TMR2 bug in FF3.6

[Mikael Nordman]

Hi Mikael,
Thanks for the update on tmr2 in FF 3.6.

I made the changes and played with tmr3 and tmr2 assembles.

Just for information, here are my results…

Putting a square wave on the o'scope, and looking at 1 sec pulses I get:

Using the command    1000 ms_test   (decimal of course).

  TMR3      ~1.012 sec high and low periods.

  TMR2      ~1.016 sec high and low periods.

ms_test  ( n - )
   begin
     porte c@ $0f and porte c!          \ turn 4 LEDs on
     dup ms                             \ pause
     porte c@ $f0 or porte c!           \ turn 4 LEDS off
     dup ms                             \ pause
     key?
   until  drop  ;

Pete

On 12/5/2010 2:25 AM, Mikael Nordman wrote:
> Hi Pete.
> Looked at this problem and found bug in WARM.
> I have not used TMR2 for a long time.
> I have not tested the correction but it shud
> be alright. You have to undo your change in the
> config file of course.
>
> 73s Mike
>
> #ifdef MS_TMR2
>         ;; Timer 2 for 1 ms system tick
>         movlw   h'7d'      ; Prescale = 4, Postscale = 16
>         movwf   T2CON, A
>         setf    PR2, A
>         bsf     PIE1, TMR2IE, A
> #else
>
> SHOULD BE :
>
> #ifdef MS_TMR2
>         ;; Timer 2 for 1 ms system tick
>         movlw   h'7d'      ; Prescale = 4, Postscale = 16
>         movwf   T2CON, A
>         movlw   tmr2ms_val
>         movwf   PR2, A
>         bsf     PIE1, TMR2IE, A
> #else
>
>
>
> On 26.11.2010 20:34, Pete Zawasky wrote:
>> Hi Mikael,
>> I am back on 3.6 and the 18F6527 for a while.
>>
>> I noticed that TMR2 gives me double the MS timing.
>> e.g. 2000 ms -- a 2 sec pause actually gives me a 4 sec pause.
>>
>> Changing the tmr2postscaler from d'16' to d'8' seems to correct the timing.
>>
>> TMR3 seems OK.
>>
>> Pete
>>
>
>

[Pete Zawasky]

Hi Mikael,
With TMR2 and TMR3 working about the same,  I decided to stay with TMR3 in the Configure file, as I believe you have.
I also took a look at  MS to tighten the timing up a bit:

Here is a slight change to get rid of an extra ms time period that shows up on my PIC18F6527 (32MHz) apps.  Note the addition of 1- in the definition of MS.

; ms  +n -      Pause for n millisconds
; : ms ( +n - )    
;   ticks +
;   begin
;     pause dup ticks  -  1-  0<
;   until drop ;
;
        dw      link
link    set     $
        db      NFA|2,"ms"
MS:
        rcall   TICKS
        call    PLUS
MS1:   
        call    PAUSE
        rcall   DUP_A
        rcall   TICKS
        call    MINUS
        call    ONEMINUS
        call    ZEROLESS
        movf    Sminus, W, A
        iorwf   Sminus, W, A
        bz      MS1
        goto    DROP

Pete

[Pete Zawasky]

But…. based on the definition of MS :

ms         ( +n - )
           Wait for at least +n milliseconds,
           but not more than +n plus 2x the resolution of the TICK clock.

I am back to your original MS code and TMR3 providing TICKs.

TMR2 will be used for other things….

Pete

[Mikael Nordman]

Good that you found the actual definition of MS.
I was going to look it up also.
I had a similiar definition in mind.

Mike