Electronic Design

Bob's Mailbox

Dear Bob:
I very much enjoyed your column, "What's All This Taguchi Stuff, Anyhow? (Part II)," which appeared in the June 10 issue. A major function of a voltage regulator is to make a steady output voltage when faced with changing input voltage.

Taguchi ignored this major function and managed to design an active voltage divider. After looking at the circuit optimized by Mr. Taguchi, me—thinks that a passive voltage divider made from a few resistors could have done the job about as well. A cheap, simple, and reliable passive design like this would reduce the "losses a product imparts to society" even more.

In the zeal for demonstrating his tool, Taguchi threw out a major function of a voltage regulator, throwing out reality and thus losing sight of what tools are for—to help design and build REAL products. It seems that the tool has become more important than its use.

He goes on to demonstrate his disconnection with reality a second time—the optimization showed the output to be much more stable than the reference, and alarm bells didn't go off in his head!

Perhaps there are useful techniques to be extracted from the Taguchi method. Too bad they are all mixed in with so much horse exhaust. Even worse, the proponents of the method don't know that anything smells.

Keep up the good work, and keep encouraging us to get our hands dirty by using the best circuit simulation tools available—the parts and a soldering iron, along with a good dollop of understanding and common sense.
- Donald Borowski, Hewlett-Packard Co., Spokane Div., Liberty Lake, Wash.

Thanks, Donald. We agree on many things. But a resistive divider would work just as badly as Taguchi's "optimized regulator" in a TV set.—RAP

Dear Bob:
Regarding your recent column on Taguchi..........

YES!
- Bob Dobkin, Vice President, Engineering, Linear Technology Corp., Milpitas, Calif.

Eloquent, man! Thanks.—RAP

Dear Bob:
Your recent article "What's All This Taguchi Stuff, Anyhow? (Part II)" was an absolute classic. I hope you've succeeded in opening the eyes of at least a few of the followers of the pious Dr. Taguchi and his acolytes.
- Frank Latos, Duo Systems Inc., Farmington Hills, Mich.

Thanks for your perceptive comments. I could never have zapped them like that if they hadn't taken themselves so seriously!—RAP

Dear Bob:
Thank you for a most entertaining diatribe, re Taguchi (Pease Porridge, June 10, 1993). On the instinctive level, it seems obvious that a regulator cannot produce a stable output voltage with such lack of dependence on the circuit's component values. How does it get to know what 115 V is, then?

Can it somehow pluck "115 V-ness" out of thin air and match its own output against that? If Taguchi were an engineer, he would have realized that this was what he's asking his "optimized" circuit to do. I would be embarrassed to even claim it.

Logically, something internal to the circuit has to be ACCURATE. A good design reduces the need for accuracy down to a minimum of well-controlled components.

These components (resistors, Zeners, etc.) may be part of a feedback loop for improved circuit "robustness" against outside stimuli and can be cheap, but they are the core of the circuit and constitute the reference. So, Taguchi can design a regulator without a reference. Hmm....

Whenever someone such as Professor Barker starts to talk about selecting transistors for Beta (or hFE), alarm bells immediately ring. This is basically the last thing you want to do. It's going to be expensive come production, and the designer will regret it. I don't mind someone specifying a minimum hFE in their design, but never a maximum.

Any design has to cope with the full gamut of parameters that the transistor's spec cares to fling at you, and it's safer to assume that a component manufacturer will aim high in terms of hFE.

Anyway, what happens to this selecting hFE policy under the conditions of varying temperature? I shudder to think what an environmental proving program would reveal on a Taguchi/Barker circuit.

The great danger of such charlatanism is that it might be believed. There are students out there with no internal reference against which to check the thinking.

Students are stuck with having to believe what they're taught. Back in my university days, I would have swallowed it.

Worse still for us engineers is the potential for gullible managements to insist such gymnastics can be performed. There are companies managed by people who are only too willing to believe in panacea, and then make it look bad for engineering if it cannot enact the miracles. Tail wags dog.

Thanks for your column—it provides a stable reference against some pretty major external perturbations. Long may you continue to unmask the perturbators.
- Paul J. Robertson, BS c.AMIEE, Principal Engineer, GEC-Marconi, Basildon, Great Britain

You're right! If Taguchi's circuit did meet the system needs, it would not be cost-effective, but VERY EXPENSIVE. I have no idea where he got the notion that tight selection of Beta saves money! In high-voltage transistors, Beta does not get high, anyhow.—RAP

Dear Robert:
I've read your articles with great pleasure for many years. I especially enjoyed your "Taguchi, Parts I and II." I had lunch with G. Taguchi in Kansas City last month during a speaking tour, which I think was to promote his latest book. He's not a terribly interesting luncheon mate, believe me (I think he was being paid by the word, and he was getting close to his budget). But he got quite animated when I asked if there was any truth to the rumor that he was going to appear on "Larry King Live" later that week.

I've often wondered why the Japanese hold the writings of these gurus in such high regard, and with such little scrutiny. An answer was suggested—of all places—by a response that Dr. Dean Edell, M.D., gave to a question posed by a caller to his daily radio talk show: "Why doesn't any significant medical research come out of Japan?" Dr. Edell made the observation that Eastern cultures foster a strong reverence for their elders, and that scrutiny and criticism, which are an integral part of the Western scientific method, are considered rude and disrespectful. This may explain why Taguchi's writings are full of errors and omissions.

For an honest, straightforward approach to optimizing designs, I would recommend Keki R. Bhote's book, World Class Quality (Using Design of Experiments to Make it Happen)—ISBN 0-8144-5053-9-AMACOM, 135 W. 50th St., New York, NY, 10020. It uses a common-sense approach to minimizing output variations and doesn't obfuscate the process with bizarre functions or orthogonal arrays. It's a good one. I should know, I've got a Master's Degree...in Science!
- Jonathan Jambor, Wichita, Kans.

I won't fault Taguchi for not being a great conversationalist, but on everything else... I agree that Mr. Bhote's book is excellent—he uses graphs, statistics, and common sense to solve problems that Taguchi's Experts never could. He talks about "knobs" that engineers can grasp to control systems. My kind of thinking.—I recommend it. —RAP

Hide comments

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.
Publish