|
EARLY TRANSISTOR
AND DIODE HISTORY AT BELL LABS Art Uhlir Jr. |
||
|
Oral History – Art Uhlir Jr. (Continued) |
|
|
|
The corrosion chemists of Building 1 were among those consulted.
They suggested applying red lead in the form of a grease. That did push the
ball in the right direction and made plastic encapsulation tolerable for some
applications like transistor radios.
Eventually it proved habit-forming for all kinds of germanium
transistors to be hermetically sealed in a desiccated atmosphere. What was your next assignment
after working with the junction transistors? I was given the responsibility for a new task under a US
Government Joint Services contract. Task 8 was to investigate if p-n junction
technology could do something for radar receivers. The Signal Corps was the most continuous sponsor of Task
8. At that time, they had furthered
the development by Philco of the 1N263 point-contact germanium diode, which
appeared to give a better noise figure than any but the most selected silicon
point-contact devices, the later having roots to work done in the 1930’s at
Bell Labs Holmdel. The Bell microwave-relay receiving mixers gave adequate performance with the Western Electric silicon point-contact diodes. But the new TH 6GHz transcontinental microwave relay system included a transmitting upconverting modulator designed around a special silicon point-contact diode that could not be reproduced. At this time relay station sites had been acquired; traveling wave tubes and other circuits were in serial production; problems with antennas were solvable. So Bell Labs was uncommonly agreeable to undertaking this government research on diodes. |
So I started with p-n junction theory, modeled a hemispherical
(P+)N junction diode, and predicted small-signal microwave impedance. The
computations would have been tedious but we saved effort by first learning to
program in Bell System Interpretive language. The IBM computer performed three floating-point operations per
second and occupied a good-sized room. We made some gold-bonded germanium diodes in cartridges used for
Western Electric point-contact diodes. The Model Lab mounted dice of the
specified resistivity on diode cartridge parts. A slightly bent whisker of
gallium-doped gold wire was mounted in another part. This wire was at hand
because it was used in bonding to the grown-junction NPN base region. We
confirmed that the 1 GHz impedances versus bias agreed well with the
theory. For the first diodes to be
tested in the system, I insisted on cutting the wire with a razor blade to
make a chisel point to give the right size after bonding. I bonded them with
one pulse from a circuit that I made with a telephone relay (from the
stockroom, of course). Go
To Uhlir Oral History, Page 7 |
|
|
COPYRIGHT
© 2004 by Jack Ward. All Rights
Reserved. http://www.transistormuseum.com/ PAGE
6 |
||