michael brown wrote:
> I really can't describe the problem in two sentences other than to say
> that different rollover preloads into TMR1H in the ISR cause
> unexplainable shifts in the timing accuracy  >60ppm by just changing the
> rollovers from twice per second to once every two seconds.

This sounds like maybe the act of writing or adding to the timer is causing
a missed count of a few instructions.  Note that timer 1 is 16 bits and
therefore can't be written in one operation.  What PIC are you using?  I
remember something about an errata regarding multi-byte timer reads/writes
for many of the 18F parts.  Did you account for the timer possibly not
getting incremented during access to it plus a few instructions after that?
I don't remember if timer 1 works that way, but timer 0 certainly does.
Also if you are clever about your period, you can arrange to change only the
high byte of the timer each overflow.

I would use the simulator to look at timer 1 values right before and after
when it's altered and make sure it didn't skip a cycle in there.

> You'd really
> have to refer to the original post for the details.

Not gonna happen.  It's your job to put all the relevant information in one
easily accessible place without too much surrounding BS that it gets
summarily ignored.


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist