I enjoyed your February 7, 1994 column about the testing controversy. I helped to develop a very sophisticated microwave instrument when I worked for the High Priced company. It sold for about $45,000 at the time (and is over $60,000 now). The instrument was very complex, so every unit was placed in a burn-in chamber for a week.
The production manager, ever searching for a way to maximize profits, saw all of this finished goods inventory sitting there, wasting floor space and representing $$$$ that were not flowing into the corporation in a timely manner. So he decided that this burn-in would no longer be required.
Luckily, we kept data (failures, test data, etc.) on all of these instruments, so I was able to show that each and every instrument failed at least once during the burn-in period! Because the standard warranty period for the instrument was one year, I was able to show that it cost the company less (due to warranty costs) to burn-in the instruments than to eliminate the tests. And we had a lot fewer unhappy customers. Logic and the bottom line prevailed when I showed this data to the division manager. (The production line was not charged for warranty costs in the accounting system, so the production manager did not see these costs in his budgets.)
The moral of this story is that complex instruments fail in complex ways, and that it is important for the manager that produces the unit to be charged for the problems.
We also used the failure data to redesign some circuits when a design problem could be identified. Many problems appeared to be simply random component failures. (If a million 1- resistors were placed in a series combination, and if the failure rate of each resistor were 1 X 10-6, what is the failure rate of the resistor string?)
Larry Martin Consulting
Gee, did Mr. Deming ever decree, "Thou shalt not burn-in your equipment before final test?" I don't think so! If those 1- resistors tend to fail short, that's not so bad. If they fail open, you'll be busy trying to keep that resistor working!----RAP
"Floobydust." I first saw that word sometime in the early 1970s. It was used in one of National's application manuals. I plowed through my library and came up with Linear Applications, Handbook 1, dated 1973, and Linear Applications, Handbook 2, dated 1977. The 1973 book is rather battered, worn, and wrinkled. The pages are yellow and marked here and there with notes written in red. The 1977 book also is beat up, but the pages are still white----a better quality paper. Sitting adjacent to Handbook 1, on its right side, used to be an application manual that concerned itself with audio. In 1978, I made the mistake of loaning the book to a kid just coming along. Always a mistake, because loaned books are lost forever. In any case, the word FLOOBYDUST was used in that book as the title of the section describing various signal distortions that could be induced, and the circuits used to create those effects.
I'll probably never have a use for that book again, but it is imperative that I have it in my library. So, if you search the archives of National and come across the book, I would certainly appreciate receiving a copy if one is available. (Actually, you can never tell. A couple of weeks ago, I had occasion to look up some data concerning a 6SN7 in the RCA Receiving Tube Manual, circa 1950.)
While I'm at it, I've been trying to drop you a note for some years now concerning one of the helpful troubleshooting tricks described in your series on analog whatever. I remember your suggestion of shunting points in physical circuitry with handheld radial capacitors. Well, sometimes oxidation prevents electrical contact at the end of the leads and whatever it is you are trying to shunt. It is best to take a pair of cutters and cut the leads at a sharp bias, say 60 degrees. This puts a sharp point on the end of the leads, and very effectively pierces any insulating coating that may exist. Blood is very corrosive, so care is required. When not in use, the capacitors can be stored with the end of the leads pushed into a cork. That's a good storage device for test leads employing steel phonograph needles in chucks as probes. I wish I could find some of those probes. Maybe you'll come across a couple of those while searching for the Manual, if you decide to look.
PETER L. KROHN
I agree, old books are valuable, and hard to keep track of. I'll send you a photocopy of the old Audio Handbook for your library. Yes, a capacitor with its leads cut diagonally does make a better probe than other components. I have a collection of old corks for that use, too.----RAP
I purchased a "GreenPlug" for my refrigerator and I am very pleased with the improvement. The refrigerator runs quieter and the automatic icemaker works better than ever.
Supposedly, it is based upon NASA technology (whatever that means). I suspect it is a simple filter, or a circuit based on fuzzy logic, etc. Could you enlighten me on its operation? Does it really save electricity? Outwardly, the claims seem too good to be true, but it is doing something.
I have a difficult time believing a device so inexpensive (less than $30) works so well. I would appreciate your comments, and perhaps a solicitation of comments from others who have used the GreenPlug.
I've been hearing about the NASA circuits for a long time. They couldn't make it work 20 years ago; maybe now they finally got it working. Ask the manufacturer for a copy of the report from the independent testing lab, which will prove that it saves you energy and money at high line and low line. Or, wait to see what "Consumer Reports" says.----RAP
I just read your "What Is All This Apples and Oranges Stuff, Anyhow?" article in the May 2 issue. Amen.
