Here is a cheap Geiger counter I purchased from eBay some time ago.
As with a lot of stuff on eBay, its origins are obscure. It has the model number "MAC.TP-909" printed on it, and claims to have been produced by "MAC Canada" (a company which, if it exists, has no presence on the web). Most likely it's the same thing as an LK3600 which is sold by Jinning Luke Testing Equipment. Which one is a clone of the other, or whether both are copies of some other device, isn't clear; often with some of these Chinese gadgets, the original product is produced by some nameless factory and different sellers stamp their own brand on it.
I decided to get this unit over several others on offer for three main reasons:
Firstly, it was cheap. It wasn't a big loss if it didn't work.
Secondly, it displayed dose rate in μSv/h to three decimal places rather than the usual two. While I'm well aware of the difference between precision and accuracy, it would stand to reason if a unit has an extra decimal place of resolution it would at least be somewhat more sensitive. The data sheet also suggested using a 10 second sample rate for measuring background levels, which also implies a high sensitivity. Unfortunately, it turned out this was most definitely not the case; it was nowhere near as sensitive as these things implied.
Thirdly, and very importantly, it looked like any other item of electrical test equipment and wasn't decorated by stupid radiation danger symbols. This meant it could be taken places, such as on-board aircraft (which I have now done many times), without exciting hysterical over-reactions from the ignorant. I don't know why Geiger counters are so commonly decorated with a trefoil radiation danger symbol. This does not make any sense; the symbol is designed to warn people of the presence of an ionising radiation source which the counters so decorated invariably do not contain.
I'm going to criticise the unit fairly heavily here, but I need to make it clear that this wasn't fraud or a ripoff. I did get a genuine Geiger counter that does (sort of) work. There was nothing dishonest about the way it was presented on eBay, the price was very good, the delivery was fast and the seller [who FWIW is no longer on eBay] appeared to be efficient and honest.
Setup and usage
The first issue I noticed was that it was surprisingly difficult to set up and many of the functions on it were impossible to understand from the broken English in the manual and/or appeared to be useless.
There was some form of total dose accumulation which I could never figure out. The thing doesn't have a clock in it and it appears that you have to manually enter date stamps to create a dose record. It would be impractical to use this thing as a dosimeter.
There is an alarm function which flashes an LED and produces a beep on a small speaker when a threshold is exceeded. Unfortunately, it only produces a momentary flash and "pip" sound (both easily missed) once at the end of each sampling interval (which might be up to 80 seconds long), even if the radiation level is over the threshold for the entirety of that interval. The makes the alarm function pretty much useless.
Even changing the sampling rate was a trial; you have to alternate between pressing the "up arrow" and "double up arrow" buttons to scroll through the available rates. There is no menu structure on the thing; you just press buttons and things happen.
The unit claims it has 1, 2, 5, 10 and 20 second sample rates, but in fact the actual update rate is exactly 4 times the claimed rate. That is to say, the "10 second" sample rate actually only updates every 40 seconds.
How could such an error occur?
One possibility is that the firmware was developed for a micrcontroller with a 16 MHz crystal, but then the device was subsequently fitted with a 4 MHz crystal to save power and the firmware was not fully updated to reflect this.
Despite having a speaker, there is no way to set it up to produce the classic Geiger counter clicks. This is very disappointing because this gives a more immediate indication of radiation levels than waiting for the display to update.
There is a back-light. This only turns on when you change the display mode. So, to illuminate the display you are currently looking at, you have to cycle through all of the display modes to get back to where you started. Also, the back-light remains on for a variable length of time from 10 seconds upwards.
Having said all of this, it does at least work for detecting radiation dose rate, which is the main thing you want to use it for.
Calibration and accuracy
The measurements of background level by this counter were jumping all over the place (a function of sample rate, more on this later) and appeared to be too low by about a factor of two. They should have been between 0.1 and 0.2 μSv/hr, but were nearly all under 0.1.
I decided to look into this further since it shouldn't have been this inaccurate. Was it a faulty tube, or was it a gross calibration error?
Because this unit does not appear to do any sort of filtering or dead-time compensation (at least at background rates), it was possible to determine how it was calibrated by observing the readings. I found that at the 80 second sample rate, you would get readings of (for example) 0.067, 0.070 and 0.072 μSv/hr, but never 0.068, 0.069 and 0.071. A little bit of observation showed that the counter was counting in 0.0026 μSv/hr increments. This is to be expected because the counter counts a whole number of pulses rather than measuring some continuous quantity, so the reading will only be able to change in steps corresponding to an integral number of pulses. More sophisticated signal processing (dead time compensation, filtering, or non-linear scaling to fit tube characteristics) would have hidden this effect, but fortunately it appears that either these don't have any effect at background rates or, more likely, the counter lacks these altogether.
