Module Lesson

3 19
Embedded Systems I/O Analog Interfacing
Version 2 EE IIT, Kharagpur 1 Version 2 EE IIT, Kharagpur 2
Instructional Objectives
After going through this lesson the student would be able to

x Know the interfacing of analog signals to microcontrollers/microprocessors

x Generating Analog Signals
x Designing AD and DA interfaces
x Various Methods of acquiring and generating analog data

Pre-Requisite Fig. 19.2 The Analog-Digital-Analog signal path with real time processing

Digital Electronics, Microprocessors Different Stages of Fig.19.2

19(I) Introduction Stage-1 Signal Amplification and Conditioning; Stage-2 Anti-aliasing Filter; Stage-3 Sample and
Hold;
Fig.19.1 shows a typical sensor network. You will find a number of sensors and actuators
Stage-4 Analog to Digital Converter; Stage-5 Digital Processing and Data manipulation in a
connected to a common bus to share information and derive a collective decision. This is a
Processor;
complex embedded system. Digital camera falls under such a system. Only the analog signals are
shown here. Last lesson discussed in detail about the AD and DA conversion methods. This Stage-6 Processed Digital Values are temporarily stored in a latch before D-A conversion;
chapter shall discuss the inbuilt AD-DA converter and standalone converters and their Stage-7 Digital to Analog Conversion; Stage-8 Removal of Glitches and Spikes; Stage-8 Final
interfacing. Low pass filtering

19(II) Embedded AD Converters in Intel 80196

Fig.19.3 shows the block diagram of the AD converter inbuilt to 80196 embedded processor. The
details of the subsystems are given as follows:
Analog Inputs
EPA or PTS
Command
VREF ANGND
Analog Mux
Control
Successive Logic
Sample Approximation
and Hold A/D
Converter
Status

AD_RESULT AD_COMMAND AD_TIME AD_TEST

† Multiplexed with port inputs

Fig. 19.1 The Analog Interfacing Network
Fig. 19.3 The block diagram of the Internal AD converter
Analog Inputs: There are 12 input channels which are multiplexed with the Port P0 and Port P1
of the processor.

Version 2 EE IIT, Kharagpur 3 Version 2 EE IIT, Kharagpur 4

ANGND: It is the analog ground which is separately connected to the circuit from where analog threshold-detection, calculate the value for the successive approximation register and write that
voltage is brought inside the processor. value to the high byte of AD_RESULT. Clear the low byte or leave it in its default state.

Vref: It is reference voltage which decides the range of the input voltage. By making it negative AD_TEST A/D Conversion Test
bipolar inputs can be used. This register specifies adjustments for zero-offset errors.

EPA: Event Processor Array AD_TIME A/D Conversion Time

This register defines the sample window time and the conversion time for each bit.
Control applications often require high-speed event control. For example, the controller may
need to periodically generate pulse-width modulated outputs or an interrupt. In another INT_MASK Interrupt Mask
application, the controller may monitor an input signal to determine the status of an external The AD bit in this register enables or disables the A/D interrupt. Set the AD bit to enable the
device. The event processor array (EPA) was designed to reduce the CPU overhead associated interrupt request.
with these types of event control. This chapter describes the EPA and its timers and explains how
to configure and program them. The EPA can control AD converter such as generating timing INT_PEND Interrupt Pending
pulses, start conversion signals etc. The AD bit in this register, when set, indicates that an A/D interrupt request is pending.

