Pic App

PIC Application Notes
Contents
Interfacing a Matrix Keypad ...................................... 5 Receiving RS-232 Serial Data....................................9 Sending RS-232 Serial Data .....................................16 Reading Rotary Encoders .........................................23 Producing Sound & Music .......................................29 Driving an LCD Display ..........................................35 Direct & Indirect Addressing ...................................42 The PIC16C71 A/D Convertor .................................47 Managing Multiple Tasks .........................................53 An External A/D Convertor .....................................58 Using Serial EEPROMs ...........................................61 Using Dynamic RAM ...............................................68 Running at 32 kHz....................................................76 Generating/mixing Sine Waves ................................83 Using Interrupts ........................................................88
APPS
TechTools PIC Application Notes Page 1
Page 2 PIC Application Notes TechTools

This document contains helpful application notes. These notes cover such topics as serial communications, driving LCDs, reading rotary encoders, and using external DRAMs. APPS For many common projects, the application notes may contain information that can save time when starting a new project; of course, they may also prove useful if youre new to PICs or to microcontrollers in general.
1: Interfacing a Matrix Keypad
Interfacing a Matrix Keypad
Introduction. This application note covers the use of matrix-encoded keypads with PIC microcontrollers. It presents an example program in TechTools assembly language for reading a 4 x 4 keypad. Background. In order to use as few input/output (I/O) pins as possible, most keypads of eight or more switches are wired in a matrix arrangement. Instead of interfacing each set of contacts to an I/O pin, switches are wired to common row and column connections, as shown in the figure. For a number of switches, this method can save quite a few pins:
No. of Switches 8 12 16 20 Matrix (rows x columns) 2x4 3x4 4x4 5x4 I/O Pins Required 6 7 8 9
APPS
The disadvantage of matrix encoding is that it requires some additional programming. The listing shows a program that reads a 4 x 4 keypad using an 8-bit I/O port and presents the number of the key pressed on a 4-bit port. How it works. Code at the beginning of the program sets the lower four bits of port RB (connected to the columns of the keypad) to output, and the upper four bits (rows) to input. It also sets port RA, which will drive LEDs indicating the binary number of the key pressed, to output. The routine scankeys does the actual work of reading the keypad. It performs the following basic steps: Assert a 1 on the current column. Does the 1 appear on the current row? > No: increment key and try the next row. > Yes: exit the subroutine. Try the next column. If the first key, key 0, is pressed, the routine exits with a 0 in the variable key, because it ends before the first increment instruction. If no key is
pressed, the variable key is incremented 16 (010h) times. Therefore, this number serves as a flag to the program that no key has been pressed. On the issue of switch debouncing: After a switch is pressed, it may take 20 milliseconds or more for it to settle into its new state. That means that one key press can register as many repeated presses. The easiest way to defeat this problem is to read the switches, and then wait a while before reading them again. Thats the purpose of the :delay loop in the main program.
470 470 470
1 18 LED LED LED LED
470
RA2
2
RA1
17
You may want to add 1k series resistors in each column (RB0 - RB3). This prevents direct shorting if two or more buttons are pressed at the same time.
Vcc
3 4 5 6 7 8 9
RA3 RTCC MCLR Vss RB0 RB1 RB2 RB3
RA0
16
220 pF
15
OSC1
PIC OSC2 16C54 Vdd RC/P

RB7 RB6
10k
14 13 12 11
Vcc
10k 10k 10k 10k
RB5
10
RB4
Modifications. In circumstances where electrical noise might be a problem, Microchips data sheet on the PIC indicates that it might be wise to move the port I/O assignments to the beginning of scankeys. The reason is that electrostatic discharge (ESD) from the users fingertips, or presumably any other strong electrical noise, could corrupt an I/O control register and switch an input pin to output. This would prevent the routine from reading one or more rows of the keypad.
A milder zap could conceivably cause a false input to the keypad. Some routines check for this condition by comparing two or more consecutive readings of the keys. Unless several readings match, no data is returned to the main program. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
; PROGRAM: KEYPAD ; This program demonstrates a simple method for scanning a matrix-encoded ; keypad. The 4 columns of the pad are connected to RB.0 through RB.3; 4 rows to ; RB.4 through RB.7. The scanning subroutine returns the code of key pressed ; (0h0Fh) in the variable key. If no switch is pressed, the out-of-range value 10h is ; returned in key (this avoids the use of a separate flag variable). keypad row1 row2 row3 row4 = = = = = rb rb.4 rb.5 rb.6 rb.7
APPS
; Variable storage above special-purpose registers. org 8 cols ds 1 key ds 1 index ds 1 ; Remember to change device info if programming a different PIC. device pic16c54,rc_osc,wdt_off,protect_off reset start ; Set starting point in program ROM to zero. org 0 start mov !rb, #11110000b mov !ra, #0 :keys call scankeys cje key, #16, :delay mov ra, key :delay nop nop djnz index,:delay goto :keys
; cols out, rows in ; ra all outputs
; Subroutine to scan the keypad. Assumes that the calling routine will delay long ; enough for debounce. scankeys clr key clr keypad mov cols,#4 ; 4 x 4 keypad setb c ; put a 1 into carry ; On the first time through the following loop, the carry bit (1) is pulled into keypad. ; On subsequent loops, the lone 1 moves across the column outputs. :scan rl keypad clrb c ; follow the 1 with zeros jb row1, press ; If a 1 is detected, quit the routine with the current value of key. If row1, column1 is ; pressed, the value of key is 0 (on the first loop). If not, increment key and try the ; next row. inc key jb row2, press inc key jb row3, press inc key jb row4, press inc key djnz cols, :scan ; Try all 4 columns. press ret ; Return with value in key. ; If no key is pressed, the value will be 10h due to execution of the final increment ; instruction. The program should interpret this out-of-range value as no press.
2: Receiving RS-232 Serial Data
Receiving RS-232 Serial Data
Introduction. This application note presents a simple program for receiving asynchronous serial data with PIC microcontrollers. The example program, written using TechTools assembly language, displays received bytes on a bank of eight LEDs. Background. Many controller applications involve receiving data or commands from a larger system. The RS-232 serial port is a nearly universal means for this communication. While the PIC lacks the serial receive function found on some more expensive chips, it can readily be programmed to receive serial data. A byte of serial data is commonly sent as a string of 10 bits; a start bit, eight data bits, and a stop bit, as shown in figure 1 below. The start and stop bits help the receiver to synchronize to the incoming data bits. In some cases, a serial transmitter will lengthen the stop bit to 1.5 or 2 times the duration of the data bits in order to ensure proper sync under noisy conditions. The speed of a serial transmission is expressed in baud or bits per second (bps). Since a complete transmission is 10 bits long, the number of bytes per second is one-tenth the baud rate. A 1200-baud signal conveys 120 bytes per second. The bit duration is 1 second divided by the baud rate. For instance, each bit of a 1200-baud signal is 833 microseconds long. RS-232 is an electrical standard for signals used in serial communication. It represents a binary 1 with a level of -5 to -15 volts, and a 0 with +5 to +15 volts. In order for a 5-volt device like the PIC to interface with this signal, additional circuitry must convert the RS-232 signal to logic levels. The 1489 quad line receiver used in the circuit (see schematic, figure 3) can convert four RS-232 signals to 5-volt logic levels; the example circuit uses only one section of the device.
APPS
START BIT
BIT 0
BIT 1
BIT 2
BIT 3
BIT 4
BIT 5
BIT 6
BIT 7
STOP BIT
Figure 1. A byte of serial data. The byte depicted is 01011010, the ASCII code for Z.
Where cost or space is a problem, the PIC can accept the RS-232 signal through a 22k resistor, as shown in the inset to figure 3. The resistor limits the input current, while the PICs internal clamping diodes (intended to protect against static electricity) clip the voltage to logic levels. The resistor method does not invert the RS-232 signal, so three minor changes to the program are required as shown in comments to listing 1. The method also gives up the noise rejection built into the 1489, and should not be used in noisy environments or over long cable runs. The example presented here does not use any of the RS-232 handshaking lines. These lines help when a fast computer must communicate with (for instance) a slow printer. When the receiving device does not use the handshaking lines, it is necessary to loop them back as shown in figure 2. That way, when the computer asks for permission to send, the signal appears at its own clear-to-send pin. In effect, it answers its own question. Although it is the most common, RS-232 is not the only serial-signaling standard. RS-422 is becoming increasingly popular because of its resistance to noise, high speed, long allowable wire runs, and ability to operate from a single-ended 5-volt power supply. Since the only difference between RS-232 and RS-422 is the electrical interface, conversion requires only the substitution of an RS-422 line receiver chip.
DB-25 PIN 2 4 5 6 8 20 7
DB-9 PIN 3 7 8 6 1 4 5 DATA RTS CTS DSR DCD DTR GND to PIC circuit GND to SERIAL IN
Figure 2. Hookups for standard 9- and 28-pin connectors. Connecting RTS to CTS disables normal handshaking, which is not used here.
How it works. The example program in listing 1 is a no-frills algorithm for receiving serial data in the popular N81 format; i.e., no parity bit, eight data bits, and one stop bit. Listing 2 is a BASIC program for sending individual bytes to the circuit. Listing 1 begins by setting up the input and output bits. It then enters a loop waiting to detect the start bit. Once the start bit is detected, the program waits one-half bit time and checks to see whether the start bit is still present. This helps ensure that the program isnt fooled by a noise burst into trying to receive a nonexistent transmission. It also makes sure that subsequent bits are read during the middle of their time slots; another precaution against noise. Once it detects and verifies the start bit, the program enters another loop that does the actual job of receiving the data. It works like this: Wait one bit time. Copy input bit to carry bit. Rotate the receive byte right. Decrement the bit counter. Is the counter zero? > No, loop again. > Yes, exit the loop. If you are unfamiliar with the rotate right (rr) instruction, you may not see how the input bit gets from the carry bit into the receive byte. Performing an rr on a byte moves its bits one space to the right. Bit 7 goes to bit 6, bit 6 to bit 5, and so on. Bit 0 is moved into the carry bit. The carry bit moves into bit 7. Once the byte is received, the program waits a final bit delay (until the middle of the stop bit), copies the received byte to the output port to which the LEDs are connected, and goes back to the beginning to await another start bit. The program can be set up for most standard data rates. The table lists PIC clock speeds and values of the bit time constant bit_K (declared at the beginning of listing 1) for a wide range of common rates. For other combinations of clock speed and data rate, just replace the delay
APPS
routines with ones that provide the appropriate timing. The footnote to the table gives general guidance. One final hardware note: Although timing isnt overly critical for receiving this type of serial data, resistor/capacitor timing circuits are inadequate. The PICs RC clock is specified to fairly loose tolerances (up to 28 percent) from one unit to another. The values of common resistors and capacitors can vary substantially from their marked values, and can change with temperature and humidity. Always use a ceramic resonator or crystal in applications involving serial communication.
RS-232 serial in
1/4 MC1489A: pins 1, 4, 7, 10 to gnd; pin 14 to +5 Vdc; rest not connected.

13 12 +5 47pF 11 1 18
RA2
2
RA1
17
RA3
3
RA0
16
ceramic resonator w/integral capacitors
RTCC
4
OSC1
15
MCLR
5
RS-232 serial in
22k
Vss
1 6
PIC 16C54
OSC2
14
+5
Vdd
13 470 470 470 470
RA2
7 8
RB0 RB1 RB2

9
RB7
12
LED LED LED LED
RB6
11
RB5
10
A 22k resistor may be used instead of the 1489. Change the program as marked in listing 1.
RB3
RB4
470 470 470 470
LED LED LED LED
Figure 3. Schematic to accompany RCV232.SRC.
Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
Values of Timing Constant Bit_K for Various Clock Speeds and Bit Rates
Serial Bit Rate (bit time) Clock Frequency 1 MHz 2 MHz 4 MHz 8 MHz 300
(3.33 ms)
600
(1.66 ms)
1200
(833 s)
2400
(417 s)
4800
(208 s)
9600
(104 s)
19,200
(52 s)
APPS
206
102 206
50 102 206
24 50 102 206
1 . bitrate
24 50 102
24 50
24
Other combinations of clock speed and bit rate can be supported by changing the bit_delay and start_delay subroutines. The required bit delay is is
1 1200
For example, at 1200 baud the bit delay
= 833s. The start delay is half of the bit delay; 416s for the 1200-baud example.
Calculate the time delay of a subroutine by adding up its instruction cycles and multiplying by
4 clockspeed
. At 2 MHz, the time per instruction cycle is
4 2 ,000 ,000
= 2s.
; PROGRAM: RCV232 ; This program receives a byte of serial data and displays it on eight LEDs ; connected to port RB. The receiving baud rate is determined by the value of the ; constant bit_K and the clock speed of the PIC. See the table in the application note ; (above) for values of bit_K. For example, with the clock running at 4 MHz and a ; desired receiving rate of 4800 baud, make bit_K 50. bit_K half_bit serial_in data_out = = = = 24 bit_K/2 ra.2 rb ; Change this value for desired ; baud rate as shown in table.
; Variable storage above special-purpose registers. org 8 delay_cntr ds 1 ; counter for serial delay routines bit_cntr ds 1 ; number of received bits rcv_byte ds 1 ; the received byte ; Org 0 sets ROM origin to beginning for program. org 0 ; Remember to change device info if programming a different PIC. Do not use RC ; devices. They are not sufficiently accurate or stable for serial communication. device reset pic16c54,xt_osc,wdt_off,protect_off begin

; Set up I/O ports. begin mov mov snb !ra, #00000100b !rb, #0 serial_in ; Use RA.2 for serial input. ; Output to LEDs.
:start_bit
; Detect start bit. Change to sb ; serial_in if using 22k resistor ; input. ; No start bit yet? Keep watching. ; Wait one-half bit time to the ; middle of the start bit. ; If the start bit is still good, ; continue. Otherwise, resume ; waiting. ; Change to jnb Serial_in,
jmp call
:start_bit start_delay
jb
Serial_in, :start_bit
:start_bit ; if using 22k resistor input. mov bit_cntr, #8 ; Set the counter to receive 8 data ; bits. ; Clear the receive byte to get ; ready for new data. ; Wait one bit time. ; Put the data bit into carry. ; Change to movb c,/Serial_in if ; using 22k resistor input. ; Rotate the carry bit into the ; receive byte. ; Not eight bits yet? Get next bit. ; Wait for stop bit. ; Display data on LEDs. ; Receive next byte.
clr
rcv_byte
:receive
call movb
bit_delay c,Serial_in
rr
rcv_byte
djnz call mov goto ; ; ; ;
bit_cntr,:receive bit_delay data_out, rcv_byte begin:start_bit
This delay loop takes four instruction cycles per loop, plus eight instruction cycles for other operations (call, mov, the final djnz, and ret). These extra cycles become significant at higher baud rates. The values for bit_K in the table take the time required for additional instructions into account. mov nop djnz delay_cntr,#bit_K delay_cntr, :loop
bit_delay :loop

ret
; This delay loop is identical to bit_delay above, but provides half the delay time. start_delay :loop mov nop djnz ret delay_cntr,#half_bit delay_cntr, :loop
APPS
BASIC Program for Transmitting Bytes via COM1

10 REM Open the serial port com1. Substitute the desired baud rate 20 REM for 9600 in this line. The parameters CD0, CS0, DS0, and OP0 30 REM serve to disable hardware handshaking. They may be omitted if 40 REM these lines are looped back as shown in figure 2. 50 OPEN com1:9600,N,8,1,CD0,CS0,DS0,OP0 FOR OUTPUT AS #1 60 CLS 70 REM At the prompt, enter a value between 0 and 255 representing a 80 REM byte of data to send out the serial port. 90 INPUT ASCII code to send: , A% 100 PRINT #1, CHR$(A%); 110 GOTO 60 120 END
3: Sending RS-232 Serial Data
Introduction. This application note covers transmission of asynchronous serial data using PIC microcontrollers. It presents an example program in TechTools assembly language that transmits a text string serially via RS-232 at speeds up to 19,200 bits per second. Hardware examples demonstrate the use of a popular serial line driver, as well as a method for using the PICs output to directly drive a serial line. Background. Many PIC applications require sending data to a larger computer system. The common RS-232 serial port is almost ideal for this communication. While the PIC lacks the onboard serial communication hardware available with some more expensive controllers, it can readily be programmed to add this capability. A previous application note (#2, Receiving RS-232 Serial Data) covered the data format and RS-232 signals in some detail. Here are the highlights: A byte of serial data is commonly sent as 10 bits: a start bit, eight data
+5 10 F
Sending RS-232 Serial Data
16
10 F
C1+
2 10 F 3
Vcc
+10 C14
MAX 232 GND

X1 out R1 in
15 14 13
10 F 10 F RS-232 serial out
C2+
5
12
C26
R1 out
11
-10
7
X1 in
10 1 18
X2 out
8
X2 in
9
RA2
+5 2
RA1
17
R2 in
R2 out
1k
RA3
3
RA0
16
RTCC
4
OSC1
15
MCLR
5 RS-232 serial out 1 Reset
OSC2
14 +5 13 12
RA2
6 7
PIC Vss 16C54 Vdd

RB0 RB1 RB7 RB6
11
The PIC's output may be connected directly to the serial input of the terminal or PC. Change the program as marked in listing 1.
RB2
9
RB5
10
RB3
RB4
bits, and a stop bit. The duration of these bits is calculated by dividing the rate in bits per second (commonly baud) into 1 second. The stop bit can be stretched to any desired duration. Under the RS-232 standards, a high (1) is represented by a negative voltage in the range of -5 to -15 volts. A low (0) is a voltage from +5 to +15 volts. In addition to connections for data input, output, and ground, most RS-232 ports include handshaking lines that help devices turn the flow of data on and off when necessary (for instance, when a printer is receiving data faster than it can store and process it). Many applications avoid the use of these hardware signals, instead embedding flow-control commands in the data stream itself. A traditional method of sending serial data is to use a parallel-in serialout shift register and a precise clock. The start, data, and stop bits are loaded into the register in parallel, and then shifted out serially with each tick of the bit-rate clock. You can easily use a software version of this method in a PIC program. Really, the most difficult part of transmitting an RS-232-compatible signal from the PIC is achieving the signaling voltages. However, the hardware example presented here shows that: There are devices that generate the RS-232 voltages from a single-ended 5-volt supply. It is possible to use the PICs output to directly drive an RS-232 input, if the cable is kept short. How it works. The PIC program in listing 1 sends a short text string as a serial data stream whenever the PIC is reset. To reset the PIC, push and release the reset button shown in the schematic . If you lack appropriate terminal software for your computer, listing 2 is a BASIC program that will help get you started. If you use other terminal software and get device timeout errors, try cross-connecting the handshaking lines as shown in figure 2 below. This has the effect of disabling handshaking. The program in listing 2 does this in software. The PIC program starts by setting port RA to output. Serial data will be transmitted through pin RA.2. The program consists of two major loops, defined by the labels :again and :xmit. The first loop, :again, initializes the
APPS
bit counter, and then gets a byte from the subroutine/table called string. Once string returns a byte in the w register, the program puts this byte into xmt_byte. After a start bit is sent by clearing the serial-out pin, :xmit takes over. It performs the following steps: Rotate the transmit byte right (into carry). Move the carry bit to serial output. Wait one bit time (set by bit_K). Decrement the bit counter. Is the counter zero? > No, loop again. > Yes, exit the loop. With the :xmit loop finished, the program sets the serial-out pin to send a stop bit, and checks to see whether it has reached the end of the string. If not, it goes back to :again to retrieve another byte. If it is done with the string, the program enters an endless loop.
DB-25 PIN 3 4 5 6 8 20 7 DB-9 PIN 2 7 8 6 1 4 5 DATA RTS CTS DSR DCD DTR GND to PIC circuit GND to SERIAL OUT
Figure 2. Hookups for standard 9- and 28-pin connectors. Connecting RTS to CTS disables normal handshaking, which is not used here.
The hardware portion of the application is fairly straightforward. The PIC produces logic-level signals that the MAX232 chip converts to acceptable RS-232 levels, approximately 10 volts. The MAX232 has onboard charge-pumps that use the external capacitors to build up the required
signaling voltages. There are other chips that provide additional features, such as additional I/O channels, higher signaling rates, lower power consumption, and the ability to work without external capacitors. Modifications. The MAX232 draws more current and costs more (in single-part quantities) than the PIC it supports. Most of the time, this is still a bargain compared to adding a bipolarity power supply to a device just to support RS-232 signaling. In fact, the MAX232 has excess current capacity that can be used to power other small bipolarity devices, such as low-power op-amps. Depending on the application, additional voltage regulation may be required. If you plan to use the serial port infrequently, consider powering the MAX chip through one of the PICs I/O pins. Your program could turn the MAX on shortly before it needed to send a serial message, and turn it back off afterward. This option is especially attractive for uses that require wringing the most life out of a set of batteries. A sample MAX232 that we tested drew 15 mA while driving two simulated serial-port loads. This is well within the PICs drive capability. A further test showed that the MAX232s output voltages rose to their full 10-volt levels about 1.5 milliseconds after power was applied to the chip. If your program switches the MAX232 on and off, program a 1.5ms delay between turning the device on and sending the first bit. In some cases, you may be able to dispense with the serial line driver altogether. Make the changes marked in listing 1 for direct connection and wire RA.2 directly to the serial receive connection of your terminal or PC. The changes in the program invert the logic of the serial bits, so that a 1 is represented by an output of 0 volts and a 0 is a 5-volt output. According to RS-232, the receiving device should expect voltages of at least -3 volts for a 1 and +3 volts for a 0. Voltages lying between +3 and -3 are undefined, meaning they could go either way and still comply with the standard. As a practical matter, though, it wouldnt be smart to let 0 volts be interpreted as a 0, since this is a serial start bit. Any time the serial input was at ground potential, the terminal or PC would attempt to receive incoming serial data. This would cause plenty of false receptions. Serial-
APPS
port designers apparently take this into account and set the 0 threshold somewhere above ground. Thats why this cheap trick works. Dont count on it to provide full RS-232 performance, especially when it comes to sending data rapidly over long cables. Keep cables short and you shouldnt have any problems (19,200 baud seems to work error-free through 6 feet of twisted pair). If your application will have access to the handshaking pins, you may be able to steal enough power from them to eliminate batteries entirely. According to the RS-232 specifications, all signals must be capable of operating into a 3000-ohm load. Since many computers use 12 volts for their RS-232 signals, each line should be capable of delivering 4 mA. In practice, most can provide more, up to perhaps 15 mA. The only problem with exploiting this free power is that the software running on the PC or terminal must be written or modified to keep the handshaking lines in the required state. One final hardware note: Although timing isnt overly critical for transmitting serial data, resistor/capacitor timing circuits are inadequate. The PICs RC clock is specified to fairly loose tolerances (up to 28 percent) from one unit to another. The values of common resistors and capacitors can vary substantially from their marked values, and can change with temperature and humidity. Always use a ceramic resonator or crystal in applications involving serial communication. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
Values of Timing Constant Bit_K for Various Clock Speeds and Bit Rates
Serial Bit Rate (bit time) Clock Frequency 1 MHz 2 MHz 4 MHz 8 MHz 300
(3.33 ms)
600
(1.66 ms)
1200
(833 s)
2400
(417 s)
4800
(208 s)
9600
(104 s)
19,200
(52 s)
APPS
206
102 206
50 102 206
24 50 102 206
1 . bitrate
24 50 102
24 50
24
Other combinations of clock speed and bit rate can be supported by changing the bit_delay and start_delay subroutines. The required bit delay is is
1 1200
For example, at 1200 baud the bit delay
= 833s. The start delay is half of the bit delay; 416s for the 1200-baud example.
Calculate the time delay of a subroutine by adding up its instruction cycles and multiplying by
4 clockspeed
. At 2 MHz, the time per instruction cycle is
4 2 ,000 ,000
= 2s.
; ; ; ; ;
PROGRAM: XMIT232 This program transmits a string of serial data. The baud rate is determined by the value of the constant bit_K and the clock speed of the PIC. See the table in the application note (above) for values of bit_K. For example, with the clock running at 4 MHz and a desired transmitting rate of 4800 baud, make bit_K 50. = = 24 ra.2 ; 24 for 19,200-baud operation @ 8 MHz
bit_K serial_out
; Variable storage above special-purpose registers. org 8 delay_cntr bit_cntr msg_cntr xmt_byte ds ds ds ds 1 1 1 1 ; ; ; ; counter for serial delay routines number of transmitted bits offset in string the transmitted byte
; Org 0 sets ROM origin to beginning for program. org 0 ; Remember to change device info if programming a different PIC. Do not use RC ; devices. They are not sufficiently accurate or stable for serial communication. device pic16c54,xt_osc,wdt_off,protect_off reset begin begin mov mov mov !ra, #00000000b msg_cntr, #0 bit_cntr,#8 ; Set port to output. ; Message string has nine ; characters; 0 through 8. ; Eight bits in a byte.
:again

mov call mov clrb call rr movb call djnz setb call inc cjbe w,msg_cntr string xmt_byte,w serial_out bit_delay xmt_byte serial_out,c bit_delay bit_cntr,:xmit serial_out bit_delay msg_cntr msg_cntr,#8, :again

; Point to position in the string. ; Get next character from string. ; Put character into transmit byte. ; Change to setb serial_out for ; direct connection. ; Start bit. ; Rotate right moves data bits into ; carry, starting with bit 0. ; Change to movb serial_out,/c for ; direct connection. ; Data bit. ; Not eight bits yet? Send next bit. ; Change to clrb serial_out for ; direct connection ; Stop bit. ; Add one to the string pointer. ; More characters to send? Go to ; the top. ; Endless loop. Reset controller to ; run program.
:xmit
:endless
goto
:endless
; To change the baud rate, substitute a new value for bit_K at the beginning of this ; program. bit_delay :loop mov nop djnz ret jmp delay_cntr,#bit_K delay_cntr, :loop
string linefeed.
pc+w
; Message string consisting of ; TechTool followed by a
retw
T,e,c,h,T,o,o,l,10
BASIC Program for Receiving Text String via COM1 10 CLS 15 REM Substitute desired baud rate for 19200 in the line below. 20 OPEN com1:19200,N,8,1,CD0,CS0,DS0,OP0 FOR INPUT AS #1 30 IF NOT EOF(1) THEN GOSUB 200 40 GOTO 30 200 Serial$ = INPUT$(LOC(1), #1) 210 PRINT Serial$; 220 RETURN
4: Reading Rotary Encoders
Reading Rotary Encoders
Introduction. This application note covers the use of incremental rotary encoders with PIC microcontrollers. It presents an example program in TechTools assembly language for reading a typical encoder and displaying the results as an up/down count on a seven-segment LED display. Background. Incremental rotary encoders provide a pair of digital signals that allow a microcontroller to determine the speed and direction of a shafts rotation. They can be used to monitor motors and mechanisms, or to provide a control-knob user interface. The best-known application for rotary encoders is the mouse, which contains two encoders that track the x- and y-axis movements of a ball in the devices underside. Rotary encoders generally contain a simple electro-optical mechanism
APPS
CLOCKWISE
Phase 1
Phase 2
COUNTER CW
Figure 1. Quadrature waveforms from a rotary encoder contain directional information.
CLOCKWISE
Phase 1 Phase 2 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1
COUNTER CW
Figure 2. Sequence of two-bit numbers output by the phases of a rotary encoder.
consisting of a slotted wheel, two LEDs, and two light sensors. Each LED/sensor pair is arranged so that the devices face each other through the slots in the wheel. As the wheel turns, it alternately blocks and passes light, resulting in square wave outputs from the sensors. The LED/sensor pairs are mounted offset relative to one another, so that the output square waves are offset 90 degrees in phase. This is known as quadrature, because 90 degrees amount to one-quarter of a full 360degree cycle.
Vcc Rotary Encoder
(Digi-Key GH6102)
1k 1k
V+ p1 p2 gnd
18
6 5 4 1
Vcc
2 3
RA2 RA3 RTCC

4 5 6
RA1
17
RA0
16
220 pF
15 NC 14 13 NC 12 470
OSC1 10k Vcc
PIC OSC2 16C54 Vdd Vss

MCLR RB0 RB7 RB6
RB1
8
a
11
470 470
RB2
9
RB5
10
b c
f a b g e d c
RB3
RB4 470 470 470 470
d e f g
Common-Cathode LED Display
This phase relationship provides the information needed to determine the encoders direction of rotation (see figure 1).
The dotted lines in figure 1 indicate a common method of reading direction. For instance, if phase 1 is high and phase 2 is rising, the direction is clockwise (CW). If phase 1 is low and phase 2 is rising, the direction is counterclockwise (CCW). For the sake of interpreting this output with a PIC or other microcontroller, its probably more useful to look at the changing states of the phases as a series of two-bit numbers, as shown in figure 2 above. When the encoder shaft is turning CW, you get a different sequence of numbers (01,00,10,11) than when it is turning CCW (01,11,10,00). You may recognize this sequence as Gray code. It is distinguished by the fact that only one bit changes in any transition. Gray code produces no incorrect intermediate values when the count rolls over. In normal binary counting, 11 rolls over to 00. If one bit changed slightly before the other, the intermediate number value could be incorrectly read as 01 or 10 before settling into the correct state of 00. Interpreting this code amounts to comparing the incoming sequence to the known sequences for CW and CCW rotation. A lookup table would do the trick. However, this approach, while easy to understand, is inefficient. The shortcut method uses an interesting property of the twobit Gray code sequence. Pick any pair of two-bit numbers from the CW sequence shown in figure 2; for instance, the first two: 10, 11. Compute the exclusive-OR (XOR) of the righthand bit of the first number with the lefthand bit of the second. In this case, that would be 0 XOR 1 = 1. Try this for any CW pair of numbers from the table, and youll always get 1. Now reverse the order of the number pair: 11, 10. XOR the right bit of the first with the left of the second (1 XOR 1 = 0). Any CCW pair of numbers will produce a 0. How it works. The schematic in figure 3 shows a typical rotary encoder connected to the lower two bits of port RA, and a seven-segment LED display to port RB. The circuit performs a simple task: the count displayed on the LED goes up when the control is turned CW, and down when it is turned CCW. The display is in hexadecimal, using seven-segment approximations of the letters: A, b, C, d, E, and F.
APPS
The program begins by setting up the I/O ports and clearing the variable counter. It gets an initial input from the encoder, which goes into the variable old, and strips off all but the two least-significant bits (LSBs). The body of the program, starting with :loop, calls check_encoder and then displays the latest value of counter on the LED display. Most of the interesting business happens in check_encoder itself. Here, the program gets the latest value at the encoder inputs, strips all but the two LSBs, and XORs the result into a copy of the old value. If the result is zero, the encoder hasnt moved since its last reading, and the routine returns without changing the value of counter. If the value has changed, the routine moves the value in old one bit to the left in order to align its LSB with the high bit of the two-bit value in new. It XORs the variables together. It then examines the bit old.1. If the bit is 1, counter is incremented; if its 0, counter is decremented. Modifications. To avoid slippage errors (where a change in encoder position does not change the counter, or results in the wrong change), check_encoder must be called at least once every 1/(encoder resolution max revs per second). For instance, if the encoder might turn 300 rpm * (5 revs per second) and its resolution is 32 transitions per turn, check_encoder must be called every 1/(32*5) seconds, or 6.25 milliseconds. For a user interface, bear in mind that generally the larger the knob, the slower the input. Substitution of a larger control knob may be all thats required to reduce the sampling rate. In circumstances where electrical noise might be a problem, Microchips PIC data sheet indicates that it might be wise to move the port I/O assignments to the beginning of check_encoder. Electrostatic discharge (ESD) from the users fingertips, or some other electrical noise, could corrupt an I/O control register. This would prevent the routine from reading the encoder. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.

; ; ; ; ;
PROGRAM: Read Rotary Encoder (RD_ENCDR.SRC) This program accepts input from a rotary encoder through bits RA.0 and RA.1, determines the direction of rotation, and increments or decrements a counter appropriately. It displays the hexadecimal contents of the four-bit counter on a seven-segment LED display connected to port RB. = = ra rb
APPS
encoder display
; Variable storage above special-purpose registers. org 8 temp counter old new ds ds ds ds 1 1 1 1
; Remember to change device info when programming a different PIC. device pic16c54,rc_osc,wdt_off,protect_off reset start ; Set starting point in program ROM to zero. org 0 start mov mov clr mov and call mov call mov goto !rb, #0 !ra, #255 counter old, encoder old, #00000011b chk_encoder w, counter sevenseg display, w :loop new, encoder new, #00000011b temp, new temp, old :return ; Set rb to output. ; Set ra to input.
:loop
chk_encoder mov and mov xor jz clc rl xor jb :down dec
; Get latest state of input bits. ; Strip off all but the encoder bits. ; ; ; ; ; ; ; Is new = old? If so, return without changing counter. Clear carry in preparation for rotate-left instruction. Move old to the left to align old.0 with new.1.
old old, new old.1, :up counter
; If the XOR resut is 1, increment ; counter, otherwise decrement.

:up skip inc and mov ret jmp retw retw counter counter, #00001111b old,new
:return sevenseg
pc+w ; display lookup table 126, 48, 109, 121, 51, 91, 95, 112 127, 115, 119, 31, 78, 61, 79, 71
5: Producing Sound & Music
Producing Sound & Music
Introduction. This application note presents a subroutine for producing tones (or musical notes) by specifying values for frequency and duration. The example program plays a brief tune through a speaker. The circuit uses two PIC outputs driven differentially to produce a 10-volt peak-topeak signal from a single-ended power supply.
0.047F Speaker
APPS
18
RA2
+5 2
RA1
17
+5
RA3
3
RA0
16 10k 4.7 pF
1k 4
RTCC MCLR
5
OSC1
15
Reset 6
Vss RB0
7
PIC 16C54
OSC2
14
Vdd
13
RB7
12
RB1
8
RB6
11
RB2
9
RB5
10
RB3
RB4
Background. Most high-level programming languages include a command for producing tones or musical notes that accepts two inputs: frequency and duration. These are the basis for programs that generate music, audible prompts, sound effects, and even Morse code. In PIC assembly language, its easy to produce a tone by toggling an output bit. Its also easy to control the duration of the tone by counting cycles. The problem with counting cycles is that the duration ends up varying with the frequency. A 1000-hertz (Hz) tone is made up of 1-millisecond (ms) cycles. If your program counts out 200 cycles, the resulting tone lasts 200 ms. But if you change the frequency to 5000 Hz, each cycle takes only 0.2 milliseconds, so 200 cycles last only 40 ms. The program presented here takes a different approach. The subroutine beep consists of a single loop that always takes 20 microseconds (s) to execute. A 16-bit value controls the number of loops and therefore the duration of the sound.
An eight-bit counter controls the frequency of the sound. If the frequency value is 100, then the routine inverts the speaker lines every 100th execution of the main loop. The larger the frequency value, the lower the output frequency. How it works. The circuit in figure 1 plays a short tune each time the PIC is powered up or reset . The speaker connects through a capacitor to two of the PICs input/output pins. Most PIC circuits switch a load between a single pin and ground or +5 volts. As a result, the largest voltage swing they can generate is 5 volts. In this case, the program ensures that RA.0 and RA.1 are always complemented; when RA.0 is 1, RA.1 is 0 and vice versa. The speaker therefore sees a 10-volt swing. This makes a noticeably louder sound, especially with piezo devices, which are more efficient at higher voltages. The program that controls the PIC begins by setting port RA to output and loading a pattern of alternating 0s and 1s into it. The program then looks up two values from ROM tables* notes and lengths. It stores these values in the variables freq and duratn, and calls the subroutine beep.
Beep accepts the values and produces a tone of the specified frequency and duration. The key to this routine is the ability of the exclusive-OR (XOR) logic function to selectively invert bits. XORs truth table is:
A 0 0 1 1 XOR B 0 1 0 1 Result 0 1 1 0
*You can stuff a series of bytes into the program ROM for later use by your program. Just set up a subroutine beginning with jmp pc+w followed by the directive retw byte0, byte1, byte2... When you need to access one of these numbers from your program, just move the number corresponding to the byte value you want into the w register, and call the table. Upon return, the byte value will be in the w register.
When B contains a 0, Result = A. When B contains a 1, Result is the inverse of A. Each time beep loops, it XORs the speaker lines with the variable tgl. Most of the time, this variable contains 0s, so the XOR has no effect. The counter variable f_temp is decremented each time the loop executes. When f_temp reaches 0, tgl gets loaded with 1s. Now XORing the
speaker lines with tgl inverts them. The routine reloads f_temp with the value freq, and the process starts again. APPS After beep returns, the main program increments the variable index and checks to see whether all the notes have played. If they havent, the program loops. If they have, it enters an endless do-nothing loop waiting to be reset or shut off. A couple of notes about using beep in other applications: To calculate the value freq for a given frequency, divide the frequency in hertz into 50,000 (4-MHz clock) or 100,000 (8-MHz clock). If freq is 0, beep will produce a silent delay of the length set by duratn. To calculate duratn, divide the desired length of the tone by 5 ms. Modifications. The frequency and duration values that make up the tune
CHARGE were programmed by ear, but you can readily program real music
by using the table below. It lists the three octaves that beep adequately covers. Note that the routine is not always right on the nominal frequency of the notes, but its generally within a few hertz. The worst-case error is about 1 percent. The frequencies listed assume a 4-MHz clock. If you use an 8-MHz clock, the values for freq listed in the table will still accurately represent musical notes, but theyll be shifted upward by one octave. If you use a resistor-capacitor oscillator, your notes could wind up being sharp or flat (or worse) because of variations in clock frequency. For applications in which its important that the notes sound just right, use a ceramic resonator or crystal. To program a tune, look up each note in the table and put the corresponding freq into notes. To calculate values for lengths, pick an appropriate duration for a whole note, and scale half-note, quarter-notes, eighths and so on correspondingly. The duration in milliseconds of the note will be approximately 5 x duratn. So, if you pick a value of 128 for a whole note, it will last 640 ms. When youre done filling in notes and lengths, count the notes and set the constant end_note equal to this number. Also make sure that the number of notes is equal to the number of lengths.
Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
Values of Variable freq Required to Create Musical Notes (4-MHz clock)

Firs t Oc t av e Nominal Actual Value of (Hz) (Hz) freq 220 247 262 294 330 349 392 220 248 262 294 329 350 391 227 202 191 170 152 143 128 Second Octave Nominal Actual Value of (Hz) (Hz) freq 440 494 523 587 659 698 784 439 495 521 588 658 694 781 114 101 96 85 76 72 64 Third Oc t av e Nominal Actual Value of (Hz) (Hz) freq 880 988 1047 1175 1319 1397 1568 877 980 1042 1163 1316 1389 1563 57 51 48 43 38 36 32
Note A B C D E F G
*To get values of freq for sharps (#), multiply the value listed by 0.9442 and round off the result. For example, the freq value for C# in the first octave would be 0.9442 x 191 = 180.
; ; ; ; ; ; ; ;
PROGRAM: CHARGE.SRC This program in TechTools assembly language plays a short tune, Charge, from data stored in a pair of tables. It uses a general-purpose routine called beep to generate the notes from frequency and duration data. The tune will play when the PIC or Downloader is first powered up, and any time the device is reset. Note that although beep uses seven file registers, five of these may be reused elsewhere in the program for counting, or any use that does not require their values to be preserved after beep is called. = = ra 6 ; Connect speaker to adjacent ra ; number of notes in the tune
spkr pins. end_note
; Variable storage above special-purpose registers. org 8 freq ds 1 ; passess frequency value to beep duratn ds 1 ; passes duration value to beep f_temp ds 1 ; temporary counter (frequency) d_hi ds 1 ; temporary counter (high byte of ; duration) d_lo ds 1 ; temporary counter (low byte of ; duration) tgl ds 1 ; temporary variable t_pat ds 1 ; temporary variable index ds 1 ; note counter used by main ; program ; Device data and reset vector. Remember to change these entries if programming

; another type of PIC. device reset

pic16c54,rc_osc,wdt_off,protect_off start
APPS
; Set ROM origin to 0 for start of program. org 0 start mov !ra, #0 mov ra, #1010b mov mov call mov freq. mov call mov call inc cjb :endless goto w,index lengths index, #0 w,index notes freq, w
; Set port ra to output. ; Use alternating bits as push-pull ; drivers. ; Start index at 0. ; Look up the frequency of next ; note. ; Put the frequency value into
:loop
duratn, w beep index index,#end_note,:loop; All notes done? If not, loop. :endless ; Sit in a loop until reset.
; Look up the length of the next ; note. ; Put the length value into duratn. ; Play the note.
Notes cies
jmp retw
pc+w
; Lookup table for note frequen-
90, 67, 52, 44, 52, 44 ; 4-MHz or RC operation.
; For 8-MHz operation (e.g., the downloader w/ built-in xtal) substitute the sequence ; below for Notes: ; retw 180, 134, 104, 88, 104, 88 Lengths jmp retw pc+w ; Lookup table for note durations. 35, 35, 35, 55, 30,127; 4-MHz or RC operation.
; For 8-MHz operation (e.g., the downloader w/ built-in xtal) substitute the sequence ; below for Lengths: ; retw 70, 70, 70, 110, 60, 255 beep A. mov mov jnz mov mov t_pat, #255 f_temp, freq :cont t_pat, #0 d_hi, duratn ; All ones to invert all bits in port ; If freq is zero, fill pattern with ; zeros. ; ..for a silent pause. ; Variable duratn goes into the
:cont

clr mov xor nop nop nop djnz mov mov :dur_lo sub jc sub sc ret done. jmp :noRoll1 clr jmp nop nop jmp :main tgl :dur_lo ; Else loop to :main. ; If f_temp is not zero, dont ; ..pluck the speaker. ; Waste time to ensure that all ; paths through ; ..the routine take 20s. :main d_lo tgl, t_pat spkr, tgl ; high byte ; ..of a 16-bit counter. ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;
:main :delay
XORing the speaker bits with 1s inverts them. Zeros have no effect. Nops pad main loop to 20s. When f_temp reaches zero, reload the freq. ..value and put tgl_pat into tgl in order to ..invert the speaker bits on the next loop. Decrement low byte of duration. If no borrow, go to noRoll2. If borrow occurs, decrement the ..high byte of duration. If the high byte ..needs a borrow, routine is
f_temp, :noRoll1 f_temp, freq tgl, t_pat d_lo, #1 :noRoll2 d_hi, #1
:noRoll2
6: Driving an LCD Display

Introduction. This application note shows how to interface PIC microcontrollers to common Hitachi liquid-crystal display (LCD) modules. The program, in TechTools assembly language, writes text to the display, reads display status, and creates custom character patterns. Background. LCD modules based on the Hitachi 44780 controller are plentiful and inexpensive, and range in size from 8 to 80 characters. While a complete description of these LCDs features and operation is beyond the scope of this application note, here are the highlights:
Driving an LCD Display
APPS
18
RA2
+5
RA1
17
TO LCD Vdd
RA3
3
RA0
16
1k 4
RTCC MCLR
5
OSC1
15
ceramic resonator
Reset 6
Vss RB0
7
PIC 16C54
OSC2
14
Vdd
13
+5
RB7
12
RB1
8
RB6
11
RB2
9
RB5
10
RB3
RB4
14 13 12 11 10 9 8 7
6 DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0
R/W RS
Vdd Vo Vss
2 3 1
10k
FROM RA.3
Hitachi LCD modules display the standard ASCII character set, plus selected Japanese, Greek, and math symbols. They operate from a single-ended 5-volt supply, and communicate with a bus or controller through 11 input/output (I/O) lines. The data lines are tri-state; they go into a high-impedance state when the LCD is not enabled. The three control lines control the LCD. The enable (E) line determines whether the LCD listens to the other control and data lines. When disabled, the LCD ignores all data and control signals. When enabled, the LCD checks the state of the other two control lines and responds accordingly.

The read/write (R/W) line determines whether the LCD reads bits from the data lines, or writes bits to them. Register-select (RS) determines whether the LCD treats data as instructions or characters. Here is the truth table for the control lines: E R/W RS 0 1 0 1 0 1 LCD disabled. LCD enabled. Write to LCD. Read from LCD. Instructions. Characters/bytes.
Writing to the LCD requires the basic steps listed below. (Reading from the LCD follows the same sequence, but the R/W bit must be set.) Clear the R/W bit. Set or clear the RS bit as appropriate. Set the E bit (E=1). Clear the E bit (E=0). When power is applied to the LCD, it resets itself and waits for instructions. Typically these instructions turn on the display, turn on the cursor, and set the display to print from left to right. Once the LCD is initialized, it can receive data or instructions. If it receives a character, it prints it on the screen and moves the cursor one character to the right. The cursor marks the next location at which a character will be printed. The LCDs internal processing is similar. A memory pointer determines where the next byte will be stored. When a new byte arrives, the pointer advances. To write to sequential locations, establish the starting address and then write one byte after another. Characters are stored in data display (DD) RAM. Regardless of the number of characters visible on the display, the LCD has 80 bytes of DD RAM. Characters in off-screen RAM can be made visible by scrolling the display. The LCD also has 64 bytes of character-generator (CG) RAM. Data in

CG RAM determines the bit maps of the characters corresponding to codes 0 through 7 (in normal ASCII, these are control codes). To download bit maps to the LCD, first set the CG RAM address to the desired starting point (usually 0), and then write the bytes to the LCD. Because the pointer increments with each write, you dont need to keep specifying addresses. Figure 2 is an example pattern definition. The program listing shows how to define custom characters.
APPS
Address in Character Generator RAM 0000 0001 0010 0011 0100 0101 0110 0111
Bit Map
Data 01111 10001 10001 01111 00001 00001 00001 00000
retw Data 15 17 17 15 1 1 1 0
Figure 2. Programming a custom character pattern. Before you can write to DD RAM after defining custom characters, the program must set a DD RAM address. The LCD reads and writes whichever RAM bank (DD or CG) was last specified in a set-address instruction. Once you have set an address in DD RAM, the next data write will display a character at the corresponding location on the screen. Until now, we have talked about reading and writing to the LCD as though it were regular memory. Its not. The LCDs controller takes 40 to 120 microseconds (s) to complete a read or write. Other operations can take as long as 5 milliseconds. To avoid making the PIC wait a worstcase delay between operations, the LCD has a 1s instruction that reads the address counter and a busy flag. When the busy flag is set (1), the LCD cannot handle a read or write. The program listing includes a subroutine (blip_E) that makes sure the busy flag is cleared (0) before talking to the LCD. The address returned along with the busy flag is either the DD or CG RAM pointer, depending on which address was last set.

Figure 3 is a list of LCD instructions for reading and writing memory. Some other useful instructions appear as constants in the beginning of the program listing. How it works. The circuit in figure 1 interfaces a PIC to an LCD module. When the power is turned on, or the circuit is reset, the PIC initializes the LCD, downloads four custom characters, and prints TechTool forward and backward (with custom mirrored characters). The potentiomenter connected to the LCDs Vo pin controls contrast. If the display is hard to read, appears blank, or is filled with black pixels, adjust this control. Power to the LCD is controlled by a PIC I/O bit. In this way, the PIC ensures that the 5-volt supply is up and stable before switching on the LCD. If the circuit used the PICs sleep mode, it could shut down the LCD to save approximately 1.5 mA. If you use this feature, make sure that the
Set DD RAM address RS R/W DB7 DB6 DB5 DB4 0 0 1 A A A
DB3 A
DB2 A
DB1 DB0 A A
Set CG RAM address RS R/W DB7 DB6 DB5 DB4 0 0 0 1 A A
DB3 A
DB2 A
DB1 DB0 A A
Read busy flag and address RS R/W DB7 DB6 DB5 DB4 0 1 Bsy A A A
DB3 A
DB2 A
DB1 DB0 A A
Write data to RAM (CG or DD, most recently set) RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 1 0 D D D D D D D D
Read data from RAM (CG or DD, most recently set) RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 1 1 D D D D D D D D
Figure 3. Commonly used memory operations. (A = address; D = data; Bsy = busy)

data and control lines are cleared to 0s or set to input before putting the PIC to sleep. Otherwise, leakage through the LCDs protection diodes might continue to power it. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
APPS
; PROGRAM: Drive Liquid Crystal Display (LCD_DRVR.SRC) ; This program initializes a Hitachi LCD module, defines a set of four custom ; characters, and displays the message TechTool forward and backward (in custom, ; mirror-reading letters). It includes a subroutine, blip_E, that handles all the required ; handshaking to send data or instructions to the LCD. LCD_pwr RS RW E data count char_cnt = = = = = = = ra.3 ra.2 ra.1 ra.0 rb 16 32 ; ; ; ; ; ; ; ; ; +5 to LCD module 0 = write, 1 = read 0 = instruction, 1 = data 0 = disable, 1 = enable Data to LCD Number of characters in demo string. Number of bytes in custom character definition
; Declare constants for common LCD instructions. To perform the functions listed ; below, clear bit RS and send #constant to the display. Remember to set RS again ; before sending characters to the LCD. clear = 1 ; Clears the display (fills it with ; blanks). home = 2 ; Returns display to the home ; position. shift_l = 24 ; Shifts display to the left. shift_r = 28 ; Shifts display to the right. crsr_l = 16 ; Moves cursor to the left. crsr_r = 20 ; Moves cursor to the right. blink_c = 11 ; Blinks whole character to indicate ; cursor position. no_crsr = 8 ; Turns off the cursor. ; Set RAM origin above special registers, declare variables, and set code origin.

temp temp2 counter ; Device data device reset start mov mov mov mov setb call mov call mov call mov call mov call setb mov :stuff mov call mov call inc cjb clrb mov call setb mov mov call mov call pic16c54,xt_osc,wdt_off,protect_off start ra, #0 rb, #0 !ra,#0h !rb,#0h LCD_pwr wait temp2,#00110000b blip_E temp2,#00001110b blip_E temp2,#00000110b blip_E temp2, #01000000b blip_E RS counter, #0 ; Write to CG RAM: start at ; address 0. ; Set RS to send data. ; Initialize ports, power LCD and ; wait for it to reset. ; set control lines to output ; set data lines to output org ds ds ds org 8 1 1 1 0 ; ; ; ; Temporary counter. Pass data or instructions to blip_E. Index variable.
; Initialize LCD: set 8-bit, 1-line ; operation.
w, counter my_chars ; Get next byte. temp2, w ; Write byte to CG RAM. blip_E counter counter, #char_cnt,:stuff RS ; Clear RS to send instruction. temp2, #10000000b ; Address 0 of DD RAM. blip_E RS counter, #0 w,counter msg temp2,w blip_E ; Send the message string to the ; LCD.
send_msg :loop

inc cjb jmp

counter counter,#count,:loop :loop2 ; Endless loop. Reset PIC to run ; program again.
:loop2
APPS
; Write data or instructions (in variable temp2) to the LCD. blip_E movb pa2, RS ; Store current state of RS in ; unused bit. clrb RS ; Clear RS to send instruction. setb RW ; Set RW to read. mov !rb, #255 ; Port RB: all inputs. :loop setb E ; Enable the LCD. nop mov temp, data ; Put LCD data (RB) into temp. clrb E ; Disable LCD. snb temp.7 ; Is the LCD busy? jmp :loop ; Yes, try again later. movb RS, pa2 ; No, send the data or instruction. clrb RW mov !rb, #0 mov data, temp2 setb E nop clrb E ret wait :loop mov djnz djnz temp2,#200 temp,:loop temp2,:loop ret pc+w ; Table of ASCII and custom ; characters to display. T,e,c,h,T,o,o,l,s,2,0,0,2,1,2,3 ; Create a delay for LCD power-on ; reset.
msg
jmp retw
my_chars
jmp retw retw retw retw
pc+w
; Table of data for custom ; backwards characters. 6,4,4,4,4,4,14,0 ; backwards l 0,0,13,19,1,1,1,0 ; backwards r 0,0,14,16,30,17,30,0 ; backwards a 15,17,17,15,1,1,1,0 ; backwards P
7: Direct & Indirect Addressing
Introduction. This application note describes direct and indirect addressing and shows a method for avoiding the gaps in the PIC16C57s banked memory. Direct Addressing. PIC microcontrollers have 32 or 80 bytes of RAM, which Microchip refers to as file registers. The first seven registers (eight in the case of the 28-pin PICs) are mapped to special functions, such as the real-time clock/counter (RTCC), status bits, and input/output (I/O) ports. Figure 1 shows a simplified memory map. The simplest way to manipulate the contents of a PICs RAM is to specify a register address in an instruction, like so: mov 010h, #100
Direct & Indirect Addressing
This instruction, which moves the decimal number 100 into register 10 hexadecimal (h), is an example of direct addressing. Most of the instructions in a typical PIC program use this addressing mode. The TechTools assembler has several helpful features for direct adAddress (hex)
0 1 2 3 4 5 6 7 81F 202F 303F 404F 505F 606F 707F indirect rtcc pc status fsr ra rb rc
Description
Reads/writes address pointed to by fsr. Real-time clock/counter. Program counter9 to 1 1 bits wide; lower 8 may be read/written. Flag bits for arithmetic operations, sleep and reset, and ROM page selects. Pointer; address of data accessed through indirect. I/O port ra. I/O port rb. I/O port rc on 55 and 57; general-purpose register on 54 and 56. General-purpose RAM (54, 55, 56) and RAM bank 0 of 57 RAM warp; reads and writes to 0F . RAM bank 1 (57 only) RAM warp; reads and writes to 0F . RAM bank 2 (57 only) RAM warp; reads and writes to 0F . RAM bank 3 (57 only) RAM Banks 13, 16C57
Figure 1. Simplified memory map of PICs 16C54 through 57.
dressing. The first of these is labeling. Instead of referring to registers by address, you may assign names to them: APPS counter = . . . mov 010h ;Program header
counter, #100
Labeled memory locations are often called variables. They make a program more understandable and easier to modify. Suppose you needed to change the location in which the counter data was stored. Without the label, you would have to rely on your text editors searchand-replace function (which might also change other numbers containing 10). With a label, you could change the counter = ... value in the program header only. You can also define variables without specifying their address by using the ds (define space) directive: org ds . . . mov 8 1 ;Start above special registers. ;One byte labeled counter.
counter
counter, #100
Using ds assigns the label to the next available register. This ensures that no two labels apply to the same register, making variable assignments more portable from one program to another. The only caution in using ds is that you must set the origin using the org directive twice; once for the starting point of variables in RAM, and again (usually at 0) for the starting point of your program in ROM. Labels can be assigned to individual bits in two ways. First, if the bit belongs to a labeled byte, add .x to the label, where x is the bit number (07). Or assign the bit its own label: LED = ra.3 ;Bit 3 of port ra controls LED.
The TechTools assembler has predefined labels for the special-purpose registers, and the bits of the status register. See your manual for a list. Indirect Addressing. The registers used in direct addressing are set forever when the program is burned into the PICs ROM. They cannot change. However, many powerful programming techniques are based on computing storage locations. Consider a keyboard buffer. If keystrokes cant be processed immediately, they are stored in sequential bytes of memory. Pointersvariables containing addresses of other variablestrack the locations of data entered and data processed. The PICs indirect addressing mode allows the use of pointers and the high-level data structures that go with them, such as stacks and queues. Using indirect addressing for the earlier example (writing 100 to register 10h) would look like this: mov mov fsr, #010h indirect, #100 ;Set pointer to 10h. ;Store 100 to indirect.
The value in the file select register (fsr; register 04h) is used as the address in any instruction that reads/writes indirect (register 00h). So
7 6 5 4 3 2 1 0 address, 0Fh
Figure 2. The 16C57 file-select register.
banked memory enable; 1 = on, 0 = off bank select, 03h unused; reads as 1
Memory Register banks:
0 1 2 3
10h to 1Fh 30h to 3Fh 50h to 5Fh 70h to 7Fh
storing 10h in the fsr and then writing 100 to indirect is the same as writing 100 to address 10h. APPS A more practical example would be to store a series of values from an I/O port to sequential registers in memory. All it takes is a loop like this: mov mov mov inc cjb pointer, #010h ;Set start address. fsr, pointer ;Put pointer into fsr. indirect, rb ;Move rb to indirect. pointer ;pointer = pointer + 1. pointer,#01Fh,:loop
:loop
This fragment assumes that a variable named pointer was declared previously (using =, equ, or ds), and that rb is set for input. The loop will rapidly fill registers 10h through 1Fh with data samples from rb. PICs with 32 bytes of RAM (54, 55, and 56) have a five-bit-wide fsr. Since all registers are eight bits wide, the highest three bits of the fsr in these devices are fixed, and always read as 1s. Keep this in mind if you plan to perform comparisons (such as the last line of the example above) directly on the fsr. It will always read 224 (11100000b) higher than the actual address it points to. The 16C57 has 80 bytes of RAM and a seven-bit-wide fsr. The highest bit of its fsr is fixed and reads as a 1. Seven bits allows for 128 addresses, but only 80 are used. The remaining 48 addresses are accounted for by three 16-byte gaps in the 57s memory map. See the RAM warps in figure 1. Because these warps map to the lowest file registers of the PIC, they can cause real trouble by altering data in the special-purpose registers. To avoid this problem, consider using a subroutine to straighten out the memory map and avoid the warps. Below is an excerpt from a program that uses the registers from 10h on up as a storage buffer for up to 64 characters of ASCII text. For the purposes of the program, address10h is location 0 in the buffer; 7F is location 63. When the program needs to write a value representing a position in the buffer to the fsr, it puts the value into the w register and calls buf_ptr (buffer pointer).

buf_ptr mov mov cjae cjae cjae jmp add add add add ret temp,w fsr, temp temp,#030h,:bank3 temp,#020h,:bank2 temp,#010h,:bank1 :bank0 fsr,#010h fsr,#010h fsr,#010h fsr,#010h
:bank3 :bank2 :bank1 :bank0
It may be more useful in some applications to treat these memory locations as register banks, as they are described in the Microchip literature. According to this model, bit 4 of the fsr enables bank selection when it is a 1. The 16-byte bank in use is then selected by bits 5 and 6 of the fsr as shown in figure 2. This model explains the warps in the memory map. Each of the three warp addresses (20h, 40h, and 60h) has a 0 in the bit-4 position. This disables banked memory, causing the PIC to disregard all but bits 0 through 3 of the address.
8: The PIC16C71 A/D Convertor
The PIC16C71 A/D Convertor
Introduction. This application note presents a program in TechTools assembly language that uses the PIC16C71s built-in analog-to-digital converter (ADC) to measure an input voltage and flash an LED at a proportional rate. Background. One of the most popular enhancements offered by the new PIC16C71 is its eight-bit ADC, which features: 20-microsecond (s) conversion time (nearly 50,000 samples per second, depending on additional processing time). Four multiplexed inputs. Built-in sample-and-hold. 1 least-significant-bit accuracy (better than 20 millivolts with a 5-volt reference). Selectable voltage reference (Vdd or RA.3). While using the ADC is fairly straightforward, it does require a series of decisions much like those required to select and use a separate ADC. The first consideration is hardware.
+5
APPS
1k 1 18
RA2/Ain2
2
RA1/Ain1
17
5k pot
RA3/Ain3
3 +5 4
RA0/Ain0
16 10k 4.7 pF
RTCC/RA4 MCLR
5
OSC1
15
OSC2
Vss
6
RB0
LED 220 7
PIC Vdd 16C71

RB7 RB6
14 13 12
RB1
8
11
RB2
9
RB5
10
RB3
RB4
Input Characteristics. The ADC produces a digital output that is proportional to an analog input. A voltage reference determines the input voltage that will produce a full-scale (255) digital output. The voltage reference can be the +5-volt power-supply rail, or some other voltage source between 3 volts and the power supply voltage + 0.3 volts. The ADC is most accurate with a reference voltage of 5.12 volts, according to the manufacturers specifications.
The specifications recommend that the analog voltage source being measured have an impedance of no more than 10k. Above this value, accuracy suffers. They also suggest that the source have not less than 500 impedance. This limits current through the PIC in the event that your program reconfigures the analog input pin as an output, or some other circuit trauma occurs.
Clock Source. The PICs ADC, like the PIC itself, requires a clock signal. The ADC performs a conversion in 10 of its clock cycles, which must be no shorter than 2s. Clock signals for the ADC can come from two sources, the PICs own clock or an on-chip resistor-capacitor (RC) oscillator exclusive to the ADC.
When the PICs clock is the source, it is divided by 2, 8 or 32, depending on the status of the ADC clock source bits (see figure 2). In order to have an ADC clock signal of 2s or longer, the PIC clock speed must not exceed 1, 4, or 16 MHz, respectively. If you plan to run the PIC faster than 16 MHz, or you want the ADC conversion rate to be independent of the PIC clock, you must use the ADCs RC oscillator. The tradeoff in using the RC oscillator is that its period can vary from 2 to 6s, depending on temperature and manufacturing tolerances.
Interrupt Enable. The ADC is relatively slowat 20 MHz the PIC can execute 100 instructions in the 20s the ADC takes to make a conversion. In some cases, it makes sense not to force a PIC program to wait in a loop for a conversion to finish. The alternative is to configure the ADC to announce conversion complete through an interrupt. To keep things as simple as possible, the example program does not take advantage of interrupt capability.
Pin Configuration and Voltage Reference. Pins RA.0 through RA.3 can serve as inputs to the ADC. One of the choices you must make when setting up the ADC is which pins to configure as analog inputs, which (if any) as digital inputs, and what to use as a voltage reference. Figure 2 shows the range of available choices.
Note that the control register containing the configuration and voltage reference bits is in register page 1. To access it, you must first set bit RP0. The program listing shows how.
APPS
Input Selection. Only one of the pins configured for analog input can
ADC Control and Status Register (ADCON0, register page 0, 08h) ADCS1
7
ADCS0 unused
6 5
CHS1
4
CHS0 go_done
3 2
ADIF
1
ADON
0
ADC power switch: 0 = ADC off 1 = ADC on Interrupt flag bit: set when conversion is complete Conversion flag bit: set to start conversion cleared when conversion is done Clock select bits: 00 = Oscillator/2 01 = Oscillator/8 10 = Oscillator/32 11 = RC clock Channel select bits: 00 = channel 0 (Ain0) 01 = channel 1 (Ain1) 10 = channel 2 (Ain2) 11 = channel 3 (Ain3)
ADC Control Register (ADCON1, register page 1, 88h) 0

7
0
6
0
5
0
4
0
3
0
2
ADIF
1
ADON
0
Remaining bits of ADCON1 are unimplemented and always read 0 Pin configuration bits: 00 = RA0RA3, analog; Vdd reference 01 = RA0RA2, analog; RA3 reference 10 = RA0/1, analog; RA2/3, digital; Vdd reference 11 = RA0RA3, digital input
Figure 2. ADC control registers.
actually sample at any one time. In other words, if you want ADC input from two or more channels, your program must select a channel, wait long enough for the sample-and-hold circuit to charge up, command a conversion, get the result, and select the next channel... How long should the program wait for the sample-and-hold? Microchip suggests a worst case of 4.67s plus two ADC clock cycles (after a conversion, the sample-and-hold takes a two-cycle break before it begins sampling again). How it works. The PIC in figure 1 accepts a voltage from a pot wired as variable voltage divider. The PICs ADC, which is set up to use Vdd as a reference, outputs a one-byte value thats proportional to the input voltage. This value controls a timing routine that flashes an LED. When the input voltage is near 5 volts, the LED flashes about once a second. When its near 0, the LED flashes very rapidly. The program listing shows how its done. Most of the code is devoted to setting up the ADC. Constants at the beginning of the program are assigned with values that, when loaded into the appropriate ADC control registers, turn the ADCs various features on and off. If you wish to change, for instance, the pin that the circuit uses for analog input, just comment out the line containing AD_ch = 0 and uncomment the desired channel (commenting and uncommenting are handy techniques for temporarily removing and restoring instructions in source code. Putting a semicolon (;) in front of a line causes the assembler to ignore it, as though it were a comment). If you accidentally leave two assignements for AD_ch uncommented, the assembler will catch the mistake, flag the redefinition and tell you the line number of the error. The assembler combines values assigned to ADC_ch and ADC_clk into a single byte by performing a logical OR (|) on the values and putting the result into another constant, ADC_ctl. This technique makes the program easier to understand and modify, and doesnt cost a thing in PIC program memory or processing time. The assembler does all the work. Program listing. This program may be downloaded from our Internet ftp
site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com. APPS
; ; ; ; ; PROGRAM: Using the 16C71s analog-to-digital converter (ADC71.SRC) This program demonstrates use of the ADC in a simple circuit that samples a voltage and flashes an LED at a proportional rate. The header contains a number of constants representing setup constants for the ADC control registers. Uncomment the constant corresponding to the desired ADC setting. set the ADC clock source and speed. Uncomment one. 0 ; Oscillator x 2 (<= 1 MHz). 64 ; Oscillator x 8 (<= 4 MHz). 128 ; Oscillator x 32 (<= 16 MHz). 192 ; RC oscillator, 26 us.
; The following constants ;AD_clk = ;AD_clk = ;AD_clk = AD_clk =
; The following constants select a pin for ADC input. Uncomment one. AD_ch = 0 ; ADC channel 0 (Ain0, ;AD_ch = 8 ; ADC channel 1 (Ain1, ;AD_ch = 16 ; ADC channel 2 (Ain0, ;AD_ch = 24 ; ADC channel 3 (Ain0, AD_ctl = AD_clk | AD_ch ; Logical OR.
pin pin pin pin
17). 18). 1). 2).
; The following constants determine which pins will be usable by the ADC and ; whether Vdd or RA.3 will serve as the voltage reference. Uncomment one. AD_ref = 0 ; RA.0-3 usable, Vdd reference. ;AD_ref = 1 ; RA.0-3 usable,RA.3 reference. ;AD_ref = 2 ; RA.0/1 usable, Vdd reference. ;AD_ref = 3 ; All unusabledigital inputs only. device rc_osc,wdt_off,pwrt_off,protect_off id ADC1 org ds ds 0Ch 1 1
counter1 counter2
; Set starting point in program ROM to zero. Jump past interrupt vector to beginning ; of program. (This program doesnt use interrupts, so this is really not needed, but it ; will be easier to add interrupts later if required.) org 0 jmp 5 org 5 start mov !ra, #255 ; Set
RA
to input.

mov mov mov setb mov clrb setb :loop call setb call clrb goto wait :not_done setb snb jmp mov !rb, #0 intcon, #0 adcon0,#AD_ctl rp0 adcon1,#AD_ref rp0 adon wait rb.0 wait rb.0 :loop go_done go_done :not_done counter2,adres

; Set RB to output. ; Turn interrupts off. ; Set AD clock and channel. ; Enable register page 1. ; Set usable pins, Vref. ; Back to register page 0. ; Apply power to ADC. ; Delay for time determined by ; ADC input. ; Turn LED on. ; Delay for time determined by ; ADC input. ; Turn LED off. ; Endless loop. ; Start conversion. ; Poll for 0 (done). ; If 1, poll again. ; Move ADC result into counter.
; The number of loops this delay routine makes is dependent on the result of the AD ; conversion. The higher the voltage, the longer the delay. :loop djnz counter1, :loop djnz counter2, :loop ret
9: Managing Multiple Tasks
Managing Multiple Tasks
Introduction. This application note presents a program in TechTools assembly language that demonstrates a technique for organizing a program into multiple tasks. Background. Like most computers, the PIC executes its instructions one at a time. People tend to write programs that work the same way; they perform one task at a time. Its often useful to have the controller do more than one thing at a time, or at least seem to. The first step in this direction is often to exploit the dead time from one taskthe time it would normally spend in a delay loop, for instanceto handle a second task. The PICs speed makes this quite practical in many cases. When several tasks must be handled at once, this approach can quickly become unworkable. What we need is a framework around which to organize the tasks. We need an operating system. The program in the listing illustrates an extremely simple operating system that runs each of eight small subprograms in turn. When the subprograms finish their work, they jump back to the system. Notice that this method does not require the call instruction, so it leaves the two-level stack free for the use of the subprograms. How it works. The circuit and program comprise an eight-LED flasher. Each of the LEDs flashes at a different rate. While this could be accomplished differently, the program is easier to understand and maintain because the code that controls each LED is a separate task. The system portion of the program acts like a spinning rotary switch. Each time it executes, it increments the task number and switches to the next task. It does this by taking advantage of the PICs ability to modify the program counter. Once the task number is loaded into the working register, the program executes the instruction jmp pc+w. The destinations of these jumps contain jmp instructions themselves, and send the program to one of the eight tasks. Not surprisingly, a list of jmp instructions arranged like this is called a jump table. Modifications. For the sake of simplicity, this task-switching program
APPS

1 18
RA2
2
RA1
17
RA3
+5 3
RA0
16
RTCC
4
OSC1
15
MCLR
5
Vss
6
PIC 16C54
OSC2
14
+5
Vdd
13 470 470 470 470
RB0
7
RB7
12
LED LED LED LED
RB1
8
RB6
11
RB2
9
RB5
10
RB3
RB4
470 470 470 470
LED LED LED LED
lacks one important attribute: fixed timing. The individual tasks are permitted to take as much or as little time as they require. In some real applications, this wouldnt be acceptable. If the system maintains a master timer (like the variable ticks in this program) it should increment at a consistent rate. With many possible paths through the code for each task, this may seem like another problem. A straightforward solution is to use the PICs RTCC to time how long a particular task took, then use that number to set a delay to use up all of the tasks remaining time. All you need to know is the worst-case timing for a given task. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.

; ; ; ;

PROGRAM: TASK.SRC This program demonstrates task switching using the PICs relative addressing mode. It handles eight tasks, each of which flashes an LED at a different rate. The flashing routines base their timing on a master clock variable called ticks. = rb
APPS
LEDs
; Put variable storage above special-purpose registers. org 8 task ticks time0 time1 time2 time3 time4 time5 time6 time7 ds ds ds ds ds ds ds ds ds ds 1 1 1 1 1 1 1 1 1 1 ; ; ; ; ; ; ; ; ; ; ; ; The task number used by the system. Master time clock, increments once for each system cycle. Timer for task 0. Timer for task 1. Timer for task 2. Timer for task 3. Timer for task 4. Timer for task 5. Timer for task 6. Timer for task 7.
; Remember to change device info if using a different part. device pic16c54,xt_osc,wdt_off,protect_off reset start ; Set starting point in program ROM to zero. org 0 start mov mov clr clr inc cjne clr inc mov jmp jmp jmp jmp jmp jmp jmp jmp !rb,#00000000b task, #7 ticks LEDs task task, #8, :cont task ticks w, task pc+w task0 task1 task2 task3 task4 task5 task6 ; Set port rb to output. ; Set task number. ; Clear system clock. ; Clear LEDs ; Next task number. ; No rollover? Continue. ; Rollover: reset task and ; increment the clock. ; Prepare to jump. ; Jump into table, and from there ; to task #.
system
:cont

jmp task0 cjne inc cjne clr xor jmp cjne inc cjne clr xor jmp cjne inc cjne clr xor jmp cjne inc cjne clr xor jmp cjne inc cjne clr xor jmp cjne inc cjne clr xor jmp cjne inc cjne clr task7 ticks, #255,:cont time0 time0, #3, :cont time0 LEDs, #00000001b system ticks, #255,:cont time1 time1, #8, :cont time1 LEDs, #00000010b system ticks, #255,:cont time2 time2, #6,:cont time2 LEDs, #00000100b system ticks, #255,:cont time3 time3, #11,:cont time3 LEDs, #00001000b system ticks, #255,:cont time4 time4, #12, :cont time4 LEDs, #00010000b system ticks, #255,:cont time5 time5, #4, :cont time5 LEDs, #00100000b system ticks, #255,:cont time6 time6, #23,:cont time6 ; ; ; ; ; Every 255 ticks of system clock increment task timer. Every 3 ticks of task timer, toggle LED, and reset task timer.
:cont task1
:cont task2
:cont task3
:cont task4
:cont task5
:cont task6

xor jmp cjne inc cjne clr xor jmp

LEDs, #01000000b system ticks, #255,:cont time7 time7, #9,:cont time7 LEDs, #10000000b system
:cont task7
APPS
:cont
10: An External A/D Convertor
Introduction. This application note shows how to interface an 8-bit serial analog-to-digital converter to 16C5x-series PICs. Background. Although the PIC 16C71 offers a built-in analog-to-digital converter (ADC), some users may prefer an external device. One thats particularly suited to use with the PIC is the National Semiconductor ADC0831, available from Digi-Key, among others. Interfacing the 831 requires only three input/output lines, and of these, two can be multiplexed with other functions (or additional 831s). Only the chip-select (CS) pin requires a dedicated line. The ADCs range of input voltages is controlled by the VREF and VIN() pins. VREF sets the voltage at which the ADC will return a full-scale output of 255, while VIN() sets the voltage that will return 0.
An External A/D Convertor
+5 1 8
CS 05V in
2
Vcc
Vin(+)
3
ADC 0831
CLK
6
Vin()
4
DO
5
GND
Vref
+5
18
RA2
2
RA1
17
RA3
470 LED +5 3
RA0
16
RTCC
4
OSC1
15
MCLR
5
Vss
6
PIC 16C54
OSC2
14
Vdd
13
+5
RB0
7
RB7
12
RB1
8
RB6
11
RB2
9
RB5
10
RB3
RB4
In the example application, VIN() is at ground and VREF is at +5; however, these values can be as close together as 1 volt without harming accuracy or linearity. You may use diode voltage references or trim pots to set these values.
How it works. The sample program reads the voltage at the 831s input pin and uses the eight-bit result to control a timing routine that flashes an LED. The LED flashes slowly when the input is 5 volts, and very rapidly as it approaches ground. The subroutine convert handles the details of getting data out of the ADC. It enables the ADC by pulling the CS line low, then pulses the clock (CLK) line to signal the beginning of a conversion. The program then enters a loop in which it pulses CLK, gets the bit on pin data, and shifts it into the received byte using the rotate left (rl) instruction. Remember that rl affects not only the byte named in the instruction, but the carry bit as well. When all bits have been shifted into the byte, the program turns off the ADC by returning CS high. The subroutine returns with the conversion result in the variable ADresult. The whole process takes 74 instruction cycles. Modifications. You can add more 831s to the circuit as follows: Connect each additional ADC to the same clock and data lines, but assign separate CS pins. Modify convert to take the appropriate CS pin low when it needs to acquire data from a particular ADC. Thats it. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
APPS
; ; ; ;
PROGRAM: AD831.SRC Program demonstrates the use of an external serial ADC (National ADC0831) with PIC16C5x-series controllers. A variable dc voltage (0-5V) at pin 2 of the ADC controls the blinking rate of an LED connected to PIC pin ra.3. = = = = ra.0 ra.1 ra.2 ra.3
clk data CS LED
; Put variable storage above special-purpose registers. org 8 counter1 ds 1

counter2 ADresult ds ds 1 1
; Remember to change device info when programming part. device pic16c54,xt_osc,wdt_off,protect_off reset start ; Set starting point in program ROM to zero org 0 start mov ra, #00000100b mov !ra,#00000010b mov blink xor call goto call mov add !rb,#00000000b ra, #8 wait blink convert counter2,ADresult counter2,#1
; Set CS to disable ADC for now. ; Make ra.1 (data) input, rest ; outputs. ; Make rb output. ; Invert bit3 of ra (LED). ; Delay depends on ADC data. ; Endless loop. ; Get AD result and put into ; counter2. ; Add 1 to avoid underflow when ; AD=0.
wait
; Time delay produced by loop depends on value of ADresult. Higher values ; produce longer delays and slower blinking. :loop djnz counter1, :loop djnz counter2, :loop ret convert clrb mov setb nop clrb clr setb nop clrb movb rl djnz setb ret CS counter2,#8 clk clk ADresult clk clk c,data ADresult counter2,:loop CS ; Enable the ADC. ; Set up for eight data bits. ; Pulse the clock line (insert ; nop if PIC clock >4 MHz). ; Clear byte to make way for new ; data. ; Pulse the clock line (insert nop ; if PIC clock >4 MHz). ; ; ; ; Move data bit into carry. Rotate carry into byte. Eight bits collected? If not, loop. End the conversion.
:loop
11: Using Serial EEPROMs
Using Serial EEPROMs
Introduction. This application note shows how to use the Microchip 93LC56 EEPROM to provide 256 bytes of nonvolatile storage. It provides a tool kit of subroutines for reading, writing, and erasing the EEPROM. (Note that EEPROMs made by other manufacturers will not work with the PIC16Cxx.) Background. Many designs take advantage of the PICs ability to store tables of data in its EPROM program memory. The trouble is that the larger the tables are, the smaller the space left for code. And many applications could benefit from the ability to occasionally update data tables for calibration or other purposes. What the PIC needs is the equivalent of a tiny disk drive.
+5
1 8
APPS
CS
2
Vcc
7
CK
3
93LC56 NC
ORG Vss
6 5
DI
4
DO
1k
1 18
RA2
2
RA1
17
RA3 +5
3
RA0
16
RTCC
4
OSC1
15
MCLR
5
Vss
6
PIC OSC2 16C54 Vdd

RB7
14 13
+5
RB0
7
12
RB1
8
RB6
11
RB2
9
RB5
10
RB3
RB4
470 (all)
LEDs
The Microchip 93C56 and 93LC56 electrically erasable PROMs (EEPROMs) are perfect for these applications. They communicate serially via a three- or four-wire bus using a simple synchronous (clocked) communication protocol at rates of up to 2 million bits per
second (Mbps). Its possible to read a byte in as little as 10 microseconds (including the time required to send the instruction opcode and address). Once a program has begun reading data from the EEPROM, it can continue reading subsequent bytes without stopping. These are clocked in at the full 2 Mbps rate; 1 byte every 4 microseconds. Erasing and writing these serial EEPROMs happens at a more leisurely pace. While the opcode, address, and data can be clocked into the chip at high speed, the EEPROM requires about 2 milliseconds to erase or write a byte. During this time, the chip cannot process additional instructions. The PIC can poll a flag to determine when the automatic erase/write programming cycle is over. As soon as this flag goes high, the EEPROM is ready for more instructions. Data stored in the EEPROM will be retained for 10 years or more, according to the manufacturer. The factor that determines the EEPROMs longevity in a particular application is the number of erase/write cycles. Depending on factors such as temperature and supply voltage, the EEPROM is good for 10,000 to 1 million erase/write cycles. This rules out its use as a substitute for ordinary RAM, since many PIC applications write to RAM thousands of times a second. At that rate, the EEPROM could be unusable in as little as 20 seconds! For a thorough discussion of EEPROM endurance, see the Microchip Embedded Control Handbook, publication number DS00092A, October 1992. How it works. The circuit in the figure specifies a 93LC56 EEPROM, but a 93C56 will work as well. The difference is that the LC device has a wider Vcc range (2.55.5 V, versus 45.5 V), lower current consumption (3 mA versus 4 mA), and can be somewhat slower in completing internal erase/ write operations, presumably at lower supply voltages. In general, the LC type is less expensive, and a better match for the operating characteristics of the PIC. The schematic shows the data in and data out (DI, DO) lines of the EEPROM connected together to a single PIC I/O pin. The 1k resistor prevents the PIC and DO from fighting over the bus during a read operation. During a read, the PIC sends an opcode and an address to the EEPROM. As soon as it has received the address, the EEPROM activates DO and puts a 0 on it. If the last bit of the address is a 1, the PIC
could end up sourcing current to ground through the EEPROM. The resistor limits the current to a reasonable level. APPS The program listing is a collection of subroutines for reading, writing, and erasing the EEPROM. All of these rely on Shout, a routine that shifts bits out to the EEPROM. To perform an EEPROM operation, the software loads the number of clock cycles into clocks and the data to be output into temp. It then calls Shout, which does the rest. If you dont have the EEPROM data handy (Microchip Data Book, DS00018D, 1991), you should know about a couple of subtleties. First, when the EEPROM powers up, it is write protected. You must call EEnable before trying to write or erase it. Its a good idea to call EEdisbl (disable writes) as soon as possible after youre done. Otherwise, a power glitch could alter the contents of your EEPROM. Also, you cannot write all locations (EEWrall) without first erasing all locations (EEwipe). Modifications. The table of constants at the beginning of the listing specifies the opcodes for each of the EEPROM operations. Although the opcodes are only three bits long, they are combined with a trailing dontcare bit. This bit is required for compatibility with the 512-byte 93LC66. With the 66, this would be address bit A8. If you want to modify this code for the 66, add a line to the read, write, and byte erase routines to copy A8 into bit 4 of temp just before calling Shout. If you want to increase capacity by adding more EEPROMs, you can bus the data and clock lines together and provide separate chip selects to each device. If you plan to run your PIC faster than 8 MHz, add one or two nops where marked in the listing. The clock must be high for at least 500 nanoseconds. The low time must also be greater than 500 ns, but the move-data, rotate, and looping instructions provide enough delay. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.

