From: Douglas C. Smith (email@example.com)
Date: Sun Feb 06 2000 - 18:53:49 PST
Thanks for the explanation. Sounds like the method is somewhat like
building up simple physical models and extending the experiment to a
real board if initial results look good as I like to do. I call that
putting the problem on the "analog" computer.
Yes, I would like to see the simulation but.....
DougS (one of the several Dougs on the list)
Ray Anderson wrote:
> You are correct, the simulation shown in Larry's spice
> deck is missing some key ingredients. But that is by design,
> not by omission.
> He has provided what we call a single-node analysis.
> All the capacitors are connected between ground and a single
> node as you have observed. We use this simple topology as a
> first step in our decoupling methodology to choose the proper
> mix of capacitors at frequencies below those where the plane
> resonance becomes an important factor. We find the single node
> analysis to be quite useful in establishing a good first cut
> of the values and numbers of caps required to meet our calculated
> target impedance.
> As a second step, after we have achieved an acceptable composite impedance
> profile as a result of the single-node analysis, we move on to a multi-node
> analysis. I believe this is what you have suggested. Our multi-node analysis
> applies the capacitors determined in the single-node analysis to a plane model
> consisting of an array of orthogonal transmission lines and other parts that
> provide an accurate plane model that incorporates copper, skin effect, and
> dielectric losses. By running the multi-node analysis we can fine tune the
> spatial placement of the caps on the plane and add or delete caps as required
> to arrive at a final decoupling solution that spans from 10's of KHz up to a
> couple of GHz.
> Larry provided the single-node analysis model for comparison
> with DougB's closed form solution. A multi-node analysis compared to
> the closed form solution would be comparing apples and oranges so to
> speak. A closed for solution for the multi-node problem would be
> prohibitively complex.
> I'd like to show you our 2-D transmission line plane
> model, but alas, lawyers, proprietary IP and the like.....
> > Hi All,
> > I think the simulation below is missing some key ingredients, the
> > variable
> > time delay between the capacitors representing their spacing on the
> > board
> > and loading of the devices. I see below caps connected between vdd and
> > 0.
> > I have observed that devices lower the Q of the whole structure.
> > The ground and power planes look to me like a 2 dimensional
> > transmission
> > line with random hi-Z and lo-Z connections. A simple simulation of
> > caps
> > in parallel may not give a useful answer unless the caps are right
> > next
> > to each other. I would like to see the spice model of the 2
> > dimensional
> > transmission line modeled to a few GHz. (lots of elements!!)
> > DougS
> **** 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
-- ----------------------------------------------------------- ___ _ Doug Smith \ / ) P.O. Box 1457 ========= Los Gatos, CA 95031-1457 _ / \ / \ _ TEL/FAX: 408-356-4186/358-3799 / /\ \ ] / /\ \ Mobile: 408-858-4528 | q-----( ) | o | Email: email@example.com \ _ / ] \ _ / Website: http://www.dsmith.org -----------------------------------------------------------
**** 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 ****
This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:35:00 PDT