Shooting for $50 a probe. They can be 1 MHz probes for all I care, my main goal is just to get 4 differential channels. If I need to do high speed probing I can use conventional methods.
If you only need up to 1 MHz, then there are several integrated instrumentation amplifiers which can be used for good performance. I would probably start with the AD8421 because it has low bias current inputs and high AC common mode rejection.
The big problem is limited input common mode range. Compensated resistive dividers can be added to the inputs but they need to be adjusted to maintain common mode rejection. If high impedance dividers are used, then the printed circuit board substrate becomes a problem; a lot of commercial high voltage differential probes have problems with printed circuit board "hook" making them essentially trash despite their high prices.
I don't understand why these things are so expensive, isn't it just a few opamps inside them?
For high voltage differential probes, the high input impedance dividers on the input make it very difficult to maintain high common mode rejection. Many commercial high voltage differential probes are junk because of printed circuit board "hook".
Feedback amplifiers made from integrated operational amplifiers have a lot of problems with overload recovery, transient response, and coupling between the inputs (1) so designs using operational amplifiers are cutting edge simply because they are a poor choice for this application.
(1) There is a reason integrated differential pairs are *not* used for high performance differential probes; coupling between the integrated transistors ruins AC common mode rejection.
I finally scored a nice 1.5KV 25MHz probe for 130 Euros, but after examining it, I think that one can build a probe quite cheaply. Unfortunately, in the end, it has to be calibrated with a Spectrum Analyzer and a tracking generator, and this is the real issue with building one.
Transient response calibration is more difficult and necessary. It requires a high voltage reference level pulse generators which is a very specialized piece of test equipment. Any RF signal generator can be used to check for resonance peaks.