Reverse engineering and repair of LeCroy HFP2500 probe

[dmderev]

This project started when I received the allegedly not functional HFP2500 probe. It was specified as 2500MHz, 100kOhm input probe with 0.7pF capacitance. When plugged into oscilloscope, it was properly recognized, but was showing only some signal at frequencies above 1GHz, and no DC or low frequency signals. I was a bit upset as I identified that the probe head is not working and I will have to find some rare RF FETs or other exotic components. I disassembled the probe head and found there the following .... The rest is in the PDF description.
<...>
I assembled the probe, and as before, there was no input signal… So what was wrong? The problem was in the receptacle from probe pin – as the probe can be easily disassembled and falls apart, someone put glue on contacting surfaces, and the pin did not get connected with probe amplifier. What a pity – spending all that time to not notice that trivial mistake? TV troubleshooting guide suggests that when TV does not work, it makes sense to check if it is plugged in power outlet. My case was very similar! Fix with scratching it out with exacto knife fixed the problem.
The final check of frequency response has shown ~1.8GHz flat frequency response within 3dB, and a 5dB peak at 2.8GHz. That may be the instrumentation/ground inductance problem.
However, since the work was done, someone curious may find this information helpful… My findings:
= It is not necessary to use FETs in high frequency probes, high quality but cheap RF transistors (especially SiGe) may give the probe bandwidths in excess of 2GHz. The impedance at high end is determined by capacitance anyway.
= Using varactors as lower leg of capacitive divider is OK especially when the signal voltage is lower than 1V. This will usually be the case in composite, dual path architectures
= LeCroy was not pushing design to the limit, having bipolar transistors in the probe makes it virtually un-destructible
= They did not connect common ground of power supply to shield of coax and t the probe. The connection happens when the probe is plugged, but it may creates stresses when probe is being plugged or used standalone. I added this connection, but IMHO, it is a design defect.
= They put a decent coaxial cable and 7 more power and signal wires in a 4mm rather flexible probe cable… I wish I could source it too. Anyone knows?

[dmderev]

The main document did not get through. So trying to re-attach here...

[dmderev]

Pictures, too...

[DaJMasta]

For such an expensive probe I'm impressed they got away with such a front end!  No exotic materials, substrates, or custom parts - I wonder if this could inform a nice little DIY design.