Dear Mr. Pease:
An ex-student of mine, Anthony Bentley, sent me a copy of your June 13 article on Fuzzy Logic. Tony was in charge of Feedback Control in the Welding Department at Sandia in Livermore and transferred a few months ago to Feedback Control Applications at Sandia in Albuquerque. He successfully designed automated machines for a variety of welding systems. He feels pretty much the same as you about Fuzzy Logic, and so do I.
You might be amazed at the large number of highly mathematical journals devoted entirely to Fuzzy Logic----mathematics for its own sake. My own research area is in feedback design for highly uncertain systems----linear, nonlinear, time-invariant or time-varying, single and multiple input-output----with emphasis on quantitative design. I have tried to judge F.L. by seeking a significantly uncertain system, showing a systematic F.L. design procedure for trying to achieve desired performance specifications.
For example, consider a 4-by-4 multiple-input multiple-output, highly uncertain, nonlinear, time-varying system with assigned performance tolerances. We have a systematic design procedure that can handle problems such as this. In fact, this particular one can be done in a few hours without even using a computer. Anyhow, after a few hours in the library (besides always being on the lookout for such), I was not able to find a single such F.L. example. This is the case also for "H-infinity," a highly mathematical so-called design procedure for linear time-invariant feedback control problems.
Fuzzy Logic, H-infinity, and neural networks and expert systems provide many opportunities for paper-publishing and government grant solicitation. There is such a huge number of so-called researchers who are "driven" by the need to write papers! These are convenient vehicles for this purpose. There is no concern for practicality. For example, the IEEE Transactions on Automatic Control, gave their best practical paper award, a few years ago, to a paper authored by two professors from the prestigious CalTech, plus one visiting professor from Sweden, and the design had no simulation whatsoever.
When a colleague of mine simulated their design, he found it to be instable. He notified them. They claimed a misprint, and gave him another design. He tried it and found it to be fantastically underdamped and with four infinite-bandwidth, high-gain (72 dB) compensating amplifiers. And this design theory is being extensively taught in graduate courses in the universities! Most Professors of Feedback Control in Electrical Engineering Departments in the universities (it is somewhat better in Mechanical Engineering) know extremely little about practical control systems, and really do not care----so long as they get government grants.
Mr. Horowitz: On the many, many Fuzzy Logic papers filled with hype, I agree with you. But soon we will see GOOD F.L. systems; more later. Meanwhile, every serious control engineer should study Mr. Horowitz's book "Quantitative Feedback Design Theory (QFT) Vol. 1"----I've bought a copy. Request a brochure from the author at (303) 499-5357 or e-mail: [email protected]
I am enclosing an article from Design News that covers the new Whirlpool fridge. It appears they are using Fuzzy Logic for adaptive defrost control. I find this somewhat humorous as I have been working in the HVAC controls business for 10 years and we have always used adaptive PID control in our designs. Does this mean we have been using Fuzzy Logic without even knowing it? I still think I could make the Whirlpool defrost work better without Fuzzy Logic. I kind of agree with you----on a lot of the applications, only the logic of their reasoning is fuzzy.
On the subject of The Audio Critic, I must disagree with their debunking of myths. I have been a high-end audio enthusiast for a long time and to claim that all amps of similar specs sound the same is absolutely false (The Audio Critic has claimed this for some time). What I have discovered is that most high-end amps, particularly vacuum-tube amps, are much more sensitive to loading of the inputs and outputs than mid-fi equipment. Therefore, the length and impedance of the cables you use to interconnect the equipment can have a profound effect on the sound. But to say that you can insert several different amps into the same system using the same cables and have them all sound the same (assuming similar specs) is without foundation. Sounds good in theory, though.
The so-called "holographic" imaging capabilities of high-end components (vacuum tubes in particular) are very real and I haven't been able to duplicate the results with any mid-fi equipment. Maybe there is "magic in the tubes". Anyway, I just thought I would throw in my own hobbyist observations for what they are worth.
I have been building my own audio equipment for 22 years and I have a BEE from Georgia Tech ('84). If you really share The Audio Critic's views, don't feel bad. I studied under W. Marshall Leach at Tech and he believes audio amps of similar specs sound the same, too. Maybe you just have to know what to listen for; until I heard the imaging magic of a system done right, I also was an unbeliever.
SAMUEL E. JONES
I don't recall saying that "audio amps of similar specs sound the same." I'm just a skeptic on all high-tech audio equipment. The only thing I'm an expert on is the splicing of speaker wires.----RAP
Dear Mr. Pease:
A few issues ago you wrote a comment about how "tube sound" could be implemented by including circuitry to produce second-harmonic distortion in a transistor circuit that would otherwise produce relatively clean sound.
I have been thinking about trying this, and I was hoping you might have a simple design you would like to share. Ideally, the distortion would be adjustable and would generate only even harmonics. Ok, Bob, what kind of tricks do you have up your sleeve this time.
Sorry, Jay, I don't know the tricks of how to start with a clean solid-state amplifier and convert it to "the vacuum-tube sound." It's not just the distortion, it's the soft damping and the clipping, too. Can any reader recommend some technical article on this?----RAP
Your article in the Feb. 7 issue on quality scored a bullseye! While in the U.S. Air Force, I was subjected to Deming's philosophy. It was just plain common sense, nothing earth-shattering.
Imagine what would happen if the "No Test" mentality is extended to other industries. Publishing companies can get rid of editors because all of the writers' work would be error-free. Computer software would never give erroneous results or fail to work. Some industries are not suited for Deming's ideas.
I was wondering if you could do an article on mean time between failure (MTBF). I recently saw a new disk drive advertised to have a MTBF of 500,000 hours. That's over 50 years. I'm interested in knowing how MTBF is determined for products. I don't believe that the 500,000 hours was determined from direct testing. Keep up the great work!
Santa Clara, Calif.
You might ask the manufacturer how the MTBF was computed. Also, you can ask how the observed failure rate compared to the estimates when they ran life tests.----RAP
All for now. / Comments invited! RAP / Robert A. Pease / Engineer
Mail Stop D2597A
P.O. Box 58090
Santa Clara, CA 95052-8090