Richard Knoppow wrote:
Bill Hewlett's original patent on the Wein bridge
oscillator is USP 2583649. It can be found on either the
patent office site at http://www.uspto.gov or on Google
patents. Google has the advantage that one can download a
PDF directly. There is likely also additional stuff on the
Hewlett-Packard Memory Project site at: HP Memory Project
I don't think they have the Stanford thesis there but I
have downloaded the IRE paper somewhere, probably Google
could help find that too.
The performance of late versions of the well known -hp-
200CD could be made much better than the specs by careful
choice of tubes and lamps. The lamps especially have a
significant effect on the distortion. Stock lamps were aged.
I now forget the exact method but it seems to me that it
involved running the lamp at red heat for a few minutes and
reducing temperature slowly. This probably relieved strains
in the filiment. In any case I was often able to get the
distortion down to perhaps a quarter of the spec. For very
special applications where very low distortion is required
one can reduce the feedback level. This will cut down the
distortion but also makes the oscillator have amplitude
instablility so there is a limit. I seem to remember an ap
note covering this but its been forty years since I worked
for -hp-.

I am VERY interested in any information anybody can come with about
using light bulbs as nonlinear elements, especially any good mathematical
models of light bulbs and any information about proper burn-in of lamps
to stabilize their characteristics.

The patent doesn't talk too much about that, and I found nothing on the
HP Legacy website.

I have been trying for the past ten years to find any original work on
light bulbs as nonlinear elements and have consistently come up dry.
Plenty of people use them, but it seems like a lot of trial and error
is involved and no real systematic modelling has been done. I'm very
surprised, especially given how good HP is about such things.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."