; ; ; ; ;
PROGRAM: EEPROM.SRC This program is a collection of subroutines for reading, writing, and erasing the 93LC56 (or 'C56) serial EEPROM. As a demonstration, it writes a scanning pattern to the 256 bytes of the EEPROM, and then reads it back to eight LEDs connected to port rb. = = = = = = = = = = ra.0 ra.1 ra.2 192 160 152 128 224 144 136 ; ; ; ; ; ; ; ; ; ; ; ; ; ; Pins DI and DO of the EEPROM Clock pin--data valid on rising edge Chip select--high = active Opcode for read Opcode for write Opcode to enable erase and write Opcode to disable erase and write Opcode to erase a byte Opcode to erase all bytes Opcode to write all bytes with specified data
D CLK CS ROP WROP EWEN EWDS ERAS ERAL WRAL
temp EEdata EEaddr clocks tick1 tick2
org ds ds ds ds ds ds
8 1 1 1 1 1 1
; Temporary variable for EE ; routines ; Passes data to EEwrite/wrall, ; from EEread ; Passes address to EErase, ; EEwrite, EEread ; Number of clock cycles for ; SHiftOUT ; Timer for Delay--not required for ; EE routines ; Timer for Delay--not required for ; EE routines
; Remember to change device info when programming part device pic16c54,xt_osc,wdt_off,protect_off reset start org 0 start mov ra,#0 ; Clear ports mov rb,#0 mov !ra,#0 ; Make all ra pins outputs initially mov !rb,#0 call EEnable ; Turn off write/erase protection mov EEdata,#1 mov EEaddr,#0 :loop call EEwrite ; Write scanning pattern to ; EEPROM

call rr ijnz call mov call mov inc call goto

Busy EEdata EEaddr,:loop EEdisbl EEaddr,#0 EEread rb,EEdata EEaddr delay :loop2
; Turn write/erase protection back ; on
APPS
:loop2
; Shift out the bits of temp to the EEPROM data line. Shout rl temp ; Rotate bit7 of temp into carry movb D,c ; Move carry bit to input of ; EEPROM setb CLK ; Clock the bit into EEPROM nop ; Clock must be high > 500 ns clrb CLK djnz clocks,Shout ret ; Read the byte in EEaddr into EEdata. EEread mov temp,#ROP mov setb call mov mov call mov mov setb movb rl clrb djnz mov mov clrb ret clocks,#4 CS Shout clocks,#8 temp,EEaddr Shout !ra,#1 clocks,#8 CLK c,D temp CLK clocks,:read EEdata,temp !ra,#0 CS
; Move the read opcode into temp ; Number of bits to shift out (op+1)
:read
; Call to unprotect after EEdisbl or power up. EEnable setb CS mov clocks,#12 mov temp,#EWEN call Shout clrb CS

ret ; Call to protect against accidental write/erasure. EEdisbl setb CS mov clocks,#12 mov temp,#EWDS call Shout clrb CS ret ; Write the byte in EEdata to EEaddr. EEwrite mov temp,#WROP mov clocks,#4 setb CS call Shout mov clocks,#8 mov temp,EEaddr call Shout mov clocks,#8 mov temp,EEdata call Shout clrb CS ret ; Erase the byte in EEaddr. Erasure leaves FFh (all 1s) in the byte. EErase mov temp,#ERAS mov clocks,#4 setb CS call Shout mov clocks,#8 mov temp,EEaddr call Shout ret ; Erase the entire EEPROM--all 256 bytes. Call before using EEwrall below. EEwipe setb CS mov temp,#ERAL mov clocks,#12 call Shout clrb CS ret ; Write the byte in EEdata EEwrall setb mov mov call mov mov to every address. Call EEwipe first. CS temp,#WRAL clocks,#12 Shout clocks,#8 temp,EEdata

call clrb ret Shout CS
APPS
; Check flag to determine whether the EEPROM has finished its self-timed erase/ ; write. Caution: This will lock up your program until D goes high. Busy nop mov !ra,#1 setb CS :wait jnb D,:wait clrb CS mov !ra,#0 ret Delay djnz djnz ret tick1,Delay tick2,Delay ; Delay routine for demo.

Introduction. This application note covers the use of dynamic RAM (DRAM) with PIC16C5x series PICs. It presents a circuit using a Motorola 1-Mb x 1 DRAM IC, and a program in TechTools assembly language showing how to refresh, write, and read the DRAMs 1,048,576 bits of storage.
+5 1k Reset RTCC VDD N/C Vss Function generator N/C RA0 RA1 RA2 RA3 RB0 RB1 RB2 RB3 RB4 MCLR OSC1 OSC2 RC7 RC6 2.2k RC5 Speaker
12: Using Dynamic RAM
Using Dynamic RAM
8MHz
Ceramic resonator w/ integral caps
PIC 16C55
RC4 RC3 RC2 RC1 RC0 RB7 RB6 RB5
D W RAS TF RB0 RB1 RB2 RB3 +5 A0
VSS Q CAS
MCM A9 511000A
A8
RC1 RC0 RB7 RB6 RB5
1Mb x 1 DRAM A7 A1
A2 A3 VCC 0.1F A6 A5 A4
RB4

Background. When a project requires a large amount of memory at the lowest possible cost and in the smallest available package, dynamic RAM (DRAM) is currently the only choice. Unfortunately, DRAM has a bad reputation for being difficult to work with. That reputation is built on the experiences of designers squeezing the greatest possible speed out of the large memory arrays used in general-purpose computers. In PIC applications, where speeds are more leisurely and 1 kByte is still a lot of memory, DRAMs can be downright mild-mannered. A review of the basic differences between DRAM and static RAM (SRAM) is in order. The basic storage unit in an SRAM is a flip-flop circuit made up of several transistors. Once set or cleared, the flip-flop remains in that state until set or cleared again, or until power is removed. This is a convenient form of temporary storage, since it requires no attention from the processor except when its being read or written. DRAM uses even simpler storage cells, consisting of just a capacitor and a transistor. The state of the cell, set or cleared, is stored as a charge on the capacitor. However, the capacitor inevitably leaks and loses its charge over time. When this happens, the stored data is lost. To prevent the loss of data, each cell in a DRAM must be periodically read and rewritten before its charge fades completely. This ensures that the storage capacitors always have a fresh charge, so the process is called refreshing the DRAM. The responsibility for refreshing the DRAM is shared between the external processor and circuitry inside the DRAM, as we will see later. SRAM and DRAM also use different addressing schemes. To access a particular cell of an SRAM, you place the bits of its address on the SRAMs address pins. The number of address pins must be sufficient for each cell to have a unique binary address. A 256-cell SRAM requires 8 address lines, since 8 binary digits are required to express 256 addresses (28 = 256). By the same token, a 64k SRAM needs 16 address lines (216 = 65,536) and a 1M SRAM, 20 address lines (220 = 1M). DRAMs use a two-dimensional addressing scheme. Each cell has two addresses, a row and a column. A 256-cell DRAM would require a 4-bit row address and a 4-bit column address (24 x 24 = 256). A 64k DRAM
APPS

would need 8 rows and 8 columns (28 x 28 = 65,536) and a 1M DRAM, 10 rows and 10 columns (210 x 210 = 1M). Therefore, a 1-Mb DRAM has a 10-bit wide address bus. To access a memory cell, you put the row address on the bus and activate the DRAMs row-address strobe (RAS), then put the column address onto the bus and activate the column-address strobe (CAS). At the expense of a two-step addressing processing, the number of address lines is cut in half. The RAS and CAS lines play an important role in the refresh process. DRAM manufacturers know that refresh is a pain in the neck, so they add features to make it easier. For example, just cycling through the DRAMs row addresses will satisfy refresh requirements. The DRAMs internal circuitry takes care of the rest. This is known as RAS-only refresh, because all the user has to do is put a sequential row address on the bus, and blip the RAS line. The MCM 511000A DRAM shown in the figure supports RAS-only refresh. Although it has 10 bits worth of row addresses, the DRAM requires that only the lower 9 bits (512 addresses) be accessed in RAS-only refresh. An even simpler refresh scheme, the one illustrated in the program listing, is known as CAS-before-RAS refresh. All the processor has to do in this case is activate CAS, activate RAS, and then deactivate both strobes. Each time it does this, it refreshes a row. In a normal access, RAS is activated before CAS, so reversing the order serves as a signal to the DRAM. When the DRAM sees this signal, it uses its own internal row-refresh counter to supply row addresses for the refresh process, ignoring the external address bus. The MCM 511000A DRAM requires refresh every 8 milliseconds, but it doesnt care whether it is performed all at once (burst refresh), or spread out over time (distributed refresh). Some applications dont require a separate refresh routine at all. Any application that accesses 512 row addresses every 8 milliseconds doesnt need any further refresh. If the application doesnt access all rows in the required time, but does read or write the DRAM 512 or more times every 8 milliseconds, a CAS-before-RAS refresh cycle can be

tacked to the end of each read or write. This is known as hidden refresh. Not all DRAMs support all refresh schemes, so its important to review the manufacturers specifications before finalizing a design. Data on the MCM 511000A for this application note came from the Motorola Memory Data book, DL113/D, Rev 6, 1990. How it works. The sample application shown in the figure and program listing accepts a stream of input bits, such as the output from a squarewave generator, and records them into sequential addresses in the DRAM. At the same time, it echoes these bits to a speaker. When the DRAM is full, the program plays back the recorded bits through the speaker until the circuit is shut off or reset. To demonstrate the circuit, reset the PIC and twiddle the tone-generator frequency knob. It will take about 52 seconds for the PIC to fill the DRAMs 1,048,576 bits of storage. When it does, you will hear the sequence of tones played back through the speaker. Because the playback routine requires fewer instructions than recording, the pitch of the tones will be slightly higher and the playback time shorter. The write loop takes up to 120 program cycles to execute; 60 microseconds when the PIC is operating from an 8-MHz clock. The exact number of cycles for a trip through the loop depends on how long the program spends in the subroutine inc_xy, which increments 10-bit row and column counters. At 60 microseconds per loop, the PIC samples the input pin approximately 17,000 times per second. If you use the circuit to record frequencies higher than half the sampling rate (approximately 8.5 kHz), youll hear some interesting squawks. The output tone will become lower as you adjust the input tone higher. What youre hearing is known as aliasing noise or foldover. The only way to faithfully reproduce an input frequency is to record two or more samples per cycle. You will hear this referred to as the Nyquist theorem or Nyquist limit. If you are interested in modifying this application to build a recording logic probe or tapeless audio recorder, keep Nyquist in mind. Modifications. This application will work fine (albeit at faster rates of sampling and playback) if you remove call refresh from the :record and APPS

:playback loops. The reason is that with call refresh removed the program accesses row addresses sequentially every 30 microseconds or so. This means that 512 rows are refreshed every 15 milliseconds. Although this exceeds the maximum refresh time allowed by the DRAM specification, DRAMs are rated conservatively. Further, you are unlikely to hear the data errors that may be occurring because of inadequate refresh!
If you want to organize data stored in the DRAM as bytes (or nibbles, 16bit words, etc.) instead of bits, consult the DRAM specs on fast page mode reads and writes. Using this method of accessing the DRAM, you issue the row address just once, then read or write multiple columns of the same row. This would be faster and simpler than eight calls to the single-bit read and write routines presented here. Program listing. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
; ; ; ;
PROGRAM: Dynamic RAM Basics (DRAM.SRC) Upon power up or reset, this program starts sampling pin ra.0 and recording its state in sequential addresses of a 1Mb dynamic RAM. When the DRAM is full, the program plays back the recorded bits in a continuous loop. = = = = = = = = = ra.0 ra.1 ra.2 ra.3 rb rc rc.2 rc.3 rc.5 ; Signal generator input ; DRAM row-address strobe ; DRAM write line (0 = write) ; DRAM data line ; Low bits of address bus ; High bits of address bus ; DRAM Q line ; DRAM column-address strobe ; Signal out to speaker
Sin RAS WR Dout adrb_lo adrb_hi Din CAS Sout
; Put variable storage above special-purpose registers. org 8 r_ctr row_lo row_hi col_lo ds ds ds ds 1 1 1 1 ; ; ; ; Refresh counter Eight LSBs of row address Two MSBs of row address Eight LSBs of column address

col_hi flags flag ds ds = 1 1 flags.0

; Two MSBs of column address ; Holder for bit variable flag ; Overflow flag for 20-bit address
APPS
; Remember to change device info when programming part. device pic16c55,xt_osc,wdt_off,protect_off reset start ; Set starting point in program ROM to zero org 0 start setb setb mov mov mov clr clr clr clr clr call call call call jnb :play call call call goto write (rb). mov AND OR clrb movb DRAM, movb mov AND OR (rc). clrb clrb setb setb WR CAS WR RAS ; ; ; ; Set up to write. Strobe in the column address. Conclude the transaction by restoring WR, RAS, & CAS high Sout,Sin adrb_lo,col_lo adrb_hi,#11111100b adrb_hi,col_hi ; and echo it to the speaker. ; Put LSBs of col addr on bus (rb). ; Clear bits adrb_hi.0 and .1. ; Put MSBs of col addr on bus RAS CAS !ra,#1 !rb,#0 !rc,#00000100b flags row_lo row_hi col_lo col_hi refresh refresh write inc_xy flag,:record refresh read inc_xy :play adrb_lo,row_lo adrb_hi,#11111100b adrb_hi,row_hi RAS Dout,Sin ; Disable RAS and CAS before ; setting ports to output. ; Make ra.0 (Sin) an input. ; Make rb (low addresses) output. ; Make rc.2 (Din) an input. ; Clear the variables.
; Initialize DRAM. ; Refresh the DRAM. ; Write Sin bit to DRAM. ; Increment row and col ; addresses. ; Repeat until address overflows. ; Refresh the DRAM. ; Retrieve bit and write to Sout) ; Increment row and col ; addresses. ; Loop until reset. ; Put LSBs of row addr on bus ; ; ; ; Clear bits adrb_hi.0 and .1. Put MSBs of row adr on bus (rc). Strobe in the row address. Supply the input bit to the
:record

setb ret read (rb). mov AND OR (rc) clrb mov AND OR (rc). clrb movb setb setb ret CAS Sout,Din RAS CAS ; Strobe in the column address. ; Copy the DRAM data to the ; speaker. ; Conclude transaction by ; restoring ; RAS and CAS high (inactive). RAS adrb_lo,col_lo adrb_hi,#11111100b adrb_hi,col_hi ; Strobe in the row address. ; Put LSBs of col addr on bus (rb). ; Clear bits adrb_hi.0 and .1. ; Put MSBs of col addr on bus CAS ; (inactive).
adrb_lo,row_lo
; Put LSBs of row addr on bus
adrb_hi,#11111100b ; Clear bits adrb_hi.0 and .1. adrb_hi,row_hi ; Put MSBs of row addr on bus
; This routine implements a CAS-before-RAS refresh. The DRAM has a counter to ; keep track of row addresses, so the status of the external address bus doesn't ; matter. The DRAM requires 512 rows be refreshed each 8 ms, so this routine must ; be called once every 125 microseconds. Changing the initial value moved into r_ctr ; will alter the refresh schedule. To refresh the entire DRAM, move #0 into r_ctr (256 ; loops) and call the routine twice in a row (512). refresh :loop mov clrb clrb setb setb djnz ret r_ctr,#8 CAS RAS CAS RAS r_ctr,:loop
; ; ; ; ;
Activate column strobe. Activate row strobe. Deactivate column strobe. Deactivate row strobe. Repeat.
; ; ; ; ;
This routine increments a 20-bit number representing the 1,048,576 addresses of the DRAM. The number is broken into two 10-bit numbers, which are each stored in a pair of byte variables. Note that wherever the routine relies on the carry bit to indicate a byte overflow, it uses the syntax "add variable,#1" not "inc variable." Inc does not set or clear the carry flag. add sc row_lo,#1 ; Add 1 to eight LSBs of row addr. ; If carry, add 1 to MSB, else ; return.
inc_xy

ret inc sb ret clr add sc ret inc sb ret clr reached setb ret flag

row_hi row_hi.2 row_hi col_lo,#1 ; ; ; ; Increment MSBs of row address. If we've overflowed 10 bits, clear row_ hi and add 1 to column address, else return.
APPS
; If carry, add 1 to MSBs, else ; return. col_hi col_hi.2 col_hi ; ; ; ; Increment MSBs of col address. If we've overflowed 10 bits, clear col_ hi and set the flag to signal main program that we've
; the end of memory, then return.

