Boris, Thanks for the update. Fantastic detective work! best regards, scott -- Scott McMorrow Teraspeed Consulting Group LLC 2926 SE Yamhill St. Portland, OR 97214 (503) 239-5536 http://www.teraspeed.com Boris Yost wrote: >Dear List: > > Q: When is unshielded twisted pair cable not shielded, not twisted, and >not paired?!? > A: When it is unshielded twisted pair RIBBON cable!!! > The twisted part is twisted, but where you put the connectors has to be >regularly and uniformly oriented and straight. Unless you are going real >real far with the cable, the percentage of cable that is not twisted is >probably 20%. That part couples like cheap ribbon cable. > So my problem had nothing to do with "coupling THROUGH a plane". As a >matter of fact, it had nothing really to do with the LVDS pair that had the >bad symbol. A few clock earlier than my error, some noisy thing couples >into the LVDS clock, probably an unlucky alignment of bits. It does a >better job coupling into the complement line than the true line. The LVDS >PLL gets nudged from where it should be and jitters. A few clocks later I >get a bit error. The error happens to be in the "worst" pair of a basically >good design. I fixed it by filtering the clock. I should go back and find >the root cause of the noise and try to make it go somewhere else. > The previously mentioned dip I believe is ground bounce and I'm still >deciding whether it is bad enough that I must be rid of it. > >Boris > >----- Original Message ----- >From: "Boris Yost" <yost@xxxxxxxxxxxxxxxxxxx> >To: <si-list@xxxxxxxxxxxxx> >Sent: Friday, February 28, 2003 4:47 PM >Subject: [SI-LIST] Coupling THROUGH a plane? > > > > >>Dear List: >> >>I'm working on a persistent annoying bit error with LVDS. We had one >>design that we made hundreds of and never had a problem. We copied the >> >> >LVDS > > >>receiver part (same parts and same layout) to a new design. Now I have a >>bit error on both ordinary and torture test patterns that shows up on 10% >> >> >of > > >>boards. I have tracked it down to one particular symbol, and it turns out >>that symbol has a dip of about 40% of signal swing in common mode voltage. >>The eye diagram of the difference signal is wide open. >>Now, while we copied exactly the LVDS, we didn't copy everything else. >>Turns out, they put about 3 ordinary CMOS traces under the victim signal. >>However, the victim and my supposed aggressors are separated by a solid >>ground plane. (That is, it is as solid as anybody's ground plane ever >>is--except for the via holes.) The LVDS traces are on the surface, and >> >> >the > > >>other traces are on layer 3 of a 6 layer board. The board that works >>doesn't have traces there. >>Should I be looking hard at this difference, or should I be looking at our >>many other potential problems. >> >>Best regards, >>Boris Yost >>Mgr Electrical Engineering >>Rainbow Displays >> >> >> > > > ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu