallel or

slave serial modes.
There are three components to Embedded VME
 ANSI C source code, which is shipped with Diamond Programmer. The
user compiles this ANSI C Source code into their target system.
 Algorithm VME File, which contains the programming algorithm for the
target FPGA. The Algorithm VME file is generated using the Deployment
Tool.
 Data VME File, which contains the data that will be programmed into the
FPGA. The Data VME file is generated using the Deployment Tool.

For all five embedded types, the Embedded VME support is comprised of C
source files that users must port into their embedded systems for the purpose
of programming Lattice devices. The porting process is also known as the
customization and compiling process. The end product of the porting process
will be the Embedded VME in compiled form, which will reside in the
embedded systems.

Programming Tools User Guide 19

EMBEDDED FLOW OVERVIEW :

Depending on the port interface, such as JTAG, SPI, or I2C, the user can
select one of the five embedded VME types.

Figure 9 shows an example of Full VME embedded file generation for the
JTAG port.

Figure 9: Full Embedded VME Flow

The programming data and programming instructions are compiled into a

binary VME file format for the driver to load into the target devices. The VME
file can be provided to the driver as a stand-alone file or linked together with
the driver.

Figure 10 shows a high-level example of a file-based embedded VME used

for field upgrades.

Figure 10: Example Embedded VME Programming Configuration

20 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW :

Porting of the JTAG VME into Embedded Systems

Porting JTAG Embedded VME is simple and the requirements are very simple
to follow:

AC Requirements:
 TCK Fmax = 25 MHz.
 TCK Rise Time and Fall Time = 50ns maximum.
 Delay function resolution and accuracy = 1 millisecond minimum.

DC Requirements:
 I/O voltage level of the driver = I/O voltage level of the VCC JTAG port of
the target devices. The VCC that power the JTAG port can be:
 VCC core (All EE based devices)
 VCCIO (MachXO devices)
 VCCJ (All SRAM based and Flash based FPGA devices)
 Programming current = 1 Ampere maximum.

JTAG Programmability of Lattice Devices

Lattice's devices can be classified into three groups based on
programmability:
 SRAM based only devices (volatile devices).
 EE based devices (non-volatile devices).
 Flash based devices (non-volatile devices).

Note
For information on configuring the Lattice iCE40 family of devices from an
embedded processor, refer to TN1248, iCE40 Programming and Configuration
Guide.

The SRAM based only devices are the easiest devices to support in terms of
Embedded VME porting for they normally do not require accurate timing.

The EE based devices are much more challenging for they require the 1
millisecond resolution and accurate timing.

The Flash based devices are the most challenging among the three types.
The delay function not only must have the 1 millisecond resolution and
accuracy, it also must be able to provide the cumulative delay time up to 150
seconds – the worst case erase time of some Flash based devices.

The EE based devices are supported by the programming flow generally

referred to as the fixed pulse width flow. The Flash based devices require the

Programming Tools User Guide 21

EMBEDDED FLOW OVERVIEW :

Figure 11: ispMACH4000 Fixed Delay Time Programming Flow

looping programming flow. The most critical requirement to ensure the

devices are programmed reliably rests on the accuracy of the delay function.

If the devices are not given the required programming delay time, the Flash
based devices will fail the verification during programming. It will be worse if
failure happens during Flash erase.

When the erase operation is terminated before completion due to insufficient

delay time, the Flash will have an unknown pattern residing in it which might
cause the device entering the contentious state. When the device is in
contentious state, it will be very hot and would not respond to further
programming commands. EE based devices may not fail verification but they
will fail to meet the 10 year data retention as specified.

All Lattice EE and Flash based devices are designed with the over-stress and
over-charge protection technology. This technology is very critical to the
superior In-System Programmability of Lattice’s non-volatile devices. The
devices cannot be damaged when given an erase delay time or programming
delay time longer than the minimum specified.

The accuracy of the delay function discussed in this document will focus only
on meeting the minimum requirement. It means that when the delay function
is called to generate, for example, 100 ms delay time (by calling the delay

22 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW :

Figure 12: MachXO Looping Programming Flow

function 100 times), as long as the resulting delay time is equal to or greater
than 100 ms, the delay time is considered accurate.

The only undesirable side effect of applying longer erase delay time or
programming delay time to the devices is on the programming throughput.
Diamond Programmer provides precise delay timing to meet the programming
specification and optimizing on throughput.

The Deployment Tool software will automatically determine whether the

device requires the fixed delay time programming flow or the looping
programming flow when generating the VME file. Also, the device specific
programming algorithm details, such as row size, the maximum loop count,
and the delay required per each loop, are all embedded into the VME file
automatically.

Programming Tools User Guide 23

EMBEDDED FLOW OVERVIEW :

If interested to find out the details of their VME files, simply convert the binary
VME files into the corresponding text based SVF file format using the
Download Debugger shipped with Diamond Programmer.

The verify flow is identical for the looping and fixed pulse programming flow.
The delay required for the verify flow is usually ~100us per row. The VME file
still requires 1 millisecond delay per row since the PC cannot provide
accurate timing if the delay is less than 1 millisecond.

Thus, the only Embedded VME implementation challenge is the development

of an accurate delay function to provide 1 millisecond minimum delay time.

Embedded VME Porting Detail

Step 1: Customize JTAG Embedded VME by modifying hardware.c

Figure 13: Map Four GPIO Pins from the CPU to the Four JTAG Pins

The pin mapping index table on the hardware.c must be revised to match with
the customer’s board layout. On the PCB that is the target for porting the
Embedded VME, it is important and a good practice to route the JTAG port to
a test header for easy access using an oscilloscope or connecting to
Programmer for debugging.

24 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW :

All VME files begin with IDCODE verification to ensure the JTAG port pins are
mapped and connected properly.

Modify the Delay Function

When porting Embedded VME to a native CPU environment, the speed of the
CPU or the system clock that drives the CPU is usually known. The speed or
the time it takes for the native CPU to execute one loop then can be
calculated.

The for loop usually is compiled into the ASSEMBLY code as shown below:
LOOP: EDC RA;
JNZ LOOP;

If each line of assembly code needs four (4) machine cycles to execute, the
total number of machine cycles to execute the loop is 2 x 4 = 8.

Usually: system clock = machine clock (the internal CPU clock).

Note
Some CPUs have a clock multiplier to double the system clock for the machine clock.

Let the machine clock frequency of the CPU be F (in MHz), then one machine
cycle = 1/F.

The time it takes to execute one loop = (1/F) x 8.

It is obvious that the formula can be transposed into one microsecond = F/8.

Example: The CPU internal clock is set to 48 MHz, then one microsecond =
48/8 = 6.

Programming Tools User Guide 25

EMBEDDED FLOW OVERVIEW :

The C code shown below can be used to create the millisecond accuracy. All
that needs to be changed is the CPU speed.

Step 2: Calibration

It is important to confirm if the delay function is indeed providing the accuracy

required. It is also important to confirm the TCK frequency. As an example, we
will estimate the minimum system clock frequency of the native CPU that
does not require the TCK to be slowed down. The TCK could be generated by
the following code.
writePort (g_ucPinTCK, 0x00);
writePort (g_ucPinTCK, 0x01);

Let the number of system clocks to execute one line of code = 8 clocks.

The total number of clock for one pulse = 2 x 8 = 16.

The total amount of time for one pulse = 1/F x 16.

Lattice devices TCK frequency max = 25 MHz.

The equation becomes: 1/25 = 1/F x 16.

The maximum frequency of the CPU: F = 16 x 25 = 400 MHz.

26 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW :

If the system clock of the native CPU is faster than 400 MHz, the TCK pulses
must be slowed down to meet the 25 MHz maximum specification.

The setup time and hold time of TDI, TMS, and TDO relative to TCK is not of
concern for Embedded VME is constructed in the fashion that it is not possible
to violate that requirement whenever the frequency of TCK is within the
specification.

Figure 14: JTAG Embedded VME Delay Calibration

The calibrate function in Embedded VME can be launched by using the –c

switch to cause the waveform as follow captured on the scope with the probe
attached to the TCK wire.

If the pulse width is found to be smaller than 1 millisecond, then increase the
cpu_frequency value until 1 millisecond delay is captured by the calibration
function.

If the TCK frequency is found to be faster than 25 MHz, then change the
sclock() function in hardware.c as shown below. The IdleTime normally is

Programming Tools User Guide 27

EMBEDDED FLOW OVERVIEW :

Figure 15: JTAG Embedded VME Delay Calibration TCK Waveforms

28 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW :

initialized to 0. If it is initialized to 1, then the TCK frequency is effectively

reduced by half. Use this technique to reduce the TCK frequency until
meeting the specification.

Step 3: Program Devices

Once the calibration is done, the Embedded VME (actually the JTAG port
driver) is ready to program the devices. The device specific programming
information is all self-contained in the VME file.

The VME file actually has six major sections:

1. Check the IDCODE,
2. Erase the device,
3. Program the device,
4. Verify the device,
5. Program the done fuse,
6. Wake-up the device.

IDCODE check failure is the most common failure when porting Embedded
VME. It is a good practice to generate a Verify IDCODE only VME file first.
Run the VME file. If it passes, then the JTAG port to GPIO mapping is
confirmed. Once the port mapping is confirmed, then the programming VME
file can be used.

Accurate timing is very critical to program devices reliably.

Using the calibration routine provided by Embedded VME will achieve the
accurate timing.

Programming Tools User Guide 29

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

JTAG Full VME Embedded

The JTAG Full VME Embedded VME software brings programming software
to embedded applications. Using Lattice Semiconductor’s Diamond
Programmer and Deployment Tools, you are provided with all necessary
capabilities for programming devices in a single or multiple device chain.
Developed to solve many programming issues facing today’s PLD users,
JTAG Full VME Embedded provides advanced features including fast
programming times, and small file sizes.

The JTAG Full VME Embedded software is a simplified version of the full
Diamond Programmer. By making it serial vector format (SVF) file centric,
JTAG Full VME Embedded is better targeted for embedded systems. Lattice
JTAG devices are supported and users are able to program competitor
devices through a simple SVF file translator. Lattice JTAG devices are those
devices that can be programmed using the IEEE 1149.1 boundary scan TAP
controller interface. Users are able to quickly and efficiently program chains of
devices using this powerful utility, thus improving productivity and lowering
costs.

An advantage of JTAG Full VME Embedded over vendor or architecture-

specific methods is that once it is developed, it supports all present and future
devices. As long as the programming flow can be described as an SVF file,
the main engine does not have to change. For embedded environments, it is
important to have deterministic memory requirements. By pre-processing the
SVF file, it is possible to know the exact resources required to implement the
programming algorithm and to store the programming data. The nature of the
SVF file also allows the resources available to determine how the file is
processed. Large shift instructions can be broken into multiple instructions if
the embedded system does not have enough RAM available to store the
entire row in one pass. Since the SVF file is serial in nature, it can be
segmented to fit available RAM, PROM or FLASH memory.

The JTAG Full VME Embedded source code is designed to be hardware and
platform independent. A VME data file, or VME file, runs on all JTAG Full VME
Embedded applications.

See Also JTAG Full VME Embedded Basic Operation

VME File Format

A VME file is simply an SVF file that has been compressed. SVF file includes
algorithm and data file in ASCII format, and VME file is the SVF file in the
optimized binary format. Compared with SVF file, VME files require minimized
memory space to store the bitstream file and has optimized code size. SVF
keywords such as SIR and SDR are replaced with the byte codes 0x11 and
0x12, respectively. This reduces the VME file by writing only one byte of data,
the byte codes, instead of writing the entire SVF keyword, which would use
more characters.

JTAG Full VME Embedded file supports compression to reduce the VME file
size by compressing the data and address streams. A looping compression is

30 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

also employed to reduce the file size even further by taking advantage of the
repeating SVF constructs. The following describes each compression
scheme.

Compressed VME Files The compression scheme is applied to the

address and data stream following SIR and SDR, respectively. These streams
will try to be compressed by 0x00, 0xFF, or by 4-bit count.

For example, consider the following line in a SVF file:

SDR 102 TDI (20000000000000000000000000);

The address stream is ‘20000000000000000000000000’. The repeating

zeros in the stream can be easily compressed by 0x00. Compression with
0xFF works in the same manner, except that instead of the data stream
containing zeros, it would contain ‘F’s.

Compression by 4-bit count works by looking for repeating patterns within the
data stream that are not zeros or ‘F’s.

For example, consider the following line in a SVF file:

SDR 80 TDI (7F97F97F97F97F97F97F9);

The repeating 4-bit count in this example would be ‘7F9,’ because it repeats
throughout the data stream. The 4-bit would be written only once in the VME
file, and would be followed by the number of repetitions found within the data
stream.

The compression scheme reduces the file size by not extrapolating repeating
information within the address and data streams. That task is left for the VME
processor.

Looping VME Files In an SVF file, repeating constructs can be observed.

The looping scheme takes advantage of these constructs by creating a
template with the repeating information, and the differentiating date is
replaced by a placeholder. The differentiating data will be written after the
construct.

For example, the following data is found in a SVF file:

SIR 5 TDI (01);

SDR 102 TDI (20000000000000000000000000);

SIR 5 TDI (02);

SDR 80 TDI (7BFFF7BFFFF7BFFF7BFF);

SIR 5 TDI (07);

RUNTEST IDLE 3 TCK 1.20E-002 SEC;

SIR 5 TDI (01);

Programming Tools User Guide 31

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

SDR 102 TDI (10000000000000000000000000);

SIR 5 TDI (02);

SDR 80 TDI (FFFF7FFFFFFFFFFFFFFF);

SIR 5 TDI (07);

RUNTEST IDLE 3 TCK 1.20E-002 SEC;

SIR 5 TDI (01);

SDR 102 TDI (08000000000000000000000000);

SIR 5 TDI (02);

SDR 80 TDI (FFFFFFFFFFFFFFFFFFFF);

SIR 5 TDI (07);

RUNTEST IDLE 3 TCK 1.20E-002 SEC;

...

...

The looping template is built based on the repeating SIR lines. Notice how the
TDI values for the SIR commands are a repeating sequence of 01, 02, and
07. In this case the resulting template would be:

SIR 5 TDI (01);

SDR 102 TDI VAR;

SIR 5 TDI (02);

SDR 80 TDI VAR;

SIR 5 TDI (07);

RUNTEST IDLE 3 TCK 1.20E-002 SEC;

VAR is written in place to hold the data that does not repeat. The non-
repeating data will get written into the VME file following each template. The
example above would look like this in the VME file:

LOOP 3

SIR 5 TDI (01);

SDR 102 TDI VAR;

SIR 5 TDI (02);

32 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

SDR 80 TDI VAR;

SIR 5 TDI (07);

RUNTEST IDLE 3 TCK 1.20E-002 SEC;

ENDLOOP

(20000000000000000000000000)

(7BFFF7BFFFF7BFFF7BFF)

(10000000000000000000000000)

(FFFF7FFFFFFFFFFFFFFF)

(08000000000000000000000000)

(FFFFFFFFFFFFFFFFFFFF)

The ‘LOOP 3’ tells the VME processor to loop the template three times. Each
time it encounters a ‘VAR’, it will grab the first available line of data following
the ‘ENDLOOP’ and replace ‘VAR’ with it. This technique reduces the file
significantly by keeping the similar constructs to a minimal, and only writing
the differences.

JTAG Full VME Embedded Flow

The JTAG Full VME Embedded System allows you to program a device using
the microprocessor in an embedded system. When you install the VM
software, a separate VMEmbedded folder containing the VME source code
and executables is installed on your hard drive. Compiling the VME source
code gives you an executable file that you can store in your system’s memory
for programming using the JTAG port.

The following figure illustrates the JTAG Full VME Embedded flow.

See Also Generating VME Files

 Testing VME Files

Programming Tools User Guide 33

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

JTAG Full VME Embedded System

Memory
The following figure illustrates JTAG Full VME Embedded system memory.

See Also JTAG Full VME Embedded Basic Operation

JTAG Full VME Embedded Basic

Operation
There are three modes of JTAG Full VME Embedded operation.

File Mode Under the file mode, data is stored in a file system such as a
hard drive or a DOS flash. The data file is accessed using C library calls, such
as fopen, fread, and fclose. The file read operations collect data into the
system memory. The system memory of the Embedded system must be able
to store the entire bitstream from the file in a contiguous block of memory. The
memory block can be allocated in one of the three locations.
 Data Segment – You can pre-determine how many bytes of data the
bitstream will require and then create an uninitialized array variable to
hold the data. This permanently allocates a portion of the Data Segment.
For example:
char programmingData[0x10000]; // allocate 64K

 Stack Segment – You can pre-determine how many bytes of data the
bitstream will require and then create an uninitialized array variable to

34 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

hold the data. Depending upon the function call sequence, this may or
may not permanently allocate a portion of the system memory. See the
example code below.
int MyFunction ( ) {
char bitstreamArray[0x10000];
}

 Heap Segment – You can determine at runtime how many bytes of data
the bitstream will require and then dynamically allocate an uninitialized
array variable to hold the data. You are responsible for freeing the
memory when it is not being used any longer. Below is an example.
char *bitstreamData;
bitstreamData = (char *)malloc(numberOfBitstreamBytes);

Static Linking Mode Under the static linking mode, the bitstream data is
converted from the file on the hard drive into a C source code file. The C
source code defines a byte array. The byte array is exactly the size of the
bitstream. The byte array can be linked into either the Code Segment or the
Data Segment. The memory allocated for the bitstream is permanently
consumed.

PROM Mode Under the PROM-based mode, the bitstream file is converted
from the file on the hard drive into an Intel HEX file. The HEX file is loaded into
a non-volatile memory using a PROM programming tool. The HEX file data is
placed in the non-volatile memory at a known address (that is, a fixed
address). The user C code initializes a pointer. The pointer is given the
starting address of the HEX byte stream. The memory used by the bitstream
is permanently allocated in the non-volatile memory.

See Also JTAG Full VME Embedded System Memory

VME Source Code

The JTAG Full VME Embedded source code is written in standard ANSI C
and is simplified with embedded applications in mind. Most embedded
applications have greater limits on program and data sizes than PC or
workstation applications. The areas most likely to differ between platforms are
the timing delay function and hardware port manipulation.

The current version of the JTAG Full VME Embedded software is available
through the Programmer installation. The installation creates a sub-directory
called VMEmbedded, where the pre-compiled executables, source code, and
readme.txt can be found.

There are four sets of embedded-related source code that are shipped with
Programmer.
 VME – The file-based VME is the programming engine that accepts VME
files as command line arguments to process the devices. By default, the
executable compiled from this source code targets Windows operating

Programming Tools User Guide 35

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

systems. You can make small modifications to make the compiled

executable accommodate other platforms.
 VME_eprom – The PROM-based VME is the programming engine that
requires compiling a HEX file, which is a C-programming file, with the
source code to create an executable engine that can be embedded onto
the embedded system.
 svf2vme – The svf2vme is a command line utility that can convert SVF
files into VME files.
 vme2hex – The vme2hex is a command line utility that can convert VME
files into HEX files.

Among all the source codes, only the hardware.c file requires user changes.
You should customize the hardware.c file according to your target platform.

JTAG Full VME Embedded

Programming Engine
The programming engine of the JTAG Full VME Embedded software is driven
by the byte codes of the VME format file. It manipulates the I/O ports and
sends commands to the customer firmware. The commands sent from the
programming engine requires the I/O system to be connected to the device's
JTAG port. The VME byte codes instruct the engine as to what sequence of
functions to follow in order to shift in instructions, move the TAP controller
state machine, shift data in and out of the device, and observe delay. The
engine has the following three layers.
 User interface layer (ispvm_ui.c) – Directs inputs and outputs.
 Processor layer (ivm_core.c) – Decodes commands, checks CRC prior to
processing, and does optional decompression.
 Physical layer (hardware.c) – Shifts data to target device. This is the only
file that you need to edit. See Customizing for the Target Platform for
details.

36 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

The following figure illustrates JTAG Full VME Embedded JTAG port
programming engine.

RAM Size Requirement for VME

To calculate the worst-case size of memory needed to program a device, in
terms of bytes, locate the size of the largest register in the device. This is
usually the data shift register. Divide that number by eight, and then multiply
the quotient by two: one for TDI and one for TDO. If the device has MASK,
multiply the quotient by three instead of two.

This method only calculates the RAM requirements for the data of the device.
It does not account for transient variables that are used to execute the
programming algorithm. Transient variables are more difficult to calculate
because they appear in and out of scope often. Also, a variable size may
depend on the microprocessor’s register size. For example, an integer
variable on a 32-bit system is four bytes while the same variable on a 16-bit
system is only two bytes.

To approximate the RAM requirement for the run-time variables, add twenty
percent to the required RAM.

