My project is starting

Printable View

Show 40 post(s) from this thread on one page

- 2nd December 2010, 03:02
cncmachineguy

I have a problem, not sure where it is. My gut feeling is its with the background things that may be happing when using DT-INT. Here is the entire program.

Code:

' Name : Interface card.pbp ' Compiler : PICBASIC PRO Compiler 2.6A ' Assembler : MPASM ' Target PIC : 16F1947 ' Hardware : PIMill Interface ' Oscillator : internal ' Keywords : ' Description : PICBASIC PRO program to interface parallel port ' to 4 stepper drives and limit switches ' ' ' ' ' Configure ports ' ' ANSELA = %00000000 'port A all digital ANSELE = %00000000 'port E all digital ANSELF = %00000000 'port F all digital ANSELG = %00000000 'port G all digital CM1CON0 = %00000000 'No comparaters CM1CON1 = %00000000 CM2CON0 = %00000000 CM2CON1 = %00000000 ' preload latches before setting direction LATB = %00000000 LATC = %00000000 LATD = %00010000 'LED's 0 for red, 1 FOR GREEN should start red LATF = %00001111 ' low nibble is drive enable. 1 disables LATG = %00000000 ' I/O direction TRISA = %11111111'step and dir inputs TRISB = %11000001'7-0 icsp,icsp,h4o,h3o,h2o,h1o,estopO,enable In TRISC = %00000000'step and dir outputs TRISD = %00000000'led6,i2c,i2c,led5,led4,led3,led2,led1 TRISE = %11111111'ls4,lsh4.ls3,lsh3,ls2,lsh2,ls1,lsh1 TRISF = %00010000'spindle dir,servo out,dac out,flood input,4en,3en,2en,1en TRISG = %11100111'x,x,icsp/mclr,rly2,rly1,Rx2,Tx2,DI 'REMOVE THIS LINE IF NOT SETTING THE CONFIGS IN CODE SPACE @ __config _CONFIG1,_FOSC_INTOSC & _WDTE_OFF & _PWRTE_OFF & _MCLRE_OFF & _CP_OFF & _CPD_OFF & _BOREN_OFF & _CLKOUTEN_OFF & _IESO_OFF & _FCMEN_OFF @ __config _CONFIG2,_WRT_OFF & _VCAPEN_OFF & _PLLEN_ON & _STVREN_OFF & _BORV_25 & _LVP_OFF 'Internal osc setup OSCCON = %01110000 'clock is 8 meg. turn on PLL for 32 meg clock DEFINE OSC 16 ' 32 meg not allowed according to book. be sure to double things that need this 'Variables LED0 VAR LATD.0 ' Assign name "LED" to PORTD LED1 VAR LATD.1 LED2 VAR LATD.2 LED3 VAR LATD.3 LED4 VAR LATD.4 LED5 VAR LATD.7 CNT VAR BYTE ENALL VAR PORTB.0 BSRSTORE VAR BYTE EN1 VAR LATF.0 EN2 VAR LATF.1 EN3 VAR LATF.2 EN4 VAR LATF.3 ASM #DEFINE EN1A LATF,0 'stuff for testing ENDASM wsave VAR BYTE $70 SYSTEM INCLUDE "DT_INTS-14.bas" ' Base Interrupt System 'INCLUDE "ReEnterPBP.bas" ' Include if using PBP interrupts ASM INT_LIST macro ; IntSource, Label, Type, ResetFlag? INT_Handler TMR2_INT, _FiveMicroSec, ASM, yes endm INT_CREATE ; Creates the interrupt processor ENDASM '5.0 uSec OR 200,000 Hz INTERRUPTS T2CON = %00000100 'Prescaler = 1:1, TMR2ON 'PRELOAD 223 "FROM MULTI-CALC" PRELOAD VAR byte PRELOAD = 239 ' FROM TESTING TO FIND THE RIGHT NUMBER TMR2 = PRELOAD @ INT_ENABLE TMR2_INT ; enable Timer 2 interrupts Main: 'DO WHATEVER YOU WANT HERE IF CNT=200 THEN TOGGLE EN2 TOGGLE LED1 CNT=0 ENDIF TOGGLE LED2 GOTO Main '---[TMR2 - interrupt handler]-------------------------------------------------- FiveMicroSec: TMR2 = PRELOAD 'THIS LINE TOGGLE EN1 'AND THIS LINE CNT =CNT+1 'WITH THIS CREATE A INT ROUTINE EVERY 4.99uS ' LATC = PORTA 'ADDING THIS LINE CHANGES THE TIME TO 5.245uS!! ' SO WHY DOES THIS CHANGE THE TIMING?@ INT_RETURN

