Two tests were performed to assess the vibrational stability of the current design.

Test #1 was run on a capacitor without the roller cage installed in the thrust bearing. It was surmised that removing the ball bearings from the thrust bearing would provide added safety factor against capacitance drift. The test was run for 90 minutes total duration, with each segment of the spectra sweeping over a 30-minute interval (i.e. 20-80Hz, 80-350Hz, 350-2kHz).

Test #2 was run on a capacitor with a standard thrust bearing installed. The test was run for 60 minutes total duration, with each segment of the submitted spectra sweeping over a 20-minute interval.

Results
Test #1. The net capacitance change after the 90 minute run without the roller cage installed in the thrust bearing was +0.2pF. With an electrode capacitance density of 1175.4pF per inch, a 0.2pF change in capacitance correlates to 0.00017 inches of linear movement of the variable electrode.

Test #2. The net capacitance change after the 60 minute run through the spectra with the roller cage installed in the thrust bearing was ­0.4pF (as delivered prototype caps). This corresponds to a linear movement of ­0.00034 inches from the variable electrode.

Analysis
Since the constant load from the bellows pulls the capacitor towards maximum capacitance, it makes sense that any drive screw drift (motion) caused by vibration would tend to increase the capacitance of the unit. Since the test with the low friction thrust bearing produced a result that defies this logic, the probability that this is what we have measured is very low.

It is much more likely that the measured change in capacitance is due to the angular displacement of the variable electrode with respect to the fixed electrode. The clearance between the variable shaft and it's guide bushing is on the order of 0.0002 to 0.0004 inches to allow for a lubricant film. This clearance permits a theoretical angular deflection of up to 0.06° or ±0.0010 inches of side to side motion at the variable electrode tip corresponding to a ±0.5pF capacitance shift due to the angular deflection alone.

Conclusion
No linear motion of the variable electrode was detected during the course of the tests (i.e. no rotational motion in the drive screw).