Programming Tools User Guide 37

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

RAM Calculation Example The following is a partial SVF for the M4A3 32/
32 device:

SIR 6 TDI (03);

SDR 80 TDI (00000000000000000040);

SIR 6 TDI (04);

SDR 202 TDI (200000000000000000000000000000000000000000000000000);

SIR 6 TDI (07);

SDR 202 TDI (200000000000000000000000000000000000000000000000000)

TDO (000410410410410410410410410104104104104104104104104)

MASK (210410410410410410410410410104104104104104104104104);

The largest data frame size in this device is 202 bits. Therefore, TDI, TDO,
and MASK each require 26 bytes, making it a total of 3*26 = 78 bytes. To
account for run-time variables, the total required size would be 78 * 1.2 = 94
bytes.

To verify that the calculation is correct, convert the SVF file to VME, and use
the VME2HEX utility to convert from VME to HEX. This utility generates the
vme_file.h file, which gives the definitive memory size requirement.

The variables that are of concern to memory are:

 MaskBuf
 TdiBuf
 TdoBuf
 HdrBuf
 TdrBuf
 HirBuf
 TirBuf
 HeapBuf
 CRCBuf
 CMASKBuf

As expected, MaskBuf, TdiBuf, and TDOBuf each requires 26 bytes. If the

device were in a chain, HdrBuf (Header Data Register), TdrBuf (Trailer Data
Register), HirBuf (Header Instruction Register), and TriBuf (Trailer Instruction
Register) would need extra bytes.

If the VME file had been generated with the looping option, HeapBuf would
require extra bytes as well. Looping requires slightly more RAM but
significantly less ROM. When the VME file has not been looped, it does not
require any additional RAM, but ROM size can significantly increase. This

38 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

trade-off is file-dependent.If the original SVF were 1532-compliant, CRCBuf

and CMASKBuf would require extra bytes as well.

ROM Size Requirement for JTAG Full

VME Embedded
To calculate the worst-case ROM size for a given device, multiply the number
of frames by the frame size. Divide that number by eight to obtain the required
ROM size, in terms of bytes.

This method assumes that the SVF file will be generated with the turbo option.
If the SVF file were generated with the sequential option, the worst-case ROM
size would be doubled.

This method only calculates the ROM requirements for the data. It does not
account for opcodes that are used to translate the algorithm of the device. To
approximate the ROM requirement for the algorithm opcodes, add twenty
percent to the required ROM.

The actual ROM requirement might be significantly less than the theoretical
worst-case requirement because SVF2VME utilizes two compression
techniques, compression and looping, to decrease the VME file. The file size
difference is file-dependent.

ROM Size Calculation Example The following example calculates the

worst-case ROM size for the device LC4128:
Frame size = 740
x Number of frames = 100
____________________________
Data ROM size = 74000 bits
Data ROM size = 74000
x 1.2
____________________________
Overall ROM size = 88800 bits

JTAG Full VME Embedded Required

User Changes
To make the JTAG Full VME Embedded or JTAG Slim VME Embedded
software work on your target system, you need to modify the following C
functions in the hardware.c source code.

Timer The engine requires the ability to delay for fixed time periods. The
minimum granularity of the delay is 1 microsecond. You can determine the
type of delay. This can be a simple software timing loop, a hardware timer, or
an operating system call, for example, sleep().

Port Initialization The firmware needs to place the port I/O into a known
state. The software assumes this has occurred.

Programming Tools User Guide 39

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Get Data Byte The engine calls the GetByte() function to collect one byte
from the JTAG Full VME Embedded or CPU bytestream.

Modify Port Register The engine, as it parses the bitstream data, changes
an in-memory copy of the data to be written onto the I/O pins. Calls to this
function do not modify the I/O pins. The engine uses virtual types (for
example, DONE_PIN) which this function turns into physical I/O pin locations
(for example, 0x400).

Output Data Byte The engine calls this function to write the in-memory
copy onto the I/O pins.

Input Status This function is used by the engine to read back programming
status information. The function translates physical pin locations (for example,
0x400) into virtual types used by the engine (for example, DONE_PIN).

Output Configuration Pins Some systems may wish to use the FPGA
CFG pins, and have the Embedded system control them. There is a separate
function call to manipulate the CFG pins.

Bitstream Initialization You must determine how you plan to get the
bitstream into your memory system, pre-compiled, HEX file based, or
dynamically installed. Whichever method you use the data structures which
pin to the bitstream need to be initialized prior to the first GetByte function call.

See Also Customizing for the Target Platform

 VME Source Code
 JTAG Slim VME Embedded Source Code

40 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Program Memory Requirement

The following figure illustrates the JTAG Full VME Embedded program
memory requirement.

Program Memory Allocation

The following figure illustrates the JTAG Full VME Embedded program
memory allocation.

Programming Tools User Guide 41

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Sample Program Size

This page provides sample program size for JTAG Full VME Embedded,
JTAG Slim VME Embedded, and sysCONFIG Embedded.

JTAG Port non-JTAG Port Total

JTAG Full VME JTAG Slim sysCONFIG

Embedded VME Embedded
Embedded

32-Bit 16-Bit 8-Bit 32-Bit 16-Bit

File Based 52KB 21KB 4.2KB 48KB 19KB As Shown

(Bitstream File External)

PROM Based 52KB 21KB 4.2KB 48KB 19KB As Shown +

(Bitstream File Integrated) VME File Size

42 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

VME File Size

The following table compares VME file sizes taking typical Lattice devices for
examples.

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

ispMACH 4000
Programming Mode: Turbo
Operation: Erase, Program, Verify

LC4032V/B/C/ZC 172 1KB 8KB

LA4032V/ZC 3KB

LC4064V/B/C/ZC-32IO 352 1KB 12KB

LA4064V/ZC-32IO 4KB

LC4064V/B/C/ZC 356 1KB 13KB

LA4064V/ZC 5KB

LC4128V/B/C/ZC 740 1KB 22KB

LA4128V/ZC 10KB

LC4256V/B/C/ZC 1592 1KB 41KB

19KB

LC4256V/B/C-160IO 1624 1KB 43KB

20KB

LC4384V/B/C 2616 1KB 65KB

31KB

LC4512V/B/C 3632 1KB 89KB

43KB

ispMACH 4000ZE
Programming Mode: Turbo
Operation: Erase, Program, Verify

LC4032ZE 172 1KB 8KB

3KB

LC4064ZE 356 1KB 13KB

5KB

Programming Tools User Guide 43

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

LC4128ZE 740 1KB 22KB

10KB

LC4256ZE 1592 1KB 41KB

19KB

ispPAC Power Manager II

Programming Mode: Sequential
Operation: Erase, Program, Verify

ispPAC-POWR607 81 1KB 9KB

2KB

ispPAC-POWR6AT6 56 1KB 1KB

1KB

ispPAC-POWR1014/14A 123 1KB 18KB

6KB

ispPAC-POWR1220AT8/2 243 1KB 39KB

13KB

LatticeEC/ECP
Programming Mode: Sequential
Operation: Fast Program

LFEC/ECP-1E 470 1KB 75KB

75KB

LFEC/ECP-3E 558 1KB 126KB

125KB

LFEC/ECP-6E 754 1KB 217KB

216KB

LFEC/ECP-10E 1094 1KB 379KB

378KB

LFEC/ECP-15E 1270 1KB 519KB

518KB

LFEC/ECP-20E 1358 1KB 642KB

641KB

44 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

LFEC/ECP-33E 1798 1KB 958KB

957KB

LatticeXP
Programming Mode: Sequential
Operation: FLASH Erase, Program, Verify

LFXP3C/E 558 1KB 437KB

117KB

LFXP6C/E 734 1KB 673KB

189KB

LFXP10C/E 2148 1KB 1087KB

340KB

LFXP15C/E 2500 1KB 1565KB

494KB

LFXP20C/E 2676 1KB 1930KB

611KB

MachXO
Programming Mode: Sequential
Operation: FLASH Erase, Program, Verify

LCMXO256C/E 192 1KB 36KB

8KB

LCMXO640C/E 320 (used BSCAN 1KB 70KB

Length)
17KB

LCMXO1200C/E 424 (used BSCAN 1KB 111KB

Length)
35KB

LCMXO2280C/E 544 (used BSCAN 1KB 195KB

Length)
62KB

Programming Tools User Guide 45

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

MachXO2
Programming Mode: Sequential
Operation: FLASH Erase, Program, Verify

LCMXO2-256ZE 504 1KB 42KB

11KB

LCMXO2-256HC 504 1KB 42KB

11KB

LCMXO2-640ZE 888 13KB 89KB

24KB

LCMXO2-640HC 888 13KB 89KB

24KB

LCMXO2-1200ZE 1080 33KB 172KB

48KB

LCMXO2-1200HC 1080 33KB 172KB

48KB

LCMXO2-640UHC 1080 33KB 172KB

48KB

LCMXO2-2000ZE 1272 41KB 243KB

68KB

LCMXO2-2000HC 1272 41KB 243KB

68KB

LCMXO2-1200UHC 1272 41KB 243KB

68KB

LCMXO2-2000HE 1272 41KB 243KB

68KB

LCMXO2-4000ZE 1560 49KB 409KB

115KB

46 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

LCMXO2-4000HC 1560 49KB 409KB

115KB

LCMXO2-2000UHC 1560 49KB 409KB

115KB

LCMXO2-4000HE 1560 49KB 409KB

115KB

LCMXO2-2000UHE 1560 49KB 409KB

115KB

LCMXO2-7000ZE 1992 127KB 700KB

198KB

LCMXO2-7000HE 1992 127KB 700KB

198KB

LCMXO2-7000HC 1992 127KB 700KB

198KB

MachXO3L
Programming Mode: Sequential
Operation: FLASH Erase, Program, Verify

LCMXO3L-2100C 1272 41KB 243KB

68KB

LCMXO3L-4300C 1560 49KB 409KB

115KB

LCMXO3L-6900C 1992 127KB

198KB

Platform Manager 2
Programming Mode: Sequential
Operation: FLASH Erase, Program, Verify

LPTM20 888 20KB 94KB

14KB

Programming Tools User Guide 47

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

LPTM21 1080 40KB 158KB

22KB

LatticeECP2/2S
Programming Mode: Sequential
Operation: Fast Program

LFE2-6E/SE 752 1KB 191KB

190KB

LFE2-12E/SE 752 1KB 365KB

364KB

LFE2-20E/SE 1018 1KB 566KB

565KB

LFE2-35E/SE 1284 1KB 802KB

800KB

LFE2-50E/SE 1643 1KB 1141KB

1140KB

LFE2-70E/SE 2032 1KB 1702KB

1700KB

LatticeECP2M/2MS
Programming Mode: Sequential
Operation: Fast Program

LFE2M20E/SE 1615 1KB 758KB

756KB

LFE2M35E/SE 2042 1KB 1259KB

1258KB

LFE2M50E/SE 2419 1KB 2023KB

2022KB

LFE2M70E/SE 2799 1KB 2537KB

2535KB

LFE2M100E/SE 3120 1KB 3282KB

3281KB

48 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

LatticeXP2
Programming Mode: Sequential
Operation: FLASH Erase, Program, Verify

LFXP2-5E 638 1KB 561KB

154KB

LFXP2-8E 772 1KB 859KB

243KB

LFXP2-17E 2188 1KB 1423KB

446KB

LFXP2-30E 2644 1KB 2308KB

731KB

LFXP2-40E 3384 1KB 3176KB

1017KB

LatticeECP3
Programming Mode: Sequential
Operation: Fast Program

LFE3-17E/EA 2584 1KB 505KB

503KB

LFE3-35E/EA 3416 1KB 883KB

882KB

LFE3-70E/EA 6724 1KB 2336KB

2333KB

LFE3-95E/EA 6724 1KB 2336KB

2333KB

LFE3-150E/EA 8380 1KB 3719KB

3714KB

ECP5U
Programming Mode: Sequential
Operation: Fast Program

Programming Tools User Guide 49

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Device Device Row Size JTAG Slim VME VME File Size
(Bits) Embedded File Size No Compression
No Compression (1K Byte = 1024 Bytes)
(1K Byte = 1024 Bytes)

Algorithm File Size

Data File Size

LFE5U-85F 1136 1KB 85KB

1883KB

ECP5UM
Programming Mode: Sequential
Operation: Fast Program

LFE5UM85F 1136 1KB 85KB

1883KB

LatticeSC/SCM
Programming Mode: Sequential
Operation: Fast Program

LFSC3GA15E 1043 1KB 543KB

LFSCM3GA15E 542KB

LFSC3GA25E 1316 1KB 701KB

LFSCM3GA25E 700KB

LFSC3GA40E 1652 1KB 1078KB

LFSCM3GA40E 1076KB

LFSC3GA80E 2156 1KB 2723KB

LFSCM3GA80E 2719KB

LFSC3GA115E 2639 1KB 3088KB

LFSCM3GA115E 3084KB

Important!
If the VME file size is more than 64K bytes, the VME Embedded driver must be
compiled as a 32-bit program. A 16-bit program can only address up to 64K bytes of
memory maximum.

50 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Using JTAG Full VME Embedded

The procedure of generating and processing the VME can be done by using
the Programmer graphical user interface.

Table 1: JTAG Full VME Program Descriptions

Program Description

JTAG Full VME Embedded The file-based JTAG Full VME Embedded is the
(file-based) programming engine that accepts VME files as
command line arguments to process the device(s).

The EPROM-based JTAG Full VME Embedded is the

programming engine that requires compiling a HEX file,
JTAG Full VME Embedded
which is a C-programming file, with the source code to
(EPROM-based)
create an executable engine that can be embedded
onto the embedded system.

svf2vme The svf2vme is a command line utility that can convert

SVF files into VME files.

vme2hex The vme2hex is a command line utility that can convert

VME files into HEX files.

Generating VME Files

A VME file is a variation of an SVF file that has been compressed into a binary
file. It allows you to program a device from the microprocessor on your printed
circuit board. The VME files can be created in Deployment Tool by selecting
Lattice Programmer-generated an XCF file. An XCF file is a configuration file
used by Diamond Programmer and for programming devices in a JTAG daisy
chain. The XCF file contains information about each device, the data files
targeted, and the operations to be performed.

In Deployment Tool, the JTAG Full VME Embedded software will then take the
device chain information and generate the VME file. If a non-Lattice device is
in the chain, you must add a JTAG-SVF device and supply the SVF file. For
chains with JTAG-SVF devices, JTAG Full VME Embedded generates two
VME files. You can use one or both files to program the device.

To generate a VME file:

1. In Programmer, create a project, and add the target devices into the chain
with the appropriate operations and data files. If a non-Lattice device is in
the chain, set the device as a JTAG-SVF device and provide the
appropriate SVF file, SVF vendor, and TCK frequency. Refer to
Programmer online help for more information on how to use Programmer.
2. Save the Programmer project (.xcf).
3. In Deployment Tool, choose Create New Deployment.
4. For Function Type, choose Embedded System.

Programming Tools User Guide 51

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

5. For Output Type, choose JTAG Full VME Embedded, then click OK.
6. In the Step 1 of 4 dialog box, select the XCF file, and click Next.
7. In the Step 2 of 4 dialog box, elect the desired file options. For detailed
option descriptions, including the option that allows you to generate a
HEX (.c) file, see the Deployment Tool online help.
8. Click Next.
9. In the Step 3 of 4 dialog box, in the Output File box, specify the location
and file name of the VME file.
10. Click Next.
11. In the Step 4 of 4 dialog box, click Generate.

Deployment Tool generates the VME file depending upon the options you
have chosen, and returns a message indicating that the process succeeded
or failed.

Testing VME Files

Use the Download Debugger to process the VME file using any of the Lattice
programming cables. Refer to Download Debugger online help for details.
This processor can run through port 0x0378 of the parallel port using the
Lattice download cable.

VME files can also be processed using the command line. See Running the
Deployment Tool from the Command Line online help for details.

Converting an SVF File to VME File

VME files can also be generated the traditional way by using the svf2vme
source code. The utility will expect an SVF file as argument.

Choosing the File-Based or EPROM-

Based Version
To generate a PROM-based VME, select the “Generate HEX (.c) File” option
in the Deployment Tool Step 2 of 4 dialog box.The programming engines of
the file-based and PROM-based processors are identical in the way they
handle the VME commands. Their difference lies in the way they interface
with VME data. For a convenient demo, the file-based version assigns a file
pointer to the binary VME file directly. The pointer is assigned based on a
command line argument. With some minor modification, this version is useful
for embedded high-level 32-bit microprocessors that can dynamically allocate
RAM and have large amounts of data and code memory. For more modest

52 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

embedded systems or smaller processors, the PROM-based version is useful

because the memory resources are completely defined when compiling the
executable.The VME file is converted to one or more C files and a header file
that are compiled with the core routines.

Customizing for the Target Platform

The main routines that will require customization are in the hardware.c file.
They include the routines for reading from and writing to the JTAG pins and a
delay routine. These routines are well commented in hardware.c and are at
the top of the file. In readPort(), a byte of data is read from the input port. In
writePort(), a byte of data is written to the output port. In ispVMDelay(), the
system delays for the specified number of milliseconds or microseconds. The
port mapping is set at the top of the hardware.c file.
The source code files are written in ANSI C. The JTAG Full VME Embedded
source codes are located in the <install_path>\embedded_source directory.

See Also JTAG Full VME Embedded Required User Changes

Advanced Issues
Since SVF files are serial in nature, many vendors have options on the type of
operations to be performed when generating the SVF files. If an SVF file is too
large for the targeted embedded application, consider removing optional
operations or breaking up the operations by creating multiple SVF files. This
approach is much better than arbitrarily dividing the VME file.

EPROM-based JTAG Full VME

Embedded User Flow
This appendix details the steps the user must take to use the EPROM-based
JTAG Full VME Embedded.

Step 1. Create Chain With Programmer

Using Programmer, add the target devices into the chain with the appropriate
operations and data files. If a non-Lattice device is in the chain, set the device
as a JTAG-SVF device and provide the appropriate SVF file, SVF vendor, and
TCK frequency. For more information on supporting non-Lattice devices, see
Programmer’s on-line help documentation.

Step 2. Generate VME File

Use the Deployment Tool to generate the VME file. Refer to the Deployment
Tool online help for more information on Deployment Tool.

Programming Tools User Guide 53

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 16: EPROM-based JTAG Full VME Embedded User Flow

Step 3. Convert VME to HEX

A HEX file can be created from a VME file by using the vme2hex source code
that is shipped with Programmer, or by selecting the Generate HEX (.c) File
option in Deployment Tool. This source code can be found in the installation
path of Programmer, under the
<install_path>\embedded_source\vmembedded\sourcecode\svf2vme. A HEX
file is a C-programming language file that has the VME byte codes converted
and stored in an array.

Step 4. Modify EPROM-based Source Code

The file hardware.c must be modified to target the embedded system. In

particular, the following functions must be changed to be able to write, read,
and observe the delay, respectively:
 void writePort( unsigned char a_ucPins, unsigned char
a_ucValue )
 unsigned char readPort()
 void ispVMDelay(unsigned int delay_time)

Step 5. Compile Source Code and HEX Files

Combine the source code and HEX files into a project to be compiled. This
may be done by using a microcontroller compiler.

54 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Programming Engine Flow

The programming engine of the JTAG Full VME Embedded is driven by the
byte codes of the VME file. The byte codes instruct the programming engine
as to what sequence of functions to follow in order to shift in instructions,
move the TAP controller state machine, shift data in and out of the device, and
observe delays.

The TAP controller is a synchronous state machine that is based on the TMS
(Test Mode Select) and TCK (Test Clock) signals of the TAP and controls the
sequence of operations of the circuitry defined by the IEEE 1149.1 standard.
The TCK signal can be driven at a maximum of 25 MHz. JTAG devices in the
chain may limit the TCK speed. Confirm the Maximum TCK for all the devices
in the programming chain.

Figure 17: TAP Controller State Diagram

In the Shift-DR state, a decoder is present to select which shift register is

enabled and connects it between TDI and TDO. The following are the shift
registers: Address Shift Register, Data Shift Register, 32-bit Shift Register,
and Bypass. The 32-bit Shift Register is used to store the ID code and

Programming Tools User Guide 55

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

USERCODE. The first bit clocked into any of the registers is placed into the
MSB, and data is shifted towards TDO as additional bits are clocked in.

Figure 18: Shift Registers

The engine core is implemented as a switch statement. The cases in this

switch statement perform specialized functions based on the byte code and
its operand(s). These functions may end up calling other switch statements,
calling the engine core recursively, setting global variable values, or
interfacing with the device directly. Once the byte code instruction has been
executed, it returns to the main switch statement to process the next byte.

The processor begins by calculating the 16-bit CRC of the VME file and
comparing it against the expected CRC. If that is successful, the processor
then verifies the version of the VME file to make sure it is supportable. The
version is an eight byte ASCII of the format ____<Major
Version>.<Minor Version>, where <Major Version> and <Minor
Version> are digits 0-9. If the version verification fails, the processor
returns the error code –4 to indicate a file error.

The Main Engine Switch calls the appropriate case statements based on the
incoming byte code from the VME. Unrecognized byte codes will result in the
program exiting with the error code –4 to indicate a file error.

STATE Case Statement

The STATE case statement expects a state following the STATE byte code to
instruct the processor to step the IEEE 1149.1 bus to the next state. The state
must be a valid stable state, which is IRPAUSE, DRPAUSE, RESET, or IDLE

SIR Case Statement

The SIR case statement begins by extracting the size of the register. The size
will be used later to indicate how many bits of data will be sent or read back
from the device. If the flow control has been set to CASCADE, then the
processor shifts the device to the SHIFTIR. The presence of CASCADE in the
flow control indicates that the SIR instruction is targeting over 64Kb of data
and has been broken down to ease the memory requirements.

56 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 19: Main Engine Switch

Figure 20: STATE Case Statement

If CASCADE has not been set, then the processor shifts the device into the
safe state IRPAUSE, and then to SHIFTIR. If HIR exists (see HIR Case
Statement), then the processor will bypass the HIR. The SIR sub-switch is a
switch that is based off of the byte codes that can potentially be found after
the SIR byte code.

The TDI byte code indicates that there is data that needs to be shifted into the
device. The data following the TDI byte will be extracted and decompressed,
if compression were selected, and held in memory until it is ready to be
shifted into the device.

Programming Tools User Guide 57

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 21: SIR Case Statement

The TDO byte code indicates that there is data that needs to be read and
verified from the device. The data following the TDO byte will be extracted
and decompressed, if compression were selected, and held in memory until it
is ready to be compared against data in the device.

The XTDO byte code indicates that the TDO data is the TDI data of the
previous frame, such as in the case of concurrent, or turbo, programming.
Data will not follow the XTDO byte code, resulting in a smaller VME. Instead,
the previous frame’s TDI data will be used as the current TDO data.

The MASK byte code indicates that there is mask data that needs to be used
when comparing the TDO values against the actual values read from the
device. The data following the MASK byte will be extracted and
decompressed, if compression were selected, and held in memory until it is
ready to be used when comparing against data in the device.

The DMASK byte code indicates that there is dynamic mask data that needs
to be used when comparing the boundary scan. The data following the
DMASK byte will be extracted and decompressed, if compression were
selected, and held in memory until it is ready to be used when comparing
against the boundary scan.

The CONTINUE byte code terminates the SIR instruction. When this byte is
encountered, it indicates that the processor is ready to send or read and verify
data from the device using the data it is currently holding in memory. If any
byte codes other than TDI, TDO, XTDO, MASK, DMASK, and CONTINUE
were encountered in the SIR Sub-switch, the program will exit with the error
code –4, indicating a file error.

If the TDO or XTDO byte code were encountered in the SIR sub-switch, then
that indicates that the SIR instruction is going to read data from the device,
else the SIR instruction is going to send data to the device.

If reading and verifying data from the device were successful, the processor
checks if the CASCADE flag has been set. If it is set, the control returns to the
Main Engine Switch. If the flag is off, the processor checks if TIR exists (see
TIR Case Statement). If it exists, then the trailer devices must be bypassed.
Next, it shifts the device to the stable state that followed the ENDIR byte code

58 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 22: SIR Case Statement Continued

(see ENDIR Case Statement). The control returns back to the Main Engine
Switch.

If reading and verifying data from the device were unsuccessful, the
processor checks if the vendor has been set to Xilinx. If the vendor is Xilinx,
repeat the read loop up to 30 times before returning an error. If the vendor is
not Xilinx, the processor bypasses the TIR if there are trailer devices. Next, it
shifts the device to the stable state that followed the ENDIR byte code. The
error code returned is –1 to indicate a verification failure.

If TDO or XTDO were not encountered in the SIR sub-switch, then the
processor sends data to the device. If the CASCADE flag has been set, the
control returns to the Main Engine Switch. If TIR exists, then the trailer
devices must be bypassed. Next, it shifts the device to the stable state that
followed the ENDIR byte code. The control returns back to the Main Engine
Switch.

SDR Case Statement

The SDR case statement works similar to the SIR. It begins by extracting the
size of the register. The size will be used later to indicate how many bits of
data will be sent or read back from the device. If the flow control has been set
to CASCADE, then the processor shifts the device to the SHIFTIR. The
presence of CASCADE in the flow control indicates that the SDR instruction is
targeting over 64Kb of data and has been broken down to ease the memory
requirements.

Programming Tools User Guide 59

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 23: SDR Case Statement

If CASCADE has not been set, then the processor shifts the device into the
safe state DRPAUSE, and then to SHIFTDR. If HDR exists (see HDR Case
Statement), then the processor will bypass the HDR. The SDR sub-switch is a
switch that is based off the byte codes that can potentially be found after the
SDR byte code.

The TDI byte code indicates that there is data that needs to be shifted into the
device. The data following the TDI byte will be extracted and decompressed,
if compression were selected, and held in memory until it is ready to be
shifted into the device.

The TDO byte code indicates that there is data that needs to be read and
verified from the device. The data following the TDO byte will be extracted
and decompressed, if compression were selected, and held in memory until it
is ready to be compared against data in the device.

he XTDO byte code indicates that the TDO data is the TDI data of the
previous frame, such as in the case of concurrent, or turbo, programming.
Data will not follow the XTDO byte code, resulting in a smaller VME. Instead,
the previous frame’s TDI data will be used as the current TDO data.

The MASK byte code indicates that there is mask data that needs to be used
when comparing the TDO values against the actual values read from the
device. The data following the MASK byte will be extracted and
decompressed, if compression were selected, and held in memory until it is
ready to be used when comparing against data in the device.

The DMASK byte code indicates that there is dynamic mask data that needs
to be used when comparing the boundary scan. The data following the
DMASK byte will be extracted and decompressed, if compression were
selected, and held in memory until it is ready to be used when comparing
against the boundary scan.

The CONTINUE byte code terminates the SDR instruction. When this byte is
encountered, it indicates that the processor is ready to send or read and verify
data from the device using the data it is currently holding in memory. If any
byte codes other than TDI, TDO, XTDO, MASK, DMASK, and CONTINUE
were encountered in the SDR Sub-switch, the program will exit with the error
code –4, indicating a file error.

60 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 24: SDR Case Statement Continued

If the TDO or XTDO byte code were encountered in the SDR sub-switch, then
that indicates that the SDR instruction is going to read data from the device,
else the SDR instruction is going to send data to the device.

If reading and verifying data from the device were successful, the processor
checks if the CASCADE flag has been set. If it is set, the control returns to the
Main Engine Switch. If the flag is off, the processor checks if TDR exists (see
TDR Case Statement). If it exists, then the trailer devices must be bypassed.
Next, it shifts the device to the stable state that followed the ENDDR byte
code (see ENDDR Case Statement). The control returns back to the Main
Engine Switch.

If reading and verifying data from the device were unsuccessful, the
processor checks if the vendor has been set to Xilinx. If the vendor is Xilinx,
repeat the read loop up to 30 times before returning an error. If the vendor is
not Xilinx, the processor bypasses the TIR if there are trailer devices. Next, it
shifts the device to the stable state that followed the ENDIR byte code. The
error code returned is –1 to indicate a verification failure.

If TDO or XTDO were not encountered in the SDR sub-switch, then the
processor sends data to the device. If the CASCADE flag has been set, the
control returns to the Main Engine Switch. If TDR exists, then the trailer
devices must be bypassed. Next, it shifts the device to the stable state that
followed the ENDDR byte code. The control returns back to the Main Engine
Switch.

Programming Tools User Guide 61

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

XSDR Case Statement

The XSDR case statement works exactly like the SDR case statement, except
that it sets the EXPRESS flag. The EXPRESS flag indicates to the processor
that the VME is performing concurrent programming. Therefore, the TDO data
shall use the previous frame’s TDI data. This reduces the VME size drastically
because the data is not duplicated.

WAIT Case Statement

The WAIT case statement expects a number following the WAIT byte code to
represent the number of milliseconds of delay the device must observe. The
delay is observed immediately. The user must update the delay function in the
source code to make the target embedded system observe the delay duration
correctly.

Figure 25: WAIT Case Statement

TCK Case Statement

The TCK case statement expects a number following the TCK byte code to
represent the number of clocks that the device must remain in the run test idle
state. The clock cycles are executed immediately into the device.

Figure 26: TCK Case Statement

ENDDR Case Statement

The ENDDR case statement expects a state following the ENDDR byte code
to represent the TAP state that the processor shall move the device to after an
SDR instruction. This state will be stored in a global variable.

62 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 27: ENDDR Case Statement

ENDIR Case Statement

The ENDIR case statement expects a state following the ENDIR byte code to
represent the TAP state that the processor shall move the device to after an
SIR instruction. This state will be stored in a global variable.

Figure 28: ENDIR Case Statement

HIR Case Statement

The HIR case statement expects a number following the HIR byte code to
represent the number of header devices. The number will be stored in a
global variable and the processor will issue this number of bypasses prior to
executing an SIR instruction.

Figure 29: HIR Case Statement

TIR Case Statement

The TIR case statement expects a number following the TIR byte code to
represent the number of trailer devices. The number will be stored in a global
variable and the processor will issue this number of bypasses after executing
an SIR instruction.

Figure 30: TIR Case Statement

Programming Tools User Guide 63

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

HDR Case Statement

The HDR case statement expects a number following the HDR byte code to
represent the number of header register bits. The number will be stored in a
global variable and the processor will issue this number of bypasses prior to
executing an SDR instruction.

Figure 31: HDR Case Statement

TDR Case Statement

The TDR case statement expects a number following the TDR byte code to
represent the number of trailer register bits. The number will be stored in a
global variable and the processor will issue this number of bypasses after
executing an SDR instruction.

Figure 32: TDR Case Statement

MEM Case Statement

The MEM case statement expects a number following the MEM byte code to
represent the maximum frame size in bits. Memory buffers will be allocated for
TDI, TDO, MASK, and DMASK data according to the maximum number.

Figure 33: MEM Case Statement

VENDOR Case Statement

The VENDOR case statement expects the vendor type following the
VENDOR byte code to represent the vendor the VME supports. Different
vendors require different programming algorithms that must be supported.

64 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

This byte notifies the embedded processor to enable the specified vendor
support.

Figure 34: VENDOR Case Statement

SETFLOW Case Statement

The SETFLOW case statement expects an instruction following the
SETFLOW byte code to instruct the embedded processor to enable certain
properties during execution. This is useful for cascading and looping VME
files, where the processor flow must change in order to take advantage of
these features.

Figure 35: SETFLOW Case Statement

RESETFLOW Case Statement

The RESETFLOW case statement works to reset the properties enabled
during the SETFLOW case statement.

Figure 36: RESETFLOW Case Statement

HEAP Case Statement

The HEAP case statement expects a number following the HEAP byte code to
indicate the size of the upcoming repeat loop. In the file-based JTAG Full
VME Embedded, this size is used to dynamically allocate memory to hold the
repeat loop. In the EPROM-based embedded, the heap array is set to point to
the heap buffer in the HEX file.

Programming Tools User Guide 65

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 37: HEAP Case Statement

REPEAT Case Statement

The REPEAT case statement is executed if the VME were generated with the
looping option. A looping VME attempts to reduce the VME size by forming
loops around similar algorithm. Following the REPEAT byte code, a number
indicating the number of repeats is extracted. The heap buffer is build by
reading the number of HEAP size (see HEAP case statement) bytes and
storing them in memory. Recursive calls are made back to the Main Engine
Switch, which will process the byte codes within the heap buffer. The
recursive calls end when the repeat size is zero.

Figure 38: REPEAT Case Statement

ENDLOOP Case Statement

The ENDLOOP byte code terminates a loop iteration and shall be
encountered only when the embedded processor is processing a repeat loop.
This byte code shall always be the last byte of the heap buffer. When this byte
code is found, the control returns back to the looping control, where the repeat
size gets decremented and the next iteration of the loop begins, unless the
repeat size is zero.

66 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Figure 39: ENDLOOP Case Statement

ENDVME Case Statement

The ENDVME case statement exits the main engine switch. This byte code is
the last byte of the VME.

Figure 40: ENDVME Case Statement

SHR Case Statement

The SHR case statement expects a number following the SHR byte code to
perform a right shift on the TDI data buffer. At this point the TDI data buffer
should store the register address. By simply right shifting the register address
to increment to the next frame instead of having the VME contain several
register address buffers, the VME size is reduced.

Figure 41: SHR Case Statement

SHL Case Statement

The SHL case statement works similar to the SHR case statement, except
that it shifts to the left.

Figure 42: SHL Case Statement

Programming Tools User Guide 67

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

FREQUENCY Case Statement

The FREQUENCY case statement expects a number following the
FREQUENCY byte code to establish the TCK frequency.

Figure 43: FREQUENCY Case Statement

VUES Case Statement

The VUES case statement sets the flow control register to indicate that the
VME is invoking the Continue If Fail feature. Under this condition, if the
USERCODE verification fails, then the embedded processor continues with
programming the data. If the USERCODE verification passes, then the
processor exits without programming.

COMMENT Case Statement

The COMMENT case statement is executed if the VME file were generated to
support SVF comments. This statement expects a number to indicate the size
of the comment. The comment is then read one byte at a time and displayed
onto the terminal. It ends when the number of bytes processed equals the
number indicating the size of the comment.

Figure 44: COMMENT Case Statement

HEADER Case Statement

The HEADER case Statement is executed if the VME file were generated with
header information. Currently, this feature is not supported.

Figure 45: HEADER Case Statement

68 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

LCOUNT Case Statement

The LCOUNT case statement is executed if the VME file targets FLASH or
PROM devices. It allows the engine to repeatedly check the status of the
device before programming the next block of data. This statement expects a
number to indicate the number of status checks before issuing a failure return
code. The engine will use an index to point to the repeated commands in a
buffer and issue them to the device. The index is reset after each iteration.
This will continue until the number of status checks gets decremented to zero,
or until the status is verified to be true.

Figure 46: LCOUNT Case Statement

LVDS Case Statement

The LVDS case statement informs the processor about the number of LVDS
pairs and which are paired. This ensures that the processor will drive opposite
values back into the pairs.

VME Byte Codes

Appendix C lists the byte codes that are found in the VME and interpreted by
the embedded processor.

General Opcode Value Description

VMEHEXMAX 60000L Sets the HEX file maximum size to 60K
SCANMAX 64000L Sets the maximum data burst to 64K

Programming Tools User Guide 69

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Formatting Opcode Value Description

CONTINUE 0x70 Indicates the end of a VME line
ENDVME 0x7F Indicates the end of a VME file
ENDFILE 0xFF Indicates the end of file

JTAG Opcode Value Description

RESET 0x00 Traverse to TLR
IDLE 0x01 Traverse to RTI
IRPAUSE 0x02 Traverse to PAUSE IR
DRPAUSE 0x03 Traverse to PAUSE DR
SHIFTIR 0x04 Traverse to SHIFT IR
SHIFTDR 0x05 Traverse to SHIFT DR

Flow Control
Value Description
Opcode
INTEL_PRGM 0x0001 Intelligent programming in effect
CASCADE 0x0002 Currently splitting large SDR
REPEATLOOP 0x0008 Currently executing a repeat loop
Indicates the next stream needs a right
SHIFTRIGHT 0x0080
shift
Indicates the next stream needs a left
SHIFTLEFT 0x0100
shift
VERIFYUES 0x0200 Indicates Continue If Fail flag

Data Type Register

Value Description
Opcode
EXPRESS 0x0001 Simultaneous program and verify
SIR_DATA 0x0002 SIR is the active SVF command
SDR_DATA 0x0004 SDR is the active SVF command
COMPRESS 0x0008 Data is compressed
TDI_DATA 0x0010 TDI data is present
TDO_DATA 0x0020 TDO data is present
MASK_DATA 0x0040 MASK data is present
HEAP_IN 0x0080 Data is from the heap
LHEAP_IN 0x0200 Data is from the intelligent data buffer
VARIABLE 0x0400 Data is from a declared variable
CRC_DATA 0x0800 CRC data is present
CMASK_DATA 0x1000 CMASK data is present
RMASK_DATA 0x2000 RMASK data is present
READ_DATA 0x4000 READ data is present
DMASK_DATA 0x8000 DMASK data is present

Hardware Opcode Value Description

signalENABLE 0x1C Assert the ispEN pin
signalTMS 0x1D Assert the MODE or TMS pin
signalTCK 0x1E Assert the SCLK or TCK pin
signalTDI 0x1F Assert the SDI or TDI pin
signalTRST 0x20 Assert the RESTE or TRST pin

Vendor Opcode Value Description

VENDOR 0x56 Indicates vendor opcode is following
LATTICE 0x01 Indicates Lattice or JTAG device
ALTERA 0x02 Indicates Altera device

70 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

XILINX 0x03 Indicates Xilinx device

SVF Opcode Value Description
ENDDATA 0x00 Indicates the end of the current SDR data stream
RUNTEST 0x01 Indicates the duration to stay at the stable state
ENDDR 0x02 Indicates the stable state after SDR
ENDIR 0x03 Indicates the stable state after SIR
ENDSTATE 0x04 Indicates the stable state after RUNTEST
TRST 0x05 Assert the TRST pin
HIR 0x06 Specifies the sum of IR bits at lead
TIR 0x07 Specifies the sum of IR bits at end
HDR 0x08 Specifies the number of devices at lead
TDR 0x09 Specifies the number of devices at end
ispEN 0x0A Assert the ispEN pin
FREQUENCY 0x0B Specifies the maximum clock rate to run the state machine
STATE 0x10 Move to the next stable state
SIR 0x11 Indicates the instruction stream is following
SDR 0x12 Indicates the data stream is following
TDI 0x13 Indicates the data stream following is fed into the device
TDO 0x14 Indicates the data stream is to be read and compare
MASK 0x15 Indicates the data stream following is the output mask
Indicates the data stream following is for simultaneous shift in and
XSDR 0x16
shift out
Indicates the data stream following is for shift in only and it must
XTDI 0x17
be stored for verifying on the next XSDR call
Indicates there is no data stream following, instead it should be
XTDO 0x18
retrieved from the previous XTDI token
MEM 0x19 Indicates the size of the memory needed to be allocated.
WAIT 0x1A Indicates the duration of the delay at IDLE state
TCK 0x1B Indicates the number of clocks to pulse to TCK
HEAP 0x32 Indicates the size of the memory needed to hole the loop
REPEAT 0x33 Indicates the beginning of a reap loop
LEFTPAREN 0x35 Indicates the beginning of the data following the loop
Indicates a place holder for the data if looping option has been
VAR 0x55
selected
SEC 0x1C Indicates the absolute time in seconds that must be observed
SMASK 0x1D Indicates the mask for TDI data
MAX 0x1E Indicates the absolute maximum wait time
ON 0x1F Assert the targeted pin
OFF 0x20 Dis-assert the targeted pin
SETFLOW 0x30 Change the Flow Control Register
RESETFLOW 0x31 Clear the Flow Control Register
CRC 0x47 Indicates which bits may be used in CRC calculation
CMASK 0x48 Indicates which bits shall be used in CRC calculation
RMASK 0x49 Indicates which bits shall be used in Read and Save
READ 0x50 Indicates which bits may be used in Read and Save
ENDLOOP 0x59 Indicates the end of the repeat loop
SECUREHEAP 0x60 Byte encoded to secure the HEAP structure
SVF Opcode Value Description
VUES 0x61 Indicates Continue If Fail option has been selected
DMASK 0x62 Indicates SVF file has DMASK
COMMENT 0x63 Support SVF comments in VME file
HEADER 0x64 Support header in VME file
FILE_CRC 0x65 Support CRC-protected VME file
LCOUNT 0x66 Support intelligent programming.
LDELAY 0x67 Support intelligent programming.
LSDR 0x68 Support intelligent programming.
LHEAP 0x69 Memory needed to hold intelligent data buffer
LVDS 0x71 Support LVDS

Return Codes Value Description

Programming Tools User Guide 71

EMBEDDED FLOW OVERVIEW : JTAG Full VME Embedded

Value returned when the expected data does

VME_VERIFICATION_ERROR -1
not match the actual data of the device
Value returned when the VME file cannot be
VME_FILE_READ_FAILURE -2
read
Value returned when the version is not sup-
VME_VERSION_FAILURE -3
ported
Value returned when an invalid opcode is
VME_INVALID_FILE -4
encountered
Value returned when a command line argument
VME_ARGUMENT_FAILURE -5
is invalid
Value returned when the expected CRC does
VME_CRC_FAILURE -6
not match the calculated CRC.

Unsupported SVF Syntax

The following are the SVF syntax not supported by the SVF2VME utility:
 TRST - The TRST command is ignored.
 PIO - The PIO command will cause SVF2VME to exit with an error.
 PIOMAP - The PIOMAP command will cause SVF2VME to exit with an
error.
 MAXIMUM - The optional parameter MAXIMUM is ignored. This may be
found in the RUNTEST command.
 SMASK - The optional parameter SMASK is ignored. This may be found
in the HDR, HIR, TDR, TIR, SIR, or SDR commands.
 Explicit state transitions in the STATE command that contain non-stable
states will cause SVF2VME to exit with an error. Only transitions between
stable states are supported in the table below.
 STATE RESET is supported. However, it is strongly discouraged to be
included into the SVF file. This statement causes the undesirable effect of
having all the devices in the entire JTAG chain to be reseted.

