Hi Paul, While I agree that there will, in real life, always be both modes on a differential pair, I disagree with the statement that it's "always advisable to terminate both modes". I strongly believe there are times when the negative effects of terminating for only the odd-mode impedance is acceptable (the resultant reflections are minimal and/or are absorbed quickly). LVDS busses, for instance, function with only a single resistor between the two signals. The best termination scheme depends on the bus. It could be very embarrassing to find out your competitor successfully uses 1/3 the resistors in their design. Jeff Loyer -----Original Message----- From: Paul Levin [mailto:levinpa@xxxxxxxxxxxxx] Sent: Monday, October 21, 2002 4:19 PM To: jeff.loyer@xxxxxxxxx Cc: 'ggafni@xxxxxxxxxxxxxxxxxxxxx'; Si-List; Simon Assouad Subject: Re: [SI-LIST] Re: AC Coupling and Differential Pair's Termination Folks, It is always advisable to terminate both modes. In general, you need the three resistor T termination to terminate the two modes. If, however, the lines are so loosely coupled that Zodd ~ Zeven, then you can use two separate R's to ground, each equal to Zodd or Zeven. Regardless of the degree of coupling, if you use a single line-to-line resistor, you may be able to terminate the odd mode correctly, but you ignore the even mode by failing to terminate it. Regards, Paul ____________________ Loyer, Jeff wrote: > My take on the 2nd question: > > A completely "bullet-proof" termination of a differential pair accounts for > termination of both the even and odd mode components of the signal. A > possible scheme for realizing this is given on page 66 of the Hall, Hall, > McCall book (and others). Call this scheme "A", and it takes 3 resistors > per differential pair. > > For a truly IDEAL differential pair, termination can be simplified. The > even mode termination can be eliminated. Assuming the signals transition at > exactly the same time and travel identically for the entire length of the > interconnect, the signals see only the odd-mode impedance, and each leg can > be terminated to ground for its odd-mode impedance (perhaps about 47 ohms) > only. This scheme "B" takes 2 resistors per differential pair. > > If you're doing this, simple superposition of the currents shows that a > single resistor (of twice the odd-mode impedance) between the two legs will > serve the exact same purpose. This scheme "C" takes only 1 resistor for > each differential pair. > > In non-ideal systems, there's skew between the 2 signals of the differential > pair. This can be a result of driver skew, or skew introduced by the > interconnect. In this case, there will be some reflections induced by > either scheme "B" or "C" (simulations I've done have shown scheme "B" to > induce less reflections than scheme "C"). Scheme "A" will have no > reflections. > > Picking the best scheme is (surprise) a compromise between the number of > resistors required and acceptable signal integrity degradation. Many > systems in the past have been able to use scheme "C", and it's the cheapest. > Scheme "B" seems to be better, signal-integrity-wise, than scheme "C", but > it's more expensive (parts and real-estate-wise). Of course, scheme "A" is > best for signal-integrity, but is pricey. > > The simulations are fairly straightforward, and should be done if you're not > dealing with a mature bus or have doubts about recommended terminations. > > Jeff Loyer > > > > -----Original Message----- > From: ggafni [mailto:ggafni@xxxxxxxxxxxxxxxxxxxxx] > Sent: Monday, October 21, 2002 12:04 PM > To: Si-List > Cc: Simon Assouad > Subject: [SI-LIST] AC Coupling and Differential Pair's Termination > > > Dear Experts, > > I would like to address this forum with two issues: > > AC Coupling > I read the recent thread about the AC coupling, and I would like to > raise couple of more generic questions related to that: > 1. When do you think that AC coupling is applicable? By saying that I > refer to more than the case where a 5V device is connected to a 2.5V > device. > 2. What are the pros and cons for AC coupling a clock (where one do not > have a DC Balancing issues) > > Differential Pairs Terminations > Generally, there are two flavors of differential pair's termination - > Parallel 1 - where a 100ohm or 120ohm resistor is applied between the > positive and the negative traces. > Parallel 2 - where two 50ohm resistors are pulled to one of the rails or > to a Vref. > 1. When do you think the first is applicable and when is the other? > 2. What are the pros and cons for deploying each one? > > Regards, > > > Gil Gafni > Silverback Systems > 695 Cambell Technology Pkwy > Suite #150 > Campbell, CA, 95008 > Phone: (408) 376 1303 > Fax: (408) 376 1399 > Email: ggafni@xxxxxxxxxxxxxxxxxxxxx > > > > > ------------------------------------------------------------------ > 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: > //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 > > > -- Paul Levin Senior Principal Engineer Logic Innovations, a Xyratex company ------------------------------------------------------------------ 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