2011-07-03
Ion Chamber Alpha Particle Counter
This project was inspired by an article by Norbert Renz I saw on Burkhard Kainka's website. I've built ion chamber detectors before, including some quite sensitive charge integrating and closed-loop ones I have not yet documented here, but never tried looking for the current pulses associated with individual ion trails inside the chamber. Charles Wenzel mentions in his excellent ion chamber page seeing meter jitters in his more sensitive circuits which he assumed were individual ion trail events.
This circuit is sensitive to alpha particles only. This is a blessing in a lot of ways, it makes it an almost ideal device to measure Radon in air without a photomultiplier and ZnS:Ag scintillator powder required to build a Lucas cell. It is extremely easy to build, and surprisingly easy to get working, even with less than perfect electrostatic and RF shielding. The mechanical construction of the chamber is a little bit critical, it must be metal and well bonded together. The JFET gate is completely floating, only its leakage at the relatively high drain current stabilises its voltage. The screening across the open end of the tube is pretty much mandatory or else the device will be too sensitive to electrostatic fields.
The metal can of the ion chamber is a Berocca tube that I cut to length with a pipe-cutter. The coating inside and out and has been sanded off. Only the inside coating really needs removing, to expose the metal, but I prefer the scientific motif than the Berocca livery. Glade air-freshener spray refill cans are almost ideal for the chamber. The older manually operated ones are about the right size and have no internal coating. They are also undecorated so you need not paint or sand them back if the looks of the final device are important to you. The 2N5484 JFET body is inside the BNC plug, its gate lead is extended with a bit of tinned copper wire up into the can, it is stiff enough to stay centred, but the end result is a bit microphonic, a hard tap will register a false count. The drain of the JFET goes to the body of the can/BNC plug. The source is soldered to the centre pin and the threaded compression cable retaining nut holds the plug to the can quite nicely. Some 20 mesh stainless steel screen was formed by forcing a circular piece into the off-cut of the Berocca can using its smooth lip and a piece of conduit a sloppy ID fit as a mandrel. A bit of tweaking is required to get a nice friction fit with the can, but it is not especially critical. Make sure you don't short the gate lead, give at least 5-10 mm between its end and the screen.
Here is the circuit diagram of the device as-built:
And here is a video earlier in the prototyping:
Notes
The circuit pulls about 7 mA, most of that through the JFET.
The background count is about 9 counts in 30 minutes - 5 mBq = 0.3 cpm. The chamber is 45 mm long and 27 mm ID - very roughly this means my ambient lab air has about 194 Bq/m^3 of alpha activity. (That's 5 pCi/l - a level approaching European action suggested levels and probably within the US ones which are a bit more paranoid.)
The circuit can take a while to stabilise when first switched on, or challenged with a strong source which shifts the chamber voltage significantly.
Ions from combustion are easily detected by this circuit, it makes a reasonable detector for burning things, especially gas flames like a butane torch.
When observed with a digital CRO the individual alpha events are quite large compared to the background noise in most cases. When a strong source is brought near the chamber the response becomes less characteristic of the RC-limited response of the chamber/JFET system and swings quickly at large amplitudes. I have little confidence that individual ion trails are being detected once their rate exceeds the ion sweep-out time of the chamber. Because of the low bias voltage it takes 100s of milliseconds for the chamber to recover fully, and 2-3 Hz counting is really the limit for true counts. As we are talking Bq not Hz it is ideal to keep the rates below 2 Bq as two close-together events will not be resolved and will be counted as one because of the dead-time.
It might be interesting to bias the chamber higher, approaching 100 volts, but I am unsure how practical this is, the JFET gate voltage can only float so high. It might be possible with capacitive feedback to hold the gate voltage low and discharge the gate capacitance periodically when the circuitry runs out of headroom. This suggests a digital solution for all the control required though. DSP for matched filtering has been suggested as an option too, Niels PA1DSP suggested this approach and I think it has great merit.
AC coupling and another gain stage would allow much simpler level-thresholding or DSP integration. The whole comparing the difference between the source voltage and an AC amplified version of it is a bit of a hack. The comparator hysteresis feedback resistor helps bias the threshold enough to keep the noise floor from triggering the comparator, depending on leakages into the comparator input for millivolt level setting is not a good idea.
There is a tiny hint of beta events near the noise floor when put near a bag of KCl. With sufficient shielding and a better JFET + higher bias voltage it *might* be possible to resolve them. I suspect cosmic ray muons are too weakly interacting to be counted in this manner, I doubt you'd be able to distinguish them from the background. It would be interesting to seal up a large chamber and see if the count drops as any entrapped Radon decays. Any remaining counts would be from alpha emissions from the chamber itself - or perhaps cosmic ray interactions... An experiment I must attempt.
2 comments.