Power Supply Design

The TI/Unitrode Power Supply Seminar is a tradition, having started in 1983 with the SEM100 seminar and reoccurring since at approximate 18 month intervals. These seminars have trained many power supply designers in their craft. TI has now done us all a service by putting the Unitrode seminar papers (minus some obsolete ones) on the web. They have also started to put power supply on-line training on their website. You can reach this material by starting at the Texas Instruments website. From there, the sequence changes from time to time but clicking on Power Management > Resources > Training or similar sequence will get you there. The new web material was announced in opening remarks at the 2001 Series seminar I attended in Irvine, California on May 10, 2001.

The best paper, viewgraphs, and presentation easily went to Laszlo Balogh, the star of this seminar.

A Design and Application Guide for High Speed Power MOSFET Gate Drives was written by Balogh and is the type of document that I search for as the end product of the problem/relevance/solvability/solution format of the SMPS Technology Knowledge Base. That is, if your problem is driving a power MOSFET, this paper is the best available description of your various solutions. The paper should be in the professional library of every power supply designer and the first thing you look at when you need to design your drive circuit. I especially liked the attention given to decoupling capacitors and the mention that ground planes are not always the best return path.

In musical terms, if the paper is the composition, the viewgraphs and the presentation are the performance. For world-class viewgraph presentations I was always taught that the presenter should always make or select the viewgraphs and should never use someone else's viewgraphs/arrangement. The reason? The presenter will always inject something into the presentation that is not in the viewgraphs -- sure to take the quality of the presentation down a notch or two. In the world-wide tour of the TI/Unitrode seminar it is not practical to always have the author, viewgraph author, and presenter be the same person at each location -- but when it does happen, magic sometimes happens. Balogh did the paper, viewgraphs, and presentation in the Irvine, California seminar and magic did happen.

To explain the magic, it helps to have some understanding of the Transactional Analysis model of the human personality. Under our skin, there are actually three people, a Parent, an Adult, and a Child. The Parent is the nurturing and judgmental part of the personality (I'm mostly being judgmental as I write this.) The Adult is where engineers spend most of their time, unemotionally processing and communicating facts. The Child has the emotions, including humor, and is also the source of creativity. When inventing, the engineer is in his/her Child.

We have all heard the advice that humor always helps a presentation, and sometimes it does, but unless well done, it often detracts and it always diminishes the flow of Adult-to-Adult information. Laszlo spent the complete presentation in his Adult -- pure communication of engineering knowledge. Never a trace of humor, nor a trace of irritation when interrupted with a question after asking they be saved for the end. The result -- the clearest presentation of engineering information per unit time of any presentation I have ever heard, and in 40+ years, I have heard a lot of viewgraph presentations. It might not have happened before, it may not happen again, but the magic did happen in Irvine.

Another paper I liked was Appendix A, The "k" Transformer Model -- An Inappropriate Abstraction, in Lloyd Dixon's paper, Magnetic Field Evaluation in Transformers and Inductors. We were all taught that k12=k21=k when we were first exposed to coupled inductors (transformers) in school, and it has gotten switching-mode power supply designers in trouble ever since -- it just doesn't apply to what we design. This has been discussed in obscure Ph.D. thesis and more or less obscure papers and books, but this reminder in an article sure to get wide circulation is most welcome.

There were other things of interest, but this is enough for one day's blog.