I think the best example of the problems of JIT for the electronic industry, and how 3- or 6-month lead times for ordering is shooting it in the foot, is my recent attempt at getting a small quantity of a part for personal use.
I got a flyer from a company (not NSC) talking about a new ADC. It would be perfect for a project I was thinking about working on, if I could only find a cheap, simple 12-bit ADC that talked to the Motorola 68HC11 without a lot of brouhaha. This part did it all for about $20.
I called the distributor. No, sorry, none in stock, three months lead time. I called several places, such as Digikey. No, sorry, none in stock, three months lead time. Well, most of the time, if I had to wait three months before I work on a project, that project doesn't get worked on, and I did want to try this one. I called the manufacturer. They surely could sell me three of the parts.
No, sorry, none in stock, three months lead time. BUT they forwarded my call to the samples department. Sure, two pieces for free as samples, no problem.
What a wonderful system. Here I was ready to give someone money for something and I couldn't get anyone to take it. If only the IRS worked on JIT. A company is actively advertising its products, but was not planning on having any ready to sell for the next three months. Amazing.
This is not the first time I have run afoul of long lead times and ADCs. That was many years ago when I blew up the only 12-bit ADC I had on hand, and learned it would put my PhD research back 6 months waiting for another. What I learned from that was that an 8-bit DAC and a comparator, along with some small bit of software, makes a dandy successive-approximation ADC, and that systems with enough noise don't need the extra four bits anyway.
IC availability comes in only two flavors: the parts are available and the customers aren't, AND vice versa. Sigh.----RAP
I was delighted with your Pease Porridge feature in the May 2 issue of Electronic Design. Our country is awash in wannabe educators who believe simplified approaches are warranted....lest we overstress the student's gray matter and turn 'em off. About 1/3 of the way through your article, I was thinking "I need to send Bob a copy of The Goal..." I could not have been more pleased to read your recommendation of this fine work later in the article.
You've reinforced a growing concern I have about our passion for simplistic models of complex systems. Case in point: Our government is spending billions in public and private money ostensibly for the enhancement of aviation "safety." An ongoing endeavor is to electronify the task of keeping airplanes from getting together in mid-air. This event is so rare statistically that rational analysis of cause and effect is very difficult. Taken as a whole, if two airplanes munched together every day, you'd still have to make a flight a day for years before standing a good chance of being an attendee at a disaster. (OOPS! Pardon my simplistic model.) The causes are so variable and intertwined that taking control of independent variables is virtually impossible.
Nonetheless, we will have some form of warning system in the cockpit. Pilots will report making all kinds of evasive maneuvers as suggested by the squawking device on the panel. Obviously, each action will narrowly avert a disaster so that politicians will deem the project a grand success and justify expenditures. However, I suggest that pilots will (1) become complacent and abrogate yet another pilot responsibility (keep your head up and eyes open) to a black box, and (2) become tired of making so many evasive maneuvers and begin to ignore the equipment. In any case, the newcomer to the cockpit will become yet another distraction and raise, not reduce, the probability of disaster.
Many people trust the ability of news "analysts" and politicians to pass judgment upon cause and effect based upon no science, no hard data, and, therefore, no real analysis. Risk assessment is never a part of the thinking. We'll spend billions (to save a human life or two) and accomplish little or nothing in aviation, while failing to spend a few millions on roads and bridges which would probably save many, many more lives.
I am working with an ad hoc consortium of designer-analysts to propose a very economical application of emerging technologies (GPS and gigahertz communications products) to give pilots a situation display in the cockpit that keeps a gray-matter CPU in the loop and the FAA out of the loop. The problem has been to get the government's sincere attention. They want to build more equipment, hire more people, and spend more money....not less!
So you see, Bob, fuzzy logic isn't really new. Clergy and politicians have practiced it for centuries. Business schools have only become infected in the past few decades. Your reference to old formulas and cookbooks gives them too much credence. The word "obsolete" implies validity in times past; many such crutches were never valid. Thank you, sir, for being a good observer, doing sound analysis upon observations and having the courage to write about it in the face of authority.
ROBERT L. NUCKOLLS III
I still think technology may help prevent certain types of aircraft collisions. But keeping your eyes peeled and your neck swiveling is still the best defense.----RAP
All for now. / Comments invited! RAP / Robert A. Pease / Engineer
Mail Stop D2597A
P.O. Box 58090
Santa Clara, CA 95052-8090