Now It doesn't make sense to me. I am running a 32mHz clock. 5 uS interupt should takes 160 clock cycles, or 40 asm instructions. my interupt routine has 3 PBP instructions when working and 4 when I add the LATC=PORTA line. I can't imagine that would be 40 lines of ASM. I have tried to do this with a pure ASM handler, but I can only get it to work if I toggle LATA instead of LATF. I am sure this is a bank select issue. I can work through that if I must, but I would really like to use PBP instead.

Maybe all the time is taken up with the entry and return? Also note the large disparity between what I thought the preload should be and what it is.
- 2nd December 2010, 03:36
mackrackit

Not much help but

Code:

OSCCON = %01110000 'clock is 8 meg. turn on PLL for 32 meg clock DEFINE OSC 16 ' 32 meg not allowed according to book. be sure to double things that need this

I think should be

Code:

OSCCON = %11110000 'clock is 8 meg. turn on PLL for 32 meg clock DEFINE OSC 32 ' 32 meg is allowed according to book. be sure to double things that need this

??
But I could be wrong
- 2nd December 2010, 04:29
rmteo

A timer preload of 239 indicates you have only 16 CPU (or ASM) cycles available - looks like 24 cycles are consumed by overhead.
- 2nd December 2010, 04:48
cncmachineguy

And I guess it makes sense that the overhead is in entering the interupt. Otherwise those cycles would be part of the "timer cycle" and not have to be adjusted for in the preload. I think at this point I have 2 options, either use pure ASM interupts and/or increase my timing to 8uS. Maybe I will look at how many instructions it takes to enter the interupt if using pure ASM. This will be the deciding factor I guess. Clearly there is some lag needed to enter the handler, but is it 24 cycles?
- 2nd December 2010, 05:16
cncmachineguy

@Dave, You know I would have bet my left pinky 32 was NOT in the list when I looked up DEFINE OSC, but low and behold it sure is. Thanks for making me look again. As for the bit 7 of OSCON, the way I read it is it is ignored if PLL is enabled in my config-which I think I did. Won't hurt to set it though.
- 2nd December 2010, 10:33
mackrackit

Quote:

the way I read it is it is ignored if PLL is enabled in my config-which I think I did. Won't hurt to set it though.

Yup, my mistake.
- 2nd December 2010, 14:02
cncmachineguy

Thanks Dave, Never hurts to have a second set of eyes on. specialy since I always forget my safty glasses.

Maybe I should bring this interupt question up in its own thread? I would sure like some input to see if there is something I should be doing different or if I have found the "limit" for DT_INT speed?
- 2nd December 2010, 15:06
ScaleRobotics

Darrel has a post in here somewhere that says how many instruction cycles an ASM type interrupt has, but I am having a heck of a time trying to find it. I remember it being substantially fewer than PBP type, but can't remember the number. I will try looking some more.
- 2nd December 2010, 21:51
mackrackit

http://www.picbasic.co.uk/forum/show...7587#post17587

70 cycles are mentioned.
- 2nd December 2010, 22:21
ScaleRobotics

Quote:

Originally Posted by mackrackit

70 cycles are mentioned.

