Power Electronics

Remembering Dr. Middlebrook Part 4

Sum: Do you regard your course on analog design has been well received?

RDM: Getting people to come to these courses is very difficult because you have to persuade two people. First you have to get the guy himself to recognize that this would be useful to him and second and more importantly, you’ve got to persuade his boss. Because his boss looks at the brochure and says ìI pay you a huge salary to know all this stuff. Why should I pay you money and give you time to go to this remedial course? So that has to be overcome, and I still don’t know how to do that properly, but that is what I am trying to do. This brochure that exists now is the result of a lot of thought and effort in how to say those things in the right way.

Going back to the first paper I presented to professors at the Frontiers in Education Conference, it went over pretty well and in talking to people afterward, comments were all favorable, but it really won’t go anywhere unless what I want to do next happens, which is as I’ve said the course as it stands is second level or it’s ‘technical therapy’.

Sum: If you were asked to set a level for the material of the analog course, where would you put it?

RDM: I’d really like to see it done this way as a first level course, which is what this other professor I mentioned who stood up and made comments he says that he wants to adopt these techniques in his own first level course. The first course in active circuits is the course done the year before mine at Caltech, and so when this professor said he was going to do that, I said “Great”. I didn’t want to do that myself. I think it ought to be done but I didn’t want to do it because I want to do animations and all that enjoyable stuff. And so I said “Great” and he was going to do it. Well, of course he couldn’t do it all by himself, in fact we even made a proposal, a joint proposal, which didn’t fly and so a few months ago I made another decision.

I’m going to run Part 1 of this course for professors and call it a Workshop, so I’m going to invite instructors to come to this and probably about half the time, well, there’s going to be two things, first I have to sell them on this approach. Second I want us to discuss how they think this approach could be used in a first level course, and so I have a letter all ready to go to a mailing list of 1300 names and I am going to be very curious as to whether anyone comes.

What I am trying to do here, and I have to be very careful about wording this, I am trying to be a missionary for this Design-Oriented Analysis approach, and I hope that the instructors who come to this are going to be disciples, sort of spread the gospel, in the hope that they will adopt it in their own teaching. Now I am very hesitant to put that high a level sounding term on it, which I don’t intend because this is obviously a very limited and special case, but that sort of process is the sales and marketing aspect: I’ve got to sell the professors first, and that’s what this workshop is for, and getting them to come to it is the hardest part. I’m pretty sure that once they come, I can sell them on it, just like the guys in the regular courses. Once they get there they like it. But getting them there is the hard part.

Design-Oriented Analysis is really methods of getting the answer in a Low Entropy form so that it can be used in reverse for design, because Design is the Reverse of Analysis. This is the Freudian regression thing, you have to reverse a lot of the things that have been ingrained in your mind even unconsciously about how you go about solving problems, because now in the conventional way, when you’re a graduate engineer, you’re out in the real world, you’re trying to solve real problems, how do you go about it?

You put everything in the model, you grind the algebra until you get stuck, then you start looking for things to throw out, and you better be able to justify what you’re throwing out, because mathematicians said ìif you’re going to change numbers later, you better not make that approximation because if you change the numbers, you may invalidate the approximation. So don’t do it.î This is ingrained in us, so we don’t, we postpone the approximation to the last possible moment and then when we’re forced to approximate, we better be sure that it’s OK. This is the conventional way of doing things, and this is what inevitably leads to the algebraic paralysis.

So the whole philosophy of the Design-Oriented Analysis approach is that, when I make a list at the end, almost everything is the exact opposite from what I just said. The way to get useful answers is put only as much in the model as you need to get the answer you are looking for. Leave everything else out, which is the same thing as approximating as soon as possible, not as late as possible but as soon as possible.

You make the approximations where they are useful whether they are justified or not, to the horror of the mathematicians. But all you have to do is when you get to the end you have to check whether it’s still OK, there’s only two possibilities, either it’s OK or it isn’t. If it is OK, you’re home free. If it isn’t OK, at worst you have to go back and not make that approximation but usually it’s even better than that because even if that approximation isn’t valid, it suggests another one, but at least you tried. And I summarize all that, which I say many times, you strew approximations and assumptions behind you like a wake, justified or not, because you can’t lose by trying. You can’t lose. You can only be better off, you can never be worse off than if you didn’t try.

That‘s a positive outlook, whereas in the conventional approach, because you’re taught not to approximate until you have to, everyone gets the feeling, maybe unconscious even, that if you have to approximate, somehow you failed. You weren’t able to do it exactly. You have to fall back on an approximation. That’s a negative attitude, based on the ingrained, and even unconscious, attitude that the exact answer is the ìrightî one. If you have to approximate, you fell short somehow.

It’s a negative attitude, it’s very bad for self-esteem. Why should electronic design engineers have to suffer under that overwhelming feeling of negativity and failure? That’s what conventional psychotherapy is about, overcoming that, going back to see what the cause of it was and once you understand the cause, then you can start moving forward in a more positive direction. That’s why I borrowed those words, ìFreudian regression.î

It‘s the same thing.Mathematicians have told us from high school you have to have as many equations as you have unknowns. Otherwise you can’t solve it. That’s a negative attitude, but engineers never have enough equations. That’s the definition of the game, there aren’t enough equations. So there’s no value in starting with that negative approach because you’re always looking for those exact equations and you’re never going to find them, so you never have a positive outlook on your job. You always have, even if unconscious, you always have the feeling that you’re trying to solve an impossible problem. No wonder we have all these difficulties.

The positive approach is Yes, we don’t have enough equations, so what else should we do? And you do everything you can, assumptions, tradeoffs, approximations, simplifications, you do anything you can and being a good design engineer is getting good at doing that.

Throwing it into the computer is not the way to go, because you have to change your definition of what is the ìrightî answer. The right answer is not the exact answer. The right answer is one that is approximate within the degree of tolerance allowed by the specifications, as quickly as possible.

That’s what the right answer is to an engineer, and you’ve got to bring to bear a lot more tools than mathematics. You have to make the math work for you as I’ve said many times, you have to be the master of the math, you have to make it work for you. You’re not the slave of the math. You don’t bow down to what the math tells you.

If you make the math tell you what you want to know, you control it. That’s what being an engineer is and that’s a positive attitude, it’s a different perspective altogether, and that’s the way you get results. If you’re always fighting this impossibility, what a dull, deadening career that would be. Algebraic paralysis is the first and most obvious symptom of this disease, this mental disease, that we all contract in high school, and that’s why you have to undergo technical therapy to turn all of that around, 180 degrees you have to turn it around. That’s all I try to do. That’s my philosophy, which is still changing every time I say it. It’s always different.

What I would like to do now is make people aware that this kind of approach is useful and beneficial to them in their job career as well as beneficial to their employer company and these are approaches and techniques which I have found useful and I hope you will also. You don’t have to take them. These are suggestions. There’s a lot of subjectivity.

But it‘s an uphill job, because when I write brochures it doesn’t carry much weight. Of course I’m going to say how great Design-Oriented Analysis is. That’s why in the new brochure, there’s two whole pages of quotations of people who have taken the course, to get an independent opinion because that carries much more weight than what I say. The more the ‘gospel’ is propagated independently, the more likelihood there is more people will say, “Yes, this might be useful”.

I’m only going to do my second-level course at Caltech for one more year. Starting in 1994-95, I’m going to take over the first-level course. Everything has sort of converged on me to the point where up until now I’ve been saying “I don’t want to develop that first level course, I want someone else to do it” but I finally have kind of been pushed into the situation that I’m going to have to do it, so after next year the present course is going to go into a ‘graceful retirement’, and I’m going to do the first level course. So I’m going to try to do what I’m telling the professors who will come to the Workshop to do: Can you use these approaches in a first level course?

In the first-level course at Caltech, I’m going to try to head off the students, to try to head off the need for later ‘therapy’. To try to do it right the first time. I don’t know whether that can be done, that’s what this professors’ Workshop is for and I certainly can’t do it alone, so I don’t know that it’s even possible because the whole format of my present course at Caltech and in industry is you’ve had it all before, but this is a better way to look at it, but now I’ve got to see can we teach it this way the first time around?

There’s always tremendous room for argument and discussion about how to go about doing this, but at one extreme, what I am proposing is nothing radical at all, it isn’t even changing much about the details of what is done in the first level course. It’s simply enclosing it in this Design-Oriented Analysis philosophy. And it really doesn’t take much to shift that point of view, to get the words in there, to set up the objectives, it doesn’t take much time. It just takes a lot of repetition. That’s the one extreme.

The other extreme is that the details of what is done could be replaced by these shortcut techniques and that’s where the difficulty comes in at the first level course, because I don’t think its the proper solution to replace the formal conventional techniques with these. I think it has to be done in addition. I don’t think you can throw out the conventional approach, but there are two things to tweak: one of them is you’ve got to add the short cuts, the other one is the whole thing has to be enclosed in this Design-Oriented Analysis philosophy.

The way my course is now, because it’s second level or ìtherapyî, I can say all the time, well you know all this basic stuff but this is a different viewpoint; but when it’s a first level course, you don’t know the basic stuff already and the question is can you do them both together or can you mix them in an effective way and that’s what I don’t know yet whether that’s even possible. I know several of my former students who have become professors elsewhere, Vatche’s only one of them, there are several others, they said they tried to put this into a first level course and have not found it successful, that they had a hard time doing it.

So there certainly is no guarantee that this objective can be achieved, but I’ve adopted that as my next goal to try to see whether it can be done, although it’s in parallel with doing the animation stuff. In closing, one should recognize that opportunities in one’s life do exist or present themselves from time to time. Many have worked hard and prepared for these opportunities and were able to take advantage of them. But others have not been prepared and were unable to take advantage of whatever opportunities they encountered.


K. Kit Sum, ed., Recent Developments in Resonant Power Conversion
published by Intertec Communications, Inc., Ventura, California

Related Articles:

Power Conversion Synthesis Part 1: Buck Converter Design

Power Conversion Synthesis Part 2: Zero Ripple Converters

SPICE Model Supports LDO Regulator Designs

Power Converter Synthesis Part 4: Near Zero Emissions

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.