On Fri, 15 Oct 2004, James Newtons Massmind wrote:

> I think I understand your suggestion, but doesn't the prescaler output cause
> the counter to stop counting? Is that what you ment by "clock out the
> prescaler"?

The prescaler is 'welded' to the rtcc counter. So when the gate closes you 
have the value in the counter and the value in the prescaler. The value in 
the rtcc you read out using movf etc since it's a register. Then you have 
the prescaler to deal with. The prescaler is 8 bits wide (in my case) and 
takes up to 256 pulses to clear out. Meanwhile, you can't assign to rtcc 
without clearing the prescaler. So this is done like:

 	movf	RTCC,w		; get current counter in w
 	movwf	Frtcch		; temp. rtcc storage, also output reg.
 	clrf	Frtccl		; it took zero pulses to increment rtcc

loop:
 	movf	Frtcch,w	; get stored value
 	subwf	RTCC,w		; compare: Z = 1 if not changed
 	btfsc	STATUS,Z	; not changed
 	goto	done_ok		; changed

 	bsf	Fgate,Bgate	; pulse rtcc once
 	nop			; allow outputs to settle
 	bcf	Fgate,Bgate	; if rtcc overflows rtcc will increment

 	decfsz	Frtccl,f	; it took 1 more pulse to inc rtcc
 	goto	loop		; keep trying

done_ng:
 	; bad things happened. hardware fault. so do something about it

done_ok:
 	; Ftemp contains the value of rtcc since before we started

Peter

PS: this code is unchecked, and from memory. My counter actually uses 24 
bits so clocking out involves an adjustment for the MSB carry which occurs 
when/if rtcc,7 goes 1. My gate time is up to 100 seconds afair (with some 
overflow at 50MHz - the point is to get it accurate for low frequency 
like 32768Hz).
_______________________________________________
http://www.piclist.com
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist