Hi Jeff, I think that there have been discussions in the past how accurate length matching need to be done. I agree totally with you, that very accurate length matching is easy to implement for layout guys and as long it has no disadvantage it should be done as accurate as possible. But I have seen designs that required to increase the distance between Controller and DRAMs in order to get the required length matching. In addition to this the meanders have been routed so close that there was a significant X-talk adder. If I have a defined length mismatch i can calculate very accurate what I'm loosing in the timing budget. But for X-talk and additional length it is not that straight forward to estimate the effect in timing budget. Therefore I prefer to e. g. spend 70ps in the timing budget by having 10mm length mismatch vs. an unknown impact of X-talk in my system. One reason why we might have different experience here is, because we are talking about different systems. I often deal e. g. with cheap DDR2 solder down systems (not terminated, low layer count, single ended DQS, no impedance control, ... ). I have not seen this in normal DIMM type applications on Desktop size systems or even servers ... Hermann **** Attention: Our Address has changed. Please change your database according to the information below ***** Our next Events: ================ Visit us on Embedded World 2013 in Nuremberg 26-28.02.2013 Hall 1, Booth 430 Check our website or contact us for details EKH - EyeKnowHow Hermann Ruckerbauer www.EyeKnowHow.de Hermann.Ruckerbauer@xxxxxxxxxxxxx Itzlinger Strasse 21a 94469 Deggendorf Tel.: +49 (0)991 / 29 69 29 05 Mobile: +49 (0)176 / 787 787 77 Fax: +49 (0)3212 / 121 9008 Am 22.01.2013 16:29, schrieb Loyer, Jeff: > Hello Herman, > This sentence took me very much by surprise. Can you share specifics on how > this could happen? > "I have seen Designs being killed by over-accurate length matching" > > Thanks, > Jeff Loyer > > > -----Original Message----- > From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On > Behalf Of Hermann Ruckerbauer > Sent: Tuesday, January 22, 2013 12:58 AM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] Re: equal length requirement for SDRAM signals > > Hello Leon, > > For SSO (Simultanious switching outputs) this does not help, as they will be > aligned to the DQS anyhow. > For the input noise on DQ this might improve situation a little bit for VDDQ > noise, but I don't think you have so much margin that this really changes a > lot. Of course this depends on your system speed, but usually timing margins > are tight and you don't want to loose too much timing margin. > On CA usually you do have a VTT plane which should not impact e. g. VDDQ too > much from a noise point of view. So also here I expect that you loose more > margin when doing non equal routing vs. SSN noise. > > ==> My expectation is, that with the small Window that you can use to > optimize SSN you will not really gain anything from a noise point of view, > but you will loose timing margin. > ==> I would not recommend to do it and stay with a reasonable length matching > accuracy. > > Hermann > > P.S. but I have seen Designs being killed by over-accurate length matching .. > Sometimes a not so perfect matching is better than a very tight matching, but > this depends on your system. > > **** Attention: Our Address has changed. Please change your database > according to the information below ***** > > Our next Events: > ================ > > Visit us on Embedded World 2013 in Nuremberg > 26-28.02.2013 > Hall 1, Booth 430 > > Check our website or contact us for details > > EKH - EyeKnowHow > Hermann Ruckerbauer > www.EyeKnowHow.de > Hermann.Ruckerbauer@xxxxxxxxxxxxx > Itzlinger Strasse 21a > 94469 Deggendorf > Tel.: +49 (0)991 / 29 69 29 05 > Mobile: +49 (0)176 / 787 787 77 > Fax: +49 (0)3212 / 121 9008 > > schrieb 李磊: >> hi, experts, >> in many vendors' SDRAM layout guide line, there is equal length requirement >> for SDRAM group signals. in order to meet this, we use snake lines to reach >> this target usually. >> in my understanding, this kind of requirement is just only from electrical >> timing point of view. >> Actually from SSN point of view, different signal length should have benefit >> on the contrary. >> so as long as signal timing constraint is not violated, it's better to make >> signal length unequal. >> >> Am I right? >> >> thanks. >> BR,Leon >> ------------------------------------------------------------------ >> 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 forum is accessible at: >> http://tech.groups.yahoo.com/group/si-list >> >> List archives are viewable at: >> //www.freelists.org/archives/si-list >> >> Old (prior to June 6, 2001) list archives are viewable at: >> http://www.qsl.net/wb6tpu >> >> > ------------------------------------------------------------------ > 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 forum is accessible at: > http://tech.groups.yahoo.com/group/si-list > > List archives are viewable at: > //www.freelists.org/archives/si-list > > Old (prior to June 6, 2001) list archives are viewable at: > http://www.qsl.net/wb6tpu > > ------------------------------------------------------------------ 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 forum is accessible at: http://tech.groups.yahoo.com/group/si-list List archives are viewable at: //www.freelists.org/archives/si-list Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu