I spent some time looking at my early 478A sensor again this evening and removed the ferrite (that I had added) and also bypassed the 2.7R damper resistor R1 fitted by HP. I then looked at the return loss on a VNA and also looked at the frequency response when the 432A + 478A was used to level a HP 83752A sweeper.?
?
The plots below show that with no damping the input return loss shows a classic zigzag shaped resonance just above 50 MHz. The VSWR is actually slightly better at 50 MHz without the damper but the VSWR obviously gets much worse at 53.5 MHz.?
?
With the 2.7R resistor R1 removed, the frequency response using the sweeper shows a narrow upwards blip of almost +0.3dB at 53.5 MHz and this is due to the undamped resonance here. The resonance draws energy away from the thermistors and this degrades the efficiency. The blip is positive and this shows that the efficiency must be lower in the dip as the sweeper has to put out a bigger signal to get the same output from the 478A sensor.?
?
The response from 10 MHz to 50 MHz is exactly as a simulation with about 0.1dB change from 10 MHz to 50 MHz. If I add a ferrite, this completely removes the blip at 50 MHz but it does degrade the efficiency down at 10-20 MHz. So there is no free lunch with the ferrite. I still haven't decided on the best choice of ferrite here and I may also try increasing the capacitance after the ferrite to be similar to the later 478A sensor.
?
I think this experiment does indicate that the 2.7R resistor R1 was added by HP to try and tame this internal resonance. The combined capacitance value of C3 and C4 may well have something to do with the operation of the 431 meter, but I think HP added some damping resistance to one of the capacitors to tame the resonance at 53 MHz and this may be partly why these parts are marked as factory selected.?
?
?
?
?
--
Regards
Jeremy
