Few people are born with a profound ability to understand things. Im
certainly not one of them. Ive learned a lot of things especially about how
to use the intelligence I have.
One thing Ive learned is that people any people, anywhere, any time have
great difficulty getting past what they believe to be true. I know it, so it
must be true. Dont disturb my complacency.
Anyone with a disruptive point of view is usually rejected as a idiot (qv,
Swifts observation about a confederacy of dunces). Unless, of course, the
person is a perceived expert. If anything Ive said had been voiced by
Doctor Floyd Toole, it would have been accepted as gospel by most of this
group. The principle that truth is truth, regardless of who says it, is
something most people cannot understand. Nor is the principle that one should
understand //why// they believe or disbelieve something, and be willing to
periodically reconsider their beliefs.
The following is about how I solved a serious problem that stymied people who
knew more than I did.
My first real engineering job was with Bendix Field Engineering. Bendix was a
principal contractor for NASAs Spaceflight Tracking and Data-acquisition
Network (STDN or STADAN).
https://en.wikipedia.org/wiki/Spacec...sition_Network
STDN used klystron transmitters to communicate with satellites and space
probes. These put out several kilowatts at around 2GHz. A klystron is a
traveling-wave tube, with multiple high-Q resonant cavities. These have to be
correctly tuned to get full power over a useful bandwidth. A single band wasnt
enough to cover all the frequencies NASA used. Retuning was often needed.
Retuning wasnt horribly difficult. You inserted a tuning wrench to rewind a
cavity to its start position. * Then, consulting a list of settings, you
turned the wrench to a particular reading on a turns counter. The process was
repeated for each cavity.
* There was a lot of backlash when adjusting a cavity.
The major difficulty wasnt the wrenching around, but the fact that most of
the transmitters were in the antennas wheelhouses. You had to climb a hundred
feet or more, sometimes in 60mph winds.
The solution was a motorized tuner. You pressed a button for the desired band
on a control panel in the station, and the tuner did the screwing and
unscrewing, while you enjoyed a cup o hot cocoa. (Actually, just a few sips,
because it worked quickly.)
The klystrons came from Varian, whod already aligned them for the bands NASA
used. After the tube was installed, a step-motor system was attached, and we
dialed in the necessary turns values for each cavity, for each band. Simple, ncest
pas?
No way. It didnt work. The klystrons didnt meet spec. They werent even
close. The bandwidth was usually too narrow, and the overall gain was too low
for reliable operation.
The klystrons were measured much as youd measure a speakers response. A
swept-frequency signal was applied to the input, and the output went to a
oscilloscope sweeping in sync. An absorption wavemeter on the input produced
dips in the displayed response, so we knew where we were, frequency-wise.
https://en.wikipedia.org/wiki/Absorption_wavemeter
I and my co-worker a buck-toothed dullard -- killed ourselves trying to get
flat response at Varians settings. We couldnt. We once spent a whole day on
just one channel, and got it kinda/sorta/maybe flat. We called in
more-experienced engineers, and none of them could figure out what was wrong.
As the installation deadline approached, it was decided to install the
klystrons and see what happened in the field. We had no better luck. At Cape
Canaveral, we had to get NASAs dispensation for some pretty wretched
alignments, which were barely usable.
Note that //many// more-experienced people quite failed to solve the problem
which turned out to be ludicrously simple. I was the one who solved it,
because I asked good questions.
The first was Why dont we talk to Varian? * I called and spoke with Dr
Goldfinger. (Yes, really.) Why arent we getting flat response? Are you
certain youre measuring the tubes correctly?
Oh, we are.
How do you know?
We cheat.
He was joking about the fact that Varian had been measuring klystrons for 30
years, and its test system had been calibrated to a gnats whisker. So the
problem must be with our test setup. The next question was -- Whats the
likely problem?
The test scope needed a DC signal representing the amplitude of the klystrons
RF output. This was obtained by attaching an HP point-contact detector to an
RF coupler in the transmitter cabinet. It turned out these detectors were
utter crap. Not only did some of them not work at all, but their output wasnt
flat, and varied from sample to sample. This caused faulty measurements that
led to our inability to align the klystons.
I suggested against some objection that we use the HP thermocouple power
meter installed in the transmitter cabinet. The thermocouple has a broad,
basically flat response. And no surprise when measured with the
thermocouple, the klystrons met spec right out of the box. (A few channels
benefited from minor touchups, which took less than a minute.)
The irony is that, had we installed the klystrons without measuring them,
there would have been no problem.
I remember an engineer yanking one of the HP detectors off the coupler, and
griping Its no good. The answer was staring me and him right in the
face, but neither of us saw it, because we assumed (theres that word!) there
were only two types of detectors working perfectly, and not working at all.
The thought that there might be an intermediate state grossly non-flat
response didnt cross our minds.
Need I say that I learned a great deal from this?
Case proven, and closed.
* I suspect no one else did, because they were afraid of looking stupid -- the
A man doesnt ask for directions attitude.