From this it could be calculated that the meter had been calibrated such that each pulse was worth 0.058 nSv.
The counter contains a single Chinese J305 tube (more on this later). It wasn't possible to find an official data sheet for the J305 tube, however in several places (e.g. here), it has a quoted sensitivity of 123 cpm/μSv/h.
This converts to one pulse per 0.136 nSv. I.e. different from the meter calibration by a factor of 2.3.
So I could conclude that the tube was in fact working correctly and problem was that the calibration factor in the meter was 2.3 times too low.
I can only speculate on how such an error came to be, but it is possible that the firmware in this meter was written for two SBM-20 tubes (which are about 20% more sensitive than the J305 and therefore two of them would have give a pulse rate about 2.4 time greater), but in actual production only a single J305 was used to cut costs.
Subsequent to doing doing this calculation, I found a document by the Japanese Consumer Protection Agency, where they did some comparative tests of some different Geiger counters. It's in Japanese, but it clearly shows my counter (badge-plated as an LK3600) being tested:
At a level of 5.13 μSv/h, my counter (shown as "No. 1" on the graph) was reading 2.2 to 2.3 μSv/h.
Which is pretty much exactly a factor of 2.3 too low. Ha! Vindicated! In fact, given the uncertainties involved, it's probably a fluke that my calculations came out so close.
Sample rate
The statistical accuracy of a geiger counter depends on the number of pulses you measure during the sampling period. At high radiation levels, it's OK to sample very fast because there are lots of pulses per second. At background levels, however, the tube in this counter is only giving a single pulse every 4 seconds, so you have to sample for a long period to get any sort of accuracy.
It turns out that the available sample rates (despite being 4 times slower than stated) are still too fast for measuring background radiation to any useful degree of accuracy.
Ignoring for the moment the calibration error in this instrument, the tube will give about 20 counts per 80 seconds at the normal background rate on 0.12 μSv/hr.
As the timing of the pulses varies randomly (which in an inherent property of a Geiger tube), calculation of the accuracy isn't straight forward, but using a +/- 3 standard deviation level, which you would normally do for a measuring instrument, you get a whopping +/-67% error! This is about +/- 0.08 μSv/hr.
Even if you slacken the requirement to +/-2 standard deviations (which in practice means about a third of the measurements will be outside of the stated error band), it still gives you a +/-45% error.
So it turns out the three decimal places of precision on the display is just window dressing. The third decimal place is totally meaningless and even the second decimal place is close to useless. It would be legitimate to display the reading to the nearest 0.05.
This counter could do much better if it had the option of a few slower sample rates. With a 1 hour sample rate, the error would be reduced to about +/-10%.
Range
It appears that this unit reads up to 999 μSv/hr (effectively 1 mSv/hr), but I have no way of verifying this. It is not clear whether it performs dead time compensation or detects fold-back; but based on what I've seen so far it probably doesn't do either. If this is the case, it may read significantly low as the rate approaches 1 mSv/h, and would be completely untrustworthy (reading very low) at higher levels.
What is 1 mSv/hr? It's about 8000 x background level, which sounds super-deadly but is actually tolerable for short periods; you'd be getting the equivalent of a full-body CT scan every 8 hours at these levels. In an emergency, an adult could work in such a radiation field for a week and most likely walk away essentially unscathed (your chance of contracting fatal cancer would be increased by about 1 percentage point).
It is not pleasing to learn that such levels are considered to be trivially low when it comes to a nuclear war. In his book Nuclear War Survival Skills Cresson H. Kearny comments about a Geiger counter:
"The highest dose rate that it can measure, one roentgen per hour, is far too low to be of much use in a nuclear war."1 R/h is in fact 10 mSv/h, so the maximum rate my counter can go to is ten times less than something that is considered to be "far too low" for a nuclear war! It would almost seem that a nuclear war is a bad thing that we should be trying to avoid.
Looking inside
 |
| [click for larger image] |
While fairly well laid out and not exactly a dog's breakfast, it isn't the best constructed thing ever. Some of the SMD components are not located squarely on their pads, indicating it was probably assembled on the cheap. There is significant corrosion on the leads of the LED (top right). You couldn't reliably solder something this corroded, so I assume the unit must have been sitting around in a humid warehouse for several years. There is also much dark flux remaining around the PIC microcontroller.
 |
| Microcontroller covered with flux |
This suggests that the microcontroller has been replaced. Most of the problems with this unit are related to the firmware, so it's somewhat suspicious that the micrcontroller, which contains the firmware, appears to have been replaced. Did they update it in a hurry and put in incorrect or untested firmware?
 |
| The Geiger tube |
I originally thought the Geiger tube was a Soviet SBM-20, since it was the same size as one, but on opening the heat-shrink that concealed the tube I found it was in fact a Chinese J305. This is a cheaper tube, especially for a Chinese manufacturer who can buy them in bulk. It is billed as an equivalent to the SBM-20, but is overall inferior (less sensitive, smaller operating temperature range and shorter operational lifetime), although not seriously so.
