I found an interesting patents that eliminates the noise producing feedback resistor and also does some magic to avoid the need of a reset circuit to prevent saturation. It's pretty straightforward and I never mind a little through hole perfboard soldering so I'll probably give it a go - why not for $10-20. The patent is expired and it claims (if I read it correctly regarding the rms electron noise) to be better than many commercial products including the Amptek A-250. I'll attach it below.
I've been studying some grainy pictures of the Hamamatsu H4083 to see if I can reverse engineer it since the part count is very low. But some of the SMD markings are a bit unclear (ok, they all are) and it looks like they may be pulling of some sort of trick on the board itself of creating capacitance with the traces themselves - there are all sorts of strange squares visible in addition to the smd parts themselves. If anyone has any high res photos of the innards of an Amptek A-250 or a Hamamatsu H4038, it would be fun to reverse engineer and see how the commercial outfits do this. I'm rather fascinated by these little hybrid modules.
As a little side note, from studying some eBay pictures of the Amptek XCR-210 it looks like they use a XC6103 (SMD Marking: 21V gives the specs) as a voltage detector to trigger the reset pulse. I think the PA-230 uses the same thing. If you look on eBay you'll see pictures and it is U3. But a lot of the part numbers and traces are blurred. I don't necessary want to copy their design, but it is still interesting to see how it is implemented.
All the best,
Soren
