Designer’s Guide Community

skippy

Junior Member

Offline

Posts: 26

Spectral Balance
Oct 13^th, 2003, 9:40pm

Tuinenga claims to have invented a means of extracting the large-signal frequency response of SPICE models, without using the FFT. He calls it spectral balance to differentiate it from Harmonic Balance (HB), where HB uses fast Fourier transforms to generate the frequency domain models. He wrote an article for the eetimes. But does not share any details whatsoever.

http://www.eedesign.com/columns/eda/OEG20031009S0021

Sounds interesting. Does anyone know anything more about this?

Back to top

IP Logged

Frank Wiedmann

Community Fellow

Offline

Posts: 677
Munich, Germany

Re: Spectral Balance
Reply #1 - Oct 14^th, 2003, 7:17am

Just guessing, but it might be a method similar to the one described in http://www.cadenceusers.org/cgi-bin/viewArchAbstract.cgi?abstractName=PaulW.Tuin... (maybe Ken or Andrew can give us some more details about this paper). To me, it looks like he found a way to extract such a model automatically and then use this model (some kind of frequency domain representation) in the simulation. The article http://www.eedesign.com/story/OEG20031010S0069 gives some background information about Tuinenga and his discovery.

Back to top

IP Logged

skippy Junior Member Offline Posts: 26	Re: Spectral Balance Reply #2 - Oct 14^th, 2003, 10:45am Very intriguing abstract. Thank you for your quick response. Is there a conference proceedings or slides available as well? Thanks
Back to top	IP Logged

Frank Wiedmann Community Fellow Offline Posts: 677 Munich, Germany	Re: Spectral Balance Reply #3 - Oct 14^th, 2003, 11:40am It appears that proceedings do exist; however, I do not have them. Maybe Ken or Andrew can help us out here.
Back to top	IP Logged

Ken Kundert Global Moderator Offline Posts: 2384 Silicon Valley	Re: Spectral Balance Reply #4 - Oct 14^th, 2003, 5:43pm I do not have the proceedings, nor do I have Paul's email address. If someone could send me his address (don't post it), I could invite him to add his comments to this thread. -Ken
Back to top	IP Logged

Paul Tuinenga

Guest

Re: Spectral Balance
Reply #5 - Oct 14^th, 2003, 6:54pm

I would be delighted to e-mail the IEEE BMAS workshop version of the Cadence User Group paper. Please note that I have already found both errors in the paper.

I have not coined the term "spectral balance" which is equivalent to the more common "harmonic balance" in overall approach and results. The key differentiation seems to be whether the device models have an analytic representation in the frequency domain, e.g. the work done by P. J. Khan and M. B. Steer.

So what you see in the workshop paper is this idea expressed at a system block level -- work which definitely prepared me for what I discovered later, but please don't expect that a straight line extrapolation of what you read will lead you there. I will be happy to field questions on the paper, but not about extracting analytic, frequency domain models from SPICE code (at least not in a public forum at this time).

Back to top

IP Logged

Paul Tuinenga Guest	Re: Spectral Balance Reply #6 - Oct 14^th, 2003, 6:57pm my e-mail: tuinenga@alumni.caltech.edu
Back to top	IP Logged

skippy Junior Member Offline Posts: 26	Re: Spectral Balance Reply #7 - Oct 14^th, 2003, 9:44pm Thanks Frank, Paul, and Ken. I am really glad this site provides a forum for these topics.
Back to top	IP Logged

Ken Kundert Global Moderator Offline Posts: 2384 Silicon Valley	Re: Spectral Balance Reply #8 - Oct 15^th, 2003, 1:47am My pleasure. My thanks go out to Paul for sharing his ideas. -Ken
Back to top	IP Logged