Conundrum wrote ...

Very interesting.. wasn't aware of the calibration issue.

would a smoke alarm chip work? these sense ridiculously small leakage currents...

I have now acquired some small geiger tubes (tested using a personal alarm as the stepup, worked fine) unfortunately this draws too much power. However during my experimentation noticed that an ultrabright LED in series with the tube (with parallel 100K resistor) flashed in time with the counts.

Could simplify the driver circuit as just need to epoxy any old surplus red water clear LED to the sanded down "detector", and coat the whole thing in black paint. Totally immune to interference and directly drives a micro's IO pin.
(tested with an infrared LED as the sensor)

Which brings me back to that pesky inverter circuit.
Someone else suggested using a piezo transformer from a dead laptop inverter (total overkill but they draw hardly any current and have the advantage of being directly driveable from any micro output) followed by a rectifier and seriesed zeners.

just needs a micro feedback loop to keep the thing stable, and maybe a PIN large area diode for backup in case the geiger gets saturated.

Interesting to note that pyrolytic graphite + off the shelf cheap CMOS camera = instant alpha detector. You need to cleave the PG down to as thin as possible then flour sandpaper down until the thickness of mica.
Works well.

-A