Hi Guys,
sorry for my late reply.
Today I was able to make some further progress. Robert was right in his post (/g/HP-Agilent-Keysight-equipment/message/92641) - the autozero problem I encountered was indeed related to the position where I connected the autozero signal. In principle, I have now the same schemativ as the 8484A diode power sensor.
After moving the autozero signal, I still had the problem that the autozero did not work properly, so I further analyzed the circuit of the 8484A and realised that it has less gain than my circuit, so I changed the base resistor of Q1 (https://hb9fsx.ch/wordpress/wp-content/uploads/2018/11/8481schem.png) to 1k Ohms and the autozero problem was fixed - pressing the autozero button perfectly allows the meter to zero the sensor. However, with this modification, my sensor had a different gain, and therefore the readings on the meter had an offset. I was able to re-adjust the gain with the trim pot and when I put a 10dB attenuator at the sensor input, I was able to have the same readings on the meter as on the signal generator I used to test the sensor. However, as I said, it is not terribly accurate, at very low power levels, it was around -1dB off, and at high power levels, it seems like the diode transitions from square law region to linear region and therefore the readings were also off a bit. But this is no big deal; one can simply add attenuators in front of the sensor such that it is ensured the diode is in its square law region.
Thanks at Orin for the 8484A sensor schematic. I have never found it, even though I was of course looking for it.
Joarez (/g/HP-Agilent-Keysight-equipment/message/92646) says that the capacitors need to be there to prevent the circuit from generating switching abberations. This is probably true because the 8484A manual tells that the capacitors were factory selected individually for each sensor, so they are probably related somehow to the JFET switches. In my sensor, I currently don't have these capacitors, and this is probably another reason why the readings are not 100% accurate. It could also be related to the JFETs I use - I chose the first ones I could get, but somewhere in my junk boxes I should have a pile of 2N4856, which are probably better. However, I would prefer SMD devices since I will make a PCB in SMD design. I got a bunch of MMBFJ310, but I have not enough of them to make a lot of experiments :-/
Thanks also to rmc321 and Rainer for the link to the connectors. I was once told that HP used their own connectors, so I didn't even try too hard to find the connectors, but very good you found them! I could have saved my valuable power sensor cable if I had known it earlier :-)
I think I am soon going to make another dead bug prototype and in parallel design a PCB and probably my own housing. I will share the design files as soon as I am ready.
Tobias HB9FSX
