Re: [SI-LIST] : Differential pairs and place splits

Mike Jenkins ([email protected])
Thu, 02 Sep 1999 13:55:54 -0700


Taking your example of 20-mil-spaced lines, imagine the TDR trace
crossing a 20-mil ground plane gap. The reflection coefficient
("rho") would be (150-100)/(150+100)=0.2. But this discontinuity
would last for only about 160 ps/inch * 0.02 inches * 2 = 6.4 ps.
If the TDR had a risetime of 50 ps, the actual, observed variation
in rho would be 0.2 * 6.4 ps / 50 ps = 0.0256. Qualitatively, you
are correct, but I believe this small a perturbation is down in the

To be fair though, the above assumes a perfectly differential signal.
Any common mode component would experience the same problems as any
single-ended signal crossing a split in the ground plane. For EMI,
that's probably bad. For signal integrity, rejecting common mode
might be good (?)


> Mike,
> If we consider a differential pair implemented with two ~50 ohm microstrip
> lines 10 mil apart, we get a differential impedance of ~90 ohm. If the traces
> were 20 mil apart, the differential impedance is ~100 ohm.
> If these two pairs were routed over a split, in that section ( if the split is
> wide enough) the impedances would be ~120 and ~150 ohm respectively. The
> discontinuity caused by the split increases as the coupling between the traces
> decreases. In terms of return currents, what you described as the U-turn
> becomes weaker with less coupling. In other words, the traces are behaving more
> as single-ended lines and the split is therefore more disruptive.
> May be we are saying the same thing. Less coupling between traces means more
> discontinuity at the split.
> Vinu

 Mike Jenkins               Phone: 408.433.7901            _____     
 LSI Logic Corp, ms/G715      Fax: 408.433.7461        LSI|LOGIC| (R)   
 1525 McCarthy Blvd.       mailto:[email protected]        |     |     
 Milpitas, CA  95035      |_____|    

**** To unsubscribe from si-list: send e-mail to [email protected] In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at ****