| The Culture of ETM <Back Home Next> Core employees come to adopt leadership's negotiating stance. This aspect of a company can fall anywhere between two polar extremes. At one end, distributive thinkers aim to"win" in negotiation. They negotiate from a position of scarcity. At the other end, integrative thinkers anticipate robustness. They negotiate from a position of abundance. Chuck Hayse, ETM's owner, was the archetypal integrative thinker. Gunnar Wik, his chief engineer, was the archetypal distributive thinker. Perhaps it was a healthy dichotomy in some ways. But winning organizations generally show a high degree of goal congruence. That term was meaningless in the context. My loyalty was to Chuck's ideals. I joked that he'd formed the ETM Federation to battle dark-forces of the Cober Empire. Cober equipment was ugly, routinely shipped-shot, to generally thwart would-be users. Gunnar Wik was from Cober. So was Al Johnson. Al's arrival gave me a tour of the Cober way. Pushy describes it best. Customer acceptance tests were fluffed up: swamped in detail. The cadence and focus of customer attention were controlled constantly. Chuck's way was to build stuff beefy, turn it on, and to see what smokes. Then he'd beef that up too. His was an instinctive, curiosity-based, exploration. The customer would show up. We'd patch things up to their liking too. Chuck was totally flexible. Customers balanced urgency and quality to taste. ETM's leadership dichotomy was agreeable enough. Then Gunnar's units fell to ETM'ers for testing. And customer, Bud Bedford, shared Chuck's philosophies. Outnumbered, Gunner's speedy-delivery was shelved. Those units were brought up to spec line by line. Bud knew his end users were brutal. He held out until his units were bludgeon-proof, as specified. I felt embarrassed when Bud called out "take her down to low-line" and I said: "Who me? Oops." Bud said: You are a power-supply company and you don't test for low line? Sheepishly, I put a Variac on there. Low-line made three of our four main high-voltage supplies fail, for zero output. Meanwhile, Gunnar was grumbling about our "codling" of customers. Chuck expected us to beat every line in a customer specification. Gunnar liked command-performance shipping. This basic conflict confounded expression of excellence. Sombody was always grumbling, be it Chuck or Gunnar. Employees had conflicted feedback and incentive structures as well. Then CEO, John Capovilla, rewarded the "B-minus effort" optimally, when giving out raises. Gunnar was always looking for engineering help with his leaning. Electronics enthusiasts were rare in that form. He made due with less. What materialized reflected such. Even the 100 KW line that he and Al Johnson pioneered ended in disappointment (by way of the "ETM crowbar curse). Our lack of engineering prowess meant Hughes lost their contract making a CWI tube-line they'd pioneered. I wondered if I was alone in noticing that breaking into doing big-jobs wasn't easy. Other engineers looked tortured. Nobody was tracking the anguish in the faces. Accepted design-artifacts grew to bedevil the best of us. The final blow was struck when Chuck had his son Tom, and Dan Nyberg scale up the cookie-cutter to become a 50 KV 20 Amp, tetrode-regulated, high-power microwave Amp. What we found was cookie-cutters won't be scaled. Worse yet, cookie-cutter development gleaned ETM's cultural identity. It's basis had vanished. People were loyal to the old ways. Why change? Why reject the beloved cookie-cutter? Instead, I ask, what's to accept? The following is my scathing review. Cookie Cutter 101 Beam current goes through the logic harness in at least one place. Lamps on the collector panel flash over in response to output-tube collector arcs. 125 Volt fuseholders blow like shotgun-shells for 277 V shorts. Resistors electrify wimpy 24 V "on" indicators--directly from 480 VAC mains. Each lamp's return-current terminates to (or hipots) it's front-panel. The beam regulator box-chassis is referenced to ground. The regulator circuitry isn't. Referenced 14 Ohms away from ground, internals "fly-up" to high-positive potential in an arc. The variac control board is in there too, similarly referenced. The logic harness anchors the innards into front-panel grounding through a web of shields on twisted pairs in the harness. The crowbar driver board delivers spurious-firing for "shutdown via ESP." Arcs do trigger it, a plus. But the snap of a high-vac switch does too, a big minus. Someone "fixed" the problem by charging the crowbar driver board cap's with vacuum-switch contact closure. Alas, this leaves the tube and operator vulnerable for the first two-seconds (board-chargup time lag). Establishing initial crowbar protection became a new engineering riddle. A jolt of 1 KV will trigger an ignitron. Short-circuit current is ample. Knowing this, Someone tapped the ignitron's cathode/ignitor-leads to bridge across the middle 10% of the output resistor, a 6.5 Ohm portion. Normal load current is insufficient to fire the device. But a 10 KV short-circuit imparts 1 KV of drop across that 6.5 Ohm portion, enough to drive the ignitor into firing the ignitron. This in turn diverts remaining discharge-current back to capacitors of origin. The new problem was: the 6.5 ohm resistance loaded-down our original firing pulse from the driver board. We wanted our pulse transformer to drive the ignitor without the burden of 6.5 Ohm loading. A diode was added. But diodes went bad. We beefed them up. Discouraged, we turned to spark gaps (A bad diode meant the crowbar driver board would only marginally trigger the ignitron). At least shorted diodes still facillitated pick-off-resistor firing, some consolation. Ironically, pass-tube gating is really what makes cookie-cutters pass a foil test, not the crowbar. The pass-tube can be cut-off in fifteen microseconds. To switch it off, a 13 V "crowbar fired" signal enters the regulator board, from the crowbar board. Botched execution neglected that the boards weren't referenced together. Body current elevates the reg-board ground-plane by an IR drop. This detracts from received "crowbar fired" signal amplitude. Worse, the reg-board "bolts" 200 Volts positive for most transient overcurrent conditions. The ground-referenced "crowbar-fired" signal is thus negated; due to the very follow-on current begging for interruption. With a normal output cap (.1) uf, the energy dumps in a microsecond. The IR drop settles to zero, with it the reg board: and the tube gating works. But with a 1 uf output cap, beam common doesn't settle as fast. Follow-on current keeps the IR drop above 15 V, never to allow gating onset. Follow-on current blows a hole in the foil before the contactor stops flow. Our staff computer whiz-kid, Steve Warner, first identified this deviant mode in SN 232. I specified an opto isolator and congratulated him for beating one of ETM's oldest scurges. The old beam boxes needed this modification the worst.They are all on the brink of dogging someone. I watched one guy swap all the parts of an old ETM crowbar for new. It still blew holes. Opto-isolated tube-gating circuitry came to offer hope. Chuck's first teledyne units dogged him every possible way. Off to a bad start, he'd accidently purchased and installed air-cooled tetrodes. But no, Teledyne specified water cooling. Then crowbars dogged him, ad nauzeum. In his first disaster he'd overlooked detection for spurrious ignitron firing. One night, unattended at Teledyne, a rash of wrath flaired up. Ignitrons spurriously fired in all the units at once. Five large variacs, five transformers, five output resistors . . . five supplies were total losses. Chuck's ignitron anodes now return to chassis-ground: where body-overload-sensing will be tripped by such spurrious firing. Alas, his new configuration always pegs the analog-body-current-meter on crowbarring. But this doesn't reflect actual energy delivery. With bigger fish to fry, foil testing detained chuck endlessly. He never did get it quite right. I don't know how he escaped. They foil-tested routinely from then on at Teledyne. We told Chuck the moment we vanquished that gremlin. Next> |