72 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

The following table indicates the paths taken between stable states.

Current State New State State Path

RESET RESET RESET (NO CLOCK)
RESET IDLE RESET-IDLE
RESET DRPAUSE RESET-IDLE-DRSELECT-DRCAPTURE-DREXIT1-DRPAUSE
RESET-IDLE-DRSELECT-IRSELECT-IRCAPTURE-IREXIT1-
RESET IRPAUSE
IRPAUSE
IDLE RESET IDLE-DRSELECT-IRSELECT-RESET
IDLE IDLE IDLE (NO CLOCK)
IDLE DRPAUSE IDLE-DRSELECT-DRCAPTURE-DREXIT1-DRPAUSE
IDLE IRPAUSE IDLE-DRSELECT-IRSELECT-IRCAPTURE-IREXIT1-IRPAUSE
DRPAUSE RESET DRPAUSE-DREXIT2-DRUPDATE-DRSELECT-IRSELECT-RESET
DRPAUSE IDLE DRPAUSE-DREXIT2-DRUPDATE-IDLE
DRPAUSE DRPAUSE DRPAUSE (NO CLOCK)
DRPAUSE-DREXIT2-DRUPDATE-DRSELECT-IRSELECT-
DRPAUSE IRPAUSE
IRCAPTURE-IREXIT1-IRPAUSE
IRPAUSE RESET IRPAUSE-IREXIT2-IRUPDATE-DRSELECT-IRSELECT-RESET
IRPAUSE IDLE IRPAUSE-IREXIT2-IRUPDATE-IDLE
IRPAUSE-IREXIT2-IRUPDATE-DRSELECT-DRCAPTURE-
IRPAUSE DRPAUSE
DREXIT1-DRPAUSE
IRPAUSE IRPAUSE IRPAUSE (NO CLOCK)

JTAG Slim VME Embedded

The JTAG Slim VME Embedded software, based on the serial vector format
file, enables you to quickly and efficiently program chains of devices, thus
improving productivity and lowering costs. The JTAG Slim VME Embedded
code is designed for microcontrollers with limited resources, such as the 8051
microcontroller.

The JTAG Slim VME Embedded software behaves the same as the JTAG Full
VME Embedded. The difference is it is geared to a 8051 processor. The C
code adds memory space keywords specific to the 8051 processor. The size
of the devices which can be programmed are limited by the amount of
contiguous SRAM available to the 8051 processor.

The JTAG Slim VME Embedded is available with installations of Diamond

Programmer. Its advantages over other embedded systems include:
 Footprint of less than 3KB ROM – The small footprint is made possible
by optimizing the JTAG Slim VME Embedded programming engine to use
the least amount of code in the most efficient fashion.
 Reduced RAM usage – The memory usage is fixed at a minimal set for
all IEEE 1532-compliant devices. The number of global and local
variables has been reduced to a minimum, and no data buffers are
required to be held in memory.
 Compressible algorithm and data – The programming data, calculated
by multiplying the frame size by the number of frames, can increase the
ROM requirement substantially. For example, the device LC51024MV(B)
has a frame size of 2624 with 388 frames, resulting in 125 KB of ROM.
Fortunately, the JTAG Slim VME Embedded can compress the
programming data into sizes that are much smaller. The compression is
performed frame by frame and is data file dependent.

Programming Tools User Guide 73

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

 Sequential chain programming – The JTAG Slim VME Embedded can

process multiple devices in the same chain, with mixed operations in
sequential mode.

See Also Using the PC-based JTAG Slim VME Embedded

 Using the 8051-based JTAG Slim VME Embedded

JTAG Slim VME Embedded Source

Code
The source code for both the PC-based and the 8051-based JTAG Slim VME
Embedded can be found in the
<install_path>\embedded_source\slimembedded directory.

Each project has the following files. The major entry point for JTAG Slim VME
Embedded is slim_vme.c.

slim_vme.c The slim_vme.c file is the only file that differs between the PC-
based and the 8051-based embedded solutions. This difference is due to the
way each of these interfaces to the VME algorithm and data files through the
entry point. This file contains the main and entry point functions.

slim_pro.c The slim_pro.c file provides the programming engine for the
JTAG Slim VME Embedded. The engine operates on the commands in the
VME algorithm, and fetches data from the VME data, if necessary. The engine
is responsible for functions such as sending data, verifying data, observing
timing delay, stepping through the state machine, decompression, and so on.

hardware.c The only file that you should modify is hardware.c. This file
contains the functions to read and write to the port and the timing delay
function. You must update these functions to target the desired hardware. The
released version targets the parallel port of the PC at address 0x0378 using
Lattice’s download cable.

opcode.h The opcode.h file contains the definitions of the byte codes used
in the VME algorithm format and programming engine.

debug.h The debug.h file will print out debugging information if the
preprocessor switch VME_DEBUG is defined in the project. This is an
optional file to include.

windriver.c and windriver.h The windriver.c and windriver.h files target the
JTAG Slim VME Embedded to Windows. These files will be compiled if the
preprocessor switch VME_WINDOWS is defined in the project file. These files
should be omitted when compiling the 8051-based JTAG Slim VME
Embedded onto an embedded platform.

See Also VME Algorithm Format

 VME Data Format

74 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Using the PC-based JTAG Slim VME

Embedded
The PC-based JTAG Slim VME Embedded is a quick and easy way to
validate the VME files and the JTAG Slim VME Embedded programming
engine by successfully processing the target chain of IEEE 1532 compliant
devices using the parallel port of the PC.

The JTAG Slim VME Embedded system uses a compressed binary variation
of SVF files, called VME, as input. Like the SVF file, the VME file contains
high-level IEEE 1149.1 bus operations. These operations consist of scan
operations and movements between the IEEE 1149.1 TAP states. However,
unlike the SVF file, where the programming algorithm of the device is
intermeshed with the programming data, the VME file is separated into a VME
algorithm file and a VME data file. This separation of the algorithm and data
allows the optimization of the JTAG Slim VME Embedded programming
engine. It also allows you to mix VME data files with VME algorithm files,
provided the chain and operations are the same.

Figure 9 shows an example of Slim VME embedded file generation for the
JTAG port.

Figure 47: Slim VME Embedded VME Flow

The JTAG Slim VME Embedded capability is enabled only if all the following
conditions are met:
 All the devices in the chain are IEEE-1532 compliant.
 Sequential mode is selected.
 Synchronize Enable and Disable setting is unchecked.
 Operation is not Read and Save or a display operation such as Calculate
Checksum or Display ID.

See Also Generating JTAG Slim VME Embedded Files

Programming Tools User Guide 75

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Using the 8051-based JTAG Slim VME

Embedded
To program embedded systems using the 8051-based JTAG Slim VME
Embedded, you must generate the VME files as HEX to create the VME
algorithm and data files as C programming files. Each file contains a C
programming style byte buffer that holds the VME algorithm or data.

The HEX files must be compiled along with the 8051-based JTAG Slim VME
Embedded source code. The source code contains handles that allow the
compiler to link the buffers of the hexadecimal files to the main source code.
The only source code file that you need to modify is the hardware.c file. You
must implement methods to write and read to the hardware port, as well as
observe the timing delay. You must modify the following functions according to
the target platform:
 readPort
 writePort
 ispVMDelay

The following are optional functions that you may wish to modify in the
hardware.c file in order to enable and disable the hardware conditions before
and after processing:
 EnableHardware
 DisableHardware

See Also Generating JTAG Slim VME Embedded Files

 JTAG Slim VME Embedded Source Code

VME Algorithm Format

The VME algorithm file contains byte codes that represent the programming
algorithm of the device or chain.

VME Symbol HEX Value

STATE 0x01

SIR 0x02

SDR 0x03

TCK 0x04

WAIT 0x05

ENDDR 0x06

ENDIR 0x07

HIR 0x08

76 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

VME Symbol HEX Value

TIR 0x09

HDR 0x0A

TDR 0x0B

BEGIN_REPEAT 0x0C

FREQUENCY 0x0D

TDI 0x0E

CONTINUE 0x0F

END_FRAME 0x10

TDO 0x11

MASK 0x12

END_REPEAT 0x13

DATA 0x14

PROGRAM 0x15

VERIFY 0x16

ENDVME 0x17

DTDI 0x18

DTDO 0x19

The byte codes perform the same operations as the SVF commands, with the
exception of BEGIN_REPEAT, CONTINUE, END_FRAME, END_REPEAT,
DATA, PROGRAM, VERIFY, ENDVME, DTDI, and DTDO.

The byte codes BEGIN_REPEAT, END_REPEAT, PROGRAM, VERIFY,

DTDI, and DTDO are used to support a repeating VME algorithm structure to
minimize the algorithm size, a feature that the linear SVF does not provide.

The byte code CONTINUE appears at the end of every SIR and SDR
instruction as a terminator.

The byte code END_FRAME appears at the end of every frame in the VME
data as a terminator.

Translation from the SVF file to VME algorithm file is done command by
command. For example, the following SVF line:
SIR 8 TDI (16);

will be converted to the following VME line, in binary:

0x02 0x08 0x0E 0x68

The VME Algorithm file is similar to the SVF file with the following differences:

Programming Tools User Guide 77

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

 VME Algorithm uses byte codes from the table below to represent SVF
commands
 Fuse data and USERCODE have been removed
 Looping algorithm

The following is an example of an EPV VME Algorithm file and the SVF
translation for the LC4064V device:
Table 2: VME Algorithm Example
VME Algorithm Format Serial Vector Format (SVF) Description
0x0A 0x00 HDR 0;
0x08 0x00 HIR 0;
0x0B 0x00 TDR 0;
0x09 0x00 TIR 0;
0x06 0x03 ENDDR DRPAUSE;
0x07 0x02 ENDIR IRPAUSE;
0x01 0x01 STATE IDLE;
0x02 0x08 0x0E 0x68 0x0F SIR 8 TDI (16); Shift in the IDCODE
instruction
0x01 0x01 STATE IDLE;
0x03 0x20 0x0E 0xFF 0xFF 0xFF SDR 32 TDI (FFFFFFFF) TDO Verify the IDCODE
0xFF 0x11 0xC2 0x09 0x01 0x80 (01809043) MASK (0FFFFFFF);
0x12 0xFF 0xFF 0xF0 0x0F
0x02 0x08 0x0E 0x38 0x0F SIR 8 TDI (1C); Shift in the PRELOAD
instruction
0x03 0x44 0x0E 0x00 0x00 0x00 SDR 68 TDI (00000000000000000); Shift all zero data into
0x00 0x00 0x00 0x00 0x00 0x00 boundary scan cells
0x0F
0x02 0x08 0x0E 0xA8 0x0F SIR 8 TDI (15); Shift in ENABLE instruction
0x01 0x01 0x04 0x03 0x05 0x14 RUNTEST IDLE 3 TCK 2.00E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x02 0x08 0x0E 0xC0 0x0F SIR 8 TDI (03); Shift in ERASE instruction
0x01 0x01 0x04 0x03 0x05 0x64 RUNTEST IDLE 3 TCK 1.00E-001 SEC; Execute RUNTEST instruction
0x01 0x01
0x02 0x08 0x0E 0x84 0x0F SIR 8 TDI (21); Shift in ADDRESS INIT
instruction
0x01 0x01 STATE IDLE;
0x02 0x08 0x0E 0xE4 0x0F SIR 8 TDI (27); Shift in PROGRAM INCR
instruction
0x0C 0x5F 0x15 N/A Begin PROGRAM repeat loop
of size 95
VME Algorithm Format Serial Vector Format (SVF) Description
0x03 0xE0 0x02 0x18 0x14 0x0F SDR 352 DTDI (DATA); Notice the forth byte is
0x18, which is actually
DTDI. DTDI instructs the
processor to send in data
from the data buffer
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x13 N/A Terminate the repeat
algorithm
0x02 0x08 0x0E 0x58 0x0F SIR 8 TDI (1A); Shift in PROGRAM USERCODE
instruction
0x03 0x20 0x18 0x14 0x0F SDR 32 DTDI (DATA); Shift in the USERCODE The
USERCODE can be found in
the data buffer.
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x02 0x08 0x0E 0x80 0x0F SIR 8 TDI (01); Shift in ADDRESS SHIFT
instruction

78 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Table 2: VME Algorithm Example (Continued)

0x03 0x5F 0x0E 0x00 0x00 0x00 SDR 95 TDI Shift in beginning address
0x00 0x00 0x00 0x00 0x00 0x00 (400000000000000000000000);
0x00 0x00 0x02 0x0F
0x02 0x08 0x0E 0x54 0x0F SIR 8 TDI (2A); Shift in READ INC
instruction
0x0C 0x5F 0x16 N/A Begin VERIFY repeat loop of
size 95
0x01 0x01 0x04 0x03 0x05 0x01 RUNTEST IDLE 3 TCK 1.00E-003 SEC; Execute RUNTEST instruction
0x01 0x01
0x03 0xE0 0x02 0x0E 0x00 0x00 SDR 352 TDI (0000000000000 Verify the frame against
0x00 0x00 0x00 0x00 0x00 0x00 00000000000000000000000000 the data in the data buffer
0x00 0x00 0x00 0x00 0x00 0x00 00000000000000000000000000
0x00 0x00 0x00 0x00 0x00 0x00 00000000000000000000000) DTDO
0x00 0x00 0x00 0x00 0x00 0x00 (DATA);
0x00 0x00 0x00 0x00 0x00 0x00
0x00 0x00 0x00 0x00 0x00 0x00
0x00 0x00 0x19 0x14 0x0F
0x13 N/A Terminate the repeat
algorithm
0x02 0x08 0x0E 0xE8 0x0F SIR 8 TDI (17); Shift in USERCODE
instruction
0x03 0x20 0x0E 0xFF 0xFF 0xFF SDR 32 TDI (FFFFFFFF) Verify the USERCODE against
0xFF 0x19 0x14 0x0F DTDO (DATA); the USERCODE in the data
buffer
0x02 0x08 0x0E 0xF4 0x0F SIR 8 TDI (2F); Shift in PROGRAM DONE
instruction
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x01 0x01 STATE IDLE;
0x02 0x08 0x0E 0x78 0x0F SIR 8 TDI (1E); Shift in DISABLE
instruction
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x17 N/A End VME Algorithm

See Also JTAG Slim VME Embedded Source Code

 VME Data Format
 Generating JTAG Slim VME Embedded Files

VME Data Format

While the VME algorithm file contains the programming algorithm of the
device, the VME data file contains the fuse and USERCODE patterns.

The first byte in the file indicates whether the data file has been compressed.
A byte of 0x00 indicates that no compression was selected, and 0x01
indicates that compression was selected. When compression has been
selected, each frame is preceded by a frame compression byte to indicate
whether the frame is compressible. This is necessary because even though
you might elect to compress the VME data file, it is possible that a
compressed frame will actually be larger than an uncompressed frame. When
that happens, the frame is not compressed at all and the frame compression

Programming Tools User Guide 79

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

byte of 0x00 notifies the processor that no compression was performed on the
frame.

When compression has not been selected, the VME data file becomes a
direct translation from the data sections of the SVF file. The END_FRAME
byte, 0x10, is appended to the end of every frame as a means for the
processor to verify that the frame has indeed reached the end.

Uncompressed VME Data Format Compressed VME Data Format

0x00 0x01
<Frame 1>0x10 <Compress Byte><Frame 1>0x10
<Frame 2>0x10 <Compress Byte><Frame 2>0x10
… …
<Frame N>0x10 <Compress Byte><Frame N>0x10

The compression scheme used is based on the consecutive appearance of

the 0xFF byte in a frame. This byte is ubiquitous because an all 0xFF data file
is a blank pattern. When a consecutive number of n 0xFF bytes are
encountered, the VME data file will have the byte 0xFF followed by the
number n converted to hexadecimal, where n cannot exceed 255. For
example, if the following were a partial data frame
FFFFFFFFFFFFFFFFFFFF12FFFFFF

the resulting compressed data would be

0xFF 0x0A 0x12 0xFF 0x03

When the processor encounters the first byte 0xFF, it gets the next byte to
determine how many times 0xFF is compressed. The next byte is 0x0A,
which is ten in hexadecimal. This instructs the processor that 0xFF is
compressed ten times. The following byte is 0x12, which is processed as it is.
The next byte is again 0xFF followed by 0x03, which instructs the processor
that 0xFF is compressed three times.

See Also JTAG Slim VME Embedded Source Code

 VME Algorithm Format
 Generating JTAG Slim VME Embedded Files

VME Required User Changes

To make the JTAG Full VME Embedded or JTAG Slim VME Embedded
software work on your target system, you need to modify the following C
functions in the hardware.c source code.

80 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Timer The engine requires the ability to delay for fixed time periods. The
minimum granularity of the delay is 1 microsecond. You can determine the
type of delay. This can be a simple software timing loop, a hardware timer, or
an operating system call, for example, sleep().

Port Initialization The firmware needs to place the port I/O into a known
state. The programming software assumes this has occurred.

Get Data Byte The engine calls the GetByte() function to collect one byte
from the VME or CPU bytestream.

Modify Port Register The engine, as it parses the bitstream data, changes
an in-memory copy of the data to be written onto the I/O pins. Calls to this
function do not modify the I/O pins. The engine uses virtual types (for
example, DONE_PIN) which this function turns into physical I/O pin locations
(for example, 0x400).

Output Data Byte The engine calls this function to write the in-memory
copy onto the I/O pins.

Input Status This function is used by the engine to read back programming
status information. The function translates physical pin locations (for example,
0x400) into virtual types used by the engine (for example, DONE_PIN).

Output Configuration Pins Some systems may wish to use the FPGA
CFG pins. There is a separate function call to manipulate the CFG pins.

Bitstream Initialization You must determine how you plan to get the
bitstream into your memory system, pre-compiled, HEX file based, or
dynamically installed. Whichever method you use the data structures which
pin to the bitstream need to be initialized prior to the first GetByte function call.

See Also Customizing for the Target Platform

 VME Source Code
 JTAG Slim VME Embedded Source Code

Program Memory Requirement

The following figure illustrates the JTAG Slim VME Embedded program
memory requirement.

Programming Tools User Guide 81

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Program Memory Allocation

The following figure illustrates the JTAG Slim VME Embedded program
memory allocation.

82 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Sample Program Size

This page provides sample program size for JTAG Full VME Embedded,
JTAG Slim VME Embedded, and sysCONFIG Embedded.

JTAG Port non-JTAG Port Total

JTAG Full VME JTAG Slim sysCONFIG

Embedded VME Embedded
Embedded

32-Bit 16-Bit 8-Bit 32-Bit 16-Bit

File Based 52KB 21KB 4.2KB 48KB 19KB As Shown

(Bitstream File External)

PROM Based 52KB 21KB 4.2KB 48KB 19KB As Shown +

(Bitstream File Integrated) VME File Size

VME File Size

Refer to “VME File Size” on page 43 for a table that compares VME file sizes
taking typical Lattice devices for examples.

Programming Tools User Guide 83

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Generating JTAG Slim VME Embedded

Files
The Slim VME files can be generated by using Deployment Tool as described
as follows.

Figure 48: Slim VME File Generation Flow

In Programmer, create a project, and add the target Lattice IEEE 1532
compliant devices into the chain with the appropriate operations and data
files. Refer to Programmer online help for more information on how to use
Programmer.
1. Save the Programmer project (.xcf).
2. In Deployment Tool, choose Create New Deployment.
3. For Function Type, choose Embedded System.
4. For Output Type, choose JTAG Slim VME Embedded, then click OK.
5. In the Step 1 of 4 dialog box, select the XCF file, and click Next.
6. To have the software check for a USERCODE match between the device
and the VME file before programming, select the Verify USERCODE,
Program Device if Fails option.

Note
Synchronize Enable and Disable has been turned on while using Sequential mode, the
software will force the VME file into Turbo mode.

7. In the Step 2 of 4 dialog box, elect the desired file options. For detailed
option descriptions, including the option that allows you to generate a
HEX (.c) file, see the Deployment Tool online help. To significantly reduce
the ROM required for storing the VME Data buffer in the embedded
system, select Compress VME File.
8. Click Next.
9. In the Step 3 of 4 dialog box, in the Output File box, specify the location
and file name of the VME algorithm and data files.
10. Click Next.
11. In the Step 4 of 4 dialog box, click Generate.

84 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Deployment Tool generates the VME files depending upon the options you
have chosen, and returns a message indicating that the process succeeded
or failed.

JTAG Slim VME Embedded Source

Code
Both the PC and 8051-based JTAG Slim VME Embedded source code can be
found in the installation path of Programmer under the
<install_path>\embedded_source\slimembedded\sourcecode directory.

Each project requires the following files:

slim_vme.c
The file slim_vme.c is the only file to differ between the PC-based and 8051-
based embedded solutions. This difference is due to the way each interfaces
to the VME Algorithm and Data files through the entry point. This file contains
the main and entry point functions.

slim_pro.c
The file slim_pro.c provides the programming engine of the JTAG Slim VME
Embedded. The engine operates on the commands in the VME Algorithm,
and fetches data from the VME Data if necessary. The engine is responsible
for functions such as sending data, verifying data, observing timing delay,
stepping through the state machine, decompression, and so on.

hardware.c
The only file that should be modified by the user is hardware.c. This file
contains the functions to read and write to the port and the timing delay
function. The user must update these functions to target the desired hardware
being used. The released version targets the parallel port of the PC at
address 0x0378 using Lattice's download cable.

opcode.h
The file opcode.h contains the definitions of the byte codes used in the VME
Algorithm format and programming engine.

Programming Tools User Guide 85

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

debug.h
The file debug.h prints out debugging information if the preprocessor switch
VME_DEBUG were defined in the project. This is an optional file to
include.windriver.c and windriver.h

The files windriver.c and windriver.h target the JTAG Slim VME Embedded to
Windows 95 and 98. These files are compiled if the preprocessor switch
VME_WINDOWS were defined in the project file. These files should be
omitted when compiling the 8051-based JTAG Slim VME Embedded onto an
embedded platform.

8051 JTAG Slim VME Embedded User

Flow
This appendix details the steps the user must take to use the 8051-based
JTAG Slim VME Embedded.

Figure 49: 8051 JTAG Slim VME Embedded User Flow

Step 1. Create Chain with Lattice IEEE 1532 Compliant Devices using
Programmer

Using Programmer, add the target IEEE 1532 compliant devices into the
chain with the appropriate operations and data files. All the devices in the
chain must be IEEE 1532 compliant. For more information on supporting non-
Lattice devices, see Programmer’s on-line help documentation.

86 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Step 2. Generate VME File

Use the Deployment Tool to generate the VME file. By checking the HEX
check box, the VME Algorithm and Data files will be generated as C-
programming files with the Algorithm and Data stored in C-style byte buffers.
Refer to Deployment Tool online help for more information on using the
Deployment Tool.

Step 4. Modify Source Code File hardware.c

The 8051-based source code files are written in ANSI C and can be found in
the installation path of Programmer under the <install_path>\
embedded_source\slimembedded\sourcecode\slim_vme_8051 directory. The
file hardware.c is the only file that is required to be modified by the user. The
user must modify the following functions according to the target platform:
 readPort
 writePort
 ispVMDelay

The following are optional functions that the user may wish to modify in order
to enable and disable the hardware conditions before and after processing:
 EnableHardware
 DisableHardware

Step 5. Compile Source Code and VME HEX Files

Combine the source code and VME HEX files generated into a project to be
compiled. This may be done by using a microcontroller development tool to
create the project. The source codes windriver.c, windriver.h, and debug.h
shall not be required to be added into the project.

Programming Engine Flow

The programming engine of the JTAG Slim VME Embedded is driven by the
byte codes of the VME Algorithm file. The Algorithm byte codes instruct the
programming engine as to what sequence of functions to follow in order to
shift in instructions, move the TAP controller state machine, shift data in and
out of the device, and observe delays.

The TAP controller is a synchronous state machine that is based on the TMS
(Test Mode Select) and TCK (Test Clock) signals of the TAP and controls the
sequence of operations of the circuitry defined by the IEEE 1149.1 standard.
The TCK signal can be driven at a maximum of 25 MHz for current Lattice
IEEE 1532 Compliant devices.

In the Shift-DR state, a decoder is present to select which shift register is

enabled and connects it between TDI and TDO. The following are the shift
registers: Address Shift Register, Data Shift Register, 32-bit Shift Register,
and Bypass. The 32-bit Shift Register is used to store the ID Code and

Programming Tools User Guide 87

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Figure 50: TAP Controller State Diagram

USERCODE. The first bit clocked into any of the registers is placed into the
MSB, and data is shifted one bit towards TDO as additional bits are clocked
in.

Figure 51: Shift Registers

The engine core is implemented as a switch statement. The cases in this

switch statement perform specialized functions based on the byte code and
its operand(s). These functions may end up calling other switch statements,
calling the engine core recursively, setting global variable values, or
interfacing with the device directly. Once the byte code instruction has been
executed, it returns back to the main switch statement to process the next
byte.

The processor begins by verifying the VME version of the algorithm file. The
version is an eight byte ASCII of the format _SVME<Major Version>.<Minor
Version>, where <Major Version> and <Minor Version> are digits 0-9.
If the version verification fails, the processor returns the error code
ERR_WRONG_VERSION, or -4.

88 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Figure 52: Main Engine Switch

The Main Engine Switch calls the appropriate case statements based on the
incoming byte code from the VME Algorithm buffer. Unrecognized byte codes
will trigger the UNKNOWN case statement.

HIR Case Statement

The HIR case statement expects a number following the HIR byte code to
represent the number of header devices. The number will be stored in a
global variable and the processor will issue this number of bypasses prior to
executing an SIR instruction.

Figure 53: HIR Case Statement

Programming Tools User Guide 89

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

TIR Case Statement

The TIR case statement expects a number following the TIR byte code to
represent the number of trailer devices. The number will be stored in a global
variable and the processor will issue this number of bypasses after executing
an SIR instruction.

Figure 54: TIR Case Statement

HDR Case Statement

The HDR case statement expects a number following the HDR byte code to
represent the number of header register bits. The number will be stored in a
global variable and the processor will issue this number of bypasses prior to
executing an SDR instruction.

Figure 55: HDR Case Statement

TDR Case Statement

The TDR case statement expects a number following the TDR byte code to
represent the number of trailer register bits. The number will be stored in a
global variable and the processor will issue this number of bypasses after
executing an SDR instruction.

Figure 56: TDR Case Statement

ENDDR Case Statement

The ENDDR case statement expects a state following the ENDDR byte code
to represent the TAP state that the processor shall move the device into after
an SDR instruction. This state will be stored in a global variable.

90 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Figure 57: ENDDR Case Statement

ENDIR Case Statement

The ENDIR case statement expects a state following the ENDIR byte code to
represent the TAP state that the processor shall move the device into after an
SIR instruction. This state will be stored in a global variable.

Figure 58: ENDIR Case Statement

WAIT Case Statement

The WAIT case statement expects a number following the WAIT byte code to
represent the number of milliseconds of delay the device must observe. The
delay is observed immediately. The user must update the delay function in the
source code to make the target embedded system observe the delay duration
correctly.

Figure 59: WAIT Case Statement

TCK Case Statement

The TCK case statement expects a number following the TCK byte code to
represent the number of clocks that the device must remain in the run test idle
state. The clock cycles are executed immediately into the device.

Figure 60: TCK Case Statement

Programming Tools User Guide 91

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

STATE Case Statement

The STATE case statement expects a state following the STATE byte code to
instruct the processor to step the IEEE 1149.1 bus to the next state. The state
must be a valid stable state, which is IRPAUSE, DRPAUSE, RESET, or IDLE

Figure 61: STATE Case Statement

SIR Case Statement

Figure 62: SIR Case Statement

The SIR case statement begins by extracting the size of the register. The size
will be used later to indicate how many bits of data will be sent or read back
from the device. The processor then shifts the device into the safe state
IRPAUSE, and then to the state SHIFTIR. If HIR exists (see HIR Case
Statement), then the processor will bypass the HIR. The SIR sub-switch is a
switch that is based off of the byte codes that can potentially be found after
the SIR byte code.

The TDI byte code indicates that there is data that needs to be shifted into the
device. The processor will set the TDI index variable to point to the location
where the TDI data begins in the algorithm buffer.

The DTDI byte code indicates that there is data to that needs to be shifted into
the device. Unlike the TDI byte code, the DTDI byte code signals that the data
will be coming from the data buffer. If the data buffer has compression turned
on, the first byte of the data frame will be checked to see if the frame was
indeed compressible.

The TDO byte code indicates that there is data that needs to be read and
verified from the device. The processor will set the TDO index variable to
point to the location where the TDO data begins in the algorithm buffer.

92 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

The DTDO byte code indicates that there is data that needs to be read and
verified from the device. Unlike the TDO byte code, the DTDO byte code
signals that the data will be coming from the data buffer. If the data buffer has
compression turned on, the first byte of the data frame will be checked to see
if the frame were indeed compressible.

The MASK byte code indicates that there is mask data that needs to be used
when comparing the TDO values against the actual values scanned out of the
device. The processor will set the MASK index variable to point to the location
where the MASK data begins in the algorithm buffer.

The UNKNOWN case statement is the default for unrecognized byte codes.
This case returns the error code ERR_ALGO_FILE_ERROR, or -5, to indicate
an error in the algorithm.

The CONTINUE byte code terminates the SIR instruction. When this byte is
encountered, it indicates that the TDI, DTDI, TDO, DTDO, and MASK indexes
are pointing to their correct locations and the processor is ready to send or
read and verify data from the device.

Figure 63: SIR Case Statement Continued

If the TDO or DTDO byte code were encountered in the SIR sub-switch, then
that indicates that the SIR instruction is going to read data from the device,
else the SIR instruction is going to send data to the device.

If reading and verifying data from the device were successful, the processor
checks if TIR exists (see TIR Case Statement). If TIR exists, then the trailer
devices must be bypassed. Next it shifts the device to the stable state that
followed the ENDIR byte code (see ENDIR Case Statement). The control
returns back to the Main Engine Switch.

Programming Tools User Guide 93

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

If reading and verifying data from the device were unsuccessful, the
processor checks if TIR exists. If TIR exists, then the trailer devices must be
bypassed. Next it shifts the device to the stable state that followed the ENDIR
byte code. The error code ERR_VERIFY_FAIL, or -1, is returned and the
program exits.

If TDO or DTDO were not encountered in the SIR sub-switch, then the
processor sends data to the device. If TIR exists, then the trailer devices must
be bypassed. Next it shifts the device to the stable state that followed the
ENDIR byte code. The control returns back to the Main Engine Switch.

SDR Case Statement

The SDR case statement works similar to the SIR. It begins by extracting the
size of the register. The size will be used later to indicate how many bits of
data will be sent or read back from the device. The processor then shifts the
device into the safe state DRPAUSE, and then to the state SHIFTDR. If HDR
exists (see HDR Case Statement), then the processor will bypass the HDR.
The SDR sub-switch is a switch that is based off of the byte codes that can
potentially be found after the SDR byte code.

The TDI byte code indicates that there is data that needs to be shifted into the
device. The processor will set the TDI index variable to point to the location
where the TDI data begins in the algorithm buffer.

The DTDI byte code indicates that there is data to that needs to be shifted into
the device. Unlike the TDI byte code, the DTDI byte code signals that the data
will be coming from the data buffer. If the data buffer has compression turned
on, the first byte of the data frame will be checked to see if the frame were
indeed compressible.

The TDO byte code indicates that there is data that needs to be read and
verified from the device. The processor will set the TDO index variable to
point to the location where the TDO data begins in the algorithm buffer.

The DTDO byte code indicates that there is data that needs to be read and
verified from the device. Unlike the TDO byte code, the DTDO byte code
signals that the data will be coming from the data buffer. If the data buffer has
compression turned on, the first byte of the data frame will be checked to see
if the frame were indeed compressible.

The MASK byte code indicates that there is mask data that needs to be used
when comparing the TDO values against the actual values scanned out of the
device. The processor will set the MASK index variable to point to the location
where the MASK data begins in the algorithm buffer.

The UNKNOWN case statement is the default for unrecognized byte codes.
This case returns the error code ERR_ALGO_FILE_ERROR, or -5, to indicate
an error in the algorithm.

The CONTINUE byte code terminates the SDR instruction. When this byte is
encountered, it indicates that the TDI, DTDI, TDO, DTDO, and MASK indexes

94 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

are pointing to their correct locations and the processor is ready to send or
read and verify data from the device.

Figure 64: SDR Case Statement

Figure 65: SDR Case Statement Continued

If the TDO or DTDO byte code were encountered in the SDR sub-switch, then
that indicates that the SDR instruction is going to read data from the device,
else the SDR instruction is going to send data to the device.

If reading and verifying data from the device were successful, the processor
checks if TDR exists (see TDR Case Statement). If TDR exists, then the trailer
devices must be bypassed. Next it shifts the device to the stable state that
followed the ENDDR byte code (see ENDDR Case Statement). The control
returns back to the Main Engine Switch.

Programming Tools User Guide 95

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

If reading and verifying data from the device were unsuccessful, the
processor checks if TDR exists. If TDR exists, then the trailer devices must be
bypassed. Next it shifts the device to the stable state that followed the
ENDDR byte code. The error code ERR_VERIFY_FAIL, or -1, is returned and
the program exits.

If TDO or DTDO were not encountered in the SDR sub-switch, then the
processor sends data to the device. If TDR exists, then the trailer devices
must be bypassed. Next it shifts the device to the stable state that followed
the ENDDR byte code. The control returns back to the Main Engine Switch.

BEGIN_REPEAT Case Statement

The BEGIN_REPEAT byte code makes it possible to loop the programming
algorithm, thus requiring less ROM to hold the algorithm. Programming each
frame requires one pass through the repeat loop. The ROM saved is
substantial when one considers that a device can have several thousand
frames. Instead of extrapolating the set of byte codes needed to program the
frame several thousand times, only one set will be sufficient.

The BEGIN_REPEAT case statement begins by extracting the repeat size.

The repeat size is typically the number of frames in the device that is to be
programmed. After the repeat size has been obtained, the next byte to extract
is the PROGRAM or VERIFY token. If the PROGRAM byte were present,
then a pointer must be set in the data buffer to designate the beginning of the
programming data. If the VERIFY byte were present, then the processor must
return to the beginning location of the data buffer. This method allows
programming and verification to use one set of data, thus reducing the ROM
required to hold the data buffer by half.

While the repeat size, or number of un-programmed frames, is greater than

zero, the algorithm index is set to point to the beginning of the repeat and a
recursive call is made to the Main Engine Switch to program the frame. When
the frame is processed, the Main Engine Switch returns the control to the
BEGIN_REPEAT case statement. The repeat size is decremented and the
process repeats until there are no frames left. The control then returns to the
Main Engine Switch. While in the repeat loop, any errors such as verification
or algorithm errors would result in the repeat loop returning the error code and
the program would exit.

END_REPEAT Case Statement

The END_REPEAT case statement works alongside the BEGIN_REPEAT
case statement. When the END_REPEAT byte code is encountered, it returns
the control to the caller, which is the recursive call made by BEGIN_REPEAT.
The END_REPEAT byte code appears at the end of the set of byte codes
needed to program a frame.

96 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Figure 66: BEGIN_REPEAT Case Statement

Figure 67: END_REPEAT Case Statement

ENDVME Case Statement

The ENDVME case statement is the only case where the program can return
a passing value. The case statement checks if HDR exists (see HDR Case
Statement). If HDR exists, then that indicates that there are still header
devices that need to be programmed, thus the control returns to the Main
Engine Switch. If HDR does not exist, the return value is returned to the caller,
which is the entry point function and the program ends.

Figure 68: ENDVME Case Statement

UNKNOWN Case Statement

The UNKNOWN case statement is the default for unrecognized byte codes.
This case returns the error code ERR_ALGO_FILE_ERROR, or -5, to indicate
an error in the algorithm.

Programming Tools User Guide 97

EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Figure 69: UNKNOWN Case Statement

VME Algorithm and Format

The VME Algorithm and Data files are created by deconstructing an SVF file.
An SVF file is an ASCII file that contains the programming algorithm and data
needed to program the device. The programming algorithm is described by
statements that control the IEEE 1149.1 bus operations. When generating the
VME files, Deployment Tool separates the algorithm and data into the VME
Algorithm and Data files, respectively.

VME Algorithm Format

The VME Algorithm file is similar to the SVF file with the following differences:
 VME Algorithm uses byte codes from the table below to represent SVF
commands
 Fuse data and USERCODE have been removed
 Looping algorithm

The following is an example of an EPV VME Algorithm file and the SVF
translation for the LC4064V device:
Table 3: VME Algorithm Example
VME Algorithm Format Serial Vector Format (SVF) Description
0x0A 0x00 HDR 0;
0x08 0x00 HIR 0;
0x0B 0x00 TDR 0;
0x09 0x00 TIR 0;
0x06 0x03 ENDDR DRPAUSE;
0x07 0x02 ENDIR IRPAUSE;
0x01 0x01 STATE IDLE;
0x02 0x08 0x0E 0x68 0x0F SIR 8 TDI (16); Shift in the IDCODE
instruction
0x01 0x01 STATE IDLE;
0x03 0x20 0x0E 0xFF 0xFF 0xFF SDR 32 TDI (FFFFFFFF) TDO Verify the IDCODE
0xFF 0x11 0xC2 0x09 0x01 0x80 (01809043) MASK (0FFFFFFF);
0x12 0xFF 0xFF 0xF0 0x0F
0x02 0x08 0x0E 0x38 0x0F SIR 8 TDI (1C); Shift in the PRELOAD
instruction
0x03 0x44 0x0E 0x00 0x00 0x00 SDR 68 TDI (00000000000000000); Shift all zero data into
0x00 0x00 0x00 0x00 0x00 0x00 boundary scan cells
0x0F
0x02 0x08 0x0E 0xA8 0x0F SIR 8 TDI (15); Shift in ENABLE instruction
0x01 0x01 0x04 0x03 0x05 0x14 RUNTEST IDLE 3 TCK 2.00E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x02 0x08 0x0E 0xC0 0x0F SIR 8 TDI (03); Shift in ERASE instruction

98 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : JTAG Slim VME Embedded

Table 3: VME Algorithm Example (Continued)

0x01 0x01 0x04 0x03 0x05 0x64 RUNTEST IDLE 3 TCK 1.00E-001 SEC; Execute RUNTEST instruction
0x01 0x01
0x02 0x08 0x0E 0x84 0x0F SIR 8 TDI (21); Shift in ADDRESS INIT
instruction
0x01 0x01 STATE IDLE;
0x02 0x08 0x0E 0xE4 0x0F SIR 8 TDI (27); Shift in PROGRAM INCR
instruction
0x0C 0x5F 0x15 N/A Begin PROGRAM repeat loop
of size 95
VME Algorithm Format Serial Vector Format (SVF) Description
0x03 0xE0 0x02 0x18 0x14 0x0F SDR 352 DTDI (DATA); Notice the forth byte is
0x18, which is actually
DTDI. DTDI instructs the
processor to send in data
from the data buffer
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x13 N/A Terminate the repeat
algorithm
0x02 0x08 0x0E 0x58 0x0F SIR 8 TDI (1A); Shift in PROGRAM USERCODE
instruction
0x03 0x20 0x18 0x14 0x0F SDR 32 DTDI (DATA); Shift in the USERCODE The
USERCODE can be found in
the data buffer.
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x02 0x08 0x0E 0x80 0x0F SIR 8 TDI (01); Shift in ADDRESS SHIFT
instruction
0x03 0x5F 0x0E 0x00 0x00 0x00 SDR 95 TDI Shift in beginning address
0x00 0x00 0x00 0x00 0x00 0x00 (400000000000000000000000);
0x00 0x00 0x02 0x0F
0x02 0x08 0x0E 0x54 0x0F SIR 8 TDI (2A); Shift in READ INC
instruction
0x0C 0x5F 0x16 N/A Begin VERIFY repeat loop of
size 95
0x01 0x01 0x04 0x03 0x05 0x01 RUNTEST IDLE 3 TCK 1.00E-003 SEC; Execute RUNTEST instruction
0x01 0x01
0x03 0xE0 0x02 0x0E 0x00 0x00 SDR 352 TDI (0000000000000 Verify the frame against
0x00 0x00 0x00 0x00 0x00 0x00 00000000000000000000000000 the data in the data buffer
0x00 0x00 0x00 0x00 0x00 0x00 00000000000000000000000000
0x00 0x00 0x00 0x00 0x00 0x00 00000000000000000000000) DTDO
0x00 0x00 0x00 0x00 0x00 0x00 (DATA);
0x00 0x00 0x00 0x00 0x00 0x00
0x00 0x00 0x00 0x00 0x00 0x00
0x00 0x00 0x19 0x14 0x0F
0x13 N/A Terminate the repeat
algorithm
0x02 0x08 0x0E 0xE8 0x0F SIR 8 TDI (17); Shift in USERCODE
instruction
0x03 0x20 0x0E 0xFF 0xFF 0xFF SDR 32 TDI (FFFFFFFF) Verify the USERCODE against
0xFF 0x19 0x14 0x0F DTDO (DATA); the USERCODE in the data
buffer
0x02 0x08 0x0E 0xF4 0x0F SIR 8 TDI (2F); Shift in PROGRAM DONE
instruction
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01
0x01 0x01 STATE IDLE;
0x02 0x08 0x0E 0x78 0x0F SIR 8 TDI (1E); Shift in DISABLE
instruction
0x01 0x01 0x04 0x03 0x05 0x0D RUNTEST IDLE 3 TCK 1.30E-002 SEC; Execute RUNTEST instruction
0x01 0x01

Programming Tools User Guide 99

EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

Table 3: VME Algorithm Example (Continued)

0x17 N/A End VME Algorithm

Customizing for the Target Platform

The source code files are written in ANSI C. The VME source codes are
located in the <install_path>\embedded_source\vmembedded directory. The
JTAG Slim VME Embedded source codes can be found in the
<install_path>\embedded_source\slimembedded directory.

