On Scott's request I put together a spice deck. It is configured for SRS. v(Vddq,Vssq) and v(termVddq,termVssq) show a few mVpp of noise. There's only a 20pF capacitor between Vddq and Vssq to handle spikes caused by the non-ideal pullup/down switches. One can comment the WSRS line, uncomment the WGndRef line and change rterm2 to 50 ohm to simulate the Vss-only referenced case. v(Vddq,Vssq) noise is much greater in this case, demonstrating the need for a robust low impedance PDN. v(sout,Vssq) is the signal waveform at the TX, v(termout,termVssq) is the signal waveform at the RX. Spice: * * Compare Symmetrically Referenced Signaling (SRS) * to conventional Vss only referenced signaling * .trans 0.01ns 10ns .print v(termout,termVssq) .print v(out,Vssq) .print v(sout,Vssq) .print v(Vddq,Vssq) .print v(termVddq,termVssq) * Two pulse sources to drive the pullup and pull down switches v1 1 0 pulse 0 1 0ns 0.01ns 0.01ns 2ns 5ns v2 2 0 pulse 1 0 0ns 0.01ns 0.01ns 2ns 5ns * Pull up switch gpu pu Vddq vcr 1 0 1000 rpu pu out 50 * Pull down switch gpd pd Vssq vcr 2 0 1000 rpd pd out 50 * On-die capacitance between Vddq and Vssq c1 Vddq Vssq 20p * Inductor in series with battery to make battery AC "open circuit" l1 Vddq battpos 1e-3 vbattery battpos battneg 2 r2 Vssq battneg 1e-3 .option post=2 post_version=2001 * Stack up for W-element .MATERIAL diel_2 DIELECTRIC ER=4 LOSSTANGENT=0.02 .SHAPE rect RECTANGLE WIDTH=10mil, + HEIGHT=30mil .LAYERSTACK stack_1, + LAYER=(PEC,1mil), LAYER=(diel_2,1200mil), + LAYER=(PEC,1mil) .MODEL tline W MODELTYPE=FieldSolver, + LAYERSTACK=stack_1, + RLGCFILE=ex1.rlgc + CONDUCTOR=(SHAPE=rect, ORIGIN= + (50mil,150mil)), + CONDUCTOR=(SHAPE=rect, + ORIGIN=(50mil,200mil)), + CONDUCTOR=(SHAPE=rect, ORIGIN= + (50mil,250mil)) rout out sout 50e-3 * W-element connected in SRS configuration WSRS Vddq sout Vssq 0 termVddq termout termVssq 0 FSmodel=tline * Same w-element connected as conventional Vss-only referenced line *WGndRef Vssq sout Vssq 0 termVssq termout termVssq 0 FSmodel=tline +N=3 l=4000mil * Termination resistors * Change rterm2 to 50 for Vss-only case rterm1 termVddq termout 100 rterm2 termout termVssq 100 * Same w-element used to verify line impedance is ~50 ohm. Wztst 0 5 0 0 0 floatout 0 0 FSmodel=tline +N=3 l=4000mil rsrs 1 5 50 .end Schematic: https://docs.google.com/open?id=0B1SXvUJqinZYSHBiNWRNeGZGYjQ Thanks, Vinu On 07/15/2012 08:33 PM, Liuluping wrote: > > Thanks Chris, your explain Is very helpful, especially the sampling > point problem. > > Also thanks to Steve ,Vinu ,Scott and all the experts, though not full > understand ,but your discussion is very wonderful, may be the > symmetrically referencing both Vdd and Vss in a push pull driver > system can¡¯t cancel the SSO to zero due to the unequal inductance , > > But it may reduce the SSO noise close to zero , I thought that¡¯s the > original purpose which a demo boards design from a Top IC vendor, > who¡¯s application notes strongly recommend the DDR3 signal should > symmetrically referencing both Vdd and Vss. > > I also search the in the web ,the oldest paper about this topic I can > get is: > > Modeling of simultaneous switching noise in high speed systems > > Sungjun Chun; Swaminathan, M.; Smith, L.D.; Srinivasan, J.; Zhang Jin; > Iyer, M.K. > > Advanced Packaging, IEEE Transactions on > > Publication Year: 2001 , Page(s): 132 ¨C142 > > Which was referd in Madhavan and A.Ege¡¯s ¡°Power integrity modeling and > design for semiconductors and systems¡±,chapter 3. > > It shows that the return current flow in the vdd plane when signal > transition from low to high (unterminal signal), > > when signal from high to low , the current in the signal line will > flow into the GND plane. > > So a intuitive thought was why not use a symmetrically referencing > both Vdd and Vss to reduce the reference plane transition, in order to > reduce the SSO noise? > > Thanks again to all the experts, and sorry to Hirshtal that borrow > your questions £º£© > > LIU Luping > > ·¢¼þÈË: Cheng, Chris [mailto:chris.cheng@xxxxxx] > ·¢ËÍʱ¼ä: 2012Äê7ÔÂ13ÈÕ9:51 > ÊÕ¼þÈË: Liuluping > ³ËÍ: si-list@xxxxxxxxxxxxx > Ö÷Ìâ: RE: Questions on Reference Planes for DDR3 signals > > Luping, > > Sorry for the delay, work has been busy. > > I do believe the high speed current loop will always come from the on > die decoupling but that doesn't mean reference plane optimization is > not important. > > How tight you control the reference plane w.r.t. the signal path > determines how strong a mutual term you can get to lower your overall > signal path loop inductance. Which is a di/dt issue, not a decoupling > issue. > > As for the odd mode vs. even mode SSO, you are assuming the ringing is > settled within one cycle and the perfect sampling point is in the > middle of UI. In simulations and as in real life, we observed the > optimal sampling point is slightly later than 1/2 UI. That makes an > odd mode "pull in" more problematic than an even mode "push out" > > HTH > > Chris Cheng > > Distinguished Technologist , Electrical > > Hewlett-Packard Company > > +1 510 413 5977 / Tel > > chris.cheng@xxxxxx / Email > > 4209 Technology Dr > > Fremont, CA 94538 > > USA > > -----Original Message----- > > From: Liuluping [mailto:liuluping@xxxxxxxxxx] > > Sent: Wednesday, June 27, 2012 10:58 PM > > To: Cheng, Chris > > Cc: si-list@xxxxxxxxxxxxx > > Subject: RE: Questions on Reference Planes for DDR3 signals > > Hi Chris: > > Sorry for resend again. > > What puzzle me is when the DDR3 run up to 2133Mbps,certainly we will > face >1GHz noise on the power plane, does the current loop at this > frequency should go through the die caps instead of the package caps? > It means that we have nothing to do to lower the noise at this > frequency ,include the reference plane optimize discussing now? > > As the odd mode pattern problem, I also do some calculations, for > Zuncoupled=(L/C)^0.5,Zodd=((L-Lm)/(C+Cm))^0.5,Zeven=((L+Lm)/(C-Cm))^0.5,if > Zuncoupled=50ohm(microstripline),Zodd~49ohm,Zevev~71ohm respectively, > seens the odd pattern should better at the reflection due to impedance > mismatch, > > but what we got is that the odd mode pattern have more noise > (~100mV@DDR3) than the even mode pattern. > > Thanks and regards, > > LIU Luping > ------------------------------------------------------------------ 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