Eric, I am also familiar with the proximity effect but I have always related = it to higher losses due to increased skin effect. However now that I = think about it I can see that it would also suffer from less dielectric = loss since air is generally less lossy than dielectric and since = dielectric loss is related to the delay not distance than Ravender's = measurement confirms that there would be less loss due to dielectric = loss. I know that as frequency increases the dielectric loss increases = faster than the skin effect loss and becomes more dominant around 1 GHz = so my question is, when comparing tightly coupled vs. loosely coupled = micro strip do we always see more or less net loss or is it dependent on = frequency and geometry. Has anyone done any research in this area? Thanks in advance, Paul -----Original Message----- From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of Eric Bogatin Sent: Tuesday, October 07, 2003 5:03 AM To: Si-List Cc: eric bogatin Subject: [SI-LIST] Diff. Pairs I have to agree with Jeremy's analysis of Ravender's measurement. There is a difference in the velocity of a differential signal as the coupling increases due to the larger percentage of fringe fields in the lower dielectric constant air. We describe this effect in terms of the odd mode velocity and the even mode velocity of a diff pair. If the dielectric material is uniformly distributed through the diff pair, so that where ever there are electric fields, there is the same dielectric constant, there will be no difference between the velocity of the signal traveling in the odd mode (a differential signal) and the velocity of the signal traveling in the even mode (common signal). This means that in a uniform stripline, the velocity of the two modes is the same, while in microstrip, the odd mode will generally be faster than the even mode. This is not a rise time effect, it is a velocity effect. The tighter the coupling, the bigger the difference in the odd and even mode velocities, in an inhomogeneous transmission line. When there is an inhomogeneous dielectric, a signal in the odd mode will travel faster than a signal in the even mode. If you send in a signal that has equal components of common and diff signal, the diff component will reach the end of the line before the common component. This is the real origin of far end cross talk. We can think of far end cross talk as either the difference between inductively and capacitively coupled current at the far end between two, single ended transmission lines with coupling, or as the result of the difference in speed between the odd and even modes in a diff pair. The details of this connection are in chapter 11 of my book. --eric ******************************************************************** Recently published by Prentice Hall, www.phptr.com Signal Integrity-Simplified, by Eric Bogatin Attend the GTL Signal Integrity University in Sunnyvale, CA - Nov 6-13 GTL 122 - Fundamental Principles of Signal Integrity GTL 250 - High Speed Board Design GTL 260 - Interconnect Models from Measurement ---------------------------------------------------------------------- --------------- Dr. Eric Bogatin CTO, GigaTest Labs 26235 w 110th Terr Olathe, KS 66061 v: 913-393-1305, f: 913-393-1306 e: eric@xxxxxxxxxxxx www.GigaTest.com ******************************************************************** Msg: #7 in digest From: "Jeremy Plunkett" <jeremy@xxxxxxxxxxxxxxx> Subject: [SI-LIST] Re: Diff. Pairs Date: Mon, 6 Oct 2003 18:40:18 -0700 Steve, Ravinder's edges propagate faster in the 1st case because tighter spacing between microstrip lines results in a greater portion of the field propagating in air above/between the traces vs in the fiberglass under the traces. The exact amount of speedup depends on the details of his trace geometry, soldermask geometry, and dielectric constants of the prepreg and soldermask. Outside of the special case of a mixed dielectric (as above), changes in coupling will not affect velocity one way or another (I mean "pure" propagation velocity here, not measured delay--see below). However, if you do not hold impedance constant, increasing coupling with make the impedance will go down, which will result in "slower" edges throughout the system (in V/ns) due to reduced signal amplitude (and vice versa, less coupling -> higher Z -> "faster" edges due to larger signal swing). Maybe not everyone agrees that there should be quotes around slower and faster in the sentence above. When I think about signals propagating on a transmission line, I prefer to keep the "delay" effects separate from the "amplitude" affects, even though they may both affect the measured delay. For example there is one effect of changing coupling even in a uniform dielectric while keeping Z constant; skin effect losses change due to the proximity effect. This will slightly affect the measured delay (if we measure at 50% of the transition) because it changes the waveshape at the receiving end of the line, but I don't consider it as changing the velocity because if we measure at the earliest recognizable point on the transition, it does not create any extra delay. There are 2nd order effects (variation of Er with frequency) that can complicate this mental separation of delay and amplitude, but I find it does a nice job of clarifying the 1st order effects (impedance changes and attenuation). I'm interested to hear any comments people have on it's usefulness or things I should watch out for. best regards, Jeremy |>--/\/\/--((((((((()--|> Jeremy Plunkett Signal Integrity Engineer Broadcom Corp www.serverworks.com |>--/\/\/--((((((((()--|> -----Original Message----- From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of steve weir Sent: Friday, October 03, 2003 4:40 PM To: Ravinder.Ajmani@xxxxxxxx; leeritchey@xxxxxxxxxxxxx Cc: si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Re: Diff. Pairs Ravinder, something seems very wrong with the physics here. I don't know if it is round-off error in your simulation, but Lenz' law agrees with Lee. Differential coupling resists any change. The more tightly you couple the two nets in a diff pair, the more it slows down the transitions. To cause a speed-up, they would have to switch in the same direction. If we could manufacture a machine that accelerated transitions in opposite directions we could solve the world's energy needs. Regards, Steve. At 01:12 PM 10/3/2003 -0700, Ravinder.Ajmani@xxxxxxxx wrote: >One obvious benefit is the reduction in EMI because of reduced loop area. >I have verified this through simulation. However, I am not able to >understand the phenomenon of edge degradation due to close coupling. I >ran some quick simulations on six inch long differential Microstrip nets >under the following conditions: >Case 1: Trace width 9 mils, separation 4 mils, differential impedance 99.3 >ohms >Case 2: Trace width 12.5 mils, separation 40 mils, differential impedance >98.8 ohms > >The driver had a rise time of 250 ps. The only difference I observed >between the two waveforms was that in the first case the propagation time >was 63 ps less than the second case. This is understandable since the >signals in the two branches of differential net have opposite polarity, >the coupling effect will speed them up. > >Next I tried the same simulation for Stripline nets. In this case, there >was practically no difference between two waveforms (less than 5 ps >difference in propagation delay). > >Am I missing something here. > >Regards, Ravinder >Server PCB and Flex Development >Hitachi Global Storage Technologies > >Email: Ravinder.Ajmani@xxxxxxxx ------------------------------------------------------------------ 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