Great movie of the PMT interface with sound.
This is the diagram, with some explanation:
OK, here we go, the ticking sound you hear happens whenever an energetic particle hits the crystal. That causes a tiny flash (or big flash) of light. The PMT detects this, and amplifies it into a small current. The PNP darlington (top middle) amplifies this current about 255 x 255 (beta x beta) times (65535 times), and that causes a current in the collector. So a positive voltage pulse appears on this collector for every light flash,
As you may have noticed, many here have fought with expensive preamps, incredible shielding, and the latest most expensive high tech , filters of all sorts, and what not, to make a reasonable usable flat pulse of long enough duration from this tiny shot of electrons from the PMT.
This solution (already published before, as Usenet patent - can't remember the exact number - is the simplicity itself. Note how the pulses stay nice and flat for the various amplitudes. The amplitude is maximum about 4 V, not nothing either, The minimum is set by the scope trigger level here (else the trace won't run).
In the final setup a hardware comparator in a PIC detects the start of the pulse, triggers an interrupt, does a sample and hold during the flat top, and then with that sample does a AD conversion (8 bits only for speed). After the conversion is done it discharges the input of the PNP darlington, that is cable capacitance, hole storage time etc. You can set the pulse width by selecting suitable transistors and the correct length of the screened cable to the PMT anode. In this setup the discharge switch is not connected yet, so you see the pulse slowly go back to zero, with on it some ripple from the HV supply that now tries to recharge the HV capacitors. In this setup the pulse width is about 40 us to 50 us, plenty of time to do the sample and hold thing. Of course the longer you make the pulse width, the lower the maximum frequency of pulses you can measure, and the bigger the chance that 2 pulses appear next - or halfway over each other,
Anyways, the pulse is also used to flash a LED, and used to trigger a piezo, so visual and audio signalling.
There are 255 bins, and depending on its amplitude the PIC will throw the measured value in one of the bins. The higher the energy of a particle, the brighter the flash of light, the higher the pulse amplitude. Auto-scaling always keeps the display in range (y axis), and there is your spectrogram. The 'threshold' trimpot sets the minimum energy required for a pulse to register, and gets rid of that annoying noise spike in the spectrogram all the way on the left, making the readout much better. This could be a potmeter too. Then you can set the start and end of the energy spectrum.
I am happy so far, indeed using components to get the maximum out of them (transistors) is great fun. It is different than I have seen anywhere else, and actually maybe even better,
Keywords: PMT preamp, prior art, patent, ice cream, YES.