tokamac wrote:That could be better for several reasons, mainly because the cavity is heated and quickly becomes hard to keep in resonance if the operating frequency bandwidth is too short.
If keeping the microwave source tuned to the resonant load is important, the resonant load needs to be part of the tuning network for the microwave source.
As a side benefit, building magnetron and cavity into a single unit to mount on the thrust balance, minimal microwave plumbing between components, fed with filtered DC, eliminates some suspects for false readings.
Eagleworks at JSC already uses filters and a Phase Locked Loop (PLL) to automatically lock onto the current resonate frequency of the cavity, and already mounts the power amps and equipment as a whole onto their thrust balance.
As for Roger Shawyer he has also developed a dynamic system based on piezoelectric actuators that slightly modifies the length of the cavity hence its operating frequency, to follow and match any change in the resonance. This helps fighting the natural tendency of the cavity to "detune".
Keep in mind another point, according to White's QFV conjecture, there is another use for the noisy, wide-band magnetrons, as Paul March said:
"These time rate to change
of energy modulations increase the Q-V density
in our model."