The main routines that will require customization are in the hardware.c file. It
includes the routines for reading from and writing to the JTAG pins and a
delay routine. These routines are well commented in hardware.c and are at
the top of the file. In readPort(), a byte of data is read from the input port. In
writePort(), a byte of data is written to the output port. In ispVMDelay(), the
system delays for the specified number of milliseconds or microseconds. The
port mapping is set at the top of the hardware.c file.

See Also VME Required User Changes

Slave SPI Embedded

Slave Serial Peripheral Interface (SPI) Embedded is a high-level
programming solution that enables programming the LatticeXP2 and future
FPGA families with built-in SPI port through embedded system. This allows
users to perform real-time reconfiguration to Lattice Semiconductor's FPGA
families. The Slave SPI Embedded system is designed to be embedded-
system independent, so it is easy to port into different embedded systems
with little modifications. The Slave SPI Embedded source code is written in C
code, so the user may compile the code and load it to the target embedded
system.

The purpose of this usage note is to provide the user with information about
how to port the Slave SPI Embedded source code to different embedded
systems. The following sections describe the embedded system requirements
and the modifications required to use Slave SPI Embedded source code.

This usage guide is updated for Slave SPI Embedded version 2.0. Major
changes includes new format of data file, and Lattice parallel port and USB
cable support. In the Slave SPI Embedded source code, there are updates in
intrface.c and core.c, but no update in hardware.c. Version 2.0 updates will be
marked with [New in version 2.0] tag.

Requirements
This section lists device requirements, embedded system requirements, and
other requirements.

100 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

Device Requirements
 Only Lattice Semiconductor's FPGA families with SPI port are supported.
 Single device support. Multiple device support is not available.
 The Slave SPI port must be enabled on the device in order to use the
Slave SPI interface. This is done by setting the SLAVE_SPI_PORT to
Enable using the Global Preferences spreadsheet in Diamond
Spreadsheet view.

Embedded System Requirements

A compiler supporting C code for the target embedded system is required.

A dedicated SPI interface that can be configured to Master SPI mode is

preferred. However, if the embedded system does not have a built in SPI
interface, the user may consider using a general peripheral I/O ports to
implement SPI functionality. In this case, minimum of four peripheral I/O's are
required, with at least one of them that can be tri-stated if needed.

Read and Save operations and display operations are not supported.

Other Requirements
The Slave SPI Embedded system requires memory space to store
programming data file. The storage may be internal or external memory
(RAM, Flash, etc.). The user may also consider storing the programming data
in an external system such as PC. In this case, the user needs to establish
communication between the external system and the embedded system.

Slave SPI Embedded Algorithm Format

The Slave SPI algorithm file contains byte codes that represent the
programming algorithm of the device or chain.

Table 4: Slave SPI Algorithm Format

SSPI Symbol Hex Value

STARTTRAN 0x10

CSTOGGLE 0x11

TRANSOUT 0x12

TRANSIN 0x13

RUNCLOCK 0x14

ENDTRAN 0x1F

Programming Tools User Guide 101

EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

Table 4: Slave SPI Algorithm Format (Continued)

SSPI Symbol Hex Value

MASK 0x21

ALGODATA 0x22

PROGDATA 0x25

RESETDATA 0x26

PRODATAEH 0x27

REPEAT 0x41

ENDREPEAT 0x42

LOOP 0x43

ENDLOOP 0x44

STARTOFALGO 0x60

ENDOFALGO 0x61

HCOMMENT 0xA0

HENDCOMMENT 0xA1

ALGOID 0xA2

VERSION 0xA3

BUFFERREQ 0xA4

STACKREQ 0xA5

MASKBUFREQ 0xA6

HCHANNEL 0xA7

HEADERCRC 0xA8

COMPRESSION 0xA9

HDATASET_NUM 0xAA

HTOC 0xAB

Slave SPI Embedded Data Format

While the SSPI algorithm file contains the programming algorithm of the
device, the SSPI data file contains the fuse and USERCODE patterns. The
first byte in the file indicates whether the data file has been compressed. A
byte of 0x00 indicates that no compression was selected, and 0x01 indicates
that compression was selected.

When compression has been selected, each frame is preceded by a frame

compression byte to indicate whether the frame is compressible. This is
necessary because even though you might elect to compress the SSPI data
file, it is possible that a compressed frame will actually be larger than an

102 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

uncompressed frame. When that happens, the frame is not compressed at all
and the frame compression byte of 0x00 is added to notify the processor that
no compression was performed on the frame.
Uncompressed Slave SPI Data Format Compressed Slave SPI Data Format
0x00 0x01
<Frame 1>0x10 <Compress Byte><Frame 1>0x10
<Frame 2>0x10 <Compress Byte><Frame 2>0x10
… …
<Frame N>0x10 <Compress Byte><Frame N>0x10

Generating Slave SPI Embedded Files

The Slave SPI Embedded files can be generated through Diamond
Deployment Tool by selecting the VME button on the toolbar menu. The Slave
SPI Embedded generation dialog allows the user to generate the file in hex (C
compatible) array or binary. The binary Slave SPI file can be used by the PC
version of Slave SPI Embedded and utilizes the extension *.sea for algorithm
files, and *.sed for data files. Also, the binary file can be uploaded to internal
or external memory of the embedded system if the user plans to implement
the system in that manner.

The hex file is a C programming language file that must be compiled with the
EPROM-based version of Slave SPI Embedded processor and utilizes the
extension *.c. The binary file is generated by default. Other options are
available through the dialog, such as data file compression, adding comments
to the algorithm file, or disable generating the algorithm or data file.

Modifications
The Slave SPI Embedded source code is installed in the
<install_path>\embedded_source\sspiembedded\sourcecode directory where
you installed the Diamond Programmer. There are two directories in the src
directory, SSPIEm and SSPIEm_eprom. The first directory, SSPIEm, contains
the file-based Slave SPI Embedded source code, and can support sending
and receiving multiple bytes over the channel. The second directory,
SSPIEm_eprom, contains the EPROM-based Slave SPI Embedded source
code, which supports the algorithm and data being compiled with the process
system.

In the files that require user modification, comments surrounded by asterisks

(*) will require the users' attention. These comments indicate that the following
section may require user modification. For example:

//*********************************************************************
//* Example comment
//*********************************************************************

Programming Tools User Guide 103

EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

Before using the Slave SPI Embedded system, there are three sets of files (.c
/ .h) that need to be modified. The first set, hardware.c and hardware.h, must
be modified. This file contains the SPI initialization, wait function, and SPI
controlling functions. If the user would like to enable debugging functionalities,
debugging utilities need to be modified in this file as well. [New in version 2.0]
hardware.c source code supports transmitting and receiving multiple bytes at
once. This approach may be faster in some SPI architecture, but it requires a
buffer to store the entire frame data. If the user wishes to use single byte
transmission, replace hardware.c and hardware.h with the ones in file
sspiem_sbyte.zip.

The second set, intrface.c and intrface.h, contains functions that utilize the
data and algorithm files. There are two sections in this file that requires
attention. The first one is data section. When the processor in Slave SPI
Embedded system needs to process a byte of data, it calls function
dataGetByte(). Slave SPI Embedded version 2.0 requires data file no
matter what operation it is going to process. Users are responsible to modify
the function to fit their configuration. The second section is algorithm section.
In Programmer, there is the option to generate both the algorithm file and the
data file in hex array format (C compatible). If the user wishes to compile the
algorithm and data along with Slave SPI Embedded system, use the source
code in the
<install_path>\embedded_source\sspiembedded\sourcecode\sspiem_eprom
directory. Users only need to put the generated .c file in the same folder as
Slave SPI Embedded system code and then compile the whole thing. If the
embedded system has internal memory that can be reached by address,
using EPROM version of intrface.c is also ideal. For users who plan to put the
algorithm and data in external storage, intrface.c and intrface.h may need
modification.

[New in version 2.0] The third file set is SSPIEm.c and SSPIEm.h. Function
SSPIEm_preset() provides the user to set which algorithm and data will be
processed. This function needs to be modified according to users'
configuration.

Below is information about functions the user is responsible to modify.

hardware.c
There is update in version 2.0 hardware.c source code. In previous version,
users are responsible to modify function TRANS_tranceive_stream().
Version 2.0 source code, which support transmitting multiple bytes at once,
includes two functions, TRANS_transmitBytes() and
TRANS_receiveBytes(), that function TRANS_tranceive_stream()
would call. Therefore, TRANS_tranceive_stream() no longer require user
attention in this configuration, but the user is responsible to modify
TRANS_transmitBytes() and TRANS_receiveBytes(). If the user wish
to implement Slave SPI Embedded so it transmit one byte at a time, then
TRANS_tranceive_stream() need to be modified.

int SPI_init(); This function will be called at the beginning of the Slave SPI
Embedded system. Duties may include, but not limited to:

104 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

 Turning on SPI port;

 Enabling counter for wait function;
 Configuring SPI peripheral IO ports (PIO);
 Resetting SPI;
 Initializing SPI mode (Master mode, channel used, etc);
 Enabling SPI.

The function returns a 1 to indicate initialization successful, or a 0 to indicate

fail.

int SPI_final(); This function will be called at the very end of the Slave SPI
Embedded system. The function returns a 1 to indicate success, or a 0 to
indicate fail.

void wait(int ms); This function takes a delay time (in milliseconds), and
waits for the amount of delay time. This function does not need a return value.

int TRANS_starttranx(unsigned char channel); This function will start an

SPI transmission. Duties may include, but not limited to:
 Pulling Chip Select (CS) low;
 Starting Clock;
 Flushing read buffer.

If the dedicated SPI interface in the embedded system automatically starts the
clock and pulls CS low, then this function only returns a 1. This function
returns a 1 to indicate success, or a 0 to indicate fail.

int TRANS_endtranx(); This function will finalize an SPI transmission.

Duties may include, but not limited to:
 Pulling CS high;
 Terminating Clock.

If the dedicated SPI interface in the embedded system automatically

terminates the clock and pulls CS high, then this function only returns a 1.
This function returns a 1 to indicate success, or a 0 to indicate fail.

int TRANS_cstoggle(unsigned char channel); This function will toggle

the CS of current channel. It will be called between TRANS_starttranx()
and TRANS_endtranx(). It first pulls CS low, waits for a short period of time,
and pulls CS high. A simple way to accomplish this is to transmit some
dummy data to the device. One bit is enough to accomplish this. All one (1) for
dummy is recommended, because usually the channel is held high during
rest, and Lattice devices ignore opcode 0xFF (no operation). The function
returns a 1 to indicate success, or a 0 to indicate fail.

int TRANS_runClk(); This function will run a minimum of three clocks on

SPI channel. It will be called after TRANS_endtranx() if extra clock are
required. If the dedicated SPI interface does not allow free control of clock, a
workaround is to enable the CS of another channel that is not being used.

Programming Tools User Guide 105

EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

This way the device will still see the clock but the CS of current channel will
stay high. The function returns a 1 to indicate success, or a 0 to indicate fail.

[New in version 2.0] int TRANS_transmitBytes (unsigned char *trBuffer,

int trCount); This function is available if the user wishes to implement
transmitting multiple bits one byte at a time. It is responsible to transmit the
number of bits, indicated by trCount, over the SPI port. The data to be
transmitted is stored in trBuffer. Integer trCount indicates the number of
bits being transmitted, which should be divisible by eight (8) to make it byte-
bounded. If trCount is not divisible by eight, pad the least significant bits of
the transmitted data with ones (1).

[New in version 2.0] int TRANS_receiveBytes (unsigned char *rcBuffer,

int rcCount); This function is available if the user wishes to implement
receiving multiple bits one byte at a time. It is responsible to receive the
number of bits, indicated by rcCount, over the SPI port. The data received
may be stored in rcBuffer. Integer rcCount indicates the number of bits being
received, which should be divisible by eight (8) to make it byte-bounded. If
rcCount is not divisible by eight, pad the most significant bits of the received
data with ones (1).

int TRANS_transceive_stream(int trCount, unsigned char *trBuffer,

trCount2, int flag, unsigned char *trBuffer2); This function is available for
modification if the user wishes to implement transmission with one byte at a
time. The function also appears in implementation of transmission with
multiple bytes at once, but the user does not need to modify it.

For single byte transmission, this is the most complex function that needs to
be modified. First, it will transmit the amount of bits specified in trCount with
data stored in trBuffer. Next, it will have the following operations
depending on the flag:
 NO_DATA: End of transmission. trCount2 and trBuffer2 are
discarded.
 BUFFER_TX: Transmit data from trBuffer2.
 BUFFER_RX: Receive data and compare it with trBuffer2.
 DATA_TX: Transmit data from external data.
 DATA_RX: Receive data from trbuffer2.

If the data is not byte-bounded and your SPI port only transmits and receives
byte-bounded data, padding bits are required to make it byte-bounded. When
transmitting non-byte-bounded data, add the padding bits at the beginning of
the data. When receiving data, do not compare the padding, which are at the
end of the transfer. The function returns a 1 to indicate success, or a 0 to
indicate fail.

(optional) int dbgu_init(); This function initializes the debugging

functionality. It is up to the user to implement it, and implementations may
vary.

106 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

(optional) void dbgu_putint(int debugCode, int debugCode2); This

function will put a string and an integer to the debugging channel. It is up to
the user to take advantage of these inputs.

SSPIEm.c
int SSPIEm_preset(); This function calls dataPreset() and algoPreset()
functions to pre-set the data and algorithm. The input to this function depends
on the configuration of the embedded system. This function returns 1 to
indicate success, or 0 to indicate fail.

intrcface.c - Data Section

Global Variables Global variables may vary due to different
implementation.

[New in version 2.0] int dataPreset(); This function allows user to set
which data will be used for processing. It returns 1 to indicate success, or 0 to
indicate fail.

int dataInit (unsigned char *comp); This function initializes the data. The
first byte of the data indicates if the fuse data is compressed. It retrieves the
first byte and stores it in the location pointed to by *comp. The fuse data starts
at the second byte. The implementation may vary due to different
configuration. The function returns a 1 to indicate success, or a 0 to indicate
fail. For implementation that uses internal memory, which can be accessed by
addressing, the following is an example implementation:
Points variable data to the beginning of the fuse data.
Resets count to 0.

int dataGetByte(int *byteOut); This function gets one byte from data array
and stores it in the location pointed to by byteOut. The implementation may
vary due to different configuration. The function returns 1 to indicate success,
or 0 to indicate fail. For implementation that uses internal memory, which can
be accessed by addressing, here is a sample implementation:
Gets byte that variable data points to.
Points data to the next byte.
Count++.

int dataReset(); This function resets the data pointer to the beginning of the
fuse data. Note that the first byte of the data is not part of the fuse data. It
indicates if the data is compressed. The implementation may vary due to
different configuration. The function returns a 1 to indicate success, or a 0 to
indicate fail. For implementation that uses internal memory, which can be
accessed by addressing, the following is an example implementation:
Points variable data to the beginning of the data array.

Programming Tools User Guide 107

EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

Resets count to 0.

Note: This section is data utilize functions. Modification of this section is

optional if the user wishes to compile the algorithm along with Slave SPI
Embedded system.

[New in version 2.0] int dataFinal(); This function is responsible to finish

up the data. The implementation may vary due to different configuration. The
function returns a 1 to indicate success, or a 0 to indicate fail.

intrface.c - Algorithm Section

Global variables Global variables may vary due to different
implementation.

[New in version 2.0] int algoPreset(); This function allows user to set
which algorithm will be used for processing. It returns 1 to indicate success, or
0 to indicate fail.

int algoInit(); This function initializes the data. The implementation may
vary due to different configuration. The function returns a 1 to indicate
success, or a 0 to indicate fail.

In our implementation, it only sets algoIndex to 0.

int algoGetByte(unsigned char *byteOut); This function gets one byte

from the algorithm bitstream, and stores it in the location pointed to by
byteOut. The implementation may vary due to different configuration. The
function returns a 1 to indicate success, or a 0 to indicate fail.

[New in version 2.0] int algoFinal(); This function is responsible to finish

up the algorithm. The implementation may vary due to different configuration.
The function returns a 1 to indicate success, or a 0 to indicate fail.

intrface.c - Sample Configurations

There may be many different options to configure Slave SPI Embedded data
file and algorithm file. The following are two possible configurations.

1. EPROM Approach

With this version, both algorithm and data are generated into C-compatible
Hex array and compiled along with Slave SPI Embedded source code. Here
is how the functions are modified to fit this configuration:
 Include both Hex arrays in the global scale.
 Pass the pointer to the arrays to SSPIEm_preset(). In this function,
pass the pointer to algoPreset() and dataPreset() functions,
respectively. Both functions store the pointer in global variables defined in
intrface.c.

108 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : Slave SPI Embedded

 In algoInit() and dataInit() functions, set the counters to zero (0).

 In algoGetByte() and dataGetByte() functions, read a byte from
the respective array, and increment the counter.
 In dataReset() function, reset the counter to zero (0).
 In algoFinal() and dataFinal() functions, set the pointer to both
array to NULL. This is optional.

Although optional, it may be a good idea to keep track of the size of both data
and algorithm arrays. The size of array may be passed to Slave SPI
Embedded through the preset functions.

If the embedded system uses internal memory that can be reached the same
way as using array, this configuration may also fit into the embedded system.

If the user plans to use EPROM approach, intrface.c will be available, and the
user may not need to modify it. The files intrface.c, intrface.h, SSPIEm.c, and
SSPIEm.h are in the
<install_path>/SSPIEmbedded/SourceCode/src/SSPIEm_eprom directory.

2. File System Approach

This approach is used when implementing Slave SPI Embedded command-

line executable on PC. If the embedded system has similar file system, it may
access the algorithm and data through the file system. Here is how the
functions are modified to fit this configuration:
 Pass the name of the algorithm and data file to SSPIEm_preset(). In
this function, pass them to algoPreset() and dataPreset()
functions, respectively. Both functions store the name of the file in global
variables defined in intrface.c.
 In algoInit() and dataInit() functions, open the file and check if
they are readable. If the file is not opened as a stream, set the counter to
zero (0).
 In algoGetByte() and dataGetByte() functions, read a byte from the
respective file, and increment the counter if needed.
 In dataReset() function, reset the counter to zero (0), if needed. If the
file is read as a stream, rewind the stream.
 In algoFinal() and dataFinal() functions, close both files.

Usage
In order to use the Slave SPI Embedded system, include the Slave SPI
Embedded system in the target embedded system by including SSPIem.h to
the header list. To start the processor, simply make this function call:

SSPIEm(unsigned int algoID);

Programming Tools User Guide 109

EMBEDDED FLOW OVERVIEW : I2C Embedded

Currently, the converter does not have algoID capability. This capability is
reserved for future use. Use 0xFFFFFFFF for algoID.

Return Codes from Slave SPI

Embedded
The utility returns a 2 when the process succeeds, and returns number less
than or equal 0 when it fails. Table 5 lists return codes from Slave SPI
Embedded.

Table 5: Return codes from Slave SPI Embedded

Results Return Code

Succeed 2

Process Failed 0

Initialize Algorithm Failed -1

Initialize Data Failed -2

Version Not Supported -3

Header Checksum Mismatch -4

Initialize SPI Port Failed -5

Initialization Failed -6

Algorithm Error -11

Data Error -12

Hardware Error -13

Verification Error -20

I2C Embedded
The physical I2C buss consists of two wires: SCL and SDA.
 SCL is the clock line. It is used to synchronize all data transfers over the
I2C bus.
 SDA is the data line.

The SCL & SDA lines are connected to all devices on the I2C bus. There must
be a third wire connected to ground or 0 volts. There may also be a 5V wire
for power distribution t he devices. Both SCL and SDA lines are “open drain”
drivers, meaning that the device can drive its output low, but it cannot drive it
high. For the line to be able to go high, you must provide pull-up resistors to
the 5V supply. There should be a resistor from the SCL line to the 5V line and

110 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : I2C Embedded

another from the SDA line to the 5V line. You only need one set of pull-up
resistors for the entire I2C bus, as illustrated below.

Masters and Slaves

The devices on the I2C bus are either masters or slaves. The master is
always the device that drives the SCL clock line. The slaves are the devices
that respond to the master. Only a master can initiate a transfer over the I2C
bus. A slave cannot initiate a transfer over the I2C bus. There can be, and
usually are, multiple slaves on the I2C bus. However, there is normally only
one master. It is possible to have multiple masters, but it is typical and not
covered in this document. For the purposes of this document, the MachXO2
or MachXO3L device is always the slave.

MachXO2 or MachXO3L Slave I2C

Programming
When the master communicates to a slave (MachXO2 or MachXO3L for
example) it begins by issuing a start sequence on the I2C bus. A start
sequence is one of two special sequences defined for the I2C bus, the other
being the stop sequence. The start sequence and stop sequence are the only
time when the SDA (data line) is allowed to change while the SCL (clock line)
is high. When data is being transferred, SDA must remain stable and not
change while the SCL is high. The start and stop sequences mark the
beginning and end of a transaction with the slave device.

