harvey

Harvey Dain

I got to work with Harvey on a line of 100 KW, 25 KV tube testers. It was an ill fated foray but I learned lots. Hughes had the crowbar test that caught our crowbar failing: we couldn't pass one-hundred successive foil-tests. On our first attempt, SN 335 passed that foil test 97 times before failing. Each behemouth unit needed ignitron replacement at the ATP to pass a Hughes 100-shot test. The problem with that is a "green" tube causes at least a hundred crowbar firings a week. I suppose we could have touted our half-week warranty. The next three years saw a dismal record forming; one of Hughes failing to produce Continuous Wave Illuminator tubes for the Navy. They produced the "gold standard" on their UVC modulator. Harv' faced spartan accomodation. He tried to nurse us through but we fell flat.
    Cober (our competition) uses repetitive firing for ignitrons. But those contraptions look like line-type modulators from Klingon Warships. They have thyratrons driving thyratrons; with UJTs driving each first thyratron. I couldn't do that in a field redesign/retrofit. Nor would I want to. In the end I used an SCR, a capacitor, and a pulse xfmr; with a switch-card to recharge such, between firings. I really do preffer the Microwave Products brand triggered vacuum gap as well. But alas, SN 335 hit the scrap-heap in the Autum of '92. My baby never got her repetitive-firing care-package.

     I'll not soon forget those hard lessons. Nor will I the sly tricks. One such trick was a method for desensitizing overcurrent detectors to switching transients. Harv' took me back to the root-source of the architecture: a 70 KV Cober cathode-modulator. He pointed to a dead-end high-voltage wire. It was many inches away from the push-pull switching output. This "gimmic-wire" led to the current-sense resistor assembly. I've reproduced it countless times since. And if you have spurious trips due to stray capacitance charging requirements, go for it. Here's how:
                              Harvey's Nulling Circuit
Cut the OC sense lead from its sense resistor--slap a 100 ohm 2-Watt variable-resistor in line (where you made the break). Take two switching-diodes and series them. Solder one end of the two-diode-assembly to one end of the variable resistance. Solder the other end of the two-diode assembly to the other end of the 100-ohm variable resistance. The midpoint of the two diodes goes to the gimmic wire (which should have some capacitive coupling, to the dynamic node.)
    There's ringing on the gimmic whenever a switching transient hits. The envelope of the ringing translates to a little bump in the voltage waveform across the 100 ohm pot. Make sure the diodes detract--not add--to the trip signal. Turn it shorted for no nulling. Increase the resistance to null out capacitive charging related spikes; to subdue undue tripping. Test overcurrent functions thoroughly afterwards. Tampering in this area can present proplems if not handled responsibly. Scrutiny should be concerned with over-desensitization at maximum chaos.

Thank you Harvey! Thank's also to the Hughes Electron Dynamics Division (Harvey's employer). Thanks ETM.

Home

Mentors

Crowbar "Rabit Ears" Nulling

Harv' took me to a little hotel-resturaunt called Del Conte's. He reminisced about how our chief engineer (Gunnar Wik) came out and stayed there as a young teck. Rolling grasslands were all that had seperated the hotel from Hughes--a far cry from the urban sprawl dividing the two today.
Gunnar was awakened in the middle of the night to hear his 65 KV baby died amid his final burn-in test. He rushed to the scene to find a bad incandescant bulb, in a fiber-optic interlock-link. It was considered a non-critical device failure so the burn-in clock wasn't reset to zero. Gunnar got to go home soon afterwards. Later Harv' showed me the unit and it was kicking strong. I'd say it looked 30 years old. So did the new.

Harv' spoke of doing a 250 KHZ? 13 KV cathode pulser for a magnetron. They used a triode for a switch. Perhaps the voltage amplitude wasn't full swing but the current sure was (nearly). That's an impressive feat, even today.

One MicroHenry per meter--all wires are the same--according to Harvey. He paralelled two 1/4" copper tubes from the output of a Rutherford pulse generator to it's ground stud. Inductance didn't change much as the two tubes were initially separated. At three inch separation, inductance dropped to half.

"Quarter 'Puff' Per Cubic Inch" was Harv's capacitance estimation for air-spaced metalic plates. (Two ea., one-inch square plates, spaced one inch apart.) He used water-filled PFNs because DI water has a dilectric constant of 60.

Harv' cracked up at Cober current transformer assemblies that use oil to seperate a torroid from its pulsed wire. Absent a potting lab, it remains an acceptable last-ditch method. It's that or plasma.

Q: How do you make a 19.5 turn winding?
A: Paralell a 40 turn winding with it's bifilar 39-turn twin. He saw it on a schematic and demanded answers. So he dug in. And that's what he found.

Scroll down for nulling circuit