2016. október 14., péntek

Building a radiation detector from scratch 4 - a stabilized detector

In this chapter I'm going to show you how to build an atmospheric ion chamber for detecting radiation. The detector will have little drift, can be set to zero, and can be used to compare the radioactivity of samples, measure weak sources, or follow the decay of short half-life sources for hours or days even.

These features all require that the detector to be stable, and the readings don't drift around with temperature.

Temperature drift is very likely the worst offender to our previous ion chamber design, so we're going to deal with that one this time.

An improved circuit

The next design will deal with the temperature drift by simply duplicating the amplifier, and use only one for detection. The other one will not be connected to anything, and will only provide the zero-point current, and temperature-dependent error.

We will simply subtract the output of the duplicated circuit from the detector circuit to give us a reading without the baseline current and the temperature error.

Since no one will be able to build two EXACTLY matching amplifiers, we'll provide a way to cancel any small residual error, and match the reading to zero at background radiation levels.

In the previous design we've used a DVM's voltage metering function to read really small currents. This can't be done simply enough with two matching amplifiers, so we're going to use 3 transistors per amplifier, a total of 6 transistors.


The image above is a snapshot from an electronics simulation application called LTSpice. It's freely available from Linear Technology: http://www.linear.com/designtools/software/.

This application helped a lot in improving the circuit, especially the response to changing input (e.g. fluctuations in radiation). It also doubles as a general schematic capture tool.

The circuit is basically a Siamese Darlington on steroids. Q1-Q2-Q3 forms a "Darlington triplet". This triplet connects to the ion chamber probe. Q4-Q5-Q6 forms a reference stage.

Now, there is a catch to Darlington pairs and triplets. Since the hFE of a transistor at very low currents is smaller than at the current it was designed for, simply connecting two transistors won't result in a gain that is exactly the product of the hFE values of the two transistors measured independently.

A three-stage "Darlington" like this will have much less amplification then the hFE specified in the datasheet on the third. Unfortunately the simulator program I've used - called LTSpice - doesn't seem to account for this effect. I've built the circuit using the simulated results nevertheless, since I had no other source of information at that time.

I've tried quite a few variations and the above version gave the best results. I've tried placing the resistors between the emitters and the ground, but it gave a very slow (simulated) response. The circuit described here gave reasonably good step response.

The graph below shows the voltage across a collector resistor while sending a 1pA (picoAmpere) current impulse that is 0.1 seconds long. It can be seen that the circuit is somewhat slow to respond, but considering the ions' flight time in the chamber this actually isn't a problem.



Because the transitors don't amplify all that much at picoampere currents, the magnitude of the response is actually an order of magnitude lower. Also because of this effect the voltage drop on the resistors will be much lower, since the currents through them will be less.

Since this device won't be used to detect single events, a fast response is not absolutely necessary. It would be great to detect single alpha particles from weak sources or single muons from cosmic radiation, and I still believe it's possible given a proper data logger and some kind of display mechanism. Maybe I can conjure up some solution for this in a future article.

Selecting transistors

Just like in the previous design, choose a bipolar NPN transistor type that has a high "hFE" or "beta" parameter, e.g. it will amplify currents pretty much. Think about 400-800. These types of transistor are cheap and popular. I'll use BC547C-s.

What is even more important is that we have to build two very similar amplifiers. To achieve this, we must select 3 pairs of transistors with very similar hFEs.

Do not try to match all 6 transistors. It'll take a LOT of work. Do your best and select 3 pairs. Each pair should be as close to each other as possible, but you don't need to match across pairs. Buy 30 or 50 transistors, so you'll have as many options as possible. Use a DVM with a hFE measurement function and measure your transistors. Pair up the similar ones. Do try and select pairs that are very similar.

Assemble the circuit so the transistors in a pair will occupy the same position in different amplifiers (one in the detector, and one in the dupe).

Without proper matching the baseline error will be huge, and even after zeroing the reading with a trimmer resistor there will be severe temperature drift.

Building the detector

The circuit was built on the non-coppered side of a small piece of PC board. The components were fixed with super glue.

Make sure the glue does not touch any lead as it might conduct slightly, and utterly ruin the readings.

Final assembly, top view


 A battery or power supply can be connected to the tin can (+) and the wire connecting the emitters (-). The DVM should be set to millivolts and connected to the terminals on different resistors where they meet the collectors of their respective transistor (on the bottom leads on this picture).


The transistor pairs were glued together so they stay close together, and therefore will exhibit very similar hFE no matter the ambient temperature. Pay attention to the transistor leads, as this arrangement the pairs are not mirrored, but rather rotated. See these photos and try spotting the difference on each side:

Final assembly, right side

Final assembly, left side



The resistors were soldered so they can easily be replaced. After taking some measurements I've decided to replace the resistors to a 10K fixed and a 15K trimmer instead of using 6.8K.

I've also decided to use two 9V batteries in series as this improved both the response magnitude somewhat.

I've simply poked a hole on the bottom of the tin can, and glued the PC board to it. I've soldered a 2.5cm / 1 inch stiff wire protruding from the board inside the chamber.

Most tin cans are coated with an epoxy resin in the inside, so I needed to grind that away with a piece of sandpaper. Don't make the mistake I did and do this before assembly. ;)

The inside of the chamber with the sanded walls and elongated base lead


The chamber must be shielded and grounded to prevent static electricity to interfere with the readings. If you're lucky and can get a really thin aluminium foil, it won't block even alpha radiation completely. According to my experiments the foil I use blocks around 75% of alpha radiation, which fine I'd say, because alpha is very energetic and it's very easy to detect with this sensitive circuit.

You can use hair-thin wires and solder it on the chamber forming a mesh. This will effectively shield static too, but won't prevent dust, humidity and UV light to get into it. All of these can generate false readings.

Turning it on for the first time

Connect a power supply (or battery) and measure the current uptake with a DVM. It should be around 100uA. Anything between 20uA-200mA is possible and acceptable, and this depends on transistor types and also the specimen used.

Double check that the chamber is shielded and is grounded. Let it sit for two minutes.

Measure the voltage between the appropriate leads of the resistors and ground. You should get two nearly identical values, and preferably around half the voltage of your power supply. A voltage drop of a few tenths of volts will provide a good range and decent sensitivity.

If the voltage is too high, use larger resistors, if too small, use smaller ones. If the voltage is too dissimilar (more than a few dozen millivolts) it either means your transistors are not properly paired up, or something conducting got onto one of the leads. This can be flux, fingerprint or glue. You might be able to remove such stains with isopropyl- or ethyl alcohol, acetone or petroleum ether.

Measure the voltage between the appropriate leads of the two resistors. Now you're practically measuring the difference of the voltage on the resistors. This should be reasonably close to zero, and you should be able to set it to near zero with a trimmer.

You'll never get a perfect zero, and the circuit will drift around a bit, but not nearly as much as a simple one would do. A few millivolts of drift is perfectly fine. I got about 3mV drift in the course of.... like... ever. :) I consider this circuit to be rock solid. :)

It's also worth mentioning that although the circuit doesn't drift much, large jumps of a couple (dozen) millivolts can be observed a few times every minute. This can be from Radon decay that is present in the air naturally. This source of error can be cancelled if the chamber is sealed and left alone so most of the Radon decays away.

Even after such sealing and "ageing" the chamber might be a bit "jumpy" and large spikes might be reported by the DVM. This can be because of gamma or beta rays are hitting the transistor junctions, or because of cosmic background radiation. Muons coming from above can easily hit the camber and cause a nice trail of ions causing these spikes.

It'd be interesting to build several chambers plus a coincidence detector, and having a very own cosmic ray observatory. Truly fascinating!

Measuring samples 

This detector provides ample sensitivity. It can detect anything from my small collections of radioactive materials. Thoriated tungsten welding rods, thoriated lantern mantles give very nice readings even through aluminium foil.

Alpha sources are more easily measured through a thin wire mesh with large holes. Alpha particles cannot penetrate more than a few centimetres / inches deep into air, and almost impossible for them to go through any solids or even liquid films. A thin aluminium foil is almost impenetrable for them.

The ion chamber is the most sensitive to alpha and muons. Beta rays also leave thin traces of ions after them, so they can be detected too, but most of their energy will be absorbed outside of the camber, or in the chamber's walls, and not in the air. This energy is lost and invisible for our chamber. Gamma rays hurtle right through the chamber leaving almost no sign of them. They are hard to detect, and only large fluxes of gamma photons can be detected.

My entire collections of materials packed closely together and inside a thick plastic box causes about 10mV of deflection. (Alpha is utterly blocked, a few beta electrons might escape, but most of the gamma radiation is glowing right through the walls of the plastic box.)

Weak samples can be placed inside the chamber which is sealed afterwards. Readings can be monitored over a length of time and an average reading may be calculated.

Be careful not to contaminate the chamber, and clean it thoroughly after placing anything inside.

Future improvements

The chamber could be supplied with high voltage to gather ions more efficiently. Proper data logger circuit may be added. Several chambers could be paralleled and the outputs could be fed into a coincidence detector to build a muon observatory. :)

We'll be back at these soon.