The primary I2C port of the Mach XO2 device can be used as a user I2C port
function or as a device programming port. When used for device
programming, the primary I2C port is a slave I2C with a default slave address

Programming Tools User Guide 111

EMBEDDED FLOW OVERVIEW : I2C Embedded

of 7’b1000000 for 7-bit addressing or 10’b1111000000 for 10-bit addressing.

The primary I2C port must be enabled in order to support the device
programming using the I2C protocol. This is done by setting the I2C_Port
preference to ENABLE in the software. The I2C programming supports single
device programming.

The sequence for device programming using the I2C follows the standard
Lattice device programming algorithm. The I2C bus hardware requirements,
the timing requirements, and the receive/transmit protocols must follow the
standard I2C specification. The definition of the SDA data time is the delay
form the SCL falling edge 30% VDD to SDA falling edge 70% VDD. The SDA
data setup time is the time requirement from the SDA falling edge 30% VDD
to the SCL rising edge 30% VDD.

All the I2C slave commands consist of one byte op-code followed by three
one-byte operand, except the ISC DISABLE command. The ISC DISABLE
command for I2C programming consists of one byte op-code followed by two
operands. I2C programming can also be done in the background. In this case,
the ISC ENABLE command (0XC6) should be replaced by the
LSC_ENABLE_X command (0X74).

Using the PC-based I2C Embedded

Programming
The I2C Embedded system for MachXO2 or MachXO3L is a quick and easy
way to validate I2C files and the I2C Embedded programming engine by
successfully processing the target XO2 device using the FTDI USB2 Cable of
the PC.

The programming algorithm of the device is separated into I2C algorithm file
and I2C data file. This separation of the algorithm and data allows the
optimization of the I2C embedded programming engine. It also allows you to
mix I2C data files with I2C algorithm files.

To access the PC-based I2C Embedded System, use the Diamond

Deployment Tool Version 1.4 or higher to add the MachXO2 or MachXO3L
device. Then, select the I2C embedded programming options from the
Generate I2C dialog box. For more information, refer to the Deployment Tool
online help.

The only source code file that must be modified is the hardware.c file. The
source files can be found in
<install_path>\embedded_source\i2cembedded\src\i2cem diamond directory.

hardware.c
The only file that you should modify is hardware.c. This file contains the
functions to read and write to the port and the timing delay function. You must
update these functions to target the desired hardware.

112 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : I2C Embedded

opcode.h
The opcode.h file contains the definitions of the byte codes used in the I2C
algorithm format and programming engine.

i2c_core.c
The i2c_core.c file provides the programming engine for the I2C embedded
system. The engine operates on the commands in the I2C algorithm, and
obtains data from the I2C data, if necessary. The engine is responsible for
functions such as sending data, verifying data, observing timing delay,
decompression, and so on.

i2c_main.c
The i2c_main.c file is the only file that differs between the PC-based and the
8051-based embedded solutions. This difference is due to the way each of
these interfaces to the I2C algorithm and data files through the entry point.
This file contains the main and entry point functions.

Using the 8051-based I2C Programming

To program embedded systems using the 8051-based I2C programming, you
must generate the I2C files as HEX to create the I2C algorithm and data files
as C programming files. Each file contains a C programming style byte buffer
that holds the I2C algorithm or data.

The HEX files must be compiled along with the 8051-based I2C System
source code. The source code contains handles that allow the compiler to link
the buffers of the hexadecimal files to the main source code. The only source
code file that you need to modify is the hardware.c file. The source files can
be found in the
<install_path>\embedded_source\i2cembedded\src\i2cem_eprom directory.

hardware.c
The only file that you should modify is hardware.c. This file contains the
functions to read and write to the port and the timing delay function. You must
update these functions to target the desired hardware.

opcode.h
The opcode.h file contains the definitions of the byte codes used in the I2C
algorithm format and programming engine.

Programming Tools User Guide 113

EMBEDDED FLOW OVERVIEW : I2C Embedded

i2c_core_eprom.c
The i2c_core.c file provides the programming engine for the I2C embedded
system. The engine operates on the commands in the I2C algorithm, and
fetches data from the I2C data, if necessary. The engine is responsible for
functions such as sending data, verifying data, observing timing delay,
decompression, and so on.

i2c_eprom.c
The i2c_main.c contains the main and entry point functions for 8051-based
I2C Programming.

I2C Algorithm Format

The I2C algorithm file contains byte codes that represent the programming
algorithm of the device or chain.

Table 6: I2C Algorithm Byte Codes

I2C Symbol Hex Value

I2C_STARTTRAN 0x10

I2C_RESTARTTRAN 0x11

I2C_ENDTRAN 0x12

I2C_TRANSOUT 0x13

I2C_TRANSIN 0x14

I2C_RUNCLOCK 0x15

I2C_WAIT 0x16

I2C_LOOP 0x17

I2C_ENDLOOP 0x18

I2C_TDI 0x19

I2C_CONTINUE 0x1A

I2C_TDO 0x1B

I2C_MASK 0x1C

I2C_BEGIN_REPEAT 0x1D

I2C_END_REPEAT 0x1E

I2C_END_FRAME 0x1F

I2C_DATA 0x20

I2C_PROGRAM 0x21

I2C_VERIFY 0x22

114 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : I2C Embedded

Table 6: I2C Algorithm Byte Codes (Continued)

I2C Symbol Hex Value

I2C_DTDI 0x23

I2C_DTDO 0x24

I2C_COMMENT 0x25

I2C_ENDVME 0x7F

I2C Data Format

While the I2C algorithm file contains the programming algorithm of the device,
the I2C data file contains the fuse and USERCODE patterns.

The first byte in the file indicates whether the data file has been compressed.
A byte of 0x00 indicates that no compression was selected, and 0x01
indicates that compression was selected. When compression has been
selected, each frame is preceded by a frame compression byte to indicate
whether the frame is compressible. This is necessary because even though
you might elect to compress the I2C data file, it is possible that a compressed
frame will actually be larger than an uncompressed frame. When that
happens, the frame is not compressed at all and the frame compression byte
of 0x00 notifies the processor that no compression was performed on the
frame.

When compression has not been selected, the I2C data file becomes a direct
translation from the data sections of the SVF file. The END_FRAME byte,
0x1F, is appended to the end of every frame as a means for the processor to
verify that the frame has indeed reached the end.
Uncompressed I2C Data Format Compressed I2C Data Format
0x00 0x01
<Frame 1>0x10 <Compress Byte><Frame 1>0x10
<Frame 2>0x10 <Compress Byte><Frame 2>0x10
… …
<Frame N>0x10 <Compress Byte><Frame N>0x10

The compression scheme used is based on the consecutive appearance of

the 0xFF byte in a frame. This byte is ubiquitous because an all 0xFF data file
is a blank pattern. When a consecutive number of n 0xFF bytes are
encountered, the I2C data file will have the byte 0xFF followed by the number
n converted to hexadecimal, where n cannot exceed 255. For example, if the
following were a partial data frame.

FFFFFFFFFFFFFFFFFFFF12FFFFFF the resulting compressed data would

be:

0xFF 0x0A 0x12 0xFF 0x03

Programming Tools User Guide 115

EMBEDDED FLOW OVERVIEW : I2C Embedded

When the processor encounters the first byte 0xFF, it gets the next byte to
determine how many times 0xFF is compressed. The next byte is 0x0A,
which is ten in hexadecimal. This instructs the processor that 0xFF is
compressed ten times. The following byte is 0x12, which is processed as it is.
The next byte is again 0xFF followed by 0x03, which instructs the processor
that 0xFF is compressed three times.

I2C Embedded Programming Required

User Changes
To make the I2C Embedded Programming software work on your target
system, you need to modify the following C functions in the hardware.c source
code.

Timer(SetI2Cdelay())
The engine requires the ability to delay for fixed time periods. The minimum
granularity of the delay is 1 microsecond. You can determine the type of delay.
This can be a simple software timing loop, a hardware timer, or an operating
system call, for example, sleep().

Port Initialization
The firmware needs to place the port I/O into a known state.

SetI2CStartCondition()
This function is used to issue a start sequence on the I2C Bus.

SetI2CreStartCondition()
This function is used to issue a start sequence on the I2C Bus.

SetI2CStopCondition()
This function is used to issue a stop sequence on the I2C Bus.

ReadBytesAndSendNACK()
This function is used to read the SDA pin from the port.

SendBytesAndCheckACK()
To apply the specified value to the SDA pin indicated.

116 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : I2C Embedded

Generating I2C Files

This section describes how to generate I2C files. An .xcf file is required for the
MachXO2 or MachXO3L FPGA.

To generate an .xcf file for the MachXO2 or MachXO3L, if the .xcf file
does not exist or has not yet been created:
1. Start the Diamond Programmer software and create a new Blank Project.
2. Select MachXO2 or MachXO3L as Device Family.
3. Select the Device Type according to your device.
4. Choose Edit > Device Properties, or right click on the device, and in the
dropdown menu, choose Device Properties.
5. In the Device Properties dialog box:

 In the Access Mode dropdown menu, choose I2C Interface

Programming.
 In the Operation dropdown menu, choose the desired operation.
 In the Programming File box, browse to your design’s .jed
programming file.

 In the I2C Slave Address box, enter the correct I2C slave address. The
default address is 0x40.
6. Chose File > Save or File > Save (filename).xcf As... and give the file a
name. Ensure that the extension of the file is xcf.

To generate I2C Files:

1. Start the Deployment Tool.
2. In the Getting Started dialog box, select Create New Deployment.
3. In the Function Type dropdown menu, choose Embedded System.
4. In the Output File Type dropdown menu, choose as I2C Embedded.
5. Click OK.
6. In the Step 1 of 4 dialog box, browse to the .xcf file you created with the
Programmer software, and select Input XCF file(s).
7. Click Next.
8. In the Step 2 of 4 dialog box, select Compress Embedded Files
depending upon the requirement, select Generate Hex(.c) Files for 8051
micro-processor usages, and click Next.
9. In the Step 3 of 4 dialog box, select the Algorithm File and Data File to
rename and change the location of the file name. Make sure the file name
has the extension .iea and .ied, respectively, and click Next.
10. In the Step 4 of 4 dialog box, click Generate to generate the files.
11. The files will be generated as shown as below.

Programming Tools User Guide 117

EMBEDDED FLOW OVERVIEW : I2C Embedded

12. The Deployment Tool project can now be saved by selecting File > Save
As. The saved file will generate the same data file and algorithm file when
loaded again.
13. Modify the Source Code File (hardware.c). The 8051-based source code
files are written in ANSI C. The file hardware.c is the only file that is
required to be modified by the user. The user must modify the following
functions according to the target platform:
SetI2Cdelay()
SetI2CStartCondition()
SetI2CreStartCondition()
SetI2CStopCondition()
ReadBytesAndSendNACK( int length, unsigned char *a_ByteRead
, int NAck)
Where
int length = Number of bits to read
*a_ByteRead = Buffer to store byte
int NAck - Option to send
0 = No
1 - Yes
int SendBytesAndCheckACK(int length, unsigned char
*a_bByteSend
Where
int length = Number of bits to send
*a_bByteSend = Buffer storing data to send
The following are optional functions that the user may wish to modify in
order to enable and disable the hardware conditions before and after
processing:
EnableHardware()
DisableHardware()
14. Compile Source Code and I2C HEX Files. Combine the source code and
I2C HEX files generated by Deployment Tool into a project to be compiled.
This may be done using a microcontroller development tool to create the
project.

Modify the Delay Function

When porting Embedded I2C to a native CPU environment, the speed of the
CPU or the system clock that drives the CPU is usually known. The speed or
the time it takes for the native CPU to execute one loop then can be
calculated.

The for loop usually is compiled into the ASSEMBLY code as shown below:
LOOP: EDC RA;
JNZ LOOP;

118 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : I2C Embedded

If each line of assembly code needs four (4) machine cycles to execute, the
total number of machine cycles to execute the loop is 2 x 4 = 8.

Usually: system clock = machine clock (the internal CPU clock).

Note
Some CPUs have a clock multiplier to double the system clock for the machine clock.

Let the machine clock frequency of the CPU be F (in MHz), then one machine
cycle = 1/F.

The time it takes to execute one loop = (1/F) x 8.

It is obvious that the formula can be transposed into one microsecond = F/8.

Example: The CPU internal clock is set to 48 MHz, then one microsecond =
48/8 = 6.

The C code shown below can be used to create the millisecond accuracy. All
that needs to be changed is the CPU speed.

Choosing the File-Based or EPROM-Based Version

To generate a PROM-based I2C Embedded, select the “Generate HEX (.c)
File” option in the Deployment Tool Step 2 of 4 dialog box.The programming
engines of the file-based and PROM-based processors are identical in the
way they handle the VME commands. Their difference lies in the way they
interface with I2C Embedded data. For a convenient demo, the file-based

Programming Tools User Guide 119

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

version assigns a file pointer to the binary I2C Embedded file directly. The
pointer is assigned based on a command line argument. With some minor
modification, this version is useful for embedded high-level 32-bit
microprocessors that can dynamically allocate RAM and have large amounts
of data and code memory. For more modest embedded systems or smaller
processors, the PROM-based version is useful because the memory
resources are completely defined when compiling the executable.The I2C
Embedded file is converted to one or more C files and a header file that are
compiled with the core routines.

sysCONFIG Embedded
The sysCONFIG Embedded software brings sysCONFIG port programming
capability to Lattice Semiconductor's suite of FPGA devices on embedded
systems. The software offers a unique virtual programming engine that can
support the different programming algorithms of each device family. The
software is developed in ANSI C and is available in source code form to allow
users to target specific hardware.

The Diamond software may be used to generate a bitstream file. The file then
must be converted to an embedded bitstream using Deployment Tool. The
significance of the embedded bitstream is that it contains control pin
instructions, which the sysCONFIG Embedded programming engine will
decode. Once the embedded bitstream is obtained, it will be provided to the
sysCONFIG Embedded software to configure the device.

While excellent support is offered for programming devices on PC and

workstation platforms, many alternate platforms are in use. The best example
is the use of embedded controllers and processors for programming. Many
embedded configurations are in use, but these configurations vary greatly
from project to project. Topics in this Help describe the sysCONFIG
Embedded software in detail to simplify modification for user-specific
applications.

120 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

sysCONFIG Embedded Flow

The following figure illustrates sysCONFIG Embedded flow.

sysCONFIG Embedded Bitstream Generation You can use Deployment

Tool to generate sysCONFIG Embedded bitstreams for the current Lattice
FPGA device chain.
 The chain must consist of a single Lattice FPGA device.
 The device must have an input data file.
 The device must be supported by the sysCONFIG Embedded software.
See the “sysCONFIG Embedded Device Support” topic in the Diamond
Programmer online help.

Parallel Mode and Serial Mode CPU Bitstream For LatticeECP/EC and
LatticeECP2 devices, the Deployment Tool software can generate either a
parallel configuration mode (PCM) CPU or a serial configuration mode (SCM)
CPU file depending on the operation type you have specified for the device in
the Device Properties Dialog Box in Diamond Programmer. For example, if
you have selected JTAG 1532 Mode under Device Access Options and Fast
Program under Operation, the CPU bitstream generator will output a parallel
mode CPU file. If you selected Serial Mode and Serial Program in the
Device Information dialog box, a serial mode CPU file will be generated.

sysCONFIG Embedded Bitstream Options

There are several options to consider before generating the sysCONFIG
Embedded bitstream. You can set these options in the Deployment Tool Refer
to Deployment Tool online help for more information.

Output Format This option, the most important, allows you to specify one of
three sysCONFIG Embedded bitstream formats for generating the embedded
bitstream: binary, C-code, and Intel hexadecimal. You can also generate a
CPU file in text format so that you can see all the commands, data, and the
size of data shifting for debugging purpose.

Include Comments Comments are decoded by the sysCONFIG Embedded

engine and displayed to the terminal to inform you of the engine's actions.
This can be very helpful for debugging, but it increases the file size.

Compress Embedded Bitstream Files This enables you to generate a

compressed CPU bitstream file with smaller size and faster performance.

Programming Tools User Guide 121

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

For Erase, Program, and Verify Operations, Skip the Verify Operation

You can skip the Verify operation while programming a device to reduce
processing time.

Specify Values for the CFG Pins The software allows you to specify the
values of the CFG pins by using configuration mode. By default, selection of
these values is disabled.

See Also Generating a sysCONFIG Embedded Bitstream

 sysCONFIG Embedded Basic Operation

sysCONFIG Embedded Bitstream

Format
The sysCONFIG Embedded software supports three embedded bitstream
formats.
 Binary. The binary format is used for the file-based sysCONFIG
Embedded. This allows you to call the sysCONFIG Embedded via
command line and pass in the binary embedded bitstream as an
argument. The file extension for this format is .cpu.
 C-code. The C-code format is used for the compile-based sysCONFIG
Embedded. This format is an actual C programming language file that has
the embedded bitstream expressed as an array. The file can be compiled
with the compile-based sysCONFIG Embedded source code to generate
an executable that has the embedded bitstream self-contained. The file
extension for this format is .c.
 Intel Hex. The Intel hexadecimal format is used for the function-based
sysCONFIG Embedded. This format is a standard Intel hexadecimal that
can be downloaded to the flash of the embedded system. A wrapper
application may be developed to call the function-based sysCONFIG
Embedded and pass to it the embedded bitstream from the flash. This
format contains an option to flip each byte of data prior to generating the
file. By default, each byte is flipped. Depending on the user's
implementation, it may not be necessary to flip the bytes. The file
extension for this format is .hex.

Besides the above formats, you can also generate a CPU file in text format so
that you can see all the commands, data, and the size of data shifting for
debugging purpose.

See Also sysCONFIG Embedded Bitstream Structure

 Generating a sysCONFIG Embedded Bitstream

122 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

sysCONFIG Embedded Bitstream

Structure
The embedded bitstream is the driver behind the sysCONFIG Embedded
software. It contains both instructions for the control pins and data for the data
pins of the FPGA device. This method encapsulates the programming
algorithm within the embedded bitstream, and generalizes the sysCONFIG
Embedded engine to increase versatility.

The following figure illustrates the algorithm and data contained in

sysCONFIG Embedded bitstream.

The algorithm section drives and verifies the control pins. A verification failure
return code is issued if a pin fails to verify.

The original bitstream section contains the bitstream broken down by frames.
If the comment option was selected when generating the embedded
bitstream, comments are inserted before frames and displayed by the CPU
engine to notify users the current frame being processed.

The entire embedded bitstream is protected by a CRC. Before the

sysCONFIG Embedded software sends any data to the device, it calculates
the CRC and verifies it against the CRC of the bitstream. If the verification
fails, the software aborts with a CRC error code.

This CRC is referred to as the “soft” CRC, not to be confused with the CRC
calculation that is performed internally by the FPGA device. See the figure
below.

Programming Tools User Guide 123

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

Example The following embedded bitstream example shall perform the

following:
1. Verify CRC.
2. Drive the PROGRAM pin high.
3. Send in pre-amble data 0x83 0xA7.
4. Issue the end byte command.

Embedded bitstream example:

0X08 0XFF 0XFF // 16-BIT CRC COMMAND
0X05 0X50 0X01 // DRIVE PROG_PIN HIGH
0X02 0X02 0X83 0XA7 // SEND PRE-AMBLE DATA
0X04 // END BYTE

sysCONFIG Embedded Basic Operation

The sysCONFIG Embedded is the software that will decode the sysCONFIG
Embedded bitstream and configure the FPGA device. It is available in three
interfaces and two configuration modes.

sysCONFIG Embedded Interfaces The sysCONFIG Embedded software

is available in source code form and allows three possible interfaces.

File-Based Interface The file-based interface expects the embedded

bitstream in the binary format, with .cpu extension. This interface is ideal for
simulating on PC. It allows you to call the software through command line and
pass the binary embedded bitstream as the argument. Note that this assumes
you have modified the source code to allow the program to interface to the
device through a driver.

Compile-Based Interface The compile-based interface expects the

embedded bitstream in C-code format. This file shall be compiled along with
the sysCONFIG Embedded source code to create an executable that has the

124 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

data self-contained. This interface is ideal for programming on embedded

systems.

Function-Based Interface The function-based interface expects the

embedded bitstream in Intel hexadecimal format. This file shall be stored in
the flash or PROM of the embedded system for later use. Unlike the file-
based and compile-based approach, the interface to the sysCONFIG
Embedded engine in this interface is through a function. The arguments to the
function shall be a pointer to the embedded bitstream residing in the flash and
the size of the embedded bitstream.

sysCONFIG Embedded Configuration Modes

For LatticeECP/EC and LatticeECP2 devices, the software can generate
either a parallel configuration mode (PCM) CPU or a serial configuration
mode (SCM) CPU file depending on the operation type you have specified for
the device in the Device Properties dialog box. For example, if you have
selected JTAG 1532 Mode under Device Access Options and Fast Program
under Operation, the CPU bitstream generator will output a parallel mode
CPU file. If you selected Serial Mode and Serial Program in the Device
Information dialog box, a serial configuration mode CPU file will be generated.

