Thanks for the excellent explanations/clarifications! Jeff Loyer -----Original Message----- From: andrew.c.byers@xxxxxxxxxxxxxx [mailto:andrew.c.byers@xxxxxxxxxxxxxx] Sent: Wednesday, September 24, 2003 3:00 PM To: Loyer, Jeff Cc: si-list@xxxxxxxxxxxxx Subject: RE: [SI-LIST] Re: impedance relation with frequency... Hi Jeff, In my reply, which is included below, I mention the following: ------------------- "On boards, typically the frequency-dependent impedance change starts leveling out at much lower frequencies. Essentially you are approaching the sqrt(L/C) impedance, because your omega*L overwhelms = your R. *Usually* by 1MHz your Zo curve is flat."=20 -------------------- So, in your case where the 400ps risetime has a frequency content of = 875MHz, you are still well above the place where the Zo is changing. Essentially what I was trying to say with that reply was that it is important to remember that Zo is not always constant - not in theory or practice. You = can look at the equation to see the frequency dependence:=20 Zo =3D sqrt[(R+jwL)/(G+jwC)]=20 This is straightforward. Then a practical aside is that for PCBs you = have to go much lower in frequency to see the changing Zo, but for lines on chip-like dimensions (say 2um by 10um cross section), the Zo curve is shifted up in frequency. Dr. Johnson's book describes this topic very = well. I just now read Scott's email: it is a great reference on all this. His point about how the TDR continues to rise due to resistive loss is one = way to see this effect. If you were to TDR a very resistive, long chip line, your high frequencies (short times) would jump up to the sqrt(L/C) Zo = level, then after that your TDR signal would slowly rise until it reaches the = time (lower frequencies ) when the DC resistance of the line is reached, when = you would then level off. This is way easier to describe with pictures - I = hope I got the words correct...=20 This lead to another interesting topic that Scott hinted at: causality = in transmission line models. Frequency-dependent R and L are inherently = linked. If they are not, they you risk generating a non-causal model. ALSO, Zo = does not increase to infinity at DC. But I think that the widely-used = W-element model does not verify these two requirements, and therefore is = non-causal. Does anyone have experience showing this in experiment? Or any = confirmation of this? thanks, Andy -----Original Message----- From: Loyer, Jeff [mailto:jeff.loyer@xxxxxxxxx]=20 Sent: Wednesday, September 24, 2003 12:40 PM To: Byers, Andrew C Cc: si-list@xxxxxxxxxxxxx Subject: RE: [SI-LIST] Re: impedance relation with frequency... Hi Andrew, I would like some help understanding the difference between board traces = and "chip lines". My experience has been that I can TDR a trace using a = 35ps risetime, or through 100 and 400ps filters, and measure the same Z0 for = that trace. This would seem to be backed up by the fact that there is no compensation made when measuring traces with different TDRs, regardless = of their risetime. I just confirmed that again, measuring the same 3" microstrip trace with = no filter, and 100ps and 400ps filters, and finding the TDR and TDT = waveforms stabilize at the same level, regardless of the risetime. Of course, = there's significant impact to the risetime of the TDT, but the Z0 of the trace = (as indicated by the DC level of the TDR trace) remains constant. This implies to me that the Z0 of the trace is constant for 10GHz , = 3.5GHz, or 875MHz (35ps, 100ps, 400ps risetimes, respectively). My experience with VNA seems to substantiate this - S11 typically = remains fairly constant (other than resonances at lambda/4, etc.) while S21 = varies with frequency due to loss effects. Is there something else I'm missing? Jeff Loyer -----Original Message----- From: andrew.c.byers@xxxxxxxxxxxxxx [mailto:andrew.c.byers@xxxxxxxxxxxxxx] Sent: Wednesday, September 17, 2003 12:05 PM To: jonpowell@xxxxxxxxxxxx; kbagga31@xxxxxxxxx; si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Re: impedance relation with frequency... Concerning Zo relation with frequency: Once again, depends where you live. On boards, typically the frequency-dependent impedance change starts leveling out at much lower frequencies. Essentially you are approaching the sqrt(L/C) impedance, because your omega*L overwhelms your R. *Usually* by 1MHz your Zo curve = is flat. But if you are modeling chip lines, your R value for the line = might be comparable (or greater) than omega*L up to a couple GHz or so. Then you cannot ignore this frequency dependent behavior. I have seen a typical = line on chip go from about Zo=3D100ohms @100MHz, to Zo=3D63ohms @1GHz, to = Zo=3D55ohms @10GHz. Measurement, simulation, theory, literature, and gut feel all = back this up.=20 So the bottom line (as it always is in the world of interconnect = modeling) is it depends on how high you go in frequency, the dimensions of line = you are using, and if you are designing in a narrow band or a wide band. HOWEVER, as Jon pointed out, you can often see greater variations due to coupling from nearby traces. Plus you have to remember that impedance control is an issue too - usually +/- 10% is as good as it gets for run-of-the-mill PCBs out there (but money talks).=20 To get a feel for the numbers I got above, you can use a 2D field solver that handles the frequency dependent behavior of R and L (ansoft = spicelink or some other flavor). Or you can dig up equations and plug them into a matlab or mathcad. Calculate your R and L and C (usually G is = non-existant or insignificant...) and crunch away. salud, Andy Byers =20 -----Original Message----- From: Jon Powell [mailto:jonpowell@xxxxxxxxxxxx]=20 Sent: Wednesday, September 17, 2003 9:09 AM To: kbagga31@xxxxxxxxx; si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Re: impedance relation with frequency... Karen, It is my feeling that the frequency related impedance changes on a = signal will be second order considerations compared to the impedance changes = caused by crosstalk from neighboring wires. These effects can be shown with = most good SI engines. Intel has often recommended (for instance) calculating = the effective impedance when the coupled wires on either side of the target = wire switch simultaneously with the target wire in both the same direction = (all going high and low) and opposite (target going high and low and coupled going low and high). hope this helps (and if I am wrong, I am sure someone will scream at me = so wait a couple of minutes). regards, jon -----Original Message----- From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of karan bagga Sent: Wednesday, September 17, 2003 2:24 AM To: si-list@xxxxxxxxxxxxx Subject: [SI-LIST] impedance relation with frequency... Hi From the telegraphic equations on Transmision lines it seems the = impedance of the Trace varies with frequency. In my design specifications it is specified that my trace should be on = (25 +/- 10%) Ohms. How will I do it ? How will I do these kind of analysis? The frequency of the signal is high and also the rise time is = significantly low. Will FFT be of some help here ? Regards Karan. --------------------------------- Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software ------------------------------------------------------------------ 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 ------------------------------------------------------------------ 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: =20 //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages=20 Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu =20 ------------------------------------------------------------------ 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: =20 //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages=20 Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu =20 ------------------------------------------------------------------ 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