Yes, 70 instructions cycles to save the registers, and 70 instruction cycles to restore them, but that is for PBP type interrupts. But for ASM type interrupts it is much shorter. I think it is something like 12 or 22 instruction cycles. But then, I can't find it in here or my brain.
- 2nd December 2010, 22:37
mackrackit

http://darreltaylor.com/DT_INTS-14/SPWM.html

???
- 4th December 2010, 17:58
ScaleRobotics

For those following this thread, see thread split here for more information about assembly DT_INTS interrupts : http://www.picbasic.co.uk/forum/showthread.php?t=14063
- 4th December 2010, 18:07
cncmachineguy

Thanks Walter, I was wanting to figure out how to include the other thread so to speak. There is heaps of good stuff in it, so I am glad its not buried in this one.
- 5th December 2010, 20:00
cncmachineguy

ADD in full swing here

Well, not that I need to be thinking about this yet, but in anticpation of "phase III" Anybody have any ideas for ram? I think I will be looking for at least 2-4K. The more the merrier. I do have scads of DDR sticks, that would be cool. Or I was thinking about SD cards, they would be way convinent. But at this point I am still just thinking.
- 5th December 2010, 20:26
mackrackit

SD is kind of slow, I here FRAM is fast.
- 5th December 2010, 21:04
cncmachineguy

Pretty cool looking stuff that F-RAM. I ordered some samples of the 5V I2C 512Kb guys. From Mouser I can buy them (Ok, $11 each for these big guys) but free samples is free! I really doubt i need this big. The 64Kb (yes boys and girls - thats kilo-bit) are only $3.xx each in small numbers.

Thanks Dave, now I can go back to finishing phase I :)
- 5th December 2010, 22:25
Ioannis

Quote:

Originally Posted by mackrackit

...FRAM is fast.

And they never wear...

Ioannis
- 12th December 2010, 20:04
cncmachineguy

Things are working, kinda

Here is my latest code. All this really does is check for switches and disable drives, and shift PORTa to PORTc every 5 uS. It also sends a servo pulse out every 20mS with a pulse time sweeping from 1-2 ms just for testing purposes.

Right now, everything seems to work, but I need to remove the RED line from the interupt. When I comment it out, it seems to hang in the interupt and never leave. Uncomment and it works again. I have been struggling with this for a week, so I figured I would just ask and get the answer.

Code:

' Name : Interface card.pbp ' Compiler : PICBASIC PRO Compiler 2.6A ' Assembler : MPASM ' Target PIC : 16F1947 ' Hardware : PIMill Interface ' Oscillator : internal ' Keywords : ' Description : PICBASIC PRO program to interface parallel port ' to 4 stepper drives and limit switches ' ' ' ' ' Configure ports ' ' ANSELA = %00000000 'port A all digital ANSELE = %00000000 'port E all digital ANSELF = %00000000 'port F all digital ANSELG = %00000000 'port G all digital CM1CON0 = %00000000 'No comparaters CM1CON1 = %00000000 CM2CON0 = %00000000 CM2CON1 = %00000000 ' preload latches before setting direction LATB = %00000000 LATC = %00000000 LATD = %00010000 'LED's 0 for red, 1 FOR GREEN should start red LATF = %00001111 ' low nibble is drive enable. 1 disables LATG = %00000000 ' I/O direction TRISA = %11111111'step and dir inputs TRISB = %11000001'7-0 icsp,icsp,h4o,h3o,h2o,h1o,estopO,enable In TRISC = %00000000'step and dir outputs TRISD = %00000000'led6,i2c,i2c,led5,led4,led3,led2,led1 TRISE = %11111111'ls4,lsh4.ls3,lsh3,ls2,lsh2,ls1,lsh1 TRISF = %00010000'spindle dir,servo out,dac out,flood input,4en,3en,2en,1en TRISG = %11100111'x,x,icsp/mclr,rly2,rly1,Rx2,Tx2,DI 'REMOVE THIS LINE IF NOT SETTING THE CONFIGS IN CODE SPACE @ __config _CONFIG1,_FOSC_INTOSC & _WDTE_OFF & _PWRTE_OFF & _MCLRE_OFF & _CP_OFF & _CPD_OFF & _BOREN_OFF & _CLKOUTEN_OFF & _IESO_OFF & _FCMEN_OFF @ __config _CONFIG2,_WRT_OFF & _VCAPEN_OFF & _PLLEN_ON & _STVREN_OFF & _BORV_25 & _LVP_OFF 'Internal osc setup OSCCON = %11110000 'clock is 8 meg. turn on PLL for 32 meg clock DEFINE OSC 32 'Variables LED0 VAR PORTD.0 ' Assign name "LED" to PORTD LED1 VAR PORTD.1 LED2 VAR PORTD.2 LED3 VAR PORTD.3 LED4 VAR PORTD.4 LED5 VAR PORTD.7 SERVO VAR PORTF.6 CNT VAR BYTE MYFLAGS VAR BYTE COUNTUP VAR MYFLAGS.0 COUNTDN VAR MYFLAGS.1 MS VAR MYFLAGS.2 ENALL VAR PORTB.0 TEMPF VAR BYTE FRAMECOUNT VAR BYTE POS VAR BYTE EN1 VAR PORTF.0 EN2 VAR PORTF.1 EN3 VAR PORTF.2 EN4 VAR PORTF.3 COUNTUP = 1 COUNTDN = 0 POS = 1 FRAMECOUNT = 0 MS = 0 CNT = 0 '5.0 uSec OR 200,000 Hz INTERRUPTS 'PRELOAD 39 PR = NUMBER OF CYCLES-1 'INT HANDLER COMPLIMENTS OF DARREL TAYLOR PRELOAD VAR byte PRELOAD = 39 ' FROM TESTING TO FIND THE RIGHT NUMBER PR2 = PRELOAD PIE1.1 = 1 ; Enable Timer2 interrupts INTCON.6 = 1 ; enable PEIE INTCON.7 = 1 ; enable GIE T2CON = %00000100 ; Start Timer2 DEFINE INTHAND _FiveMicroSec '---[TMR2 - interrupt handler]-------------------------------------------------- FiveMicroSec: LATF = LATF ^ 1 ; TOGGLE EN1 CNT =CNT+1 PORTC = PORTA PIR1.1 = 0 ; clear the interrupt flag @ NOP @ RETFIE Main: 'DO WHATEVER YOU WANT HERE IF ENALL = 0 THEN GOTO NOTREADY ELSE GOTO READY ENDIF READYRETURN: IF CNT >= 200 THEN ' TOGGLE EN2 TOGGLE LED5 CNT=0 FRAMECOUNT = FRAMECOUNT + 1 MS = 1 ENDIF IF FRAMECOUNT >=20 THEN FRAMECOUNT = 0 MS=0 HIGH SERVO IF COUNTUP THEN POS = POS + 1 IF COUNTDN THEN POS = POS - 1 ENDIF ' TOGGLE LED2 IF PORTE != $FF THEN SWITCHHIT GOTO SERVOCHECK GOTO Main SERVOCHECK: IF MS =1 THEN IF CNT >= POS THEN LOW SERVO MS =0 IF (COUNTUP = 1) AND (POS=199) THEN COUNTUP = 0 COUNTDN = 1 ELSE IF POS = 1 THEN COUNTUP = 1 COUNTDN = 0 ENDIF ENDIF ENDIF ENDIF GOTO MAIN SWITCHHIT: PORTF = PORTF | $0F IF (PORTE = 253) OR (PORTE = 254) THEN LED0 = 0 IF (PORTE = 247) OR (PORTE = 251) THEN LED1 = 0 IF (PORTE = 223) OR (PORTE = 239) THEN LED2 = 0 IF (PORTE = 127) OR (PORTE = 191) THEN LED3 = 0 ' ELSE PORTE = PORTE |$0F GOTO MAIN NOTREADY: PORTD = PORTD & $F0 PORTF = PORTF | $0F IF ENALL = 0 THEN NOTREADY PORTD = PORTD | $0F PORTF = PORTF & $F0 GOTO MAIN READY: PORTF = PORTF & $F0 PORTD = PORTD | $0F GOTO READYRETURN
- 13th December 2010, 01:48
cncmachineguy

feeling a little lost

I am used to using the "watch" part of MPLAB to step through a program and look at SFR's and variables. But with PBP program, I see 1 step seems to go several (I assume because of the number of instructions required for a line of code). So as I step through my program, PIE1.1 doesn't change when it clearly should.

I just don't know HOW to debug using PBP. In this aspect, right now ASM seems easier. Someone PLEASE keep me from going back to the darkside!!
- 13th December 2010, 16:08
Darrel Taylor

With Timer2 firing off every 5uS, you're not going to be able to do any meaningful single stepping.

The interrupt flag will be set again while you are looking at the last "step".
- 13th December 2010, 19:09
cncmachineguy

well if Debug with ICD3 does not hold the timer to inc only on an instruction cycle, then that makes sense. Now to the real problem, any idea why I can't rem out the

Code:

LATF = LATF ^ 1

line? With it in the int routine, all seems to work. but if I rem it, it doesn't work. I am checking with a scope and not by single stepping. Also led5 doesn't blink. So I am assuming CNT is not being incremented, which to me says it is not getting out of the int.
- 13th December 2010, 19:37
Darrel Taylor

In your code from post #58, execution will "Fall into" the ISR, never making it to the Main loop.

Either put the handler after the main loop, or add a GOTO that jumps over the ISR.
- 13th December 2010, 20:29
cncmachineguy

Well color me silly!! Ok, I will try that tonight when I get home, Thanks Darrel. I will post my success or failure, what ever the case may be.
- 14th December 2010, 02:06
cncmachineguy

Well here is the modification:

Code:

DEFINE INTHAND _FiveMicroSec GOTO MAIN '---[TMR2 - interrupt handler]-------------------------------------------------- FiveMicroSec: LATG = LATG ^ 4 ; TOGGLE EN1 CNT =CNT+1 PORTC = PORTA PIR1.1 = 0 ; clear the interrupt flag @ RETFIE Main: 'DO WHATEVER YOU WANT HERE

As you can see. the only change between last post and this is
addition of goto main and changed LATF to LATG. The goto main does seem to make it start better, but still can't rem the LATG - LATG ^4.
- 14th December 2010, 17:05
Darrel Taylor

I do not see that problem here.
It doesn't make any difference when the LAT line is commented.

Please tell me that this 5uS doohicky is not running a hobby servo????
- 14th December 2010, 18:12
cncmachineguy

Quote:

Originally Posted by Darrel Taylor

I do not see that problem here.
It doesn't make any difference when the LAT line is commented.

Thats too bad. OK, I will keep struggling with it. The way I am deciding it doesn't work is like this. A) LED5 does not flash anymore, and B)even though you have pointed out I can't single step with TMR2, It does seem to when LAT is not remed. When it is remed, single step makes it appear to stay in the handler. But I have a few more ideas to try. Thanks for looking at it.

Maybe I am not reming correct. I am putting a ' in the beginning of the line. such as this

Code:

' LATG = LATG ^4

Quote:

Please tell me that this 5uS doohicky is not running a hobby servo????

Would it be TOO much overkill?? No, The servo is just to check 1 output. The entire card is a 4 axis controller with a control for a brushless spindle motor driven from a hobby ESC. The 5uS int is to make sure no steps are missed from the controlling computer. When all is done, the machine is really a 6 axis machine with 2 of these cards running 3 axis each. All this just to cut foam airplanes!!
- 15th December 2010, 13:26
cncmachineguy

Update

Well I have no idea whats wrong, and seemingly no way to figrue it out, so for now I will just leave the LAT in. I have re-defined what output it flips so as not to be in the way of my actual use. So far I have tried rem ('), and erasing the whole line. on both tests I just compile and program, board stays in interupt. (Or just sits somewhere). un rem or put the line back in and volia it runs again. I did this about 20 times just to be sure it wasn't something I was doing. I even replaced it with a @NOP just to see if it was some kind of memory placement hic-cup.

One thing I didn't think to try was changing it to toggle a variable instead of an output. I will try that tonight.

On the good news side of things, The board and program work as desired (as long as LAT line is there). Now for some cleanup things and add in the serial com to change things like spindle speed. I am sure there will be more issues when this is attempted, but hey, thats all the fun :)

Phase 1 almost complete!
- 15th December 2010, 16:14
Darrel Taylor

Quote:

Originally Posted by cncmachineguy

On the good news side of things, The board and program work as desired (as long as LAT line is there).

That's too bad.
I was going to suggest doing it a different way.

