Dave (& Ann) Scott wrote:

> Actually it's the only way I knew to build it with 16F84s only having
> one timer.  Also, I couldn't figure out how to get to a 5-30 min.
> timer AND drive a servo with a single operating frequency/crystal.

  Here's the problem.  It's called "programming".  Perhaps we can help;
this discussion seems to have become bogged down on absurdities such as
slaving one chip from another rather than dealing with the problem of
performing the task in question.

  To perform these two tasks, you write a program loop using what is
called "isosynchronous" code to alternately count out your PWM interval
and your timing interval on each iteration of the loop.

  It depends on how fine your servo control needs to be.  If you need
finer control than can be achieved that way (and I begin to think you
may with a 4 MHz PIC) then you use a counting loop to generate the servo
pulse and TMR0 with the prescaler to count time.

  This is made much easier by using a "binary" crystal such as 4.096
MHz (assuming you want clock accuracy).  Set prescale to 16 and you time
out every 4 ms.  That's plenty of time in which to perform a servo pulse
counted in software; indeed I'm not sure you want to generate servo
pulses too often anyway.  Usually, every 10 ms or so.  On each TMR0
overflow you count down from 125, when you reach zero that's one second
and you count them to obtain your minutes, etc...

  So, immediately after each second, third or fifth such rollover, you
run the 1-2 ms servo pulse routine, then go back to waiting for TMR0
rollover.  If you have to perform some routine to obtain the value for
the servo output, you do that during the other intervals.

  Note also that for the time counting, it is sufficient that you
process the overflow anytime before the *second* overflow (with leeway),
so if you use 4 ms overflows (prescale 16, 4.096 MHz crystal), providing
you start a task immediately after an overflow, you have up to 7 ms to
complete it before returning to the "wait for overflow" loop.

  Perhaps you could explain your task better and we can see how to fit
this algorithm to it?
--
  Cheers,
        Paul B.