was the master vacuum tube technician at the top of the food chain when
I came to J. R. Pierce’s department at Bell Labs in June, 1952.
He had no degrees, but he was as smart as anyone around so far as
I was concerned. He was
also responsible for the discovery of 2 new elements, heretofore
unrecognized by others, Crud and Gunk, and their compound
Crudnium-Gunkate. He was
also a master glass blower and most encouraging to me with regard to my
interest in it. This was in
sharp contrast to my boss’ attitude, who saw no virtue at all to
engineers catering to manual skills in any respect.
of G.H.’s setups was a scheme for taking what he called “a thermal
fingerprint” of a cathode. The
main business of the department was TWT (Traveling Wave Tube) research
and each tube was a hand-crafted creation considered as a rare work of
art by G. H. and others. The
responsible engineer was primarily interested in the electrical
performance of the device, but G.H. was also concerned with all other
characteristics such as the residual gas species and their partial
pressures and the temperature distribution of all elements under all
conditions. He had a
hand-held spectrometer based on a glass prism and he claimed to be able
to identify species from the display.
The only spectral lines I was ever able to discern, however, were
Sodium lines due to phosphorescence of some Sodium bearing glass.
Years later I had occasion to have another look at “residual
gasses” using some much more sophisticated instruments.
scheme for making a “thermal fingerprint” of a cathode assembly was
a masterpiece of simplicity as well as engineering, in my opinion.
I have used it on a number of subsequent occasions and always
found the data very valuable and inspiring, but some of my superiors
have shown only scorn. Their
loss. The basic idea is
that the heater wire is, or could be, a primary resistance thermometer.
We can always know the heater temperature if we can measure its
resistance. (Most heaters
are pure Tungsten, or sometimes Tungsten + Rhenium 3%).
There may be some question regarding the resistance of the
“legs”. These are the
short lengths of heater wire between the cathode pocket proper and the
heavy leads through the vacuum envelope.
And there may be some issues depending on whether the heater is
“potted”, or not, within the cathode pocket.
But these are minor concerns.
We can account for these secondary effects somewhat if we care,
but when taking “fingerprints” we usually have larger effects to
was primarily concerned with the cooling transient.
In one configuration he measured the heater resistance using a
very low amplitude 1000 cps exciting current and recorded the 1000 cps
voltage at the external heater terminals.
In another configuration he used a very low level DC exciting
current and recorded the DC voltage during the cooling cycle.
My recollection is that both methods gave the same results, but
the 1000 cps excitation could be used during the heating cycle as well
as the cooling cycle. Of
primary concern was the temperature difference between the heater and
the cathode under operating conditions.
Since the thermal mass of the heater was much less than the
thermal mass of the cathode and associated hardware, this difference
could be determined quite accurately from the cooling transient data
immediately after the heater power was turned off and the cooling cycle
began. The heater
temperature would quickly relax to the cathode temperature and slowly
track the cathode cooling cycle.
particular interest was how this cooling “fingerprint” might vary
from tube to tube and from day to day for any given tube.
Of particular interest was the “change of state” in any
Nickel parts. The exact
temperature at which Nickel passed thru the Curie Point (changed from
magnetic to non-magnetic) was well known and became an absolute measure
of temperature at one point. This
transition was usually apparent in the ‘fingerprint’ if one looked
closely and was alert enough.
also explained to me how one could determine the “Electronic Work
Function” averaged over the surface of the cathode, with this setup.
The idea was that when current was drawn the heat removed from
the cathode, in Watts, was just the product of the current, in Amperes,
and the work function, in Volts. The
cathode temperature would dip slightly when current was drawn and the
heater power required to bring the temperature back to its original
value could be measured. I
recall that our boss was not very enthusiastic for this theory, but G.H.
routinely made the measurements and kept the results to himself.
I never took the data for my own edification while I was at Bell,
but I could see the possible benefits.
1989 I was hired at LEDD (Litton Electron Devices Division) to work on
the development of a new klystron.
Congress (in its infinite wisdom) appropriated funds to
“Improve the reliability of certain military systems and to
incorporate new technology not available when the system was first
brought on line”. Among
the ‘new technology’ was Finite Element Analysis FEA.
The U.S. Air Force had made a major advance by means of
installing “sensors” throughout an aircraft to register thermal,
mechanical, or electrical stress and alert the pilot whenever any stress
was deemed to be excessive. Advances
in computer technology had made it feasible to make similar calculations
in all manner of hardware.
main problem with FEA was that GIGO (garbage in garbage out) was
dominant unless you had a pretty good idea that your results were at
least “in the right ballpark”.
Then the results tended to be given significant credence.
I was involved in several such projects and it was always the
case that “old fashioned hand-calculations” were required to make
sure that the FEA approach gave reasonable results.
This was where I came in. In
the case of the new klystron I was working on, the first results from
FEA were clearly out of the ballpark.
I started thinking about G.H. every night when I went to bed.
After a few such evenings I wrote a QuickBasic computer routine I
called “10NODES.BAS” wherein I described the thermal network of our
cathode in terms of 10 Nodes and the thermal linkages between them.
After taking data of the cooling transient, by recording the
(pure W) heater resistance after turnoff, I felt that I had a pretty
good model of the entire thermal network.
When I gave my results to the FEA investigator we eventually
agreed remarkably well. One
of the critical issues was the “fast warmup” characteristic.
It was a system requirement that the tube be in full operation
within a very short time after application of heater voltage.
At turn-on, an over-voltage would be applied for a short time.
Then the heater voltage would be reduced to a steady state value.
had occasion to take such “fingerprints” on 6 prototype tubes over
several years and was pleased to notice that there were no thermal
anomalies… at least so far as I could detect.
This has not always been the case as friends working for other
companies have informed me. One
of the first indications of manufacturing process troubles has been
anomalies in the cathode cooling thermal transients.