If the inputs on PORTA were moved to PORTB, you could use the IOC interrupts to transfer everthing to PORTC.
Then it only interrupts when the pins change, leaving much more time for the main program.

Oh well, too late.
- 15th December 2010, 18:06
cncmachineguy

There was some reason I didn't want to use PORTB. Maybe cuz I like to leave the ICSP pins for just only that. In hindsight, and maybe moving forward, I may change it. For now it's just fine. In the current configuration there really is not much for the MAIN to do, and its not time critical. This phase is really just a stepping stone to the planned use for this. somewhere back in this thread, Walter had asked why move the step/dir through the chip at all. Short answer was as a stepping stone.

When I get to phase 3 of this project, the 5uS interupt won't be needed at all. :) (post 39 shows my plans)

Thank you for your help Darrel.
- 19th December 2010, 21:29
cncmachineguy

I think I found the problem

I think I have found the problem with interupt. When the interupt hangs and won't leave, it seems BSR=5, when this happens PIR1.1 doesn't get cleared so it just re-enters the int. So the question is, why doesn't PBP make sure the BSR=0 before it tries to PIR1.1=0?

Or the bigger question is why does it work for Darrel???

I am posting the program as it is right now, I am using MPLAB, chooseing MPLAB SIM from the debugger menu. then choose the >> not > to animate the program. With this you can see it enter the interupt and leave. I left it running for about 30 min without issue. rem the LAT line and instant hang. sometimes it takes a few times through to get stuck.

PLEASE, can I get a few people to try this so at least I will know I am not going crazy!!

Code:

