|
EARLY TRANSISTOR
AND DIODE HISTORY AT BELL LABS Art Uhlir Jr. |
||
|
Oral History – Art Uhlir Jr. (Continued) |
|
|
|
When I plugged in the diode, it immediately gave the 5 or 6 dB
conversion loss that would serve the system requirements. When the technician
asked, “Are we done now?” I said, “Try for less loss.” That was easily
accomplished with a little more tuning. So I said, “Try for gain.” I was satisfied by a few decibels. This
was in 1953/54. The microwave relay thus could proceed as planned, with
germanium if necessary. But our team moved on to diffused silicon diodes as
the final form. I came to realize
later how idyllic it was for a device developer to have an eager customer
with a complete instrumented test set warmed up and waiting to test whatever
you made. The “transmod” worked at relatively high signal levels and noise
was not a concern. Under Task 8, I developed a noise theory for frequency
converters that used an ideal p-n junction diode. It showed that the shot
noise vanished as the diode impedance became capacitive, regardless of the
circuit. It also showed that if the junction design favored a resistive
impedance, low-noise frequency conversion could be obtained (without gain) if
and only if the local- oscillator drive made the diode impedance transition
quickly between forward and reverse extremes. [REF 3]. This paper was
widely overlooked and rederived a decade later in the same department [REF
7]. |
The prediction of very low noise was greeted with skepticism
when presented to the Device Research Conference: “We at *** have tried it and it is very noisy!” That was a
great motivation for setting up a varactor upconverter from UHF to X-band and
converting the X-band back to a low frequency with a conventional resistive
down-converter. The combination gave a lower UHF noise figure than the best
production vacuum tubes. Varactor
parametric amplifiers were quickly prototyped at Bell Labs and many other
places. Diffused-silicon mesa diodes were made for these prototypes. Beyond doubt, the skeptic's
problem had been testing the capacitive diode in a conventional receiving
down-converter. When tuned for gain (or even low loss), such a converter
greatly amplifies noise generated in the following intermediate-frequency
amplifier. The same negative result was gotten in the wartime down-converter
experiments with North's welded-contact germanium diodes. [REF 8]. The problem thus was in the circuit, not in germanium which
could have been better than silicon for cryogenic operation. Except that
gallium arsenide was found to be superior to both in temperature range and
high-frequency electrical performance. Go To Uhlir Oral History, Page 9 |
|
|
COPYRIGHT
© 2004 by Jack Ward. All Rights
Reserved. http://www.transistormuseum.com/ PAGE
8 |
||