[ibis-macro] Re: Question on seeting the EMD direction
- From: "Brad Brim" <bradb@xxxxxxxxxxx>
- To: "'IBIS-ATM'" <ibis-macro@xxxxxxxxxxxxx>
- Date: Thu, 26 Jun 2008 22:42:32 -0700
Todd,
(1) Your last paragraph is extremely relevant
(2) Some of today's discussion centered around "time constants". You mention
yet another time constant which did not get discussed today, which is the
period of 10**17 bits.
The combination of (1) and (2) helps move the discussion forward.
Are we considering a time constant of 1 nanosecond (for a single bit) or are
we considering time constants on the order of 10**17 bits, which is 10**8
seconds or 36 months?
If we can't agree on a time period of consideration for our models, then how
will we ever agree on what effects might play a significant role?
I wish to make a few comments concerning your discussion of modeling as it
relates to passive interconnect:
(A) you missed measurement-based modeling
I am certain my former colleagues at Agilent will also notice this omission.
(B) you imply we aren't addressing numerical simulation [or measurement]
with the ICM spec as well as more recent discussions of EMD and netlisting
I don't agree with the same bounds of application you seem to perceive.
Measurements and numerical simulation results are quite analogous in form so
I will deal with as one topic. Any network parameter (S/Y/Z/etc), whether or
not frequency dependent, can be the results of simulations and measurements.
Including such network parameters includes measurements and simulations.
Those network parameters may be processed in some manner to generate
equivalent circuits - whether RLCK, TLs, controlled sources, pole-zero -
only then implying the processed results are in the bounded domain you
perceived.
(C) you invest effort to distinguish between "accuracy" and "precision"
"Accuracy" is the objective. One may perceive "precision" as a means to
enable or achieve accuracy. Maybe, but maybe not. One can have access to
every infinitely precise aspect of a circuit representation (topology,
components, etc) and still produce an inaccurate model. In fact, one can
have access to the same precision and still *never* be able to produce a
sufficiently accurate circuit model - short of realizing the macroscopic vs.
microscopic relationship between circuit analysis and EM analysis, and end
up actually writing an EM simulator disguised as a circuit model.
Over many years of passive modeling experience I have observed too many
exceptionally bright engineers who generate models mistakenly come to
perceive high complexity (and/or precision) in their model corresponds to
high accuracy.
For example, consider a relatively simple package interconnect. If the
frequency range of interest is less than 1/10 wavelength of phase delay
through this interconnect, then one can model it with an arbitrarily precise
circuit topology and likely never get a better result than from a simple DC
RLC measurement or a static EM solver (assuming the measurement and
simulation are each properly done). Now, let's say this interconnect becomes
a more complex 10Gbps diff pair in a modern ASIC package. Such interconnect
can have 3 wavelength or more of thru phase at the upper frequencies of
interest. One can generate an arbitrarily precise model and "might" get a
reasonable representation of the electrical behavior - then again they might
not. My experience indicates one would be quite lucky to get an acceptably
accurate broadband equivalent circuit model through hand-generation. If
there are any non-ideal return path issues (or a power plane resonance),
then how will a circuit model those with an implicitly assumed ideal return
path? If a 3D full-wave EM simulation is run that does not include the
entire power planes, how will it ever consider the plane resonance. Access
to "precision" for a circuit model may not provide much value for this case,
when generating an equivalent circuit through even arbitrarily precise
"geometric parsing/netlisting" of the geometry.
Don't get me wrong here. I am absolutely not arguing against the "precision"
EMD or netlisitng might bring to the table. I perceive such precision is
valuable. I am however asking caution from all of us to not get lost in the
"means" of precision but to keep in mind the "end" of accuracy ... and not
to equate the two.
regards,
-Brad
Brad Brim
Sigrity, Inc.
4675 Stevens Creek Blvd, Suite 130
Santa Clara, CA 95051
408-260-9344 ext 147
503-628-6230 (alternate)
503-799-9917 (mobile)
bradb@xxxxxxxxxxx
> -----Original Message-----
> From: ibis-macro-bounce@xxxxxxxxxxxxx
> [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Todd Westerhoff
> Sent: Thursday, June 26, 2008 8:53 PM
> To: 'IBIS-ATM'
> Subject: [ibis-macro] Re: Question on seeting the EMD direction
>
> Arpad,
>
> Not sure what you're getting at with your last point.
>
> I think the first challenge we have with an interconnect
> modeling spec is defining the problem we're trying to solve,
> the intended applications and what assumptions we're willing
> to make. Drawing from an example earlier today, we need to
> state our assumptions about how things like temperature and
> process are represented.
>
> There are two fundamental approaches to modeling interconnect
> as I see it:
>
> 1) An electrical equivalent modeling approach, providing a
> simulation model that is precise and accurate enough for an
> intended purpose
>
> 2) A physical description approach, describing the actual
> materials and geometries so that simulation tools can extract
> their own electrical equivalent models as needed
>
> We've been talking about the first approach with EMD and ICM.
>
> One challenge with simulation is the tradeoff between speed
> and accuracy (actually, the tradeoff is between speed and
> precision, but that's another discussion). Model detail is
> one of the key enabling elements of increased *precision*,
> and thus always gets caught up in the argument. The usual
> challenge to a given level of modeling detail is "yeah, but
> your model doesn't represent such and such a physical effect,
> and therefore it isn't accurate enough". Here as before, we
> would be better served if we distinguished between precision
> and accuracy.
>
> The level of detail required is case-dependent, which is what
> makes these discussions so difficult.
> If you have a system that's close enough to the edge,
> physical effects that don't matter in 9 out of 10 systems
> will still be the difference between success and failure in
> that one case. If you try to create the system (or language)
> that can represent any possible case, you'll probably end up
> with a system (or language) that no one wants to use (which
> is what I think Mike was getting at).
>
> So now what?
>
> If I've followed the discussion correctly, thus far we've
> said we're interested in created a spec for modeling
> interconnect at the electrical level. I think we should
> spend some more time hashing out our mission statement - are
> we modeling interconnect for any purpose at all? Over what
> frequency range? For what purpose (for example, if we said
> we wanted to model serial links for periods < 10**17 bits,
> how big a factor would temperature variation be)?
>
> The practical aspect here is that the more succinctly we can
> define our mission statement, and the more clearly we can
> articulate intended applications and fundamental assumptions,
> the greater our chance for success will be. If we can't
> clearly articulate these things, then we're taking on a very
> big task indeed.
>
> Todd.
---------------------------------------------------------------------
IBIS Macro website : http://www.eda.org/pub/ibis/macromodel_wip/
IBIS Macro reflector: //www.freelists.org/list/ibis-macro
To unsubscribe send an email:
To: ibis-macro-request@xxxxxxxxxxxxx
Subject: unsubscribe
Other related posts:
