Editor’s Note: When my Geiger counter kit refused to count (Click Hereto see details), my friend David Ashton in Australia kindly offered to take a look at it for me. He also kindly offered to document what he did in the hopes that it will help others with the same kit. So now let’s hand over to David…
When I received the counter I noticed it had a wire broken on the speaker. Max advised me that this had been connected when he assembled it, so I assumed it had broken off due to jarring in transit. I did not immediately reconnect it, which in retrospect was a bad idea! I first set about probing around with my digital multimeter (DMM) and oscilloscope. First let’s look at the schematic of the counter as illustrated in Figure 1.
Let’s start with IC1, which is a 555 Timer. This chip is intended for use in a variety of timer, pulse generation and oscillator applications. As an aside, the design for the 555 was proposed in 1970 and the first production chips were released in 1971, so this chip design has been gamely working away for 40 years now, which is rather impressive when you come to think about it.
Anyway, IC1 is a 555 in astable mode. It feeds Q1 which drives transformer T1. The secondary voltage of T1 is rectified and via R3 supplies the Geiger tube GM1. This tube will have a very high impedance in the absence of any radiation, but when any beta or gamma radiation is present this will ionize the low pressure gas inside it and the tube will “strike” much like a neon bulb. This will result in a short negative-going pulse at the + terminal of GM1. This pulse is used to perform two actions:
Switch on Q2 to produce a click in the speaker SPK
So far so good, but on close examination I realized that there are a few strange things about this circuit as follows:
The 555, though in astable mode, does not employ the standard astable configuration, which uses a resistor Ra from the supply to pin 7, another resistor Rb from Pin 7 to pins 6 and 2, and then a capacitor to ground. In this circuit, Rb is effectively zero. Rb governs the Off time (negative-going) pulse time, so with no Rb this time will be of extremely short duration.
The output of T1 is rectified by diode D1 but there is no smoothing capacitor at its output, so the GM tube will receive a pulsed DC.
Although Q3 and Q2 are referenced to the negative side of the battery, the high voltage part of the circuit is not DC-referenced to the low voltage part of the circuit. There will be some capacitive coupling from the T1 secondary to T1 primary, however, and the LED drive circuit is capacitively coupled to the GM tube anode by a 10 pF capacitor.
There is no limiting resistor in series with the LED as is common practice. Possibly the LED is a current limited one and does not need one, but another correspondent with a slightly different model counter had a place for one on his PC board.
Max’s initial complaints were as follows:
The speaker did not click at all, even when the circuit was brought close to a radioactive source.
The LED stayed on all the time.
Max had confirmed that he got a “click” from the speaker when he shorted the tube (simulating a radiation “event”). My initial theory was that there was not a high enough voltage to get the tube to ionize properly, so my first test was to check the voltage across the tube.
The result was around 79 volts, which is far too low. I had noticed the unusual configuration of the 555 oscillator so I tried putting in an Rb resistor to make the duty cycle of the pulses a bit longer. This entailed cutting some tracks on the PCB around the 555, but when I did it the voltage on the tube looked even lower. Looking at T1 output with a scope showed a very short duration “ringing” pulse output. The circuit description said that T1 provided a high voltage output due to its turns ratio. I am inclined to think that there is more of a “flyback” effect here, as the pulses are extremely short and lengthening them seems to give less voltage at the output.
Figure 2. Scope trace of T1 Output
Hor: 2mS / div, Vert: 100V / div (10v + x10 probe)
This was taken after I had put a 1nF smoothing cap on the other side of D1, which accounts for the difference between the positive and negative values.
The 555 operates at about 300 Hz. At one point I connected my DVM while my scope was already connected at T1 output and the amplitude decreased severely. I came to the conclusion that because of the high impedances of the tube and the very low duty cycle of the oscillator, even a 1 Megohm load from a DVM was probably enough to overload the HT supply to the tube and drag it down.
One of my theories all along was that the HT supply should be smoothed, so I tried various capacitors at the Cathode of D1. All seemed to improve matters – the HT went up to a bit over 100 volts on my DVM – but by this time I was convinced that my DVM was too much of a load for the HT circuit. About this time the one remaining wire on the speaker gave up the ghost and the speaker fell off.
Editor’s Note:The speaker fell off? The SPEAKER FELL OFF? My lower lip is quivering and a little tear is rolling down my cheek … my poor little baby!
I attached it to the board with double-sided tape and re-soldered it to the board with new wires. Once I had done this, and without any test equipment or smoothing capacitor in the circuit, I switched the counter on.
It made a click when I switched it on… then a few seconds later I got another click… then another… then some more at random intervals. It seemed to be working! The LED was still on all of the time, but it did flash slightly brighter when a click came. If I connected my DVM or scope onto the HT circuit, the clicks stopped. This confirmed my suspicion that even a 1 Megohm DVM or Scope input loads the HT line too much. Coming from the valve (vacuum tube) era, I should have known better!
The circuit was now operating without any modifications – I had undone my changes to the 555 circuit and did not have a smoothing capacitor on the HT. So why had it not worked for Max? He was getting clicks when he shorted the tube, so the speaker must have been connected. Some correspondents have reported high sensitivity to humidity, and possibly (being in Alabama in storm season) this may have been a factor. More than that I cannot say, I found no obvious cause that would have stopped it working.
I still had the problem with the LED being on all the time. One of my theories for this was that the unsmoothed HT pulses were getting through to the LED driver circuit and switching it on. So a smoothing capacitor should improve this. Other circuits have used a variety of capacitors – 2.2 nF being a common value. I had a 1 nF and 4.7 nF. Both improved the LED condition – it got a bit dimmer – but it did not go off. There did not seem to be much difference between to two caps, so for size reasons I stuck with the 1nF. The fact that the circuit works with no smoothing cap shows the high impedance nature of the GM tube – its own capacitance and stray capacitances in the circuit do smooth the HT to quite a large extent. The following trace is with a 1nF capacitor.
Figure 3. Cathode of D3.
Hor: 2mS / div Vert: 100 v/div (10v / x10 probe).
Note that even this loads down the HT considerably and these are not the true voltages or waveforms in the circuit.
My other theory was that possibly – since the LED had no current limiting resistor – even short pulses of high current in the LED were causing it to be much brighter than it would be if the current was limited. So I put a 470 ohm resistor in series with the LED. There was a very slight – if any – improvement, but obviously this was not the problem.
I also tried increasing the value of R3 which supplies the HT to the GM tube. Most circuits I have seen use anything between 2 and 10 Megohms for this, rather than the 100K used here. I changed R3 to 4.7 Megohms. It did not change the LED operation much, if anything, but it did improve the intensity and quality of the “click” from the speaker.
The LED driver transistor Q3 is connected to the GM tube anode via a 10 pF capacitor and a 10 Megohm resistor. So any positive-going pulse of more than a few volts would turn the transistor – and the LED – on to some extent. I experimented with a few megohm-value resistors across the Base-Emitter of Q3. There was an improvement but not a complete turn off of the LED. I also tried – separately – a couple of capacitors – a 10 pF and a 39 pF I think – again with not much improvement.
To have a look at the waveforms I was dealing with, I put my scope across the BE of Q3. There was an immediate and dramatic improvement. The LED stayed off and only flashed when the speaker clicked. It was then a simple matter to measure my scope probe – it was 1 Megohm and about 72 pF. I tried a 1 Meg resistor and a 68 pF capacitor across the BE of Q3 and it worked as well as the scope probe. I managed to capture an event along with the power supply pulses and there was a significant difference in both the amplitude and duration of the power and event pulses:
Figure 4. B-E of Q3. Power supply plus event pulses.
Hor: 2mS / div Vert: 1v / div.
This is with 1 Megohm + 68pF already across the BE of Q3 so the scope effectively makes this 500K + 140 pF approx, which will probably attenuate both pulses lower than actual values. The event pulse does have a strong negative component as well, which is not visible in this trace.
My one large problem came in testing the circuit. I tried all sorts of things that might have some level of radioactivity. I had an old glow-in-the-dark compass but that did nothing. I tried my smoke detector – but these emit primarily alpha particles which have very little penetrating power (they have difficulty getting through a sheet of paper, let alone the glass of a GM tube).
Eventually I took it to a Nuclear Medicine practice in a neighboring town (they do radiotherapy and various diagnostic tests using radioactive substances) and they very kindly let me have access to a radioactive source. Below you can see a short movie I took of this. In my workshop at home I was getting 10-20 counts (clicks / flashes) per minute which I gather is pretty typical for background radiation. When I first switched the counter on in the lab of the nuclear medicine place, I was getting 2-3 times that. As I approached the radioactive source (which I was told was an 8 GigaBequerel Technetium source giving primarily gamma radiation) the clicking and flashes rapidly increased. At the nearest point, the clicks had merged into a buzz, and the LED almost extinguished. I presume this was due to the short time between event pulses causing them to reduce in amplitude as the tube did not have time to recover fully (“quench” is the technical term) in between pulses. At any rate. you’d certainly be aware of it long before the radiation got to this level!