PTS: Peripheral Transaction Server A/D Converter Operation

The microcontroller’s interrupt-handling system has two components: the programmable An A/D conversion converts an analog input voltage to a digital value, stores the result in
interrupt controller and the peripheral transaction server (PTS). The programmable interrupt the AD_RESULT register, and sets the A/D interrupt pending bit. An 8-bit conversion provides
controller has a hardware priority scheme that can be modified by the software. Interrupts that go 20 mV resolution, while a 10-bit conversion provides 5 mV resolution. An 8-bit conversion takes
through the interrupt controller are serviced by interrupt service routines that you provide. The less time than a 10-bit conversion because it has two fewer bits to resolve and the comparator
upper and lower interrupt vectors in special-purpose memory contain the interrupt service requires less settling time for 20 mV resolution than for 5 mV resolution. Either the voltage on
routines’ addresses. The peripheral transaction server (PTS), a microcoded hardware interrupt an analog input channel or a test voltage can be converted. Converting the test inputs is used to
processor, provides high-speed, low-overhead interrupt handling; it does not modify the stack or calculate the zero-offset error, and the zero-offset adjustment is used to compensate for it. This
the Processor Status Word. The PTS supports seven microcoded routines that enable it to feature can reduce or eliminate off-chip compensation hardware. Typically, the test voltages are
complete specific tasks in lesser time than an equivalent interrupt service routine can. It can converted to adjust for the zero-offset error before performing conversions on an input channel.
transfer bytes or words, either individually or in blocks, between any memory locations; manage The AD_TEST register is used to program for zero-offset adjustment. A threshold-detection
multiple analog-to-digital (A/D) conversions; and transmit and receive serial data in either compares an input voltage to a programmed reference voltage and sets the A/D interrupt pending
asynchronous or synchronous mode. bit when the input voltage crosses over or under the reference voltage. A conversion can be
started by a write to the AD_COMMAND register or it can be initiated by the EPA, which can
provide equally spaced samples or synchronization with external events.
Analog Mux: Analog Multiplexer
Once the A/D converter receives the command to start a conversion, a delay time elapses
It selects a particular analog channel for conversion. Only after completing conversion of one before sampling begins. During this sample delay, the hardware clears the successive
channel it switches to subsequent channels. approximation register and selects the designated multiplexer channel. After the sample delay,
the device connects the multiplexer output to the sample capacitor for the specified sample time.
The associated Registers After this sample window closes, it disconnects the multiplexer output from the sample capacitor
so that changes on the input pin will not alter the stored charge while the conversion is in
AD_COMMAND register progress. The device then zeros the comparator and begins the conversion. The A/D converter
This register selects the A/D channel, controls whether the A/D conversion starts immediately or uses a successive approximation algorithm to perform the analog-to-digital conversion. The
is triggered by the EPA, and selects the operating mode. converter hardware consists of a 256-resistor ladder, a comparator, coupling capacitors, and a 10-
bit successive approximation register (SAR) with logic that guides the process. The resistive
AD_RESULT ladder provides 20 mV steps (VREF = 5.12 volts), while capacitive coupling creates 5 mV steps
within the 20 mV ladder voltages. Therefore, 1024 internal reference voltage levels are available
For an A/D conversion, the high byte contains the eight MSBs from the conversion, while the for comparison against the analog input to generate a 10-bit conversion result. In 8- bit
low byte contains the two LSBs from a 10- bit conversion (undefined for an 8-bit conversion), conversion mode, only the resistive ladder is used, providing 256 internal reference voltage
indicates which A/D channel was used, and indicates whether the channel is idle. For a levels. The successive approximation conversion compares a sequence of reference voltages to

