[EMAIL PROTECTED] wrote:

>
> I'm skeptical. Even with source terminations, you've got a nasty mismatch at
> the destination,

Right.  The reflection bounces off the far end, back toward the source, and is
absorbed as completely as possible by the source impedance.  If you need incident

wave switching, you can't do it that way, you need the far-end termination.

> assuming that you're using standard logic chips which are
> nearly infinite input impedance. Keep in mind, too, that for digital logic
> it's not the clock rate that matters, but rather the edge rates. A general
> rule of thumb is that the frequency content of a digital signal stays really
> significant out to at least 10x the nominal rate, and that any trace longer
> than 1/20 wave at the highest frequency must be treated as an RF element. For
> your 133 MHz clock, this very rough rule of thumb would have you consider
> frequencies out to 1.3 GHz. On FR-4 the wavelength is about 5", so anything
> over a quarter of an inch long is potentially trouble - not that you can't
> create troubles enough at even shorter lengths. But if you've got good
> healthy margins, and your design is all-digital and not too susceptible to
> noise, you might get away with it. As I said before, I'd be a bit skeptical.

Well, for clock lines, this is absolutely true.  They have to be clean down to
a pretty small percentage, or you have a problem.  But, data lines can have a LOT

of trash on them, as long as they settle when the data is SAMPLED!  If all the
chips sampling these lines sample at about the same time, and the reflections,
incident wave, etc. have settled down by then, it will work OK.  That is why
you can get by with much more than 1/4" at these rates, but why 5" is about where

the limit shows up.

Jon

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
*
* To leave this list visit:
* http://www.techservinc.com/protelusers/leave.html
*                      - or email -
* mailto:[EMAIL PROTECTED]?body=leave%20proteledaforum
*
* Contact the list manager:
* mailto:[EMAIL PROTECTED]
*
* Browse or Search previous postings:
* http://www.mail-archive.com/proteledaforum@techservinc.com
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to