Hi Jeff, Let me see if I can summarize this in simple physical terms. The scattering matrix deals with power values. In a reciprocal network, the 'through' power is the same regardless of the direction we go. The node voltages, on the other hand, depend on the local impedance levels, which are related to S11, S22 and source and load reflection coefficients. So probably a more generalized condition is that we can shuffle around reciprocal building blocks inside a cascaded network, and the resulting eye diagram will stay the same as long as we do not change S11 and S22. Going back to your simulations: if you calculate the pulse response of your example circuits strictly from S21 of the network, I agree, it will not change, since S21 will not change as you move the reciprocal building blocks around. Your eye diagram, however, should be usually calculated as Vout/Vsource (and for sake of simplicity, we usually assume linear driver and receiver impedances, where their linearity does not change this argument). This voltage transfer ratio can be analytically calculated if you wish, from the S-parameter flow graph. The resulting formula is symmetrical in S12 and S21, but asymmetrical in S11, S22 as well as source and load reflection coefficients. If you simulate the transfer response or the pulse response of Vout/Vsource in HSPICE, you should see the change as soon as the electrical symmetry is changed. Regards, Istvan Novak SUN Microsystems Loyer, Jeff wrote: >Hi Istvan, >Could you point me towards more information on the "voltage transfer >ratio", and the difference between it and s21/s12? When I did the pulse >response of a system where S21 =3D S12 (though S11 and S22 were very >different, and the reflections were significant), it came out equal (p21 >=3D p12, see my posting of 9/15). Thus, I think the eye diagram will = >come >out equal if S21 =3D S12. =20 > >Thanks, >Jeff Loyer > > >-----Original Message----- >From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] >On Behalf Of Istvan Novak >Sent: Saturday, September 29, 2007 12:38 AM >To: Chris.Cheng@xxxxxxxx >Cc: lifeatthesharpend; signalintegrity@xxxxxxxxxxx; ron@xxxxxxxxxxx; >si-list@xxxxxxxxxxxxx >Subject: [SI-LIST] Re: AC Coupled Signals > >Chris and All, > >I think the source of misunderstandigs might stem from the fact that in >linear, time-invariant, reciprocal networks S21=3DS12 regardless of the >sequence of the smaller building blocks within the network, BUT, the eye >diagram shows us voltage transfer ratio between source and load, and it >is NOT S21; the voltage transfer ratio DOES depend on the sequence of >contributing blocks. This happens also with linear source and load and >even if the source and load are perfectly matched.=20 > >The only case when the voltage transfer ratio remains the same in spite >of moving building blocks around, if/when reflections at the boundaries >of the particular building block are negligible. > >Thanks > >Istvan Novak >SUN Microsystems > > > >Chris Cheng wrote: > > >>I think the point is tuning S22 for the non-ideal load. S12 or S21 =3D = >> >> > > > >>remains being equal (symmetric). >> >>-----Original Message----- >>From: lifeatthesharpend [mailto:lifeatthesharpend@xxxxxxxxx] >>Sent: Friday, September 28, 2007 2:44 PM >>To: signalintegrity@xxxxxxxxxxx; ron@xxxxxxxxxxx; Chris Cheng >>Cc: si-list@xxxxxxxxxxxxx >>Subject: Re: [SI-LIST] Re: AC Coupled Signals >> >> >>What is the total loss at each point? (reciever, driver side of cap, = >> >> >=3D > > > >>receiver side of cap) in dB=3D20 >> >>It seems to stand to reason that if you lose xxx mV when the signal is >> >> > > > >>=3D stronger at the driver and some smaller portion of that when the=20 >>caps =3D are placed closer to the recever is normal since loss has=20 >>ocurred in the =3D media. The total signal loss of the system could=20 >>still be the same.=3D20 >> >>Leonard.=3D20 >> >> >> >>----- Original Message ----=3D20 >>From: Stephen Zinck <signalintegrity@xxxxxxxxxxx>=3D20 >>To: ron@xxxxxxxxxxx; Chris.Cheng@xxxxxxxx=3D20 >>Cc: si-list@xxxxxxxxxxxxx=3D20 >>Sent: Friday, September 28, 2007 12:19:22 PM=3D20 >>Subject: [SI-LIST] Re: AC Coupled Signals=3D20 >> >> >>Hello SI-LISTers,=3D20 >> >>I thought for my part in this discussion, I should do some due=20 >>diligence =3D on=3D20 this AC coupling capacitor placement location=20 >>question.=3D20 >> >>Scott McMorrow, Steve Weir and I had some off-line discussions that = >> >> >=3D=20 > > >>tended=3D20 to suggest my position dependency results may have been=20 >>caused by local=3D20 resonances from other impedance discontinuities = >> >> >in=20 > > >>the system I was=3D20 simulating. Based on this, I set out to develop = >> >> >a=20 > > >>simulation model that =3D had=3D20 a minimum of discontinuities (no=20 >>backplane vias/connectors/trace, etc.). =3D >> >> >>I used:=3D20 >> >>- Spice models of non-linear 3.125Gbit/s silicon (driver and=20 >>receiver)=3D20 >>- S-parameter based package models for both driver and receiver.=3D20 >>- A 0.01uF capacitor and its associated parasitics (via, trace, pad, = >> >> >=3D > > > >>mount,=3D20 component).=3D20 >>- 2D lossy W-Element transmission line (with di-electric and skin=20 >>effect =3D >> >>losses included).=3D20 >> >>I made the capacitor model such that I could "slide" it up and down a=20 >>15 =3D >> >>inch trace between the driver and receiver. I iteratively simulated=20 >>for =3D the=3D20 following length combinations:=3D20 >> >>- 500 mil trace from driver to AC coupling capacitor with 14500 mil = >> >> >=3D=20 > > >>trace to=3D20 receiver.=3D20 >>- 5000 mil trace from driver to AC coupling capacitor with 10000 mil = >> >> >=3D > > > >>trace=3D20 to receiver.=3D20 >>- 10000 mil trace from driver to AC coupling capacitor with 5000 mil = >> >> >=3D > > > >>trace=3D20 to receiver.=3D20 >>- 14500 mil trace from driver to AC coupling capacitor with 500 mil = >> >> >=3D=20 > > >>trace to=3D20 receiver.=3D20 >> >>The results show around 125 mV (differential) difference between=20 >>the=3D20 capacitor at the source versus the capacitor at the=20 >>destination, with =3D the=3D20 benefit going to the capacitor placed=20 >>closest to the receiver. 125 mV is =3D a=3D20 lot to give away...=3D20 >> >>I am not going to pretend to understand the physics behind these=20 >>results =3D but=3D20 I thought it worth while to at least show the = >> >> >basis=20 > > >>for my statements.=3D20 >> >>I would be happy to evolve the simulation environment if someone has=20 >>a=3D20 suggestion...=3D20 >> >>I have put together a document that I can post to an ftp site or email >> >> > > > >>=3D if=3D20 anyone would like a copy...=3D20 >> >>Kind regards,=3D20 >>Steve=3D20 >> >>Stephen P. Zinck=3D20 >>Interconnect Engineering Inc.=3D20 >>P.O. Box 577=3D20 >>South Berwick, ME 03908=3D20 >>Phone - (207) 384-8280=3D20 >>Email - szinck@xxxxxxxxxxxxxxxxxxxxxxxxxxx=3D20 >>Web - www.interconnectengineering.com=3D20 >> >> >> >> >> >> > > > ------------------------------------------------------------------ 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 technical documents are available at: http://www.si-list.net 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