Version 2 EE IIT, Kharagpur 5 Version 2 EE IIT, Kharagpur 6

the analog input, performing a binary search for the reference voltage that most closely matches
the input. The ½ full scale reference voltage is the first tested. This corresponds to a 10-bit result IN3 1 28 IN2
where the most-significant bit is zero and all other bits are ones (0111111111). If the analog
IN4 2 27 IN1
input was less than the test voltage, bit 10 of the SAR is left at zero, and a new test voltage of ¼
full scale (0011111111) is tried. If the analog input was greater than the test voltage, bit 9 of IN5 3 26 IN0
SAR is set. Bit 8 is then cleared for the next test (0101111111). This binary search continues IN6 4 25 ADD A
until 10 (or 8) tests have occurred, at which time the valid conversion result resides in the IN7 5 24 ADD B
AD_RESULT register where it can be read by software. The result is equal to the ratio of the
START 6 23 ADD C
input voltage divided by the analog supply voltage. If the ratio is 1.00, the result will be all ones.
The following A/D converter parameters are programmable: EOC 7 22 ALE
• conversion input — input channel 2-5 8 21 2-1MSB
• zero-offset adjustment — no adjustment, plus 2.5 mV, minus 2.5 mV, or minus 5.0 mV OUTPUT ENABLE 9 20 2-2
• conversion times — sample window time and conversion time for each bit CLOCK 10 19 2-3
• operating mode — 8- or 10-bit conversion or 8-bit high or low threshold detection
VCC 11 18 2-4
• conversion trigger — immediate or EPA starts
VREF (+) 12 17 2-8LSB
19(III) The External AD Converters (AD0809) GND 13 16 VREF (-)
2-7 14 15 2-6
START CLOCK

8-BIT A/D
Fig. 19.5 The signals of 0809 AD converter
CONTROL & END OF
TIMING CONVERSION Functional Description
(INTERRUPT)
8
CHANNELS
8 ANALOG
MULTIPLE- Multiplexer
XING
INPUTS S.A.R
ANALOG
SWITCHES
TRI-
The device contains an 8-channel single-ended analog signal multiplexer. A particular input
COMPARATOR
STATE channel is selected by using the address decoder. Table 1 shows the input states for the address
OUTPUT
LATCH
8-BIT lines to select any channel. The address is latched into the decoder on the low-to-high transition
OUTPUTS
BUFFER of the address latch enable signal.
SWITCH TREE
TABLE 1
3-BIT ADDRESS ADDRESS
LATCH SELECTED ANALOG ADDRESS LINE
ADDRESS
AND CHANNEL C B A
DECODER
LATCH ENABLE 256R REGISTOR IN0 L L L
LADDER
IN1 L L H
VCC GND REF(+) REF(-) OUTPUT IN2 L H L
ANABLE IN3 L H H
IN4 H L L
Fig. 19.4 The internal architecture of 0809 AD converter
IN5 H L H
IN6 H H L
IN7 H H H

The Converter
This 8-bit converter is partitioned into 3 major sections: the 256R ladder network, the successive
approximation register, and the comparator. The converter’s digital outputs are positive true. The

Version 2 EE IIT, Kharagpur 7 Version 2 EE IIT, Kharagpur 8

256R ladder network approach (Figure 1) was chosen over the conventional R/2R ladder because
of its inherent monotonicity, which guarantees no missing digital codes. Monotonicity is 111 +1/2 LSB INFINITE R
FULL-SCALE 111 TOTAL PERFECT CO
particularly important in closed loop feedback control systems. A non-monotonic relationship 110 IDEAL CURVE UNADJUSTED

A/D OUTPUT CODE

ERROR = 1/2 LSB

A/D OUTPUT CODE

110 IDEAL 3-BIT CODE
can cause oscillations that will be catastrophic for the system. Additionally, the 256R network 101 ERROR
101
does not cause load variations on the reference voltage. 100
100 -1 LSB
011 ABSOLUTE
NONLINEARITY = 1/2 LSB 011
CONTROLS FROM S.A.R. ACCURACY
010
NONLINEARITY = -1/2 LSB 010 -1/2 LSB
REF(+) 001 QUANTIZATION
001
ZERO ERROR = -1/4 LSB ERROR
000 VIN 000
0/8 1/8 2/8 3/8 4/8 5/8 6/8 7/8 VIN
0/8 1/8 2/8 3/8 4/8 5/8 6/8 7/8
1½ R VIN AS FRACTION OF FULL-SCALE VIN AS FRACTION OF FULL-SCALE

R Fig. 19.7 The 3-bit AD Converter Resolution

R Interface to a typical Processor

256 R TO
COMPARATOR Fig.19.8 shows the layout for interface to a processor with 16-address lines(AD0-AD15), read
R
INPUT and write lines and 8-data lines (DB0-DB7). The address lines are divided into two groups. AD0-
AD2 are used to select the analog channel. The ALE signal of the ADC is used to latch the
address on the lines A0-A2 for keeping a particular channel selected till the end of conversion.
R
The other group (AD3-AD15) are decoded and combined with Read and Write signals to
generate the START, ALE and OE (output enable) signals. A write operation starts the ADC.
½R The EOC signal can be used to initiate an interrupt driven data transfer. The interrupt service
subroutine can read the data through DB0-DB7 and initiate the next conversion by subsequent
write operation. Fig.19.9 shows the timing diagram with system clock (not the ADC clock).
REF(-)

Fig. 19.6 The 256R ladder network READ

INTERRUPT
The bottom resistor and the top resistor of the ladder network in Fig.19.6 are not the same value 500 kHz CLK 0E
as the remainder of the network. The difference in these resistors causes the output characteristic ADDRESS 5.000V VREF (+) E0C INTERRUPT
DECODE
to be symmetrical with the zero and full-scale points of the transfer curve. The first output (AD4 – AD15)* 0.000V VREF (-)
transition occur when the analog signal has reached +1⁄2 LSB and succeeding output transitions 2-1 DB7 MSB
occur every 1 LSB later up to full-scale. The successive approximation register (SAR) performs START 2-2 DB6
8-iterations to approximate the input voltage. For any SAR type converter, n-iterations are ALE 2-3 DB5
WRITE
required for an n-bit converter. Fig.19.7 shows a typical example of a 3-bit converter. The A/D 2-4 DB4
AD0 A 2-5 DB3
converter’s successive approximation register (SAR) is reset on the positive edge of the start ADC0808
conversion (SC) pulse. The conversion is begun on the falling edge of the start conversion pulse. AD1 B ADC0809 2-6 DB2
-7
AD2 C 2 DB1
A conversion in process will be interrupted by receipt of a new start conversion pulse.
2-8 DB0 LSB
Continuous conversion may be accomplished by tying the end-of-conversion (EOC) output to the 5V SUPPLY
SC input. If used in this mode, an external start conversion pulse should be applied after power
VCC In7 VIN 8
up. End-of-conversion will go low between 0 and 8 clock pulses after the rising edge of start
GND
conversion. The most important section of the A/D converter is the comparator. It is this section
which is responsible for the ultimate accuracy of the entire converter. GROUND 0-5V ANALOG
INPUT RANGE

In0 VIN 1

Fig. 19.8 Interface to a typical processor

Version 2 EE IIT, Kharagpur 9 Version 2 EE IIT, Kharagpur 10
The timing Diagram (Fig.19.9)
The address latch enable signal and the start conversion are almost made high at the same time as NC (NOTE 2) 1 16
COMPENSATION
per the connections in Fig.19.8. The analog input should be stable across the hold capacitor for
the conversion time(tc). The digital outputs remain tri-stated till the output is enabled externally GND 2 15
VREF(-)
by the Output Enable(OE) signal. The comparator input changes by the SAR counter and switch
tree through the ladder network till the output almost matches the voltage ate the selected analog VEE 3 14
VREF(+)
input channel.
I0 4 13
Important Specifications VCC
DAC0808
8- time-multiplexed analog channels
Resolution 8 Bits MSB A1 5 12
A8 LSB
Supply 5 VDC
Average Power consumption 15 mW A2 6 11
A7
Conversion Time 100 ȝs
A3 7 10
A6
19(IV) The DA Converter DAC0808 A4 8 9
A5

