RE: [SI-LIST] : Regarding plane splits

About this list Date view Thread view Subject view Author view

From: Dunbar, Tony (tony_dunbar@mentorg.com)
Date: Tue Jan 02 2001 - 09:31:16 PST


Dennis,

On the face of it, this is a "mildly interesting result", to use your own
words. Of course, so far, you have presented very little supporting
information but if I was asked to make a judgement about which I thought
would be the better stackup using ONLY the information you have presented so
far, I would have to say the second one. It seems I would have been wrong.

But, you also say number 1 was better by (only) a few dB across the board
(pun intended, or not, I don't know). So, can you please now provide some
extra detail that might help explain the result? A few things I'm interested
in are:

1) layer-layer separation (i.e. dielectric thicknesses),
2) a reasonable breakdown of major signal characteristics
   (i.e. logic class, freq, edge-rates, etc.)
   - by reasonable, I mean that info you can disclose and
     that doesn't require you fill more than a few lines and
     doesn't occupy TOO much of your time,
3) from a results perspective, where in the spectrum do you
   see any significant contributions from any particular
   signal types, and how do those particular contributions
   vary between the two cases,
4) layer-pair sharing of signal routing and how that changes
   between the two cases (particularly how this might have
   affected the g and psplit return paths of signals).
5) Of course, what did you find to be the major contibutory
   factors?

Just lookin' to understand it a little better!
Regards,
Tony Dunbar

-----Original Message-----
From: Dennis Yarak [mailto:yarak@apple.com]
Sent: Monday, January 01, 2001 2:29 AM
To: Itzhak Hirshtal; si-list@silab.eng.sun.com
Subject: Re: [SI-LIST] : Regarding plane splits

I haven't been following threads recently (too busy) but I do have a mildly
interesting result to report in re: split planes.

After discussions with our EMI experts we decided to do two different stack
ups with the same gerber files. This is a 10 layer/50 ohm board with
s-g-s-s-g-pslit-psplit-s-g-s, and the alternative was
s-g-s-psplit-g-psplit-s-g-s.

The first stack optimized for signal return path (only one signal layer sees
the power split, and layer 8 was routed carefully) while the second
optimized for VDD decoupling to ground.

Even though this was as controlled an experiment for EMI as we could do, the
specifics of the situation determine whether the result would be applicable
in general. We didn't have the luxury of trying to reroute layer 3 to be
better behaved across the new plane splits it sees in the alternate stack,
for example.

We found the first stack was a few dB better pretty much across the board.

Dennis Yarak

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:31 PDT