Introduction. Example programs show how to receive RS-232 serial data at 300 and 1200 baud with a PIC running at 32.768 kHz. Background. One of the PICs greatest virtues is its tremendous speed. With instruction cycles as short as 200 nanoseconds and an inherently efficient design, the PIC leaves comparably priced micros in the dust. This makes it easy to forget about the LP-series PICs, whose virtue is their ability to go very slow. At 32 kHz, these devices draw as little as 15A; good news for battery- or solar-powered applications. Life in the slow lane requires some slightly different techniques than multi-megahertz PIC design. First of all, youll need to get acquainted with the 32.768-kHz quartz crystal. This is the crystal of choice for LP PICs. Why 32.768 kHz? Its the standard timing reference for electronic watches. It resonates exactly 215 times per second, making it easy and inexpensive to build a chain of counters to generate 1-pulse-per-second ticks. Because of common use in watches, 32.768-kHz crystals are inexpensive (often less than a buck in single quantities) and accurate [20 parts per million (ppm), compared to the 50 ppm common in garden-variety clock crystals, or 3000 ppm of ceramic resonators). At a clock rate of 32.768 kHz, the PIC executes 8192 instructions per second, with instruction cycles taking about 122.1 microseconds each. Whether or not this is slow depends on your perspectivemany applications spend most of their time in delay loops anyway.
13: Running at 32 kHz
Running at 32 kHz
The first circuit (figure 1a) is normally constructed with a comparator, because of the fast switching requirements imposed by a clock oscillator. But for the pokey 32-kHz crystal, even an internally compensated opamp like the CA5160 works just fine. If you substitute another part for the 5160, you will have to play with the values of the feedback resistor (10k) and capacitor (0.001F) to get the circuit to oscillate. If you use a comparator with an open-collector output, dont forget to add a pullup resistor.

10k
The second oscillator (figure 1b) uses a CMOS inverter as its active element. In this case, the role of the inverter is played by one of the NOR gates of a 4001 chip. Adjusting the values of the 20-pF capacitors slightly will finetune the frequency of oscillation. Do not omit the 47-k resistor on the output. The capacitance of the OSC1 pin, or other connected logic (say a CMOS inverter or buffer) will pull the oscillator off frequency. It may even jump to a multiple of 32.768 kHz. This will throw your timing calculations way off.
CA5160
2
+5
0.001F
3
7 6 4
To Downloader OSC1 terminal
APPS
1a
1k 32,768-Hz XTAL
10M 1/4 4001 To Downloader OSC1 terminal 47k
1b
100k 32,768-Hz XTAL 20pF
20pF
With the help of one of these Figure 1. 32.768-kHz clocks. oscillators, you can have Downloader convenience in the development of LP applications. How it works. The application this time is a variation on application note number 2, receiving RS-232 data. The circuit shown in figure 2 receives RS-232 data at 300 or 1200 baud and displays it in binary form on eight LEDs connected to port rb. The baud rate depends on the program used. Listing 1 runs at 300 baud while listing 2 runs at 1200. The previous fast-clock RS-232 application used a counter variable to determine how many trips through a djnz loop the PIC should take. Each loop burns up three instruction cycles, so the best resolution possible with this type of delay is 3*(oscillator period/4). When the oscillator is running at 4 MHz, resolution is a respectable 3 microseconds. However, at 32.768 kHz, the resolution of a djnz loop is 366.3 microseconds! Theres another method that provides 1-instruction-cycle resolution: the

nop table. To use this approach, you create code like this:
mov jmp nop
RS-232 serial in
w,delay pc+w
;Put length of delay into w. ;Jump w nops into the table ;The nops. Each does nothing
22k
18
RA2
2
RA1
17
RA3
+5 3
RA0
16
32,768-Hz XTAL
15- to 30-pF caps
RTCC
4
OSC1
15
MCLR
5
Vss
6
PIC 16C54
OSC2
14
Vdd
13
+5 470 470 470 470 LED LED LED LED
RB0
7
RB7
12
RB1
8
RB6
11
RB2
9
RB5
10
RB3
RB4
470 470 470 470
LED LED LED LED
nop ... nop ; The program continues here.
;for 1 program cycle.
The number in w represents the number of nops to be skipped, so the larger the number the shorter the delay. If you need delays of varying sizes at several points in the program, set the table up as a callable subroutine. Dont forget to account for the program cycles used by call and return; two cycles each. For long delays, it would also make sense to use a two-step approach that creates most of the delay using loops and then pads the result with nops. Otherwise, you could end up filling most of your code space with nops.

The RS-232 reception routine in listing 1 uses nop-table delays of 0, 12, and 15 nops. If the program were expanded, other code could make use of the nop table, too. When writing such a program, make sure not to jump completely over the table! Listing 2 uses the same circuit to receive 1200-baud serial data. A bit delay at 1200 baud is 833.33 s, which is 6.8 instruction cycles at 32.768 kHz. Theres no instruction that offers a 0.8-cycle delay, so the routine gets bit 0 as early as possible after the start bit is detected (7 instruction cycles) and then uses a 7-cycle delay between subsequent bits. The program is written so that it samples the first bit early, and spreads the timing error over the rest of the reception of the byte. This approach is a little risky. It probably would produce errors if the line were noisy or the timing of the serial transmitters clock were off. Still, its a useful example of straight-line programming. In a real-world application, the serial transmitter would have to be set for 1.5 or 2 stop bits in order to give the program time to do anything useful with the received data. During the time afforded by an extra stop bit, the PIC could stuff the received by into a file register for later processing.
APPS
; ; ; ; ;
Listing 1: SLOW1.SRC (300-baud serial receive for 32-kHz clocks) This program receives a byte of serial data and displays it on eight LEDs connected to port rb. Special programming techniques (careful counting of instruction cycles, use of a nop table) allow a PIC running at 32.768kHz to receive data at 300 baud. = = = = = 14 3 0 ra.2 rb ; Executes 1 nop in table. ; Executes 12 nops in table. ; Executes 15 nops in table
half_bit bit stop serial_in data_out
; Variable storage above special-purpose registers. org 8 bit_cntr ds 1 ; number of received bits rcv_byte ds 1 ; the received byte ; Org 0 sets ROM origin to beginning for program. org 0 ; Remember to change device info if programming a different PIC.

device pic16c54,lp_osc,wdt_off,protect_off reset begin ; Set up I/O ports. begin mov !ra, #4 ; Use ra.2 for serial input. mov !rb, #0 ; Output to LEDs. :start_bit sb serial_in ; Detect start bit. jmp :start_bit ; No start bit? Keep watching. mov w,#half_bit ; Wait 1 nop (plus mov, call, and ; ret). call nop_delay ; Wait half bit to the middle of start ; bit. jnb Serial_in, :start_bit ; Continue if start bit good. mov bit_cntr, #8 ; Set counter to receive 8 data ; bits. :receive mov call movb rr djnz mov call mov goto nop_delay jmp nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop ret w,#bit nop_delay c,/Serial_in rcv_byte bit_cntr,:receive w,#stop nop_delay data_out, rcv_byte begin:start_bit pc+w ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; w w w w w w w w w w w w w w w w = = = = = = = = = = = = = = = = 0executes 15 nops. 1executes 14 nops. 2executes 13 nops. 3executes 12 nops. 4executes 11 nops. 5executes 10 nops. 6executes 9 nops. 7executes 8 nops. 8executes 7 nops. 9executes 6 nops. 10executes 5 nops. 11executes 4 nops. 12executes 3 nops. 13executes 2 nops. 14executes 1 nop. 15executes 0 nops ; Wait one bit time (12 nops). ; Put the data bit into carry. ; Rotate carry bit into the receive ; byte. ; Not eight bits yet? Get next bit. ; Wait 1 bit time (15 nops) for stop ; bit. ; Display data on LEDs. ; Receive next byte.
; If w > 15, the program will jump into unprogrammed code memory, causing a reset.
; ; ; ; ; ;
Listing 2: SLOW2.SRC (1200-baud serial receive for 32-kHz clocks) This program receives a byte of serial data and displays it on eight LEDs connected to port rb. Straight-line programming allows a PIC running at 32.768kHz to receive data at 1200 baud. Since timing is so critical to the operation of this program, the number of instruction cycles required for each instruction appears in () at the beginning of most comments. = = ra.2 rb ; RS-232 via a 22k resistor. ; LED anodes.
APPS
serial_in data_out
; Variable storage above special-purpose registers. org 8 rcv_byte ds 1 ; The received byte. ; Org 0 sets ROM origin to beginning for program. org 0 ; Remember to change device info if programming a different PIC. device pic16c54,lp_osc,wdt_off,protect_off reset begin ; Set up I/O ports. begin mov !ra, #4 ; Use ra.2 for serial input. mov !rb, #0 ; Output to LEDs. :start_bit sb jmp nop sb jmp nop nop movb nop nop nop movb nop nop nop movb nop nop nop movb nop nop serial_in :start_bit serial_in :start_bit ; (2) Detect start bit. ; (2) No start bit? Keep watching. ; (1) ; (2) Confirm start bit. ; (2) False alarm? back to loop. ; (1) ; (1) ; (4) Get bit 0. ; (1) ; (1) ; (1) ; (4) Get bit 1. ; (1) ; (1) ; (1) ; (4) Get bit 2. ; (1) ; (1) ; (1) ; (4) Get bit 3. ; (1) ; (1)
rcv_byte.0,/serial_in

nop movb nop nop nop movb nop nop nop movb nop nop nop movb mov jmp ; (1) rcv_byte.4,/serial_in ; (4) Get bit 4. ; (1) ; (1) ; (1) rcv_byte.5,/serial_in ; (4) Get bit 5. ; (1) ; (1) ; (1) rcv_byte.6,/serial_in ; (4) Get bit 6. ; (1) ; (1) ; (1) rcv_byte.7,/serial_in ; (4) Get bit 7. data_out,rcv_byte ; (2) Data to LEDs. :start_bit ; (2) Do it again
14: Generating/mixing Sine Waves
Generating/mixing Sine Waves
Introduction. This application note illustrates methods for generating audio-frequency waveforms using lookup tables. APPS Background. Digitally generating sine waves or other complex waveforms may seem like wasted effort compared to using a few discrete components or an IC. If the goal is just to make a pure tone, that view is probably true. However, if you need precise control over the frequency, duration and phase of a waveform, digital generation becomes downright cheap and simple compared to the analog alternatives. The basic procedure for generating a sine wave is easy enough: At precise time intervals, the program looks up a value from a table, delivers it to a D/A converter, then waits another interval to put out the next value. At the output of the D/A converter, the signal is a connect-the-dots version of a sine wave. The addition of a low-pass filter smoothes the jaggies, and we have a real sine wave. Other waveforms can be made in a similar way, if their patterns repeat within a time interval that can be stored in a reasonably sized lookup table, or computed on-the-fly. If they cannot, then they must be synthesized by combining multiple sine waves. The program in the listing demonstrates both techniques. When the frequency constants freq1 and freq2 are set to the same value, the circuits output is a pure sine wave. Thats because the two pointers phase1 and phase2 move through the table of sine values at the same rate. However, if the freq constants are different, the circuit produces a mixture of two sine waves.
RA2 +5 RA3 RTCC MCLR RA1 RA0 OSC1 OSC2 8-MHz ceramic resonator w/integral caps
PIC
Vss 16C54 VDD RB0 RB1 RB2 RB3 RB7 RB6 RB5 RB4
+5 20k 20k 20k 20k 20k 20k 20k 20k
Output to audio amplifier or O scope 10k 10k 10k 10k 10k 10k 10k 10k 20k 0.001F
How it works. The resistors in figure 1 are connected in an array known as an R-2R ladder. This arrangement has an interesting property. It divides each voltage present at its inputs by increasing powers of two, and presents the sum of all these divided voltages at its output. Since the PIC is capable of driving its outputs from rail to rail (0 to +5 volts), the R2R ladder converts the binary number present on port B into a proportional voltage from 0 to 5 volts in steps of approximately 20 millivolts. Most commercial D/A converters work on this same principle, but have internal voltage regulators, logic, and latches. Since our demonstration doesnt require any of those things, we can use the resistor array alone. Unfortunately, it can be difficult to find a prefabricated R-2R ladder at the volume parts houses. Figure 2 shows how to construct one from 25 surface-mount 10k resistors and a tiny printed circuit pattern. Alternatively, you may adapt the program listing to drive your favorite packaged D/A converter.
analog voltage out
RB7
RB0
Actual-size pcb pattern
Parts placement
Figure 2. Constructing an R-2R digital-to-analog converter. The program that drives the D/A converter begins by setting port rb to output and clearing its variables. Next it enters this loop: Add freq1 to acc1 > If acc1 overflows (generates a carry), increment phase1 Put the sine value corresponding to phase1 into temp Add freq2 to acc2 > If acc2 overflows (generates a carry), increment phase2 Add the sine value corresponding to phase2 into temp Copy temp to the output Do it again.
The loop always takes the same amount of time to execute, so how do freq1 and freq2 control the frequency? If the value of freq1 is 1 and acc1 starts out at 0, the program will complete 255 loops before acc1 overflows, generating a carry. Phase1 will increment once in every 256 loops. This will produce a low output frequency. If freq1 is 100, acc1 will overflow on the third trip through the loop, so phase1 will increment much faster, producing a higher frequency. This scheme, while compact, is not perfect. Take that example of freq1 = 100. If acc1 starts off at 0, the sequence goes as shown in the table to the right. The rate at which carries occur is, on average, proportional to freq1. However, the interval between carries can be 1, 2, or 3 trips through the loop. At higher frequencies, these variations become noticeable as a thickening or jittering of traces on the oscilloscope screen. If your application requires particularly pure output, keep this in mind. Modifications. Play with the values of freq1 and freq2 and observe the results on the oscilloscope. Compare the results to a textbook discussion of frequency mixing. acc1 carry
0 100 200 44 144 244 88 188 32 132 232 76 176 20 0 0 0 1 0 0 1 0 1 0 0 1 0 1
APPS
Try removing the resistor/capacitor low-pass filter from the output of the R-2R ladder and look at the change in the output waveform. If you are listening to the output through an audio amplifier, you may find that you prefer the sound of the unfiltered D/A converter. The harmonics it produces give the tone more sparkle, according to some listeners. Perhaps the most useful modification is this: Set freq1 and freq2 to the same value, say 100. Run the program and listen to the tone or observe it on the scope. If you use a scope, make note of the peak-to-peak amplitude of the tone. Now substitute mov phase1, #7 for clr phase1 at the beginning of the program. Run the program again. Do you hear (see) how the amplitude of the tone has fallen off? Try once more with mov phase1,#8. No tone at all. This is a graphic illustration of how out-ofphase sine waves of the same frequency progressively cancel each
other. The 16 sine table entries represent amplitudes at intervals of 22.5 degrees. When phase1 is initialized with a value of 8, while phase2 is zero, the phase shift between the two is 22.5 x 8 = 180 degrees. Two sine waves 180 degrees out of phase cancel each other totally. Smaller phase shifts produce proportionally less attenuation. Given the PICs limited math capabilities, this phase shifting technique is an easy way to control the overall amplitude (volume) of the tones you generate. Program listings. This program may be downloaded from our Internet ftp site at ftp.tech-tools.com. The ftp site may be accessed directly or through our web site at http://www.tech-tools.com.
; Program: Synthesizing sine waves (SINE.SRC) ; This program mixes two sinewave signals and outputs them to port b for ; conversion to analog by an R-2R ladder D/A converter. output freq1 freq2 = = = rb 100 50 ; Ladder DAC on port b. ; Sets frequency 1. ; Sets frequency 2.
; Set aside variable space above special-purpose registers. org 8 acc1 ds 1 ; Accumulator for phase 1. acc2 ds 1 ; Accumulator for phase 2. phase1 ds 1 ; Position in sine table, p1. phase2 ds 1 ; Position in sine table, p2. temp ds 1 ; Temporary variable. ; Device data and reset vector (remember to change if programming a different part). device pic16c54,xt_osc,wdt_off,protect_off reset start org 0 start mov clr clr clr clr add addb AND !rb,#0 acc1 acc2 phase1 phase2 acc1,#freq1 phase1,c phase1,#00001111b ; Make rb an output. ; Clear the accumulators. ; Clear phase pointers. ; Add freq1 to acc1 ; If carry, increment phase1. ; Limit to 0-15.
:loop

add addb AND mov call mov mov call add mov goto

acc2,#freq2 phase2,c phase2,#00001111b w,phase1 sine temp,w w,phase2 sine temp,w output,temp :loop ; Do the same for ; phase2. ; Look up sine value in table. ; Store 1st sine value in temp. ; ; ; ; Get value from table. Add sines together. Output to D/A. Do forever.
APPS
; Notice that the sine values in the table below are scaled so that no pair adds to ; more than 254. This prevents overflowing the variable temp when the phases are ; added together. If you add more phases, adjust the maximum values in this table ; appropriately. The values are in increments of 22.5 degrees (0.3927 radians). sine jmp pc+w retw 64,88,109,122,127,122,109 retw 88,64,39,19,5,0,5,19,39

Introduction. This application note shows how to use the interrupt capabilities built into the PIC 16Cxx series controllers with a simple example in TechTools assembly language. Background. Many controller applications work like a fire department. When theres no fire, they mend hoses, polish the fire truck, and wait for something to happen. When the fire bell rings, they swing into action and handle the emergency. When its over, they return to waiting. If there were no fire bell, the job would be much different. The fire crew would have to go out and look for fires, and somehow still make sure that the hoses were fixed and the truck maintained. In their scurry to get everything done, they might respond late to some fires, and miss others completely. The newer PIC 16Cxx controllers have interrupts, which work like the fire bell in the first example. When an interrupt occurs, the PIC stops what its doing and handles the cause of the interrupt. When its done, the PIC returns to the point in the program at which it was interrupted. The second example, with no fire bell, is an example of polling. This is the approach used with the 16C5x PICs, which lack interrupts. For a fast PIC, polling isnt nearly as bad as in the example. The PIC can often get everything done with plenty of time to spare. But there are times when interrupts are the simplest way to do two or more things at once. How interrupts work. For the examples here, were going to talk about the 16C84. The same principles apply to the other interrupt-capable PICs, but registers and interrupt sources may vary. First of all, the 16Cxx PICs awaken from power-up with all interrupts disabled. If you dont want to use interrupts, dont enable them. Its that simple. The PIC 16C84 can respond to interrupts from four sources: A rising or falling edge (your choice) on pin RB0/INT. Changing inputs to pins RB4 through RB7.
15: Using Interrupts
Using Interrupts

Timer (RTCC) overflow from 0FFh to 0. Completion of the data EEPROM programming cycle. APPS You can enable any combination of these sources. If you enable more than one, it will be up to your code to determine which interrupt occurred and respond appropriately. More on that later. Lets take a simple example. We want to use the RTCC interrupt to generate a steady 1-kHz square wave on pin ra.1. Every 500 s the RTCC will interrupt whatever the PIC is currently doing, toggle ra.1, reload the RTCC with an appropriate value and return. The first consideration is where to put the interrupt-handler code. When the 16C84 responds to an interrupt, it jumps to location 04h in program memory. So well use an org statement to place the handler at 04h. If you look at the program listing, youll see that we actually have two orgs: the first positions the code jmp start at 0, which is where the 84 looks for its power-on startup code, and the second positions the interrupt handler at 4. The programs startup code configures the RTCC for its initial 500-s timing period, and enables the interrupt. To do so, it turns on the first two bit switches shown in figure 1. For any interrupt to occur, the globalinterrupt enable (GIE) bit must be set. For the timer interrupt to occur, the RTCC interrupt enable (RTIE) bit must be set. INTERRUPT Once those two Global Timer Timer Interrupt Interrupt Interrupt switches are closed, Enable Enable Flag only one switch re(GIE) (RTIE) (RTIF) mains before the interrupt alarm bell goes offthe RTCC interrupt flag (RTIF). When the interrupt occurs, the PIC clears GIE to disable further interrupts, pushes the program counter onto
Figure 1. Logic of the interrupt-enable and flag bits for the RTCC.

the stack, then jumps to location 4. This process takes four instruction cycles for an internal interrupt like the RTCC, five for an external interrupt like RB0/INT. Once at location 4, the PIC executes the code there until it encounters a ret, retw, or reti instruction. Any of these will pop the value from the top of the stack and cause a jump back to the point at which the program was interrupted. However, the normal way to return from an interrupt is the reti instruction. It automatically re-enables interrupts by setting GIE. The other two returns do not. In the program listing, you will notice that the first thing the interrupt handler does is clear RTIF. Whenever the RTCC rolls over from 0FFh to 0, the PIC automatically sets RTIF. It does this regardless of the state of the interrupt bits. And it never turns RTIF back off. So its the responsibility of the interrupt handler to clear RTIF. This is true of all of the interrupt flags. If the handler doesnt clear the appropriate flag, the same interrupt will occur again as soon as interrupts are re-enabled. The resulting endless loop is what the Microchip documentation calls recursive interrupts. With a single interrupt source, you can think of an interrupt as a hardware version of the call instruction. The primary difference is that this kind of call can occur anywhere in your program, whether or not the program is ready for it. This can pose a problem. Lets say your program is executing the TechTools instruction add sum,delta when the interrupt occurs. That add instruction actually consists of two instructions; one that loads the variable delta into the w register, and a second that adds w to sum. Imagine that right after delta is loaded into w, the interrupt takes over. It performs, for instance, the instruction mov rb,data. This instruction loads the variable data into w, then moves w into rb. When its done, w contains a copy of data. When the handler returns, the PIC completes the interrupted addition. But w contains data, not the intended delta, causing an error. There are two ways to prevent this. One is to disable interrupts before

a two-part instruction, then enable them again afterwards. If the interrupt event occurs while interrupts are disabled, the PIC will set the interrupt flag, but wont jump to the handler until your program re-enables interrupts. The PIC will not miss the interrupt, it will just be slightly delayed in handling it. To use this method, the example above would be changed to: clrb add setb GIE sum,delta GIE ; Interrupts disabled. ; w = delta: sum = sum+w. ; Interrupts enabled.
APPS
The alternative approach is to begin your interrupt handler with code that saves a copy of the w register. Then, just before reti, move the copy back into w. This takes care of compound instructions that use w, but leaves another group of instructions vulnerable; the ones that use the status register. Conditional jumps and skips like jump-if-zero (jz) and compare-andjump-if-equal (cje) and their many cousins are probably obvious. More subtle are the rotate instructions rr and rl (which pull the carry bit into the most- or least-significant bit of the byte being rotated). If the interrupt handler affects any of the status bits, it may alter the outcome of the interrupted instruction. To protect w and status, you must make copies of them at the beginning of a handler, then restore those copies at the end. In order to prevent the action of restoring w from affecting status, you can use a sneaky trick. Moving a file register into w will set or clear the zero bit, but moving a nibble-swapped copy of the register into w does not. Neither does swapping the nibbles of a file register. The program listing shows how to use these loopholes to accurately copy both w and status. Keep your interrupt handlers as short and simple as possible. Examine them carefully for their effect on other portions of the program that might be executing at the time the interrupt occurs. If necessary, protect sensitive code by bracketing it with instructions that temporarily disable interrupts. Also, keep in mind that test running your program may not catch all possible interrupt-induced bugs. Use the PSIM simulator and a sharp eye to detect potential problems.

Once you understand how the mechanism works with a single interrupt source, youll be relieved to know that its not that much more difficult to handle multiple interrupt sources. Figure 2 shows the alarm-bell diagram for multiple interrupts on the 16C84. If you have all of the 16C84s interrupts enabled, your handler at 04h should begin with something like: jb jb jb jb INTF, rb0_edge RBIF, rb_change RTIF, timeout EEIF, EE_wr_done ; If INTF then handle rb.0 interrupt. ; If RBIF then handle change on rb.4-7. ; If RTIF, then handle timer rollover ; If EEIF, then handle write complete.
Code at each of those labeled locations (rb0_edge, etc.) would then deal with that particular type of interrupt. Remember that each of the handlers must clear its corresponding flag; for example, rb0_edge must include EEIF EEIE the instruction clrb INTF. What happens if an interrupt occurs while the PIC is already handling an interrupt? At that time, nothing. Remember that the PIC clears GIE automatically in response to an interrupt, then sets it when rti executes. Any interrupt event that occurs in the meantime will set the appropriate flag. That interrupt will be delayed until after the current one is finished.
RTIE RTIF
INTERRUPT
RBIE GIE INTE
RBIF
INTF
INTE/F = RB0 interrupt enable/flag bits RBIE/F = RB47 change interrupt enable/flag bits RTIE/F = RTCC overflow interrupt enable/flag bits EEIE/F = EEPROM write complete interrupt enable/flag bits
Figure 2. Logic of all four 16C84 interrupts.
Program design. Interrupts are not a cure-all for the difficulties of handling multiple tasks. In fact, you may just end up trading a difficult

programming job for an even more difficult debugging job. Experts suggest using interrupts only as a last resort. APPS
; ; ; ; ; Program: INTRPT.SRC (Interrupt demonstration) This program illustrates the use of the RTCC interrupt in the PIC 16C84. The foreground program blinks an LED on pin ra.0, while an interrupt task outputs a 1-kHz square wave through pin ra.1 (assuming a 4-MHz clock).
; Device (16c84) and setup options. device pic16c84,xt_osc,wdt_off,pwrt_off,protect_off ; Equates for LED, tone pins. Connect the LED through a ; 220-ohm resistor. Connect a speaker or earphone through ; a 1-k resistor. LED = ra.0 SPKR = ra.1 tone = 6 ; Load into RTCC for 500-us delay. ; Allocate space for some variables. Notice that in the 84 ; variable start at 0Chhigher than in the 16C5x series. org 0Ch w_copy ds 1 s_copy ds 1 countL ds 1 countH ds 1 ; ; ; ; On startup, the PIC looks at address 0 for its first instruction. Since the interrupt handler begins at address 4, well just jump over it to get to the startup routine. org 0 jmp start ; Beginning of main program.
; ; ; ; ; ; ;
Next is the interrupt handler, which must begin at address 4. This handler copies restores w and the status register. Because a normal mov w,fr alters the z bit of the status register, this routine uses mov w,<>fr, which does not. The routine actually swaps the byte twice, resulting in the correct value being written to w without affecting the z bit. org 4 handler clrb RTIF ; Clear the timer interrupt flag. mov w_copy,w ; Make a copy of w. mov s_copy,status ; Make a copy of status.

XOR mov mov swap mov reti ; ; ; ; ra,#2 rtcc,#tone status,s_copy w_copy w,<>w_copy ; Toggle bit ra.1. ; Reload rtcc for 500-us delay. ; Restore status register ; Prepare for swapped move. ; Swap/move to w, status unaffected. ; Return to main program.
Heres the startup routine and the main program loop. In the line that initializes intcon, bit 7 is GIE and bit 5 is RTIE. Writing 1s to these enables interrupts generally (GIE) and the RTCC interrupt specifically (RTIE).
Start mov setb clr mov clrb mov ; ; ; ; ; ; ; ; !ra,#0 ; Make ra pins outputs. rp0 ; Switch to register page 1. wdt ; Assign prescaler to rtcc. option,#0 ; Set prescaler to divide by 2. rp0 ; Restore to register page 1. intcon, #10100000b ; Set up RTCC interrupt.
If the interrupt handler were to alter w, the LED would stop flashing or flash erratically, since the routine is written to rely on the value of w remaining 1 in order to toggle bit ra.0. The routine also relies on reliable operation of the status register because of the two skip-if-not-zero (snz) instructions. Although this structure is a little strange, its an effective canary-in-a-coalmine demonstration that the interrupt handlers save/restore instructions do preserve both w and status. mov inc snz inc snz XOR jmp w,#1 countL countH ra,w :loop ; ; ; ; ; ; Bit in 0 position to toggle ra.0. countL=countL + 1 IF countL=0, THEN countH=countH+1 IF countH=0, THEN toggle LED.
:loop
APPS

Pic App

Uploaded by

Copyright:

Available Formats

Pic App

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pic App

Uploaded by

Copyright:

Available Formats

PIC Application Notes

TechTools PIC Application Notes Page 1

PIC Application Notes

Page 2 PIC Application Notes TechTools

PIC Application Notes

TechTools PIC Application Notes Page 3

PIC Application Notes

Page 4 PIC Application Notes TechTools

1: Interfacing a Matrix Keypad

PIC Application Notes

Interfacing a Matrix Keypad

TechTools PIC Application Notes Page 5

PIC Application Notes

1: Interfacing a Matrix Keypad

RA3 RTCC MCLR Vss RB0 RB1 RB2 RB3

PIC OSC2 16C54 Vdd RC/P

10k 10k 10k 10k

Page 6 PIC Application Notes TechTools

1: Interfacing a Matrix Keypad

PIC Application Notes

; cols out, rows in ; ra all outputs

TechTools PIC Application Notes Page 7

PIC Application Notes

1: Interfacing a Matrix Keypad

Page 8 PIC Application Notes TechTools

2: Receiving RS-232 Serial Data

PIC Application Notes

Receiving RS-232 Serial Data

TechTools PIC Application Notes Page 9

PIC Application Notes

2: Receiving RS-232 Serial Data

Page 10 PIC Application Notes TechTools

2: Receiving RS-232 Serial Data

PIC Application Notes

TechTools PIC Application Notes Page 11

PIC Application Notes

2: Receiving RS-232 Serial Data

1/4 MC1489A: pins 1, 4, 7, 10 to gnd; pin 14 to +5 Vdc; rest not connected.

ceramic resonator w/integral capacitors

RB0 RB1 RB2

LED LED LED LED

LED LED LED LED

Figure 3. Schematic to accompany RCV232.SRC.

Page 12 PIC Application Notes TechTools

2: Receiving RS-232 Serial Data

PIC Application Notes

For example, at 1200 baud the bit delay

. At 2 MHz, the time per instruction cycle is

TechTools PIC Application Notes Page 13

PIC Application Notes

2: Receiving RS-232 Serial Data

djnz call mov goto ; ; ; ;

bit_cntr,:receive bit_delay data_out, rcv_byte begin:start_bit

Page 14 PIC Application Notes TechTools

2: Receiving RS-232 Serial Data

PIC Application Notes

BASIC Program for Transmitting Bytes via COM1

TechTools PIC Application Notes Page 15

PIC Application Notes

3: Sending RS-232 Serial Data