The DAC0808 is an 8-bit monolithic digital-to-analog converter (DAC). Fig.19.9 shows the
architecture and pin diagram of such a chip.
Fig. 19.9 The DAC 0808 Signals
MSB LSB
A1 A2 A3 A4 A5 A6 A7 A8 The pins are labeled A1 through A8, but note that A1 is the Most Significant Bit, and A8 is the
Least Significant Bit (the opposite of the normal convention). The D/A converter has an output
current, instead of an output voltage. An op-amp converts the current to a voltage. The output
RANGE
I0
current from pin 4 ranges between 0 (when the inputs are all 0) to Imax*255/256 when all the
CONTROL CURRENT SWITCHES
inputs are 1. The current, Imax, is determined by the current into pin 14 (which is at 0 volts).
Since we are using 8 bits, the maximum value is Imax*255/256. The output of the D/A converter
takes some time to settle. Therefore there should be a small delay before sending the next data to
R-2R LADDER BIAS CIRCUIT GND the DA. However this delay is very small compared to the conversion time of an AD Converter,
therefore, does not matter in most real time signal processing platforms. Fig.19.10 shows a
typical interface.
VREF (+)
NPN CURRENT VCC
SOURCE PAIR
VREF (-)
REFERENCE COMPEN
CURRENT AMP

VEE

Version 2 EE IIT, Kharagpur 11 Version 2 EE IIT, Kharagpur 12

 VCC = 5V Question Answers
Q.1. What are the possible errors in a system as shown in Fig. 19.2?
13
5.000k Ans:
5 14
MSBA1 10.000V = VREF
6
A2 5k Stage-1 Signal Amplification and Conditioning This can also amplify the noise.
7 15
A3 5.000k
8 2
DIGITAL A4 Stage-2 Anti-aliasing Filter Some useful information such as transients in the real systems
9 DAC0808
INPUTS A5 cannot be captured.
10 4
A6 -
11 Stage-3 Sample and Hold The leakage and electromagnetic interference due to switching
A7
12 16 LF351 V0
LSB A8 OUTPUT
Stage-4 Analog to Digital Converter Quantization error due to finite bit length
3 0.1 ȝF +
Stage-5 Digital Processing and Data manipulation in a Processor: Numerical round up errors due
to finite word length and the delay caused by the algorithm.
VEE = -15V Stage-6 Processed Digital Values are temporarily stored in a latch before D-A conversion: Error
Fig. 19.10 Typical connection of DAC0808 in reconstruction due to zero-order approximation
LF351 is an operational amplifier used as current to proportional voltage converter. The 8-digital
inputs at A8-A1 is converted into proportional current at pin no.4 of the DAC. The reference Q.2 Why it is necessary to separate the digital ground from analog ground in a typical ADC?
voltages(10V) are supplied at pin 14 and 15(grounded through resistance). A capacitor is
connected across the Compensation pin 16 and the negative supply to bypass high frequency Ans: Digital circuit noise can get to analogue signal path if separate grounding systems are not
noise. used for digital and analogue parts. Digital grounds are invariably noisier than analog grounds
Important Specifications because of the switching noise generated in digital chips when they change state. For large
±0.19% Error current transients, PCB trace inductances causes voltage drops between various ground points on
Settling time: 150 ns the board (ground bounce). Ground bounce translates into varying voltage level bounce on signal
Slew rate: 8 mA/ȝs lines. For digital lines this isn't a problem unless one crosses a logic threshold. For analog it's just
Power supply voltage range: ±4.5V to ±18V plain noise to be added to the signals.
Power consumption: 33 mW @ ±5V

19(V) Conclusion
In this lesson you learnt about the following
The internal AD converters of 80196 family of processor
The external microprocessor compatible AD0809 converter
A typical 8-bit DA Converter

Both the ADCs use successive approximation technique. Flash ADCs are complex and therefore
generate difficult VLSI circuits unsuitable for coexistence on the same chip. Sigma-Delta need
very high sampling rate.

Version 2 EE IIT, Kharagpur 13 Version 2 EE IIT, Kharagpur 14