Continuing on the same topic - I was wondering if you would say the same thing for 10G. At 10G, I think stripline will not be as effecient because you will have to use vias and unless you make 50ohm vias, the distortion from the vias will be too significant. I have always used microstrip at 10G. Do you agree? On a related note, someone had mentioned if you are using ac-decoupling caps at 10G, the shunt capacitance of the pads counteracts the inductance of the cap body so increasing the cap, rather than decreasing it might be more helpful. He suggested simulating this effect. Anybody has any data on this or how I can simulate it - getting any data on the height of cap and putting it into ADS. thanks ----- Original Message ----- From: <MikonCons@xxxxxxx> To: <si-list@xxxxxxxxxxxxx> Sent: Wednesday, October 02, 2002 10:03 AM Subject: [SI-LIST] Re: Differential microstrip with coplanar ground traces > NOTE: This is a re-send of an earlier transmittal as I did not see it get > posted. Perhaps I deleted the original posting inadvertently. If so, just > dump this one. > > Mike Conn > ************* > Bob: > > You appear to have a few misconceptions as evidenced by your trace > descriptions. Without knowing trace lengths or dielectric materials, here are > a few initial observations for you to consider. > > "Our digital signals are 2.5Gb/s with the fastest rise time at 100ps." > > The fundamental is 1.25 GHz and, at your noted rise time, greater than 90% of > the signal energy is contained within (1/PI*Tr) = 3.18 GHz. (100 ps sounds a > bit slow for this signal; I would expect ~70 ps.) The minimum harmonic > frequency you should design for is the third harmonic, and 3*1.25 GHz = 3.75 > GHz. > > For effective performance in field capture, lowered crosstalk, and shielding > protection (for lower jitter susceptibility), the guard traces should have > vias to the ground plane at about (1/20) lambda (wavelength) at 3.75 GHz. > Assuming an Er of ~4, the 0.5" spacing you noted is >(1/10) lambda at the > fundamental and is ~(1/3) lambda for the third harmonic.... Bad News! > > "We expected the coplanar grounds to better contain the fields, and allow > closer pair to pair spacing without the risk of large crosstalk." > > For surface traces/microstrip, you have multi-mode propagation at different > speeds; i.e., most of the energy travels in the dielectric medium, but some > travels in the air above the surface. The wavelengths are different and the > composite signal at the end/destination point suffers rise time degradation > which increases jitter. But perhaps worse, the surface fields are not > effectively captured/terminated by the coplanar guard traces (per the above > comments) and will easily couple to other surface traces to yield substantial > crosstalk. > > Relative to microstrip, use of stripline construction will slow the > propagation speed of the wavefront; however, you will experience > substantially less rise time degradation as only one speed of propagation is > present, thereby preserving the signal fidelity. This effect keeps any > contribution to signal jitter to a minimum. The use of stripline also results > in closer coupling of all signal fields to the image/reference planes and > thereby produces lower crosstalk than microstrip construction. Using guard > traces with the stripline pairs further increases crosstalk reduction (per my > experience, ~4-8 dB added reduction), and the ground via spacing becomes much > less critical. > > Good luck on your next board spin. > > Mike > > Michael L. Conn > Owner/Principal Consultant > Mikon Consulting > Cell: (408)821-9843 > > *** Serving Your Needs with Technical Excellence *** > > > ------------------------------------------------------------------ > 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