WORD bug

[yettie]

to all,

use of word in a for/next statement.

	init:
		decideInterval		var		WORD
		TIMEINTERVAL		con		600
	main:
		IF decideInterval = TIMEINTERVAL then
			decideInterval = 0 
			(any code methods)
		ENDIF
		GOTO	MAIN
	
		END

It satisfies the IF condition unexpectedly. Changing the decideInterval type to BYTE and of course TIMEINTERVAL to max 255(or any 8Bit arbitrary value) will work correctly. Im using a PIC16F628 running 8MHz crystal oscillator. any ideas on this? thanks very much

regards,
yettie

[Bruce]

This is no BUG. Uninitialized variables will contain random values.
Try assigning a value to your word variable, and then see how it works.

init:
	decideInterval var WORD
	TIMEINTERVAL con 600

main:
	IF decideInterval = TIMEINTERVAL then
	   decideInterval = 0
	   HIGH 0   ' LED on  RB0 ----|>|---/\/\/----gnd
	ELSE
	   decideInterval = 600
	   LOW 0  ' LED off
	ENDIF
	pause 500
	GOTO   MAIN

	END

If your LED blinks on/off, it's working properly.

[yettie]

dear bruce,

i failed to initialized the WORD but in my program. also my bad, this is my code.

        init:
		decideInterval		var		WORD
		TIMEINTERVAL		con		600
                decideInterval = 0

	main:
                decideInterval = decideInterval + 1
		IF decideInterval = TIMEINTERVAL then
			decideInterval = 0                           ;resets the decideInteval
			(a command output to LCD display)
                ELSE
		ENDIF
		GOTO	MAIN
	
		END

i verified it that it enters the IF/THEN statement with the decideInterval = 11 or other pseudo-randomly values. sorry for the incomplete info a while ago. but this post seems to be complete now.

thanks

regards,
yettie

[Bruce]

Sorry, but I still do not understand why you think this is a bug?

With the PIC operating at 8MHz, you have an execution speed of 2 million single cycle instructions per second. That's fast..!

If you increment decideInterval on each pass through your loop, of course it will = 600 at some point. And very quickly.

Here's another example;

init:
      decideInterval  var  WORD
      TIMEINTERVAL	  con  600
      decideInterval = 0

main:
      decideInterval = decideInterval + 1
      IF decideInterval = TIMEINTERVAL then
          decideInterval = 0
          GOTO Match
      ENDIF
      GOTO  MAIN

Match: ' End up here in approximately 11.4mS
      GOTO Match
	
		END

From Main, through the loop once, then back to Main at 8MHz only takes around 19uS. Now take 19uS * 600 for ~11.4mS.

It takes only ~11.4mS before you increment decideInterval from 0 to 600, and then the IF/THEN statement evaluates as TRUE, and program flow jumps to Match.

Note: I verified this (PBP v2.46). It does not jump to match at 11. Only once
decideInterval = 600.

[yettie]

bruce,

to make things clear. i've written it in a program to have a 31.25ms timer at a prescale of 1:8,yielding a 250ms to overflow TIMER1. i verified it in a oscilloscope that it is accurate.

define  debug_reg    	porta		'nevermind my serial LCD display debugger
define  debug_bit   	0
define  debug_baud   	2400
define  debug_mode   	1

define  inthand     	checkNow
 
define OSC 8
CMCON=7

time	var	word
timeL	var	time.byte0
timeH	var	time.byte1
decideTimeInterval	var	word
TIMEINTERVAL    	con       600

;-----------------------
setup:
  debug   12,"Starting",10,13,"   Program",4
  pause   2000
  
  intcon  = %11000000   ;setting up gie and peie bits
  time    = 3036          	;calibrated to have 31.25ms scale
  tmr1L   = timeL         	;65536 - time (preload value)
  tmr1H   = timeH

  flag    	= 0
  
  PIR1.0  = 0
  PIE1.0  = 1

  T1CON   = %110001    ;setting up to 1:8 prescale
                          	;31.25ms x 8 = 250ms/count
;--------------------------------
main:
  if flag = 4 then        	;(250ms x 4 = 1s) second ticker
     flag = 0
     decideTimeInterval = decideTimeInterval + 1
     gosub  updateTime
  endif

  if decideTimeInterval = TIMEINTERVAL then		;this one is the notorious routine. it's just a reset condition.
     decideTimeInterval = 0				;It resets unexpectedly.  In contrast to this,
  endif							;changing the decideTimeInterval to BYTE and of course
							;TIMEINTERVAL <= 255 will function properly


  goto main
;----------------------------
updateTime:
  DEBUG "Time:",#decideTimeInterVal
  return
       


;------------------------
;interrupt handler     
@checkNow            
  pie1.0  = 0
  flag    = flag + 1
  time    = 3036
  toggle = portb.3		;my monitor to oscilloscope, it accurately toggles at 250ms cycle
  TMR1L   = timeL
  TMR1H   = timeH
  PIR1.0  = 0
  PIE1.0  = 1
  resume

i hope now that everything clear for you now. the decideTimeInterval counter now increments every second avoiding fast incrementing, like you said.

any ideas?

thanks and I appreciate very much

regards,
yettie

[Bruce]

Hi Yettie,

I couldn't get your code to compile without errors. Try this;

@ DEVICE MCLR_OFF,LVP_OFF,HS_OSC,WDT_OFF
DEFINE  OSC 8
define  debug_reg    	porta		'nevermind my serial LCD display debugger
define  debug_bit   	0
define  debug_baud   	2400
define  debug_mode   	1

define  inthand     	checkNow
 
CMCON=7
TRISB.3=0

'Variables for saving state in interrupt handler
wsave	           VAR	BYTE $70 system		' Saves W
ssave	           VAR	BYTE bank0 system	' Saves STATUS
psave	           VAR	BYTE bank0 system	' Saves PCLATH
fsave	           VAR	BYTE bank0 system	' Saves FSR
time	           var	word
flag               var  byte
decideTimeInterval var	word
TIMEINTERVAL       con  600

;-----------------------
setup:
  debug   12,"Starting",10,13,"   Program",4
  pause   2000
  
  intcon  = %11000000   ;setting up gie and peie bits
  time    = 3036          	;calibrated to have 31.25ms scale
  tmr1H   = time.highbyte
  tmr1L   = time.lowbyte   	;65536 - time (preload value)
  flag    = 0
  decideTimeInterval = 0
  PIR1.0  = 0
  PIE1.0  = 1

  T1CON   = %00110001    ;setting up to 1:8 prescale
                          	;31.25ms x 8 = 250ms/count
;--------------------------------
main:
  if flag = 4 then        	;(250ms x 4 = 1s) second ticker
     flag = 0
     decideTimeInterval = decideTimeInterval + 1
     gosub  updateTime
  endif

  if decideTimeInterval = TIMEINTERVAL then		;this one is the notorious routine. it's just a reset condition.
     decideTimeInterval = 0				;It resets unexpectedly.  In contrast to this,
  endif							;changing the decideTimeInterval to BYTE and of course
							;TIMEINTERVAL <= 255 will function properly
  goto main
;----------------------------
updateTime:
  DEBUG "Time:",#decideTimeInterVal
  return
;------------------------
;interrupt handler
asm
checkNow
    ; Save FSR, PCLATH, STATUS and W registers
    movwf	 wsave          ; Save W
    swapf	STATUS, W   ; Swap STATUS to W (swap avoids changing STATUS)
    clrf	STATUS	     ; Clear STATUS
    movwf	ssave	     ; Save swapped STATUS
    movf	PCLATH, W   ; Move PCLATH to W
    movwf psave	     ; Save PCLATH
	
	; isr routine
    bsf STATUS,RP0  ; bank 1
    bcf PIE1,0
    bcf STATUS,RP0  ; bank 0
    incf _flag,f    ; increment flag
    movlw 0x8
    xorwf PORTB,f   ; toggle portb.3
    movf _time,W
    movwf TMR1L
    movf _time+1,W
    movwf TMR1H 
    bcf PIR1,0
    bsf STATUS,RP0  ; bank 1
    bsf PIE1,0
    bcf STATUS,RP0  ; bank 0
    
    ; Restore FSR, PCLATH, STATUS and W registers
    movf	fsave, W      ; retrieve FSR value
    movwf	FSR	     ; Restore it to FSR
    movf	psave, W	     ; Retrieve PCLATH value
    movwf	PCLATH	     ; Restore it to PCLATH
    swapf	ssave, W	     ; Retrieve the swapped STATUS value (swap to avoid changing STATUS)
    movwf	STATUS	     ; Restore it to STATUS
    swapf	wsave, F	     ; Swap the stored W value
    swapf	wsave, W     ; Restore it to W (swap to avoid changing STATUS)
    retfie		     ; Return from the interrupt    
endasm      

    end

Look in the manual, and at some of the MeLabs examples for assembler interrupts to find out why it works.