Well I told you that I would tell you some more about radiation meters which work by the scintillation method. One classic is a machine which was made by EMI which has some advantages over a simple GM tube. The EMI device was used in the spy program spooks by a Soviet sleeper agent who digs up an atom bomb at the end of one series with the intention of blowing himself up with it outside the US embassy in London. He uses a larger probe than the DP2/4A, but I suspect that it is also an alpha/beta probe. I think the spy had the wrong tool for the job, if I was looking for a decades old atom bomb which was several feet underground I would use a NaI or BGO based counter to look for the gamma rays from the fuel in the bomb.
If the agent had paid attention in A-level physics he would know that alpha and beta is not able to pass through 12 inches of soil, but back to real life.
The GM tube is a cheap and cheerful detector, but it is very fragile and it can not distinguish between beta and alpha. If you leave the protective cover on a GM tube it might still detect high energy beta (Sr-90 and P-32) but it can no longer detect alpha and low energy beta. But as alpha emitters are so much more radiotoxic I think it is an advantage to be able to detect alpha and ignore the beta background.
The DP2/4A probe made by EMI uses two materials which convert the energy of radiation to light, one is a plastic (BC400) which is loaded with anthracene (Here is a page of a another company which makes such plastic) this has a very short decay time while the other solid is an activated zinc sulphide which has a longer decay time. The idea is that inside the DP2/4A probe a layer of the BC400 plastic has a thin coating of zinc sulphide, this is covered with a very thin metal film to keep out the light. The idea is that an alpha particle can pass through the metal foil into the zinc sulphide layer and make a flash of light, while a beta particle will pass through the zinc sulphide layer and make a flash of light inside the plastic.
The light from both layers is viewed by a photomultiplier tube (PMT), the idea of a PMT is that a photon of light hits a surface and causes a few electrons to be emitted, these electrons are accelerated in an electric field towards another metal surface where it hits the surface spalling off more electrons. The new electrons are then accelerated towards another metal surface to give another amplification of the electron pulse. This process is repeated many times to create a large pulse of electrons which hits the final anode thus making a pulse which can be detected with “normal” electronics.
Here is a photo of the insides of a broken PMT tube from a box of scrap bits used for teaching. you can see the curved dynodes where the electron multiplication occurs. Sorry it is a little out of focus.
Oh bother the software has just lost my text, but I will have a go at rewriting it. The problem with the multiple stages of amplification is that if the amplifiers are slightly non linear then the effect of a series of stages will distort the system greatly away from being linear. I am going to look at the typical NaI detector now from the point of view of a person who has been trained to design and build radio equipment.
It is well known that for all amplifiers that the following equation is true, for a linear amplifier we try to make all the K terms other than K1 as close to zero as possible. But we know that it is never possible to make the perfect linear amplifier, anyone who tells you that they have a perfect design is a liar ! I have a close to perfect – 3 dB amplifier in my basement (it is a pair of resistors which attenuate signals) but that it not a true amplifier.
y = k1 x + k2 x2 + k3 x3 + k4 x4 + k5 x5 …….. etc etc
While a class A or AB amplifier can come close to a perfect linear system it will have a upper limit to the signal level which is dictated by supply voltages, also the effect of a series of slightly non linear stages can be dramatic. I did some maths for a system where K1 was 1, K2 was 0.001 and K3 was 0.00005 and here is the results which I got in the graph at the bottom of the page.
I imagined having one stage of this system and a series of chains which have more than one stage in cascade. You should be able to see that the systems with longer chains have smaller dynamic ranges over which it is possible to estimate the input from the output. Also as we reach close to the limits of the dynamic range our ability to make an accurate estimate of the input becomes worse and worse.
I have no experience of working with the circuits with PMTs, if any PMT users are reading it might be interesting to know how linear a single dynode stage is. If we could get the equations for a single stage we could then model the whole tube.
My other worry about NaI systems is how reproducible is the transfer of photons from the NaI crystal to the PMT, if the process of converting the energy of gamma photons to light photons hitting the first dynode is subject to random effects then it will increase the line width of any gamma spectra obtained from the NaI counter.
Looking at the NaI counter and a high purity Ge detector, I was able to use my background in electronics to make an educated guess. The Ge counter has fewer physical processes between the absorption of a gamma photon and the delivery of a pulse of charge to the first amplifier. Also the amplifier chain can be shorter and made using solid state electronic parts which are easy to build good linear amplifiers from.
If I was given the job next week of creating the first ever amplifier chain for a Ge counter I would be likely to choose a low noise GaAs FET for the first stage, I would choose a FET which is suitable for service in a HF or VHF radioset as a preamp. Then I would be likely to choose a series of VHF bipolar transistor amplifiers to increase the size of the voltage pulse. I would finish off that part of signal chain with a fast operative amplifier such one intended for use in video systems.
The amplifier chain would be hard to build, I suspect that I would be unable to use some of the methods used in radio equipment such as AGC to tame the chain. The reason is that I suspect that the timescale of events would be too fast to allow normal AGC.
Then I would need to choose a ADC, I think I will save my thoughts on ADC design for another day.