One correspondent suggested that the “No-salt Salt” that you can buy (which replaces a lot of the Sodium Chloride with Potassium Chloride) is mildly radioactive. I tried this just before I sent the counter back to Max and it does indeed work. What I bought was “Lite” salt which is about 50% NaCl and about 50% KCl.
Figure 5. Geiger counter under test with potassium chloride salt.
I took three 1-minute readings with just background radiation followed by three with the (cardboard) Lite salt container right against the tube. The results are as shown in Table 1. As we see, there’s not a huge difference, but there is definitely an increase from the background and does provide a cheap and easy (and tasty) test.
Table 1. Geiger counter under test with potassium chloride salt.
Click Herefor quite a good summary of common radioactive materials.
Humidity: Some users reported high sensitivity to high humidity. This was difficult to test as my workshop was usually around 10 degrees C, so breathing on the counter probably resulted in condensation on the components and did stop it functioning. However, the problem usually cleared in a couple of minutes; faster if hot air was blown on it. Non-condensing humidity should not be a problem.
Note: This counter will detect Beta Particles (electrons) and Gamma rays. Alpha particles (helium nuclei) will not pass through the glass and metal walls of this tube and will not be detected. Smoke detectors emit mainly alpha radiation and mine did not register on this counter at all. Some readers say that they have been able to detect smoke detector radiation, though.
I have documented the modifications I made below in case anyone wants to try these on their units. Chaney make some similar kits and if yours is not the same as Max’s (C6979) you should be able to work out the changes for your kit using the schematics below.
Below is a copy of the schematic with my modifications shown in RED:
Figure 6. Modifications to Geiger counter Schematic as described in text.
Existing and modified component values are shown below.
C4 is soldered to D1 cathode and the Jumper wire as shown. C5 and R7 are mounted in new holes on the board as shown in the PCB diagram below. R7 may be bent over R6 as shown to avoid it sticking up above other components. R3 is changed from 100K to 4M7. The current limiting resistor for the LED, R8, is mounted on the track side of the board as shown below.
Figure 8. Modifications to PC Board.
Two pairs of holes are drilled adjacent to the copper tracks coming from the Base and Emitter of Q3 as shown, and C5 and R7 are mounted in these holes (C5 adjacent to Q3). Careful siting of the holes enables the component leads to be soldered to the tracks, or the leads may be bent over onto the tracks.
The track from the collector of Q3 to the LED is broken as shown and a 470 ohm resistor is soldered to the 2 pads and mounted horizontally on the board as shown. This modification does not directly eliminate any problems and may be omitted if desired and/or if operation is satisfactory without it.
Testing with varying battery voltage
Voltages higher than the nominal 9V battery voltage (up to around 11V) showed no change in operation. Lower voltages (down to around 7V) also showed no appreciable changes. Voltages lower than 7V resulted in some power supply buzz in the speaker, though the counter did continue to work.
My aim was just to get this Geiger counter working, though the temptation to make major changes was great! Having read up a bit on Geiger-Muller tube operation, most tubes need a voltage of around 400V for optimum operation, with a “Plateau” about 40V either side of this where operation will be satisfactory and repeatable.
Although it is not really possible to measure the voltage at which this particular unit operates, it is fairly obvious that the voltage on the tube will vary with varying battery voltage. Maxim produces an IC that can be used to generate a suitable regulated supply (Click Here for details) and were I to build a counter from scratch I would use something like this. If I was doing a lot of this, I’d be tempted to get or develop and make some test equipment that could measure high voltages at very small loads. With the very high input impedance of Intersil’s ubiquitous ICL 71x6 chips this shouldn’t be too difficult.
Most correspondents who had the same or similar kits had problems with the LED staying on, and this is the main area in which the folks who designed this kit appear not to have done their homework. The fix is fairly simple and wouldn’t increase the cost of the kit much (obviously for this kit that is a prime consideration). That said, once the kit is working properly its performance is pretty good – it is fairly sensitive and works satisfactorily over varying battery voltages.
Many readers commented on Chaney’s poor support and refusal to acknowledge any problems. With what I have learned from this, I’d rather build my own I think. Or maybe I should get into the kit business!
