I suspect the reason it's not working, is something I don't currently have the tools to measure.
With an OK reflected light microscope I could work out whether there's a glass or clear epoxy coating on the silicon. With an alpha spectroscope, I could characterize the source better. Tools are cheap in Asia, but the space to put them costs a fortune...
So I'm going to shelve this for now and maybe try to build a BJT amplifier for a PIN photodiode detector. I've etched some boards. Fingers crossed.
the smart thing of course would be to buy a scintillator crystal, but I hate the inelegance of it. It shouldn't be necessary.