' Name : Interface card.pbp ' Compiler : PICBASIC PRO Compiler 2.6A ' Assembler : MPASM ' Target PIC : 16F1947 ' Hardware : PIMill Interface ' Oscillator : internal ' Keywords : ' Description : PICBASIC PRO program to interface parallel port ' to 4 stepper drives and limit switches ' ' ' ' ' Configure ports ' ' ANSELA = %00000000 'port A all digital ANSELE = %00000000 'port E all digital ANSELF = %00000000 'port F all digital ANSELG = %00000000 'port G all digital CM1CON0 = %00000000 'No comparaters CM1CON1 = %00000000 CM2CON0 = %00000000 CM2CON1 = %00000000 ' preload latches before setting direction LATB = %00000000 LATC = %00000000 LATD = %00010000 'LED's 0 for red, 1 FOR GREEN should start red LATF = %00001111 ' low nibble is drive enable. 1 disables LATG = %00000000 ' I/O direction TRISA = %11111111'step and dir inputs TRISB = %11000001'7-0 icsp,icsp,h4o,h3o,h2o,h1o,estopO,enable In TRISC = %00000000'step and dir outputs TRISD = %00000000'led6,i2c,i2c,led5,led4,led3,led2,led1 TRISE = %11111111'ls4,lsh4.ls3,lsh3,ls2,lsh2,ls1,lsh1 TRISF = %10010000'manual ready,servo out,dac out,flood input,4en,3en,2en,1en TRISG = %11100111'x,x,icsp/mclr,rly2,rly1,Rx2,Tx2,SPINDLE ON 'REMOVE THIS LINE IF NOT SETTING THE CONFIGS IN CODE SPACE @ __config _CONFIG1,_FOSC_INTOSC & _WDTE_OFF & _PWRTE_OFF & _MCLRE_OFF & _CP_OFF & _CPD_OFF & _BOREN_OFF & _CLKOUTEN_OFF & _IESO_OFF & _FCMEN_OFF @ __config _CONFIG2,_WRT_OFF & _VCAPEN_OFF & _PLLEN_ON & _STVREN_OFF & _BORV_25 & _LVP_OFF 'Internal osc setup OSCCON = %11110000 'clock is 8 meg. turn on PLL for 32 meg clock DEFINE OSC 32 DEFINE NO_CLRWDT 1 'Variables 'Port assignments '------------------port B ENALL VAR PORTB.0 ESTOPO VAR PORTB.1 HO1 VAR PORTB.2 HO2 VAR PORTB.3 HO3 VAR PORTB.4 HO4 VAR PORTB.5 '------------------port D LED0 VAR PORTD.0 LED1 VAR PORTD.1 LED2 VAR PORTD.2 LED3 VAR PORTD.3 LED4 VAR PORTD.4 LED5 VAR PORTD.7 '------------------port F EN1 VAR PORTF.0 EN2 VAR PORTF.1 EN3 VAR PORTF.2 EN4 VAR PORTF.3 DIGIN VAR PORTF.5 SERVO VAR PORTF.6 MANREADY VAR PORTF.7 '------------------port G SPINON VAR PORTG.0 RLY1 VAR PORTG.3 RLY2 VAR PORTG.4 '---------------------Bytes CNT VAR BYTE FRAMECT VAR BYTE POS VAR BYTE MYFLAGS VAR BYTE PRELOAD VAR BYTE SPEED VAR BYTE DUMRUN VAR BYTE '---------------------Bits COUNTUP VAR MYFLAGS.0 COUNTDN VAR MYFLAGS.1 MS VAR MYFLAGS.2 LIMIT VAR MYFLAGS.3 DRIVEON VAR MYFLAGS.4 HOMED VAR MYFLAGS.5 '---------------------INIT VARIABLE VALUE COUNTUP = 1 COUNTDN = 0 POS = 127 FRAMECT = 0 MS = 0 HOMED = 0 CNT = 0 SPEED = 0 PRELOAD = 39 'PRELOAD 39 PR = NUMBER OF CYCLES-1 '5.0 uSec OR 200,000 Hz INTERRUPTS 'INT HANDLER COMPLIMENTS OF DARREL TAYLOR PR2 = PRELOAD PIE1.1 = 1 ; Enable Timer2 interrupts INTCON.6 = 1 ; enable PEIE INTCON.7 = 1 ; enable GIE T2CON = %00000100 ; Start Timer2 DEFINE INTHAND _FiveMicroSec GOTO START START: GOSUB NOTREADY Main: 'DO WHATEVER YOU WANT HERE IF DRIVEON = 0 THEN GOSUB CHECKREADY ENDIF IF (ENALL = 0) AND (DRIVEON = 1) THEN GOSUB NOTREADY ENDIF IF (MANREADY = 0) AND (DRIVEON = 1) THEN GOSUB NOTREADY ENDIF IF (LIMIT = 1) AND (PORTE = $FF) THEN GOSUB READY ENDIF IF SPINON THEN SPEED=POS RLY1 = 1 ELSE SPEED = 0 RLY1 = 0 ENDIF IF DIGIN THEN RLY2 = 1 ELSE RLY2 = 0 ENDIF READYRETURN: IF CNT >= 200 THEN TOGGLE LED5 CNT=0 FRAMECT = FRAMECT + 1 MS = 1 ENDIF IF FRAMECT >=20 THEN FRAMECT = 0 MS=0 HIGH SERVO ENDIF IF PORTE != $FF THEN GOSUB SWITCHHIT ENDIF GOSUB SERVOCHECK GOTO Main SERVOCHECK: IF MS =1 THEN IF CNT >= SPEED THEN LOW SERVO MS =0 ENDIF ENDIF RETURN SWITCHHIT: IF HOMED=0 THEN GOTO HOMESWITCH ENDIF ESTOPO = 1 PORTF = PORTF | $0F IF (PORTE = 253) OR (PORTE = 254) THEN LED0 = 0 IF (PORTE = 247) OR (PORTE = 251) THEN LED1 = 0 IF (PORTE = 223) OR (PORTE = 239) THEN LED2 = 0 IF (PORTE = 127) OR (PORTE = 191) THEN LED3 = 0 LIMIT = 1 DRIVEON = 0 RETURN HOMESWITCH: IF PORTE = 253 THEN HO1=1 IF PORTE = 247 THEN HO2=1 IF PORTE = 223 THEN HO3=1 IF PORTE = 127 THEN HO4=1 IF PORTE = 85 THEN HOMED = 1 RETURN NOTREADY: PORTD = PORTD & $F0 PORTF = PORTF | $0F DRIVEON = 0 HOMED = 0 RETURN CHECKREADY: IF ENALL = 0 THEN RETURN IF MANREADY = 0 THEN RETURN IF LIMIT = 1 THEN RETURN READY: PORTD = PORTD | $0F PORTF = PORTF & $F0 DRIVEON = 1 LIMIT = 0 RETURN '---[TMR2 - interrupt handler]-------------------------------------------------- FiveMicroSec: LATF = LATF ^ 32 ; TOGGLE DAC CNT =CNT+1 PORTC = PORTA PIR1.1 = 0 ; clear the interrupt flag @ RETFIE
- 19th December 2010, 21:45
cncmachineguy

Don't know why I need it, but this works:

Code:

FiveMicroSec: ' LATF = LATF ^ 32 ; TOGGLE DAC CNT =CNT+1 BSR = 0 PORTC = PORTA PIR1.1 = 0 ; clear the interrupt flag @ RETFIE

I mean, I get why setting BSR=0 fixes it, just don't know why I need to.
- 20th December 2010, 23:34
mackrackit

Just a shot in the dark....

Code:

CNT VAR BYTE BANK 0
- 21st December 2010, 00:49
cncmachineguy

OHH, great idea!! Thanks Dave, I will try this now.

So sad, got an error : Bad BANK number
- 21st December 2010, 02:35
mackrackit

Have you tried any of this in real life? Or just SIM?
- 21st December 2010, 03:15
cncmachineguy

its all in real life. I only use the MPLAB SIM when I have a problem I can't reason out, like why does the int hang. for that i need to see registers. But it showed up in real life first by my LED not flashing.
- 21st December 2010, 03:16
Darrel Taylor

Further explanation after my reply in the "Bug" thread.

When there was ...

LATF = LATF ^ 1

LATF is is BANK7, and since PBP thought it was in BANK0, it would have changed to BANK7 to access LATF, and at that point it knew it was in BANK7.
So then it changed back to BANK0 to access CNT, PORTA, PORTC etc.

When the LATF statement was commented, it thought it was still in BANK0, and didn't try to change the bank, not knowing what bank it was really in at the start of the interrupt.

Remember that this type of ASM interrupt using PBP statements, goes against the standard rules of interrupts.

I'm still embarrassed though. :o
- 21st December 2010, 03:27
mackrackit

Quote:

Originally Posted by Darrel Taylor

I'm still embarrassed though. :o

Why???

So
BSR = 0
at the beginning of the ISR will do what with LATF?

I thought it had something to do with being in the wrong bank, but like you said CNT was already in BANK 0.

Quote:

Remember that this type of ASM interrupt using PBP statements, goes against the standard rules of interrupts.

How should this be done? I am getting more confused than normal.
- 21st December 2010, 03:55
cncmachineguy

I think PBP assumes its in BANK0 unless it does something in a different bank. Since LATF is not in BANK0, PBP will clear the BSR after executing LATF line. When I comment LATF line, sometimes the Interupt occurs during an instruction where BSR !=0, so in the interupt PBP assumes BANK0 for the BANK0 stuff.

Does this make sense? Setting BSR=0 upon entering the ISR will make sure PBP is not confused :)

Thank you Darrel for letting me keep some of my hair so I can pull it out on the next problem.
- 21st December 2010, 04:01
cncmachineguy

Quote:

Remember that this type of ASM interrupt using PBP statements, goes against the standard rules of interrupts.

Well I am not too sure where the standard rules for interupts can be found, but in this instance, I am using PBP statements but ignoring all PBP context when entering and exiting the ISR. If I had actually written this in ASM, I MAY not have had this problem because I MIGHT have remembered to set BSR to what I needed. But at least I would NOT have thought it was a PBP issue.

At the end of the day, it is really my lack of understanding how PBP thinks. Now that I realize PBP assumes BANK0 all the time (maybe I even read that in the manual), I see why this is broke and why it is my fault as the programmer.
- 21st December 2010, 04:04
cncmachineguy

Quote:

When the LATF statement was commented, it thought it was still in BANK0, and didn't try to change the bank, not knowing what bank it was really in at the start of the interrupt.

How did Protus make this work???

Show 40 post(s) from this thread on one page