Forums

How to distribute a clock source?

Started by amerdsp September 13, 2007
If you have more than one (3 or 4) component that you want to run on
the same clock.  What clock source is used in this case and what is
the best way to distribute it?

Thank you,

A

"amerdsp" <amerdsp@hotmail.com> wrote in message 
news:1189722757.505930.298800@50g2000hsm.googlegroups.com...
> If you have more than one (3 or 4) component that you want to run on > the same clock. What clock source is used in this case and what is > the best way to distribute it? > > Thank you, > > A >
What are your requirements? Do you need a synchronous system (same frequency and phase)? Is a meso-synchronous system okay (same frequency but phase independent)? What frequency? What are your jitter requirements? Do you need single-ended or differential clocks? I'm sure there are lots of questions I'm not asking. TC
On Sep 13, 9:21 pm, "Mindspring Newsgroups" <no...@nowhere.com> wrote:
> "amerdsp" <amer...@hotmail.com> wrote in message > > news:1189722757.505930.298800@50g2000hsm.googlegroups.com... > > > If you have more than one (3 or 4) component that you want to run on > > the same clock. What clock source is used in this case and what is > > the best way to distribute it? > > > Thank you, > > > A > > What are your requirements? > > Do you need a synchronous system (same frequency and phase)? > Is a meso-synchronous system okay (same frequency but phase independent)? > What frequency? > What are your jitter requirements? > Do you need single-ended or differential clocks? > > I'm sure there are lots of questions I'm not asking. > > TC
Let me check all these questions and I will get back to you. Thank you for the insight.
amerdsp wrote:
> "Mindspring Newsgroups" <no...@nowhere.com> wrote: >> "amerdsp" <amer...@hotmail.com> wrote in message >> >>> If you have more than one (3 or 4) component that you want to >>> run on the same clock. What clock source is used in this case >>> and what is the best way to distribute it? >> >> What are your requirements? >> >> Do you need a synchronous system (same frequency and phase)? >> Is a meso-synchronous system okay (same frequency but phase >> independent)? What frequency? What are your jitter >> requirements? Do you need single-ended or differential clocks? >> >> I'm sure there are lots of questions I'm not asking. > > Let me check all these questions and I will get back to you. > Thank you for the insight.
You can reduce the problem to 'what needs to be timed, and how closely does it have to match with what'. For example, you may have a ripple counter, advanced by the clock, and want to take a sample of its count on the same clock. Then you want to advance on one clock edge, and sample with the other clock edge. The problem then reduces to "Is the clock 1/2 period long enough to cover the propagation time in the counter". And you have to answer such questions with the worst case variation of the various (possibly) buffered clock lines, etc. Don't forget to include the delay effects of distance, and the distortion effects of unmatched distribution lines, both for the clock and for the signals (ringing, etc.). If the clock is operating at 1 Mhz you will have few problems. If it is at 1 Ghz you will have many. Don't forget that a foot is about 1 light nanosec. (Or about 3 light nanosec per meter.) Signals (and clocks) travel at about 1/2 to 7/8 of light speed. Have fun. -- Chuck F (cbfalconer at maineline dot net) Available for consulting/temporary embedded and systems. <http://cbfalconer.home.att.net> -- Posted via a free Usenet account from http://www.teranews.com