interviews

It seems like such a simple thing to do that it's incredible that it wasn't the norm.

Oh, I've seen faulty hardware but it's inconceivable that, unless it was a very obscure bug, it got into the wild.

I remember a lot of failures on early digital ICs of all sorts, like RTL and early TTL, and especially pmos things, like eproms and shift registers (in video terminals) and early CPUs. Linear parts weren't as bad.

Nowadays, part failures are rare.

Well, that's the story so far, I suppose. Come back in 100 years time and just maybe it will have changed.

Like many laymen, I am suspicious of final theories because they always seem to turn out to be not quite final.

Cheers

I see a lot of Mostek ROM failures in the older Tek scopes and also in HP equipment from the 80's and early 90's.

tm

Let's agree to have a beer in a century or so and discuss that. ;)

Cheers

Phil Hobbs

cal by the symmetry they display, and the result is the entire structure of matter, for a start.

fermions e.g. electrons and bosons e.g. photons, are different from classic al Maxwell statistics precisely because the particles are identical. That means that if you exchange any two particles, the observables have to remai n the same, which means that the wave function has to remain the same excep t for a phase shift. (Observables aren't directly sensitive to the phase s hift, though it can be measured.)

rdinary density, owing to the Pauli principle. Pauli showed that the antis ymmetry of fermion wave functions under exchange (i.e. they change sign whe n you exchange electrons) leads to the wave function vanishing if you try t o have more than one electron per state. (Two if you count spin.)

en Bose-Einstein statistics, everything would collapse to nuclear density i f not higher. (The density of the nucleus is also set by the Fermi statist ics of the nucleons.)

My point exactly, though I didn't express it nearly as well.

nd just maybe it will have changed.

Theories changes, but the experimental results that they explain don't.

Non-identical electrons seem unlikely.

Perfectly correctly. But - like most laymen - you don't understand how they change. Some concepts are a bit more fundamental than others, and the basi c idea that fundamental particles are identical has had a long and producti ve history. Democritus (c. 460 - c. 370 BC) seems to have been the first to get to the basic idea and Lucretius (c. 99 BC - c. 55 BC)got quite a bit f urther.

Ernst Mach (February 18, 1838 - February 19, 1916) was the last serious phy sicist to be sceptical about the existence of atoms, and his failure to gra sp Einstein's "atomic" explanation of Brownian motion suggests that he wasn 't all that serious, as physicists go.

Gee Syd, any engineer who can't appreciate approximations, and the usefulness of rules of thumb within those constraints, must be a bad engineer. :)

That seems like a misconception. Theories that fully explain their corresponding observations -- that is, given their experimental error -- never stop being correct for any experiement of the same caliber. Newton's mechanics and optics are still used to this day, and haven't gotten any less accurate over time.

The experiments have gotten more precise, allowing for such observations as relativistic time dilation, which makes Galliliean relativity invalid to that level of precision. But the proceeding relation is no less true; it is embedded within it, for example approximating c-->infty.

Maxwell's equations have yet to be invalidated. But they have been modified, first into QED (E&M interacting with charged-matter-as-we-know-it; relevant to chemistry), then electroweak (only relevant to nuclear reactions), then QCD (the barest building blocks of matter-as-we-know-it, as well as a whole lot of structure intimately related to, but utterly absent from, matter-as-we-know-it-daily). There may be adjacent or higher level truths to be found, but they will never invalidate these theories, which have already been proven to high confidence levels at a present-technology level of precision.

Tim

You think perhaps the semi companies have learned a thing or two in the last fifty years or is it just because PMOS sucks?

If you look at microscope shots of early ICs, they were terribly cruddy, gross defects everywhere. Half of CMOS is PMOS, and it's very reliable these days.

Early mosfet oxides were dirty too, with nasty popcorn noise, threshold shifts, punch-throughs. And electromigration, oxidation, purple plague, bad wirebonds, bad epoxy encapsulation, all kinds of problems.

It takes an enormous amount of discipline to run a good semiconductor fab. Those process engineers are seriously compulsive types.

A silicon boule is sort of bumpy and lumpy, and it's sliced and ground down into uniform-sized round wafers. In the early days, Intel debated whether to use oddball-shaped wafers, so as not to waste any of the silicon.

National Semi eeproms are failing in Tek SD-series sampling heads, which were early/mid 90's vintage.

EEPROMS? 20-year retention? Sounds reasonable.

ay in many applications.

Way back in the 1960's I saw a course in " Vacuum Tube Technology " and th ought I ought to take that course. Vacuum tubes were going to be obsolete in another year or so.

Well it turned out the course was taught by a man from Varian and everyone in the class was employed by a vacuum tube manufacturer, but none of them had anything to do with receiving tubes. We learned about TWT's BWO's, Mul titrace CRT's ( Like a CRT with 21 gun's ), klystrons, Magnetron's.

As one of my college professors said, one tin plating plant uses more RF po wer than all the commercial broadcasting stations.

Dan

speaking of MOSFETS, did you say the inputs of your Keithley electrometer were MOSFET? I hought MOSFET were too unstable for that kind of signal detection. (Though it always made sense to me bec. of the low leakage current.)

jb

An interesting feature of QM theory is consciousness - what it is, what role it plays in QM. If there is a tie-in, as some theoretical physicists think, that might require some changes in theories. The QM aspect of the syntax is an 8 angstrom channel that houses a neurotransmitter. But how quantum coherence could be maintained over large structures like the brain seems unexplained.

Yes, small-signal mosfets, a matched pair.

It is not interesting in the slightest for QM. Consciousness plays no part whatsoever. Only the fringe physicists are deluded as to this matter.

To wit, a quote from my QM ensemble interpretation paper:

"This one is, frankly, quite daft. It makes no difference whatsoever, whether the physicist observes a double slit experiment or not. If he is outside smoking a cigarette, rather than watching his equipment dials, it makes not the slightest difference to the result, and never has such observer created reality ever occurred. The "observer" is the physical setup of the equipment, not the conscious observer."

Consciousness, as first proposed by David Chalmers, is a new property of the universe, not derivable from physics. My take on that is here:

Kevin Aylward B.Sc.

Yes, mosfets or back then more often called IGFET. Check the number of front panel "zero" adjustments: coarse, medium, fine.

piglet

just using "double" floats.

Yes, small-signal mosfets, a matched pair.

Any more info on which ones? I could prolly get schematic from Keithley... JB

R Lee Ermey, USMC Sargent on the faithfulness of dogs:

"If you lock your wife and a dog in the trunk of your car for an hour, the dog will still be friends with you."