From: "Brill Pappin"
Subject: Re: Interesting problem to solve
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Date: Fri, 18 Oct 2002 08:10:32 GMT
NNTP-Posting-Date: Fri, 18 Oct 2002 04:10:32 EDT
"Don Pearce" wrote in message
> On Fri, 18 Oct 2002 07:49:36 GMT, "Brill Pappin"
> >I guess that would work best with square waves... simple to implement
> >though... the sign waves could be converted to square waves... only I
> >also have to measure the peaks to get an accurate match between the two
> >I could even measure very short differences between the signals by
> >increasing the delay by a set amount... which would be fairly easy with
> >square waves... but how would I delay a sign wave?
> >- Brill Pappin
> Series R, then shunt C. It will shrink it somewhat as well, but if you
> are then converting to square waves, as would seem sensible, then it
> won't matter. Not entirely sure why you want to put in this added
> delay though - it is just an extra inaccuracy on top of the real delay
> you want to measure.
Was thinking that I'd then be able to measure shifts shorter than the uC
> With a square wave the peaks don't matter. You are triggering the
> counter from the edges.
Yah, the problem with that is that I will only ever see one pulse, so if my
delay is 3 pulses long, I won't see it. I would need some way to
discriminate between pulses... maybe two parts, first measures the leading
peak, then into the counter as square... when an incoming peak that matches
the leading peak comes in on the other input, the uC is told to stop on the
There would be a lot of room for error though, as its possible you'd get two
peaks the same in a row... so it would only work where the peaks where
distinguishable. if I can measure a few of them, and try to match, I'd get
better accuracy.... only problem with that is the uC is now getting into the
10s of uS at least...
- Brill Pappin