Tony Duell wrote:
Actually, this reminds me of another
application... the audio oscillator.
As is well-known, it's very difficult to make an adjustable LC oscillator
at audio frequencies. An RC oscillaotr is possible, but stabilising it
(so you get a reasonably non-distorted sine wave output) is the problem.
Now many yeats ago, a guy working for Prof Terman (you do know who I
mean, right...) solved this problem. He used a Wien bridge circuit as the
'resonant element', in the feedback loop of a *3 amplifier. The gain of
the latter was stabilised by a light bulb as a non-linera resistor with
just about the right time constant.
Anyway, I read somewhere (possibly in one of Bob Pease's columns) that
somebody had tried to improve upon this circuit using a more modern
approach, an FET to cotnrol the gain, a carefully-designed control loop,
etc. The result was a an oscillator with a more distorted output than the
simple light-ulb-stailised one...
This particular subject is near and dear to my heart.
The incandescent light bulb linearization technique is attributed to
Bill Hewlett, and was presented in his Master's thesis at Stanford
That's why I said 'you do know who I mean'. Of course it was William Hewlett.
University in 1938. The design was commercialized
into
Hewlett-Packard's first product, the model 200A audio oscillator, in
1939. (was Hewlett studying under Terman at the time?)
I believe he was. I think there;s a famous paper with both names (and
others) on that describes the technique.
Incidentally, Prof Termna's books, particularly 'Measurements in Radio
Engineering' are well worth tracking down and reading. I wish I'd had a
copy to throw at an idiot physics teacher at school... (I have it now).
OK, I'd better explain that last remark. When I was at school, we were
asked, for homework, to describe a method of measuring a capacitor. Now,
there was a totally useless method taught for this, involving connecting
the capactior to a virating reed swtich (energised at a known frequency),
in one position the capactiro is charaged from a known voltage, in the
other, it is discharged through a milliameter. The idea is that from the
current flow, you can work out how much charge is transfered per cycle,
then since you know the voltage, you can work out the capacitance. Now,
this method makes the dubious assumption that the capacitor is fully
charged and discharged each time, and that the frequency, charging
voltage, and milliamter caliration are all known accurately.
Of course I didn't describe that method. I picked one of the many AC
bridge circuits, and descried that. The teacher objected, saying that it
was a compariston and not a measurement.
Now, I can't think of _any_ measurement (in the scientific context) that
doesn't involve compariston to a standard. And there is no requirement
for that standard to be of a different type to the quantity you're
measuring (in fact in many cases it's better to measure something agaisnt
a standard of the same type). But anyway...
Prof Terman's book regards bridges as measuring circuits.... And any (UK
school) physics teacher who thinks he knows mere about electrical
measurements that Prof Terman is almost certainly deluded.
Later, well-known analog electronics god Jim Williams (staff
scientist at Linear Technology) developed a modernized equivalent of
that circuit which actually does manage to outperform it by a
considerable degree. It uses several very high-tech op-amps, along with
an analog optoisolator (an LED optically coupled to a CdS photocell), in
a fairly elaborate circuit. It is far, far more complex than Hewlett's
original oscillator, but it does work amazingly well.
Taht is proably the thing I was thinking of. I certainly remember a
comment that at least one 'improved' circuit, although more complex than
the light bulb, gave an inferior performance. But equally, I can believe
there is one that does better.
Williams presents this circuit (and many pages of
very interesting
reading, as he describes the process by which he arrived at his final
result...ideas, testing, failures, successes, optimization, etc) in the
absolutely fantastic book entitled "Analog Circuit Design: Art, Science,
and Personalities". I strongly recommend that book.
Thanks. I will certainly check it out. I've never heard of said book before.
Bob Pease may have done something like this, but I suspect you're
thinking of Jim Williams' work.
Almost certainly. Maybe Bob refered to it, maybe I am just getting
confused (the latter is very likely...)
-tony