sdr16

My earliest recollection, working in Final Test, is of Chuck working on a dud that had bounced back: an Electron Beam Analyzer for Varian. That single-bay-unit was packed with mysteries. Chuck was fixing the pulse, which had a portion missing in the middle. What I saw was two successive pulses. I was seeing it all through the wide-eyes of a child. The unit vanished and I always wondered about it.
     Ten years later I was called to fix the pulse top meter on it. It needed another of Richard-Ziskowski's pulse-top-metering correction schemes. The root of this metering problem centers around how switch-tube voltage drop is stabilized in the design. Intuitively, a tube requires voltage-drop if it is to pass current. If it had a known drop, a battery of equal voltage could be inserted in series to add lost voltage. Approaches diverge from here.
    Old ETM tube-type grid modulators feature seriesed 150 VDC floating supplies, stacked inline, with "pulse" switch tubes, in anticipation of tube conduction-drop. With pulse onset, output heads for "boosted" anode potential. Diode-clamping prevents actual output from exceeding the pulse-top rail. More precisely, pulse output exceeds rail-voltage slightly, because of clamp-diode conduction-drop. Perhaps negligible originally, Cober only had three diodes in series. But our's started with 15 in series. To make things worse, our's needed an upgrade. Richard noticed we only had 750-volts worth of diodes, but we routinely subjected them to 1.15 KV.
    Then a rash of wrath saw Richard doubling-up diode-clamp boards. Diffirence between metered-value and actual-output doubled, to 18 Volts. A 10 volt discrepancy was widely tolerated for grid modulators. But 18 volts was going to be a problem. Richard seriesed an identical number of diodes, so to mimic diode-clamp forward drop. A pullup resistor from the pulse-top-boost-supply sources current that forward-biases the diode-string, returning to the pulse-top rail. Richard then moved the pulse-top meter-divider lead from from the pulse top rail, to the high side of his new diode string. Metering came to agree with external handleld meters. But alas, with time, Richard quit liking the method. Bad diodes in the diode-clamp assembly show up as meter error. Still, it may provide the best pulse-top volt-meter-accuracy available with switch tubes. The Varian Electron Beam Analyzer got it. And the operator was happy.                                Thank you Richard.

I was relegated to cookie-cutter testing out of necessity. I watched as tecks who were not electronics enthusiasts, took shots at doing the big jobs. They came and went. Chuck Rivers, a "superteck" from Stanford Linear Accellerator, was hired to expedite the Crane 50 K. His superteck facade quickly shattered. He was a Scientologist. Not a lifetime electronics enthusiast, he seemed more predisposed to belief of the occult. Quite likable, he didn't move that program forward.
    We got more than our fair-share of Scientologists at ETM for a reason. Van Willhite, in addition to doing engineering and hiring, was a reverend of Scientology on his spare time. Most Scientologists didn't last. Nor did anyone I'd introduced. Nor would Van. Forestry engineer by pedigree: he must have been "barking up the wrong tree," to use Jakes words.
    Out of necessity, I redesigned a board Van had done (the Stepper-Motor-Translator board). The new board had to fit an existing slot, after assimilating ~50 additional parts. PCB designer Dave Chan objected. He was a "whiner" but Chuck was curious. So I started showing him what Van had done: like driving three-times rated current into three paralelled fiber-optic transmitters. All three were blown. What could I do but use three resistors instead of one. Nobody paralells diodes unless compelled to do so. Van was not an electronics enthusiast. TTL "high" failed to register at recipient TTL inputs, lost in passing, through multiple coupling diodes. This is not rocket science. The overall design was elegant. It did a lot with a little. I like Van. But jake was right: he may have been barking up the wrong tree.

The stepper motor drive system, as conceived, is small and simple; can be used to retrofit most any rotary control; affords nearly unlimited isolation; gives operators an "override" (by grabbing the knob), and can be set up to run on 7.5 Volt AC primary power. The stepper-motor-translator board allows for adjustment of all parameters by a customer computer. Step and direction fiber-optics branch out from it, to stepper-motor driver boards, located with respective control-pots.
     Driver boards use an Airpax SAA-1027 driver chip, delivering 12 volt four-phase output to matching Airpax stepper-motors. We made "customized" 10-turn pots to accomodate motor-drive from the rear. A ball-plunger type set-screw gives a mating-coupler slip-clutch action. On-board 5 V and 12 V power-supplies are grouped with a 74123 multivibrator. Used as a frequency-source to the driver-chip for motor rotation: off-board contact-closure enables it. Reverse-mode is also externally selectable by off-board contact closure. Optionally, these connect to an external three-position momentary DPDT bat-handle switch. One set of contacts enables the frequency source when the switch is flipped, regardless of direction. The other set only provides contact closure when "drive-down" is selected. Caution: SAA-1027 chips are only available in batch quantities.

Van headed our machine shop for a time. He resigned amid scandalous rumor and inuendo. My friend, Roger Ogle, worked for him later. It seems he did take one lucrative-contract with him, on leaving: a job machining pelicle-frames with a CNC mill. He settled about 125-miles north-east, in Lodi. Roger forecasts Van endlessly milking pelicle need.

Early ETM: the formative years

In my spare time, I'd designed a solid-state high-voltage linear-regulator for use with HeNe LASERs, using cascaded devices. An improved version of an older design, I saw promise for it regulating rails for grid modulators and more. FETs aren't hampered by lack of drive-current, a traditional limitation. The group of seriesed FETs are driven by taps along a voltage divider. The bottom device is in series with the other FETs in the string and serves as the "master," to which the others are slaved. High FET input impediance advances control range from marginal to favorable. Chuck thought it was "cute." Nothing came of it. Steve Braddom made my little circuit into a variable load for testing 4 KVDC Vac-Ion-Pump power supplies. Years later, Daniel Goluzek expanded upon the design achieving ~0 to 10 KV anode regulation range, for CW klystrons; used in a high-energy-physics lab called CBAF. I thought the design to be futuristic at the time.
Then I finally got to work on the Varian Beam Analyzer. To my surprise, cascaded bipolar-transistors were exactly where I'd hoped to use cascaded FETs: for regulating grid-box rails. Not futuristic: and already at work in one of ETM's oldest designs, the method was scrapped--in favor of tubes--not for lack of range. Power dissipation requirements forced the choice.