For reference, here is an SBM-20.
But one feature that jumped out at me was the anode wire (the pink wire in the photo). It runs the whole length of the tube and was in fact heat-shrinked to it. Exactly what you don't want as this maximises capacitance to earth. Even a small amount of extra capacitance on the anode can affect the accuracy of the tube and more than a little can impact on its lifespan.
The location of this wire shows that there's something seriously amiss with the design. By simply installing the tube the other way around, you could eliminate this problem at no extra cost. It looks like it was designed in a hurry by someone, possibly in response to the demand for such devices after Fukushima.
Power consumption
I measured the power consumption of this unit.
Turned on without the back-light it draws about 7.2 mA
 |
| Power Consumption: Turned on without backlight |
This is unnecessarily high for a device like this and it looks like the design hasn't been optimised for power consumption. It is powered by two AA batteries, and with alkaline cells it should last about a fortnight at this rate. I suppose that's not unusably terrible, but you could do a lot better.
With the back-light on, it draws an extra 1.6 mA
 |
| Power Consumption: Turned on with backlight |
This is considerably lower than I would have thought and means it must have a very efficient back-light.
Turned off it still draws 1 mA from the batteries.
 |
| Power Consumption: Turned off |
This is about a thousand times what I would have expect and is a puzzle. Some Geiger counters still keep counting (I believe the Gamma-Scout does this) with the power off, so you can use them as a personal dosimeter if you need to, but there is no evidence this counter does this.
This is a terrible design, and will flatten a pair of alkaline cells in about 3 months. It's something you really don't want with a device that might be stored in a cupboard until needed - I known you're meant to take the batteries out, but most people don't do this.
Some tests
I don't have any radioactive sources to test this device against, but did manage to get access to (it was in a laboratory I visited, I do not have this in my possession) a jar of uranyl acetate.
Here is the reading I obtained. Taking into account the calibration factor of the meter, this is actually 4.8 μSv/h, which is 40 x background level. This is surprisingly low given that the substance is (by weight) composed almost entirely of uranium and the meter is sitting right against the jar.
I think the issue is that this stuff is manufactured from depleted uranium. This emits mostly alpha radiation which is not picked up the by the counter (and wouldn't even penetrate the glass bottle), with only a small amount of relatively feeble gamma radiation. It is possible the counter is in fact mostly picking up gamma and beta emissions from the Thorium-234 daughter product that is created as the uranium decays.
By the way, it would not be hazardous to work in the vicinity of this jar. Radiation decreases with distance according to the inverse square law and even about a foot distant the rate is only only a little over background. Don't eat it or breath in the powder though; alpha emitters are bad news inside the body and uranium is chemically toxic as well.
Here is a photo of the counter on board an aircraft in flight. I did not want to draw attention to the counter by using a flash, so the photo is dim and blurred.
The reading translates to 3.6 μSv/h (30 x background level).
Interestingly, the Bureau of Meteorology, in their Guide to Space Radiation give an example of the expected radiation dose for a Melbourne-Perth flight, and this averages out to rate of 4 μSv/h, so it looks like the counter is doing the right thing.
Conclusion
This counter sucks in a lot of ways but does actually work. If you compensate for the calibration error it seems to be moderately accurate.
It's not useful at background levels and can't pick up natural events like radon washouts. It is highly unlikely to be able to detect the low levels of fallout we might experience from an isolated nuclear accident in the northern hemisphere.
On the other hand, the range is not high enough to be usable in the event of a near-by nuclear explosion. If you were in Alice Springs and Pine Gap was attacked, the expected peak fallout levels would be in excess of one thousand times the maximum range of this counter.
Nonetheless, it has its place.
If there is a radiological disaster far enough away that you aren't immediately overwhelmed by the fallout it is sensitive enough to warn you before radiation levels get high enough to be dangerous and could be used to navigate your way to safety. If you were worried about having a contaminated item in your possession (e.g. radioactive jewellery, which sometimes crops up), it will in most cases be able to detect this too.
I wouldn't recommend buying this particular unit, but similarly priced eBay units are worth considering if you feel you need something like this.
No comments:
Post a Comment