Re: [SI-LIST] : An Interesting Presentation

Mikhail Khusid (Mikhail_Khusid@notes.teradyne.com)
Wed, 21 Apr 1999 13:16:22 -0400

Hello everyone,

Dr. Prince is right -- the 10% risetime to clock period ratio is a good
generalization for "regular" bus architectures, and there are certain
exceptions to that rule. I just want to recognise an important exception
which is a point-to-point architecture. The 10% rule is especially
not applicable to a long interconnect. In fact, for a long
interconnect, the degraded risetime at the receiver can be more
than the clock period itself, while the eye-pattern will be just
fine and the system will be working. Therefore, the generalization
of the risetime to clock period ratio can result in quite erroneous results
for many point-to-points connections.

Michael Khusid

prince@ece.arizona.edu (Dr. John L. Prince, III) on 04/21/99 12:21:47 PM

Your reasoning is interesting to me. For years I have taught that risetime
for a well designed system would be around 10% of the clock period. Very
much slower and you have practically a sine wave for a clock waveform; very
much faster is usually not necessary and may cause any of a multitude of
problems. I am also aware that any time one makes a generalization there
will immediately surface numerous exceptions, etc etc. Using this
"principle" many phenomena (eg,SSO noise ), can be related in some way to
clock frequency, which is conceptually often useful in today's digital
environment.

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