Note: This prototype is too unstable and not recommended.
First Prototype demonstrated at Dorkbot Meeting
My first detector prototype was not that dissimilar to the CERN example, except the fluorescent tubes are placed between three metal plates. The outer plates are connected together by bolts and connected to the Negative rail of the supply and the centre plate is insulated by the fluorescent tubes and connected to the Positive rail of the supply. So far I have found the best result with small 6W fluorescent tubes is around 650V DC
First Prototype Built
Like the CERN example, when a muon flys through the fluorescent tube, the gas inside ionizes due to the high voltage field across the plates. As a result of the ionization the resistance across the plates will fall slightly and so it should be possible to measure this as a change in current flow in the high voltage source.
Schematic of first tests.
The reason for two rows of fluorescent tubes is to sense the crude presents of coincidence occurring in the top and bottom rows of fluorescent tubes due to a muon flying through both. I'm speculating that the resistance in the detector should be half compared with only one row detecting something, due to terrestrial noise. If the output is feed into a data logger and also speculate that over time the difference between cosmic and terrestrial detections could be filtered.
Off on a tangent again.
I couldn't help noticing the similarity with flash tubes and other types of gas filled trigger electronics like a Thyratron thermionic valve. Basically these tubes are biased at a voltage below ionization and when a high voltage trigger is applied briefly in the gas path between the Cathode and Anode, the gas to within the tube ionizes, the resistance to falls rapidly between the Cathode and Anode and like SCR current flows until power is removed.
Consequently I tried biasing the individual fluorescent tubes using their standard electrodes with a DC voltage somewhere below their point of ionization ~70V through a high impedance RC network. The RC network preventing sustained ionization, so producing just a pulse.
Schematic of trigger experiments various component values where tried
However, to my surprise I got quite the opposite, as I measured a voltage spike across the electrodes rather than a dip and so it would seem biasing may not be required as a strong positive spike can be clearly observed on a CRO without any biasing.
Schematic of experiment first tried and demonstrated at a Dorkbot meeting.
Nevertheless, even though "something" is causing clear observable pulses on a CRO in all variations tested above, it is difficult to confirm they are actually due to Cosmic Rays or Terrestrial Radiation over something like coronal discharges within the tube itself.
All attempts to find RFI sources have drawn a blank as pulses disappeared when the high voltage supply was switched off, other Electrical Interference has also been ruled out shielding inside a metal box. I also ruled out the supply itself without the detector and could not find any other interference sources.
I should also note that early in my building and testing of these ideas, I found that most HV supplies I built had quite allot of noise or ripple present, specially the type often recommended for Geiger Counters, so I spent quite a bit of time trying to eliminate this, with improved voltage regulation and a good bank of capacitors.
Tests with an xray source have confirmed the system dose detect radiation, however once the gas inside the tube ionizes, spurious pulses re-occur randomly after, which I suspect is caused by photons being emitted inside the tube causing new avalanches occur. Increasing the impedance of the high voltage supply and placing a discharge resistor in circuit does reduce some of the problem, but this also decreases the output signal. Also I have moved away from using the filament electrodes of the lamp, although this also detects radiation successfully with a high output voltage it also significantly increases the problem of oscillation and other spurious pulses.
So I have moved to a new improvement prototype with better coupling and RFI controls see: Prototype 2 for details.