I learned about 8877/3CX1500A7 manufacturing woes from a friend who
worked for Eimac back then. This is a summary of what I learned:
In 1968 Bill Sain of Eimac-San Carlos developed test models of zero bias
triodes with segmented-oxide cathodes and radial bar grids as reported
in internal report TR69-2. These tubes were made from existing tube
parts initially, to prove concepts of very high gain (for a cathode
driven triode) and low grid interception current. Initially Sain used
segmented cathode tetrode parts and connected control and screen grid to
ground, to make a high mu triode. Then they developed a triode grid for
several variants, based around using 4CX350 and 4CX1500B cathode
assemblies. The former tube was useful, while the higher power tube had
problems. Eventually effort was focused on the 1-2 kW tube and the 8877
was developed in a record 10 months. It was a very good tube from a
design standpoint, originally,
There were several divisions within Eimac and the Salt Lake City
operation competed with San Carlos in some broadcast and medical tube
products. Salt Lake engineers changed the material used for the cathode
heat dam, a disc at the top of the cathode cylinder, as a result of a
cost reduction program. It saved an estimated 35 cents per tube
according to former employee. It was never tested with RF, only in DC
current division tests. Kovar has iron and is very lossy in RF, compared
to the original nickel part. The result was that the part would heat and
expand, shorting the grid to cathode. ETO was making superb rf
amplifiers for MRI systems and private labeling for GE Medical at the
time. A pair of 3CPX1500A7s ran in parallel for 15kW pulsed at ~41 MHz.
Both tubes had to be fairly close in gain, another problem so that
special tube type YU-158 was sold, a selected 8877 for low grid current.
Around 1988-89 tube failures were common in the amplifiers. After about
10 to 20 minutes of RF operation the tubes would short internally
between grid and cathode. The staff at Salt Lake City had not been able
to pinpoint a cause. 8877's returned under warranty tested good in
Eimac's DC current division tests at the factory and many bad tubes were
being sent back to customers without the defect being identified. It was
a vicious circle. Meanwhile, Erbtec, a Colorado competitor, knew about
the problems from GE Medical and developed a single tube amplifier for
the same application using a new San Carlos tube, the YC-156 or YC-179,
both versions of the 3CPX5000A7 triode with much larger cathode area.
Erbtec eventually won the contract with GE Medical, and ETO took a big
loss, essentially because of a small tube change. That really hurt them,
and it was due to a small internal change at Eimac that never was tested
in high power RF situation, only in DC.
Initially the QA specification was for a maximum grid current of 50mA
with 1 amp of plate current (pulsed measurement). The 8877 had to be
assembled with great care under a microscope in order to make this spec.
The alignment of the grid bars with the beams from the segmented cathode
was critical to having a low grid current zero bias triode. While
visiting the Salt Lake City plant, one of the San Carlos technicians
noticed many of the production 8877's were measuring higher than 50 mA
and he asked about it. He was then informed that the test specification
for grid current had been changed to improve manufacturing yield. It was
later discovered that Salt Lake City had a 55 gallon drum full of
scrapped tubes and was not accounting for that poor yield. It was said
that this led to the Salt Lake City division president leaving, although
there could have been other reasons as well. Tubes that shorted were
probably in the 1988-89 production runs.
John Lyles
K5PRO
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|