> From: Bill Merson <MERSONBILL@AOL.COM>
> [cut]
> My problem is with a little device which when turned on
> starts flashing a LED at once a second.  When four interrupts are seen on the
> int port, the flashing stops and a minutes timer begins which is determined
> by bcd switches on the portb lines (up to 127 minutes since portb,0 is
> occupied with the interrupt) Reasonably good accuracy is needed so I am using
> a 32.768 khz xtal. It seems like a straight forward case of   setting up tmr0
> getting the one second from btfss tmr0,4, decrementing a counter which has 60
> stored in it, and the end of that sequence, decrement the minutes storred in
> 'time'. The program begins fine, after four interrupts the 60 sec timer
> counts down, and the time counts down.  The problem is when I return to begin
> the LED flashing again, the program jumps back to the seconds and minutes
> timer. Here is some code:
> delay   clrf      tmr0
> again   btfss    tmr0,4
>            goto      again
>            decfsz    count,f   ;count is loaded with b'01111000'
>            return
>             decfsz    time,f   ;switch one is on for one minute
>              return
>             bcf         status,0
>             rrf          time,f      ;to move it one bit from interrupt
> iserv      incf      icount,f     ; interrupts
>              bcf       intcon,1
>              retfie

A few things...  if you are using RB1-7 as a minutes count input, then
surely you should be using RLF instead of RRF if you are going to
compare with the port setting.  Also, your indentation is confusing
(to me, not the assembler).  The convention is to indent only the
single instruction after a conditional skip.

You are probably not using interrupts correctly, since you are mixing
RETURN and RETFIE.  Without seeing the rest of your code, this would be
extremely unusual.  What is even more unusual is that you don't
restore the current context before RETFIE.  If you read the datasheet,
it will show the correct way of saving and restoring W and status.

I suggest you try using the MPSIM program to see what happens.  It is
good enough to ferret out your problem for this simple application.  If
you are unfamiliar with programming, then there is almost zero
probability of your code working if you do not use MPSIM (or an ICE, if
you have one).

Personally, I would not keep resetting TMR0.  Far better to let it run
free, and capture the timer overflow events in an interrupt routine.
Since your xtal period is 1 second divided by a power of 2, then it
would be a piece of cake to get 1 second interrupts if you use the
prescaler correctly.  Without a prescaler, TMR0 will clock over every
1024/32768 seconds.  Set the prescaler to divide by 32 and it will
clock over every second, which is what you want.  Only clear TMR0 when
you want to start the timer.  The interrupt routine will get called
every second, in which you decrement a counter which is initialised to
60.  When this counter decrements to 0, you should set a flag which the
mainline code can continuously poll.  The interrupt routine would also
reset the counter to 60.  When the mainline routine sees the flag being
set, it knows that 1 minute has passed.  It then resets the flag and
increments the minute count, comparing it to port B (shifted).

Regards,
SJH
Canberra, Australia