Parallel Configuration Mode (PCM) The parallel programming engine

reads the CPU format files generated by Deployment Tool and manipulates
the I/O port. The engine requires the I/O system to be connected to the
parallel programming port of the device. For Lattice devices this is the
sysCONFIG parallel port.

Serial Configuration Mode (SCM) The serial programming engine

manipulates the I/O port. It requires the I/O system to be connected to the
serial programming port of the device. For Lattice devices this is the
sysCONFIG DI/DO port.

See Also sysCONFIG Embedded Flow

sysCONFIG Embedded Source Code

sysCONFIG Embedded source code is developed using ANSI C to ease
portability among the many different platforms. The source code files can be
found in the
<install_path>\<version_number>\embedded_source\cpuembedded
directory. The following lists the available source code files.
 main_f.c
 main_e.c
 main_tag.c
 cpu_core.c
 cpu_hard.c
 cpu_sim.c
 cpu_code.h

Programming Tools User Guide 125

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

There are five components to the sysCONFIG Embedded source code: main,
engine, hardware, simulator, and definition.

Main There are three different main files depending on whether you want to
use the file-based, compile-based, or function-based interface. Each interface
is available in main_f.c, main_e.c, and main_tag.c, respectively. When using
the compile-based interface, the embedded bitstream must be generated as
C-code and compiled with the rest of the source code.

Engine The sysCONFIG Embedded Engine can be found in the

cpu_core.c file. This file is responsible for decoding and executing the
embedded bitstream. You should not modify this file.

Hardware The hardware related functions can be found in the cpu_hard.c

file. This file is responsible for writing to and reading from the ports that
connect to the device. You may modify sections of this file to target specific
hardware. These sections are preceded by the comment “Note: user
must re-implement to target specific hardware” to alert you
that it requires modification. If you include this file, you will have to remove the
cpu_sim.c file from your project. These two files are not compatible.

Simulator The simulation function could be found in the cpu_sim.c file.

This file is used for generating an output file with the resultant code. This
could be done if you want to simply compare outputs without having a device
plugged in. The simulator will take the following argument: “.cpu file
<output file>”. If you use this file, remove the cpu_hard.c file from your
project.

Definitions The definitions for the embedded bitstream, sysCONFIG

Embedded engine, and return codes can be found in the cpu_code.h file.
You should not modify this file.

sysCONFIG Embedded Engine

The sysCONFIG Embedded software brings sysCONFIG port programming
capability to Lattice Semiconductor's suite of FPGA devices on embedded
systems. The software offers a unique virtual programming engine that can
support the different programming algorithms of each device family.

The sysCONFIG Embedded engine is based on the sysCONFIG Embedded

bitstream. It processes the bitstream until the END_BYTE_COMMAND or an
error is encountered and returns with an appropriate return code.

The sysCONFIG Embedded engine supports the following different run

operations generated from Deployment Tool:
 Erase, Program, Verify
 Erase, Program, Verify, and Secure
 Erase Only
 Verify Only
 Verify ID

126 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

 SRAM Program, Verify

 SRAM Verify Only
 XSRAM Verify Only

The sysCONFIG Embedded engine has three layers as follows:

 User interface layer (cpu_main.c) – Directs inputs and outputs.
 Processor layer (cpu_core.c) – Decodes commands, checks CRC prior to
processing, and does optional decompression.
 Physical layer (cpu_hard.c) – Shifts data to target device. This is the only
file that you need to edit.

The following figures illustrate sysCONFIG Embedded sysCONFIG port

programming engine in parallel and serial configuration modes.

Parallel Configuration Mode.

Programming Tools User Guide 127

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

Serial Configuration Mode.

Sample Program Size

This page provides sample program size for JTAG Full VME Embedded,
JTAG Slim VME Embedded, and sysCONFIG Embedded.

JTAG Port non-JTAG Port Total

JTAG Full VME JTAG Slim sysCONFIG

Embedded VME Embedded
Embedded

32-Bit 16-Bit 8-Bit 32-Bit 16-Bit

File Based 52KB 21KB 4.2KB 48KB 19KB As Shown

(Bitstream File External)

PROM Based 52KB 21KB 4.2KB 48KB 19KB As Shown +

(Bitstream File Integrated) VME File Size

128 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

Generating a sysCONFIG Embedded

Bitstream
You can generate sysCONFIG Embedded bitstream using the Deployment
Tool for programming sysCONFIG Embedded supported devices.

For LatticeECP/EC and LatticeECP2 devices, the Deployment Tool can

generate either a parallel programming mode CPU or a serial programming
mode CPU file depending on the operation type you have specified for the
device in the Device Properties Dialog Box in Diamond Programmer. For
example, if you have selected JTAG 1532 Mode under Device Access
Options and Fast Program under Operation, the CPU generator will output a
parallel programming mode CPU file. If you selected Serial Mode and Serial
Program in the Device Properties dialog box in Programmer, a serial
programming mode CPU file will be generated.

There are two important options to consider when generating the embedded
bitstream. The first option is whether or not to generate the file with built-in
comments. These comments are decoded by the sysCONFIG Embedded

Programming Tools User Guide 129

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

engine and displayed to the terminal to inform the engine's actions. This is
helpful for debugging purposes, but the file size will increase slightly.

The second, and more important, option to consider is which of the three
formats should the embedded bitstream be generated with. See sysCONFIG
Embedded Bitstream Format for details.

See Also sysCONFIG Embedded Flow

sysCONFIG SPI Port AC Parameters

The board layout and waveform diagrams are shown here to illustrate when
the configuration AC parameters can be characterized for the following
configuration modes:
 SPI (Serial Peripheral Interface)
 PCM (Parallel Configuration Mode)
 SCM (Serial Configuration Mode)

SPI (Serial Peripheral Interface)

Board Layout.

130 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

Waveform

PCM (Parallel Configuration Mode)

Board Layout

Waveform

Programming Tools User Guide 131

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

132 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

SCM (Serial Configuration Mode)

Board Layout

Waveform

sysCONFIG Interface
This page provides interface diagrams and description on the following CPU
sysCONFIG solutions.
 CPU programming solution
The SRAM fuses in LatticeXP and LatticeECP/EC and the FLASH fuses
in LatticeXP can be programmed by using the sysCONFIG port connected
to a CPU. This page will describe SRAM fuse programming and present a
66MHz programming solution for LatticeECP/EC devices.

Programming Tools User Guide 133

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

If both the DONE pin and the INIT pin are high after programming, or
refresh, the device then must be programmed successfully.

Note
Only if the FLASH DONE fuse is programmed can the FLASH pattern be
downloaded into the SRAM fuses of the device. If the FLASH DONE fuse is not
programmed, the download is blocked. FLASH DONE fuse is the last fuse to be
downloaded from FLASH to SRAM. The FLASH is downloaded to SRAM in a
protected and shielded environment and therefore it is not subject to the effect of
external noise or ground bounce which is the common problem when downloading
from external FLASH memory devices. The DONE pin is a very reliable indicator
on the programming status. This renders the readback un-necessary.

 CPU read back solution

The SRAM fuses in LatticeXP and LatticeECP/EC and the FLASH fuses
in LatticeXP can be read back by using the sysCONFIG port connected to
a CPU. This page will describe SRAM fuse read back and present CPU
read back solution for LatticeECP/EC devices.
Read back of SRAM fuses is presented here for completeness. There is
very little to gain on actually performing read back.
Read back of SRAM fuses can only be done when the following
conditions are met:
 The device has already been programmed successfully;
 The device has already waken-up and is in user mode;
 All EBR data must be ignored;
 All distributed RAM data must be ignored.

Note
Only the JTAG port of the device supports readback in both user mode and
programming mode. If the device has not yet been programmed successfully, it
will not respond to the READ_INC command sent to it on the sysCONFIG port.

Subjects included in this page:

 CPU Interface Basic
 The Bitstream Format
 CPU Parallel Programming Interface
 CPU Serial Programming Interface
 CPU Read back Interface
 CPU Byte-Wide Programming Flow
 CPU Byte-Wide Read back Flow
 CPU Bit-Wide Programming Flow

CPU Interface Basic Lattice's CPU interface supports the command,

operand, and data bitstream format.

134 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

The Bitstream Format

Bitstream Type Bitstream Orientation Data (Binary) Data (Hex)

D0..D7,D0..D7...

Preamble Code Bit[15..0] 1011110110110011 BDB3

Command Bit[31..0] Bit[31..24] = Opcode;

Bit[23..0] = Command
Information

Data IDCODE Bit[0..31] Bit[0..31] = IDCODE ECP-20-IDCODE=C20A2480

Control Register Bit[0..31] Bit[0..31] = Control Register

Config. Bit[N-1..0] Configuration Data

CRC Bit[0..16] Bit[0..16] = CRC

Usercode Bit[0..31] Bit[0..31] = Usercode

Programming Tools User Guide 135

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

Command 32-bit Command Write Read Description

Bit 31...24 Bit 23...0

Opcode Operand

RESET_ADDRESS hE2 Don't Care X X Set address to row 0 (first row).

CLEAR ALL hF0 Don't Care X Clear all SRAM fuses.

WRITE_INC hC1 # Of Frames X Program a frame then increment the row

address.

READ_INC h81 # Of Frames X X Read a frame then increment the row

address.

WRITE_USERCODE hC3 Don't Care X Program 32 bits usercode

READ_USERCODE h83 Don't Care X X Read out 32 bits usercode.

WRITE CTRL 0 hC4 Don't Care X Program the 32 bits program flow control
register 0.

READ CTRL 0 h84 Don't Care X X Read out from the 32 bits control register 0.

PROGRAM_SECURIT hF3 Don't Care X Program the security fuses to disable

Y readback.

PROGRAM_DONE / hFA Don't Care X X Terminate the programming or readback

END_READ flow.

READ_IDCODE h87 Don't Care X X Read out the 32 bits JTAG IDCODE of the
device.

VERIFY_IDCODE hC7 Don't Care X X Compare the 32 bits data against the 32 bits
JTAG IDCODE.

BYPASS hFC Don't Care X X Send data to Dout for the next device in
chain.

FLOW_THROUGH hFD Don't Care X X Drive CSO low to select the next device in
chain.

NOOP hFF None X X Put a device in a wait state for extra delay.

Bit 31 of the opcode determines if the command and data is included in CRC.
The opcode shown does not select CRC inclusion. The bitstream generated
by the Diamond software makes that decision automatically. The read
command is not recommended to include CRC.

136 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

CPU Parallel Programming Interface

Only four pins, PROGRAMN, INITN, DONE, and CCLK, are dedicated. The
other pins are dual purpose pins. The DI pin has no function on parallel
programming.

CPU Serial Programming Interface

Only four pins, PROGRAMN, INITN, DONE, and CCLK, are dedicated. The
other pins are dual purpose pins. Only the DI pin and DOUT/CSO pin are
used on serial programming.

Programming Tools User Guide 137

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

CPU Read back Interface

Only four pins, PROGRAMN, INITN, DONE, and CCLK, are dedicated. The
other pins are dual purpose pins. The read back is turned on by selecting the
Persistent On option when generating the bitstream.

The DI pin has no function on read back. Same as other dual purpose pins, it
cannot be recovered as user IO if read back is selected.

The PROGRAMN pin must be left alone during read back. Pulsing it will
terminate the read back immediately and start re-configuration.

The device responds to read back command only if it has already been
configured successfully.

CPU Byte-Wide Programming Flow Below is the CPU byte-wide

programming flow:
1. Toggle PROGRAMN pin to set the devices into programming mode.
2. Wait 1 microsecond then check if the DONE pin is low. If it is low then
continue. If it is high then report failure.
3. Pulse several clocks on CCLK to make sure the devices are ready.
4. Wait 20ms then check INIT pin. If it is high, then continue.
5. Drive CSN, CS1N and WRITEN to low to activate the byte-wide D[0..7]
interface to receive data.
6. Send the first byte bit[7..0] from the bitstream file to D[0..7] then pulse the
clock to clock it in.
7. Repeat step 6 till the whole bitstream file is sent to the device.
Programming is then complete.
8. Check if INIT pin is high. If it is high then continue. If it is low, then report
failure.
9. Pulse 100 clocks on CCLK to ensure the devices are waking up.
10. Check if the DONE pin is high. If it is high, then programming is
successful. If it is low, then report failure.

138 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

11. Drive CSN, CS1N, WRITEN back to high to terminate.

Note
1. The device captures data on the rising edge of CCLK.
2. The maximum rate sending data to the device is 66MHz.
3. The bitstream file could be for one device or merged for two or more devices.

Board Layout

Waveform

CPU Byte-Wide Read back Flow Below is the CPU byte-wide read back
flow. Before performing CPU byte-wide read back, the persistent fuse must be
programmed already to enable reading.
1. Check if both DONE and INITN pins are high. If they are high then
continue. If not then report failure.
2. Drive WRITEN, CSN, and CS1N to low to enable the D[0..7] interface to
receive command.

Programming Tools User Guide 139

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

3. Write several bytes of 0xFF to D[0..7] and clock them into the device to
clear the bus.
4. Write the pre-amble code 0xBD, 0xB3 and clock them into the device to
enable the command decoder.
5. Write the READ_IDCODE command 0x87 to D[0..7] and clock them into
the device then follow with 3 bytes of 0x00 as operand. Reading IDCODE
is necessary to ensure communication is established with the device.
6. Drive CSN from low to high (or CS1N, not both) then drive WRITEN from
low to high. Pulse one clock on CCLK to change D[0..7] from input port to
output port.
7. Drive CSN back from high to low (or CS1N if it was driven high at step 6).
Pulse one clock on CCLK to present one byte of data on D[0..7].
8. The BUSY pin tracks the CSN (or CS1N) pin. The data on D[0..7] is not
valid when it is high.
9. Read the first byte of the IDCODE from the D[0..7] of the first byte bit[0..7]
of the JTAG IDCODE.
10. Pulse CCLK then read the next byte bit[8..15] of the JTAG IDCODE.
11. Repeat step 10 till bit[16..23] then bit[24..31] is read from D[0..7].
12. Compare the 32 bits IDCODE read from the device against the expected
IDCODE of the device. If they match, then continue. If not, then report
error.
13. Repeat Step 5 to 11 to read the USERCODE. The READ_USERCODE
opcode is 0x83. It is a good practice to put the fuse checksum on
usercode to indicate that the device has been programmed correctly to
the pattern when reading back.
14. Drive CSN from low to high then drive WRITEN from high to low. Pulse
one clock on CCLK to change D[0..7] back to an input port.
15. Drive the first byte of the RESET_ADDRESS command, 0xE2, to D[0..7].
Pulse one clock on CCLK for the device to read it in.
16. Drive D[0..7] and clock CCLK to send 3 bytes of 0x00 as dummy
operands to the device. The address of the device is now set to the first
frame.
17. Drive the first byte of the READ_INC command, 0x81, to D[0..7]. Pulse
one clock on CCLK for the device to read it in.
18. Drive to D[0..7] and clock CCLK the number of frames to be read from the
device expressed in 24 bits hex number, or 3 bytes.
Example: If read 3 frames, then 3 bytes of operand is 0x00, 0x00, 0x03. If
read 256 frames, then 3 bytes of operand is 0x00, 0x01, 0x00.

Note
The number must not be larger than the maximum number of frames in the
device.

19. Drive CSN from low to high then drive WRITEN from low to high, then
pulse CCLK to set D[0..7] up as an output port.

140 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

20. Drive CSN from high to low then pulse CCLK for the device to drive the
first byte from the first frame to D[0..7].
21. Pulse CCLK then read the next byte till all the bytes in the first frame are
read. The device will increment the address automatically when all bytes
in the current frame are read.
22. Continue to pulse CCLK till all the frames or up to what is specific on the
operand sent to the device at step 18.

Note
Do not read beyond that number, otherwise the device will drive INIT low to
indicate over-read.

23. Drive CSN from low to high then drive WRITEN from high to low. Pulse
one clock on CCLK to change D[0..7] back to an input port.
24. Drive the opcode, 0xFA, of the END_READ command to D[0..7]. Pulse
CCLK for the device to read it in.
25. Drive 0x00 to D[0..7] then pulse CCLK 3 times as the dummy operand to
complete the command. Read back is then terminated.
26. Drive CSN, CS1N, and WRITEN back to high to disconnect the
sysCONFIG port of the device.

Note
1. If the command is under shift, then the device will enter error state.
2. If the command is over shift, then the device will also enter error state.
3. If the data is under shift, there is no error state.
4. If the data is over shift, dummy data is presented to D[0..7]. The device will not
enter error state.
5. Driving both CSN and CS1N to high will reset the device.
6. The read back clock frequency is much slower than 66MHZ due to the time
required to switch the polarity of D[0..7] from input to output.
7. The data is shift out from the device on the rising edge of CCLK.
8. If the security fuse is programmed, the usercode and JTAG IDCODE still can be
readback.
9. The opcode sent to the device for reading purpose are not recommended to
include CRC. Hence bit7 of the opcode is 1.
10. The number of frames put on the operand must be less than or equal to the
number of frames the device actually has.
11. If the number exceeds the actual number of frames, the device will enter error
state.

CPU Bit-Wide Programming Flow Below is the CPU bit-wide

programming flow:
1. Toggle PROGRAMN pin to set the devices into programming mode.
2. Wait 1 microsecond then check if the DONE pin is low. If it is low then
continue. If it is high then report failure.

Programming Tools User Guide 141

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

3. Pulse several clocks on CCLK to make sure the devices are ready.
4. Wait 20ms then check INITN pin. If it is high, then continue.
5. Drive bit7 of the first byte from the bitstream to DI then pulse the clock to
clock it in.
6. Repeat step 5 till bit6..0 of the first byte is sent.
7. Repeat step 6 till the whole bitstream file is sent to the device.
Programming is then complete.
8. Check if INIT pin is high. If it is high then continue. If it is low, then report
failure.
9. Pulse 100 clocks on CCLK to ensure the devices are waking up.
10. Check if the DONE pin is high. If it is high, then programming is
successful. If it is low, then report failure.

Note
1. The device captures data on the rising edge of CCLK.
2. The maximum rate sending data to the device is 66MHz.
3. The bitstream file could be for one device or merged for two or more devices.

Board Layout

142 Programming Tools User Guide

 EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

Waveform

Programming Tools User Guide 143

EMBEDDED FLOW OVERVIEW : sysCONFIG Embedded

144 Programming Tools User Guide

Numerics
8051
generating slim VME files 84
using the 8051-based slim ispVME 76

B
bitstream
generating CPU embedded bitstream 129

C
CPU Embedded
bitstream format 122
engine 126
flow 121
source code 125
CPU generating 129

D
Deployment Tool 1, 13
device programming
see programming devices
Diamond Programmer 1, 5
Download Debugger 2

E
Embedded System
RAM size requirement for ispVME 37
ROM size requirement for ispVME 39
Embedded, I2C 19
Embedded, JTAG, Full VME 19, 30
Embedded, JTAG, Slim VME 19, 73
Embedded, Slave SPI 19, 100
Embedded, sysCONFIG 19, 120

F
file generation
CPU embedded bitstream 129
file processing
VME 52
file size
program size 42
FPGA
generating a CPU embedded bitstream 129
Full VME Embedded, JTAG 19, 30

G
generating
CPU embedded bitstream 129
slim VME 84

I
I2C Embedded 19
ispVM Embedded
RAM size requirement for ispVME 37
ROM size requirement for ispVME 39
ispVME
engine 36
flow 33
source code 35
ispVME system memory 34

J
JTAG Full VME Embedded 19, 30
JTAG Slim VME Embedded 19, 73

Programming Tools User Guide 145

M
memory allocation 41
Model 300 2

P
processing
VME 52
program memory allocation 41
program memory requirement 41
Programmer
using 8
Programmer, Diamond 1, 5
programming devices
using Programmer 8
programming engine
CPU Embedded 126
ispVME 36
Programming File Utility 2

R
RAM size requirement for ispVME 37
resource requirements
program memory 41
program memory allocation 41
sample program size 42
ROM size requirement for ispVME 39

S
Slave SPI Embedded 19, 100
slim ispVME
generating slim VME 84
slim ispVME source code 74
using the 8051-based slim ispVME 76
using the PC-based slim ispVME 75
VME algorithm format 76
slim ispVMEVME data format 79
Slim VME Embedded, JTAG 19, 73
source code
CPU Embedded 125
ispVME 35
Slim ispVME 74
SPI, Slave, Embedded 19, 100
sysCONFIG Embedded 19, 120
system memory, ispVME 34

U
user changes, ispVME 39
using
8051-based slim ispVME 76
PC-based slim ispVME 75
slim ispVM Embedded System 75

V
VME
algorithm format - slim ispVME 76
data format - slim ispVME 79
format 30
processing 52
RAM size requirement 37
ROM size requirement 39

146 Programming Tools User Guide