2009-04-28
Background Radiation Monitoring
One day many moons ago, when I was in High School, the science lab was throwing out all its old equipment. Amongst the junk was an end-window Geiger tube. Unfortunately it did not come with a PSU/counter unit, just the tube head. Quite a nice head though, it has an thin Aluminium case with a brass screen to protect the Mylar end-window. Surrounding the Geiger tube there is a brown coloured liner (appears to be a polymer of some type) forced into the Aluminium tube - I am unsure of its purpose, it might simply be to centre the tube, or it might be for energy levelling or to affect some Neutron sensitivity?
In any case, it has taken me until now to finally build a supply and integrate it to the PC for logging purposes.
Hardware
The PSU is based closely on Charles Wenzel's moulded choke boost-converter multiplier design. The tube connector is an unusual vacuum tube-style plug, I'll probably replace it with a BNC at some point. The cable is just a length of figure-8 zip-cord. The Geiger tube itself is fairly small and quite old, perhaps older than me! I have no idea of its properties, but I run it at about 400 volts, much below 300 and its sensitivity drops off rapidly, it is essentially dead at 280 volts.
The count impulses are buffered by a transistor and fed to an Atmel tiny13 running code vaguely similar to the sampler. It counts the number of impulses in 10 second windows and sends back the data as ASCII text over RS-232 to the PC for logging and processing. There is also a piezo transducer for the traditional Geiger clicking sound.
So far the hardware is spread across my bench and computer desk. It is a rather fragile lash-up with a tie-pair snaking across the room to carry the impulses back to the MCU. I'll document it more once I build it in a more permanent form.
Results
The first test was of course counting alpha particles from the 37 kBq Americium 241 source I used in the ion chamber experiment. The distance from the source to the counter window was not well controlled, but as expected the counts drop with distance fairly sharply. Even the thinnest of paper or polymer films reduce the count enormously, towards the values seen by putting the source against the side of the tube where only the weak gamma emissions of the Americium make it into the tube.
Initially the background radiation data looked pretty uninteresting, just random noise:
But then it started raining:
At first I was a bit confused and wondered if the tube was OK, but eventually it became obvious that the background radiation can increase quite dramatically when it rains. This is likely the precipitation washing the natural Radon daughters out of the air and concentrating them nearer the detector.
It rained briefly on the 29th, the first time I noticed the effect. Later on the 30th-5th it rained quite a lot on and off. The sharp peak after lunch on the 2nd was just me checking the system with the Americium-241 source.
Potassium-40 Source
On the evening of the 4th I brought up a bag of natural Potassium Chloride from the garage and put it under the sensor. I also removed the outer Aluminium tube to allow weaker betas into the Geiger tube. Natural Potassium contains about 0.0117 % Potassium-40, a beta and gamma emitter.
On the morning of the 6th I put the Aluminium tube back on, the effect is quite obvious. The beta radiation from the KCl was then only able to reach the tube through the end-window without a fair amount of attenuation.
On the morning of the 7th I removed the KCl from under the tube and on the 8th I took the KCl out of the room entirely. Later that day I put the Geiger head into a die-cast Aluminium box to affect a better beta shield. From that point the great majority of the counts should be from gammas and cosmic ray muons.
The weather has been fair since, so how much of the radiation from rain-related events is gamma vrs beta & alpha remains to be seen.
Notes
Why the initial background was much calmer than is now observed is a bit of a mystery. It is possible that the tube needed some waking up from its long sleep. I am unsure if there is the potential for ageing effects or absorption of the fill gasses. Maybe some sputter-pumping took out any stray polyatomic species other than the quench gas?
The raw data varies quite a bit, so I am using a exponential smoother on it (with α = 0.01). Interestingly there are some very large spikes in the count data - some more than 20 times the average background for only one window period. I am unsure what causes these, but you can sometimes hear a flurry of activity from the piezo clicker as well. The record for the natural background is 117 counts in 10 seconds. The average over the entire experiment is about 32.8 counts per minute.
Some of the trends and peaks are still mysterious. My shack is in the attic and is frequently closed up for days on end, so natural Radon build-up may explain them. The attic is also surrounded by concrete and brick on all sides, so nuclides in them may be contributing to the background and/or shielding those from outside. Once I have the lash-up built properly into a box I can try measuring the background in different locations.
Lots more experimental work to be done, but the general approach seems to work quite well.
I may push the smoothing into the MCU code, it would return raw counts too, but the smoothed value could drive a calibrated display if I get a tube with a known calibration at some point. This basically means I can make a direct-reading instrument with an LED or LCD display fairly easily.
Leave a comment on this article.