birchoff wrote:Tom Ligon wrote:Also sounds like they were running at high vacuum. Turbopumps, outgassing for a couple of days ... I'll read in more detail.
This part is a bit suspect. the NTRS Abstract claims they didnt test at high vacuum. The paper describes how they tested at high vacuum. Then the "Summary and Forward Work" says due to not having a vacuum capable RF Amp they didn't test at vacuum. I personally have interpreted this inconsistency to mean they haven't done a hard vacuum test. Would be interested to hear what you think.
I'll have to review the whole paper.
Depending on the model, a microwave oven magnetron likely WILL operate in a vacuum chamber. We found a model (tearing down an discarded microwave from RWB's kitchen) that was capable of vacuum sealing its antenna end into a chamber. I had a 4.5" CF cover plate machined to accept it, and added a reflecting cone. We launched microwaves into PXL-1 with ti quite nicely. There's no fine tuning a microwave oven magnetron, though ... you would have to fine tune the antenna, cavity, etc.
They were using coax, so could have used hermetically sealed coax connectors, and put the electronics into a gas-filled box, in the chamber.
I want a close look at the torque data. A torsion beam apparatus has an enormous moment of inertia relative to the torsion filament's torque. I'm curious about the dynamics. The pulses look about 30 seconds long, the instrument responds in maybe a second. How much movement is involved and what does this torsion element look like ... need to check. It seems amazing that the reading can stabilize in a second. But the calibration scheme sounds credible.
One of my jobs in the UAV biz was moment of inertia tests of UAVs to get data for flight dynamics calculations. We used a 2-filament support scheme (bifilar pendulum method). You measure the period of oscillation, and even with two filaments these things oscillate a long time. A single filament offering
almost zero torque seems like it ought to oscillate like crazy with an extremely long period. Realizing they use magnets and a conductor to dampen it, still, it has to move for that to work. I know torsion filaments have been around for a long time and do work, but my gut says 1 second responses are on the fast side. The instrument is probably fine, but I'd want to think hard on it.
I've made strain gage torque-measuring devices, but not for micronewtons. MAYBE that would work. To get sensitivity up usually also means raising power dissipation in the strain gages, and heat dissipation would be worse in a vacuum. Working with tiny gages is not conducive to sensitivity due to power density. Working with very thin-walled small diameter tubing also limits heat dissipation. I can datalog at 18 bits, and 24 bit ADCs are common now ... one could imagine one has the sensitivity but then actually try it and find out that any signal is buried in noise and overcome by drift from 10 sources. I'd be happier with newtons. 0.4 N at a kW would be great.
Notice that they got interference feeding DC to the apparatus due to shifts in the magnetic field around the leads.