With all of the variables perhaps it would be good to start with a multichannel monitoring system so that, at one's leasure, you can review what each of the critical elements were doing (gas flow, pressure, voltage, current, neutron output, etc.) at any given time during various runs.
Over time you can then see where the "sweet spots" are and what you might want to actually control vs monitor and where the feedback should be.
Multichannel dataloggers are readily available at reasonable prices and many can be used with older vintage computers (I happen to be partial to some of the stuff from Pico).
Steve
Created on Saturday, March 10, 2001 1:24 PM EDT by Steve Hansen