I must apologies to Max and his many correspondents for the length of time taken to get this information – and the kit – back to him. I had 10 days off around Easter in April and hoped to have the counter to get stuck into during that time. Thereafter I was very busy at work with some projects and could not devote the time to it that I would have wished. It arrived a few days after I was back at work, of course. Max’s shipping really was by ship I think!
My thanks – and admiration – must go to all the correspondents to Max’s pages on the subject, There were some very detailed analyses of the kit – far better than I have achieved here. I tried not to be initially influenced by these posts, though most of them seem to have come up with very similar – possibly better – solutions than me. Thanks all for the inspiring input.
Here are some links to useful information and other posts:
Click Here for a summary of tube characteristics for several Russian tubes (although this appears at first glance to be in Russian, English translations are provided also).
EETimes reader pchow documented his mods – very similar to mine – very nicely (Click Here). Nice work, Peter.
EETimes reader finevlad gave links to other schematics (Click Here and Here). These were very useful in seeing how totally unrelated circuits drove the tube.
EETimes reader spyderjacks documented his excellent debugging (Click Here), including great scope prints – I wish I had a scope like that! I’m fairly ashamed of my antique….but it works…
Thanks again to EETimes reader gwasic for the suggestion about the No-Salt Salt.
And to all the readers who took the trouble to comment on Max’s original blog… my apologies once more for keeping you in suspense for so long.
I did use a x10 probe on the HV side, and it is 10 Megohms - on the BE of Q3 I had it on x1 and it was 1 Meg. Ref DMMs - I had it in mind that mine were 1Meg but I will check that now. In any case this circuit is demonstrably sensitive to even a x10 scope probe, I think something with 100 Meg would be better. It should be possible to build a 100:1 divider using (say) 22 Meg resistors which are easily obtainable, and couple it to a standard DMM...
Ref building it differently... sometimes you see a circuit and think "That has been really well designed". This is NOT one of those...
and ref uS, us, thanks for the correction.
I'm very surprised: you work with a scope probe with 1 Mohm resistance? I do that only when I need to work at 100 MHz or above (wide band probes) or when the signal amplitude is very low. In all the other cases I use a 10:1 probe with 10 Mohm input resistance.
And a multimeter with 1 Mohm input resistance? Even the cheapest one I have has 10 Mohm!
"if I was building one from scratch I'd do it a bit differently."
Were it a single case in which this isn't true? :-)
N.B. uS = microsiemens; us = microseconds
Good point Vlad. I think some of it WAS background as it did seem to vary a bit depending where I was - in building or out, etc.
Funny story - when I was at the nuclear medicine place, one of the staff said "Come and test me!" From a background count well above the usual 20 CPM, when I held the counter against him I did not get one click for about 20 seconds. "I reckon you're dead" I said. "No", he replied "I'm just made of lead!"
The other point here is that as the supply to the tube is unregulated, it might go a bit above the "Plateau" region, at which point apparently the tube will frequently self-ionise.
It's refreshing to see that, with all of the multi-GHz processors, DSPs, MEMs and everything else complex and advanced, basic electronics theory still exists. Maybe it's a little in short supply at the original designer of this kit, but here at eeTimes, it's not.
What's more is that it's a great story of troubleshooting methodology. I think some people forget that the fundamentals are still important. Very nice job.
David and Max,
Thank you both for the clarification and update.
BTW, I was also able to verify last night that Morton Salt Substitute (Potassium Chloride, w/o NaCl) gave similar results as in David's article. The 3-1/8 oz. salt container, when laid on top of the G-M tube, produced 2-3 times the CPM compared to normal background. e.g. background CPM was ~20, and the salt substitute was 40's to low 60's CPM.
Thanks again for the great investigative work, and the EE Times support of this interesting subject!
Great to hear it is working! My note is about the background count of the tube. Don't be confused trying to link it to background radiation. This should be mostly the self discharge of the tube itself.
Peter - a further comment ref your point 1. There is also a good AC capacitance coupling between the HV and LV sides through primary to secondary capacitance of T1. I think this is also responsible for introducing some AC voltage as well - at one stage I shorted the tube and still got fairly strong power supply pulses at the BE of Q3. I tried putting a 0.01 uF cap between the primary and secondary "ground" ends of T1's windings, but it did not make much difference. I was reluctant to short it because then the HV DC would have been directly coupled to Q2 and would probably have introduced further problems....
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.