Ground/power plane (GPP) resonance is more of an issue that most folks
tend to believe, particularly in mixed-signal systems where
digital components can generate enough noise on the GPP structure
to saturate and/or induce significant noise into the analog systems.
It can also be a significant SSN/SI issue when the frequency
content of the major switching signals are at a natural resonant
frequency of the board. As an example, we've modeled and measured
resonances with 40+ dB (10,000X power) noise variations across a
board.
Internally, we have been developing tools to model GPP resonance
and inter-chip noise, but they are far from release. We have also
built up several test structures to emperically characterize
the effects. However, the most practical tool we've found so
far in analyzing the resonances is a home-build field strength
prober that 'sniffs' the signal strength above an active board.
Using software to control servo motors (to physically sweep the
probe over the board) and a spectrum analyzer to measure the
field strength near the board (we have several probes, so
we can measure normal/tangential E and H fields), we generate
relative field strength across the board. This allows us
to 'visuallize' the resonances and the effects of decoupling.
The equipment is very similar to the equipment used in measuring
near-field antenna patterns. If you have a local university
with such equipment, I suggest getting access to it; you will
learn quite a bit.
In terms of available modeling tools, I've tried a quite a few,
and the best (although not great) solution I've found so far
is HP EEsof Momentum. This is a 3D planar EM simulator, which
can model multiple layers of complex planar structures. Using
this tool, we were able to match simulation to measurement
very well on passive structures. The biggest shortcomings
we ran into included: 1) no direct compatability with SPICE;
and 2) all ports/connect points had to be along the periphery
(i.e., no internal monitoring points; this was as of a year ago,
which I've heard the latest release has gotten around this).
Pat Zabinski
Mayo Foundation
On Sep 15, 9:34am, Mike Mayer wrote:
> Subject: Re: [SI-LIST] : Power-plane resonance article
> = As a matter of curiosity I was wondering how many of the si-list
subscribers
> =involved with high performance systems are currently modeling the effects of
the
> =power distribution system in their design work. Any comments why or why not
?
> =What kind of tools do you use ? Difficulties encountered ?
>
> This is a second try -- the "From" line from our domain is different
> than when I first subscribed to this group.
>
> I am looking into trying to model the power planes in an upcoming
> design. I have no trouble modelling the capacitors and the planes, but
> getting a "good" model for the power supply and the devices is not as
> easy to do. I'll probably end up using what I can find and trying to
> correlate the simulated results with the actual board.
>
> One item I picked up in an IBM app note was the idea of putting the
> pads in place for an SMA connector connected directly to the
> planes. It should make it easier to connect to the planes without
> picking up noise.
>
>
> --
> =============================================================================
> Mike Mayer Artesyn Communication Products, Inc
> Madison, WI
> http://www.artesyn.com/cp
> =============================================================================
>
>
> **** 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 http://www.qsl.net/wb6tpu/si-list ****
>
>-- End of excerpt from Mike Mayer
-- Pat Zabinski ph: 507-284-5936 Mayo Foundation fx: 507-284-9171 200 First Street SW [email protected] Rochester, MN 55905 www.mayo.edu/sppdg/sppdg_home_page.html
**** 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 http://www.qsl.net/wb6tpu/si-list ****