I want to work on a timer that triggers interrupts about every 1/100000 of a second.  I know how to set the timer speed, so suppose I have the counter set to 1 and divisor set to CPU speed / 16 (That is, 937500 HZ in fast mode).

But I'm a little lost for the rest of the steps.  What do I need to do to make sure that my interrupt routine will always run every 1/100000 of a second?  Assume the routine starts at Start_Routine, and the routine ends at End_Routine.


Start_Routine:

;Code
;Code
;Code

End_Routine

This is kind of a bad idea in general, because you won't be able to respond to an interrupt in 16 clock cycles.

Oh, did I get my computations wrong?  I'm trying to get an interrupt every 150 cycles or so

I wouldn't trust this as I've never tried it before but what I got for setting the timers at cpu speed /16 was this:

Code: [Select]

ld A,$88
out ($30),A


Granted, but like I mentioned in the first post I know how to set the timers. 

So you're saying that 9, not 1, should be outputted to the counter, right?

I may be wrong, but couldn't you get the exact number of cycles you wanted (up to 255 or maybe 256) by outputting the desired number of cycles to the counter port and setting the timer speed to the CPU clock speed?

Well, in general, you can't have an event timed more precisely than one instruction (i.e., 4 to 23 clock cycles), except by writing code to explicitly delay for that long.  Interrupts can't "fire" in the middle of an instruction, only between one instruction and the next.

If you use the auto-repeating mode, though, the timer keeps ticking after it has expired, so you can use it for accurate long-term timekeeping.  (I.e., you'll still be off by 4 to 23 clock cycles each time, but the errors won't accumulate.)  If you want to do this, you do have to be sure that you acknowledge the interrupt (by writing a value to port 31/34/37) every time the timer expires, and before it underflows.  Is that what you were originally asking about, Hot_Dog?

In answer to Runer112's implied question about whether you can count up to 256: no, the timers don't really work right if you set the initial value to zero.  But if you want an interrupt every 256 clock cycles, you can switch to (say) the CPU/2 mode and set the initial value to 128.

If you wanted to use the auto-repeat mode as I've described (i.e., timer keeps running and fires every N ticks, regardless of what the CPU is doing), then you'd want to:
- select the desired frequency (ticks per second) using port 30
- set port 31 to 3
- set port 32 to the desired number of ticks per interrupt period

Then, each time a timer interrupt occurs, you'd write 3 to port 31 again.  You wouldn't touch ports 30 or 32 after that, except to turn the timer off when your program finishes, and perhaps you might want to read port 32 to determine precisely how much time has elapsed.  If port 31 bit 2 ever gets set, you're probably doing something wrong.

If you wanted to use the non-auto-repeat mode (i.e., timer fires N ticks after the previous interrupt was acknowledged), then you'd want to:
- select the desired frequency using port 30
- set port 31 to 2
- set port 32 to the desired number of ticks until the next interrupt

and do that again each time an interrupt occurs (though you could omit the re-initialization of port 30 if you like.)

I think that's basically all there is to it.  There's also some weirdness (seemingly a hardware bug) where the programmable timers won't generate interrupts if you turn off regular timer interrupts (port 3 bit 1) and the CPU is halted.  Don't do both of those things at once, and you should be fine.