sdr19

Hughes CWI Boondoggle

Hughes rep Harvey Dane brought very diffirent philosophies to bear with his specifications. He liked to float the tube-body off ground, by a few ohms. This meant the waveguides needed dilectric joints, bypassed by spark-gaps. Much of the RF gear was floating with the body. Al johnson took the lead. SN 335 moved quickly to the acceptance testing stage.
Then Hughes specifications made our crowbar fail reliably. Hughes' specification required we pass one-hundred successive foil-tests. The first time, we made it to 97 shots. Then we blew dime-sized holes every time. We tried another ignitron: and we got 93 shots out of it. In the background, people were pouring over the ignitron data-sheet. One line touts life expectancy ranging from ten's of shots, to millions. Huh? Richardson Electronics does riggorous acceptance testing to uphold high standards for prompt failure. Actual mileage may vary. The brouhaha died down when Al got one-hunderd good shots in a row. So it shipped.
That was SN 335, 25 KV 3.5 ADC, 25 KV-swing anode modulator, solenoid supply, 100 KW cathode and collector pass tubes, 189 KVA peschel variable voltage transformer and raw-supply tank. It was a big set, 25' feet long.

Knowing what I know now, I would have gated the pass tubes "off" with crowbarring. Jake only had 30 KW pass tubes and gated them. Tubes arced new but cleaned right up. You hear a report from the socket when the tube "spits." Al's big hundred KW tubes would be fine. We surround our tubes with resistors. Our sockets are as soft as practicable. At minimum, anode series resistance is set up to handle application of full raw supply voltage without dilectric break down.
SN 335 was designed on the premise that "pickoff-diodes" couldn't survive the output-resistor environment. Al "armed" his crowbar-driver board with appearance of regulated system-480 VAC. ETM's famous "Shut-down-via-ESP" came in "handy" on SN 335. Anode-output had been overlooked in terms of arc-speed protection. But it shut down, on arcs, soft as a baby's butt. Harvey required a hard-wired reason for anode arcs to trip our crowbar. So he had us use back-to-back zeners; inline with the center-core of a BNC cable. One end tapped the anode-viewing current-transformer output. The other end tee'd into the crowbar-trigger-input jack (both backside the panel).

The Hughes CWI tube is sensitive to destruction by under-voltage. Defocusing causes RF-structure overheating and oscillation that does other damage.

Al and Gunnar couldn't use a vacuum relay to energize the tube. It would have triggered the "spurrious-firing" board.

High-voltage step-starting would be too slow. The transition period would kill the tube. We needed a semi-fast cathode rise-time. We ended up with a 100 us rise time.

Gunnar and Al decided to use a fairly huge raw-supply cap, 12 uf. They'd charge it fully, and then gate the pass-tube on gradually. A fairly small output cap (.1 uf) accepted coming up to full charge without detracting much from the 35 KV raw supply. It ramped up, and on: it was clean. We did a lot of work to get the ripple down on that unit. I think I saw less than 90 millivolts on 25 KV at full load. Thanks goes to personal friend, Steve Mannas, for getting us checked out on the precision 10 VDC reference source. We also had to go to NP0 caps on the feedback divider. Traditional ceramic caps had bad repeatability when we were setting rise-time overshoot/undershoot. The crowbar fired to take the tube out of operation. A holding circuit kept the ignitron alive for a half second after triggering. The tube didn't gate-off for crowbarring; rather it had to be "on" to dump the raw-supply capacitor-bank, through the ignitron. Gating the pass tube meant "flashing" the mechanical-crowbar dumping resistors, at each turn-off. Warm glow alerted you promptly. A second offense gives you a fresh whiff of sweet redglass smell.

Ken Lillis was looking at capacitive charging-current through ground; both with and without a "newfangled" series-resistor-bank Gunnar'd added. It isolated the oil-tank from the main machine. Ken's investigation turned up the curse blighting that arrangement. Gunnar and Al inadvertently violated ETM tradition regarding "over and under" shields on secondaries. We normally reference them to a "B-" bus of low inductance strap, also connecting raw P.S. and output capacitors. Gunnar's new resistor was presenting voltage excursions to the B- shield. Right at the beginning of the cathode-supply turn-on rise-time, there's a twitch: The tube (with +38 KV on its anode) over-shot "on" initially. Then the feedback came into control. The initial "twitch" electrically shifted winding-masses so fast that capacitive charging current to the transformer exceeded our overcurrent trip level. Jumping Gunnar's resistor out enabled running Hughes' tube bodies grounded. Perfecting this did require addition of Harvey's nulling-circuit on SN 335, but odly, not later units. We generally ground tube-bodies: that's "ETM Blue."

When Ken Lillis first jumped Gunnar's resistor, we had an arc. The 12 uf cap-bank let-loose in our faces. It was ferrocious. Lugs were blown off the ends of our jumper. The assembler went to strip it, there was nothing in it. He cut it in the middle: empty--no center conductor. The wire had two layers of insulation. The metal had blown out of the first jacket, forming blops beneath the second. But Ken didn't have his data. We fashioned a new jumper. Naturally it blasted too. This time, individual strands of the wire seemed to repell away from one-another, at two inch intervals, breaking through silicone jacketing to do so. There was no discoloration to the metal. But all the strands blew like fuses at one place. Our ears rang. Engineer Mike McFaul rushed to his car and drove away. The receptionist said he was going to have his ears checked. Ken let slip: "He'd probably gone to check his shorts." We all busted up laughing.

First we retooled Teledyne MEC, then Varian, Litton, and Hughes, in that order.

Chuck did a switch box for a unit by Universal Voltronics (another rival). It was for the first Hughes CWI test set. The operator, Harry Mikilowski, said It went to 40% duty. Going beyond smokes the screen supply. As a new-hire, I canabalized the spare driver board, for 1,700 Volt transistors. The U.V. CWI unit uses 15 KW air-cooled switch tubes.