| Dave Harpe |
| Dave looked slow to the average stuff-shirt. But he was methodical to the point of being technically impeccable. And that low-stress look--it's just a facade. I found that out when he took me sailing. We got out into the shipping lanes in the San Fransisco bay and the weather took a turn for the worst. The swells were way bigger than his boat. He was really in his element confronting those forces. He's secretly a real mover and shaker. |
| We'd had trouble driving tetrodes with our stock switch-cards. The card floats within the box. The box floats at 25 or 50KV. Troubleshooting in-circuit was out of the question. I'd been muddling through by capping and strapping. I eventually noticed some boards were better than others. This traced back to variances in the opto-isolators we were using. In passing, I mentioned this to Dave. Dave set up a test-jig to subject optoisolators to fast-risetime common-mode interference. Low and behold, he identified a batch with poor noise immunity. Hewlett packard took great interest in Dave's work. They developed a new high noise-immunity optoisolator to pass his tests. It's called the 2611 (it replaces the old 2601). |
| Dave had a boondoggle dropped on him: someone designed a tetrode switch-tube grid-drive circuit with no regard for what the tube grids needed, or how they behaved. The provisions were sparse. There was only a 300V screen supply where 1500V was normally needed. Dave dodged one bullet after another, finally settling on strapping the grids together, thus forming a triode. He applied tri-state grid drive. He hit both grids with +300V for fast risetime. He delivered about +50V to both, to sustain conduction. And he hammered them both to -500 to cut the tube off. The design was quite elegant as it used the original parts. Our tradition of switch-tube oscilation problems never surfaced for Dave. This is partly because the triode configuration has less gain, thus less tendency for parasitic oscilation. Having successive grids at the same potential encourages laminar flow of electrons through the grid region. This in turn eliminates the normal tendency for electron orbit oscilation around the grid wires. (Barkhausen oscilation) High control grid current acted to clamp the already low screen voltage, thus preventing dynatron oscilation. The resulting 15KV squarewave output looked really clean, even up close. |
| Good work Dave! |
| Dave did switching supplies for a hobby. He made a line of motor-controllers for radio-controlled airplanes. At two inches square and a half inch thick, they packed a wallop, putting out 40 Amps continuous, for five minutes. |
| Dave liked to quote former president Nixon whenever engineering botched an assignment. I can still hear him saying: "Mistakes were made." Really bad engineering prompted him to say "Get me a bucket." |