From: Dunbar, Tony (Tony_Dunbar@mentorg.com)
Date: Fri Dec 17 1999 - 15:39:04 PST
I would say that you probably missed the greatest "it depends" of all. That
is, emissions or the potential for them at least, depends (greatly) on the
deviation from a "perfect" differential case I have in my "real"
differential case. It's the deviation from differential-mode that creates
the common-mode. Some of these causes of deviation are largely under your
control when it comes to how the routing and other physical aspects of the
system come about but you have no control over such things as matching the
complementary switching of the outputs. If the switching speed from
high-to-low is different than from low-to-high - and they are likely to be
so - then I have common-mode contribution during switching. Be sure to try
and factor these things into your test-cases. In addition, some theorists
say (well, I heard of it from one, at least) that there is no need to route
a diff pair as it is traditionally done, i.e. side-by-side, and, as long as
the paths are the same electrical length and they have the same good
coupling to a plane, then all should be well. Maybe try that, too.
From: Chris Padilla [mailto:firstname.lastname@example.org]
Sent: Friday, December 17, 1999 3:29 PM
Subject: [SI-LIST] : Inner vs. Outer Layer Routing
More specifically, this is in regards to EMI and the use of
hi-speed differential signaling.
By hi-speed, I would say sub-nanosecond rise/fall times.
In terms of the likelihood to radiate (or be susceptable to),
it seems logical to me that microstrips would tend to radiate
more than (balanced) striplines. Looking at this single-ended,
it would seem true.
So the next question would be how does a differentially routed
signal help the cause to "justify" (to me) routing hi-speed
signals on outer layers rather than inner layers.
(Lets try and put off the SI implications of my thoughts here. I
know that signals travel faster on outer layers. I also know that
there is greater dispersion of the signal when placed on outer
layers. Lengths of traces are also important here, terminations,
vias, layer jumping, decoupling, etc., etc.) [Hmmm, now that I
write all those down, they may well be important!]
This leads me to ask exactly how much of a signal traveling down
this transmission line is differential-mode and how much is
common-mode? We know that radiation is often of the common-mode
Putting some numbers to this, let's assume 5-mil traces (w),
10-mil separation (s), and 4-mil dielectric height(s) (h). Is
trace height (t) an important factor here?
//////////////////////////////////////// Ground for stripline case
5 10 5
//////////////////////////////////////// Ground plane (infinite)
I have plans to run some very basic cases using Ansoft's HFSS--mostly
because that is what I am familiar with AND have access to. I am
posting this on the SI-LIST to gather any insightful thoughts on this
I have a feeling the overall answer to my question is: It depends!
It depends on components near the traces, vias, layer jumps, length
of the lines, environment these traces will be placed in (outside, in
a chassis, on a PC, etc.)
Thanks for your time.
Chris Padilla C i s c o S y s t e m s
EMC Engineer __________________________
170 West Tasman Drive | |
San Jose, CA 95134 | |
Email: email@example.com | | | |
Work: 408.525.4989 | !|! !|! |
Pager: 408.870.5410 | .!|||!. .!|||!. |
Fax: 408.525.9150 |.:!!!!!!!!!:..:!!!!!!!!!:.|
"I think the best evidence that intelligent life exists
elsewhere in the universe is that none of it has tried
to contact us...." Calvin of Calvin & Hobbes
**** To unsubscribe from si-list: send e-mail to firstname.lastname@example.org. 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
This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 11:39:09 PST