[Swlug] DIY Geiger counter
James R. Haigh (+ML.LUG subaddress)
JRHaigh+ML.LUG at Runbox.com
Mon Mar 10 20:55:01 UTC 2025
At Z+0000=2025-03-09Sun21:44:56, Rhys Sage sent:
> My latest development is that using the circuit as it was, I replaced the 1uF elecrolytic with a 0.1uF electrolytic and raised the voltage from 3v to 4.7v. I put an eBay buck converter and powered the buck converter from my onboard solar system (14.2v).
>
> I noticed that the voltage range on the oscilloscope from the circuit was varying by 0.01V. That looks like I'm at least refining what I have.
That could just be noise getting into the circuit though all the loop area, or the external power supplies. I wouldn't read too much into a 0.01v reading because you can get this just by moving your hands around sometimes -- unless there is a really obvious correlation with what you are detecting... But even then, how do you know whether it is because the detector is detecting the change or because you moved your hands while moving the radioactive source to or from your detector crystal, affecting your circuit's reception to noise through its many loop antennae? If you think how easy it is to affect the sound on an AM radio just by waving your hand near its antenna like a theremin, or walking around it, this gives an insight into how much your own movement can be affecting the electrical activity in your circuit if you have a lot of antennae in your circuit.
> The amplifier circuit came from Instructables and I put it as it was: https://www.instructables.com/BC547-Double-Transistor-Audio-Amplifier/ - I don't know whether the link will be scrubbed [...]
Got it. Okay, so I see that your photo does match the photos on that page. The thing that through me off is how you are using your breadboard rails. I see now that, of your 4 breadboard rails, you only have power (from the button cell) on the outer 2, whereas your inner 2 are both the input signal from your crystal, arriving on the green wire connected via the green 2 x 1 pinstrip and then jumped with the black wire to the other inner rail where it inputs to your amplifier by the electrolytic used as a decoupling capacitor, possibly reverse polarising it... Hmm. I wonder whether that is meant to be a nonpolarised capacitor, but the Instructables one has the same kind of capacitor, so at least you did copy it correctly.
One thing that I notice is different between yours and the Instructable is that the Instructable happens to have much tighter loop area on the signal path thanks to the twisted cable on the input, even though this is not strictly necessary for audio frequencies as it seems to be aimed at. The twisted pair on the input will help to minimise the loop area of the 1st loop, which means that it would pick-up less hum, and have slightly less distortion of the audio signal.
Most of the loop area in that Instructables amplifier is along the power supply, especially that last bit where the crocodile clips spread apart. This could be truncated by placing a low-ESR electrolytic across the ends of the crocodile clips, or in your case, across your breadboard rails -- but definitely quite using rails far apart and use adjacent ones, as this is going to massively reduce loop areas of any current loops that involve the power supply.
For your tens-of-nanoseconds pulses from your crystal, you're talking about ultrasonic frequencies, even radio frequencies, so having your input signal jump across the breadboard with the black jumper I reckon is going to be way too much loop area for the kind of frequencies it needs to be able to carry in order to get a crisp signal.
Otherwise, what will likely be happening is that you might be spreading each 20ns pulse across, say, a millisecond or so, reducing its power by a few orders of magnitude to be too difficult to detect and distinguish it from noise, especially seeing as your loop area will mean that you will be collecting a lot more noise in that same millisecond.
> I do know that the BBS in use won't do photos or videos.
You mean Mailman? My experience on other LUGs on this Mailman server is that it does support small attachments, but most are set to a tiny file-size, e.g. 200KB or 1MB. So a small PNG or GIF is usually fine but yeah, you'll not be able to send HD photos or videos.
> HTML gets scrubbed too. It would not surprise me if there were mistakes in the circuit design - repositories like GitHub and Instructibles are very much caveat emptor.
I wonder whether the electrolytic should be a nonpolarised capacitor, e.g. a ceramic. It might work for a while, but electrolytics can be permanently damaged by reverse polarity, and its use in that circuit as a decoupling cap suggests that it could be subject to reverse polarisation. The resistor there is being used to DC-bias the input of the amp, i.e. the base of the 1st transistor. Due to the 2 voltage drops of the transistors, that biasing resistor will mean that that node at the 1st base will be kept higher than about 1.2v. Conceivably the input on the upstream side of the capacitor could be higher or lower than 1.2v, which means that the capacitor would be reverse polarised in 1 of those 2 cases. It has to go higher than 1.2v to activate the amp, so it can't be placed with its positive side towards the amp. But it also can't have its positive side towards the source, because if the source drops below 1.2v, it will again be reverse polarised.
So I think that might be a mistake. The capacitor used in that Instructable as a decoupling capacitor should probably be a nonpolarised sort, not an electrolytic. The electrolytic would be better-placed across the supply to truncate the effective current-change loop area of the supply.
> After checking with the power removed and with the power applied and with both my solar 14.2 and with a 9v PP3 I can confirm the graph on the scope was still flat so I'll agree that it does look like one of the transistors isn't doing very much.
My comment about the 1st transistor looking like it was acting as a diode was wrong. I had misidentified the red croc clip as you input, when actually that is presumably another source of power. So it looks like you are using 2 different voltage sources. I see now, by matching with the Instructables circuit, that the electrolytic is where the input comes in, so that far-side inner rail, which is connected to your near-side inner rail via the black jumper, that is your input, so the green and yellow wires connecting to the near-side rails in your photo must be coming from your crystal. The yellow wires are your button-cell's positive supply, so your crystal is presumably effectively dropping its resistance when it detects activity, such that the voltage on that green wire next to the yellow rises whenever it detects something. It will rise towards your button-cell voltage, and is only pulled-low in absence of current by the amplifier circuitry.
Okay, so this makes me think that this source can't easily fall below 1.2v (only when the button cell is flat but your other voltage source is still present would it present reverse polarity to the electrolytic), so your electrolytic should have its positive side facing the crystal connected to the positive of the button cell.
At least it won't get damaged while the button cell maintains its voltage, but I can see a reason why it won't work much at all. As soon as you have any activity that charged the capacitor, it has no way to discharge in your circuit. In an audio circuit, it can discharge, either because the current is alternating, or if high-side only (as is usually the case, I think), you need DC biasing resistors. In this case, you would need to add a pull-down at least, so that the decoupling capacitor can discharge again -- but as soon as you add that, you will get the reverse-polarisation of the electrolytic again, so I still think it should not be an electrolytic there.
> That'll be the next adjustment. What I saw today was much more encouraging.
>
> I would use a lower value capacitor but my lowest electrolytic is 0.1uF. I do have some metal film capacitors down to 0.01uF.
I would avoid using electrolytics as decoupling capacitors, so the metal film capacitors are probably a better choice because I don't think they have any polarity that you have to adhere to. However, given that this is not an audio application, we don't really need to decouple any DC component of the input anyway, so I would try without it. But replace it with a resistor to protect the transistors from over-current. Such a current-limiting resistor may be unnecessary if the crystal never exceeds the current of the transistors, but I don't know enough about the crystal to know whether this is the case, so I would include a protection resistor until I know otherwise.
> I'm very much blundering around in the dark, researching and asking questions. I have my goal and I'm trying to get there despite a sketchy knowledge of electronics and rusty programming skills.
I understand electronics very well, and also computer science. My level of programming depends on the programming language and context -- patchy in places but often very precise. I am learning Forth and assembly, and find that I can achieve even better speed and exactness in these languages.
I'm wondering why you didn't try my envelope detector circuit idea. However, looking at it again, I realise that if the crystal is only connected to button-cell positive anyway, current will only ever be flowing in 1 direction, so the Schottky would be unnecessary. A very small-value capacitor may still be useful there, though, to allow more current to pass through the crystal while its resistance is momentarily lowered. Ideally it would be best to place it a close as possible to the crystal, to minimise the loop area with the crystal. I don't think it matters too much which rail this capacitor goes between, so if you choose the positive rail that the crystal already connects to, you can make a really tight loop there by connecting the capacitor directly across the crystal.
That could be a useful improvement to your circuit. Try one of your 10nF metal-film caps directly across your crystal. Then want happens is that as it is connected to your button-cell, the capacitor will charge to a steady-state voltage that is the voltage across the crystal, probably something like 3v - 1.2v = 1.8v , or something like that. It could be a polarised cap here, but I think an electrolytic would be way too big, because what would happen when the crystal detects a pulse is that it would short the capacitor -- very quickly, passing as much current as possible through the crystal in that tiny 20ns window of your pulse. Still a tiny amount of energy, but perhaps a couple of orders of magnitude higher than if your 20ns pulse is trying to drive all the inductance in your circuit caused by all the loop area.
With the capacitor shorted, it can take as many microseconds as it wants to recharge, and as it does so, it would allow current to pass from your button cell, through itself, and into the base of your 1st transistor, out of its emitter and into the base of the 2nd transistor, out of its emitter and back to the button cell. That is a long loop with lots more loop area and therefore inductance, but thanks to the capacitor across the crystal, it can take much longer than 20ns to equalise and pass this current, and as it does, you will get the amplified current loops moving as well.
This is what I am thinking that you should try next (set your plaintext font to monospace to see the following schematic):-
+ + +
| | |
| | >
C .---o--. > < R
R |xxxxxx|+ < R > 3
Y:::PHOTO xx--- 10nF > 1 <
S:::DIODE xx--- < | OPEN-COLLECTOR
T |xxxxxx| R2 | C AMPLIFIED
A '------o------\/\/\--o----B C-- OUTPUT (INVERTED)
L E--B TO PULLED-UP ADC OR
E CURRENT-LIMITED LED
|
-
Minimise the area of the crystal--capacitor loop marked with cross-hatching.
Resistance values of the 3 resistors above...
* R1 probably quite high, e.g. in the range 10kΩ to 100kΩ. Yes, so the Instructables circuit has a 10kΩ here, so keep that here as it is.
* R2 quite low, e.g. 1kΩ.
* R3 quite low, e.g. 1kΩ.
R2 & R3 are calculated based on the upper-bound voltage across them, to limit the current to what the transistors can cope with. The calculation goes like this:-
* Lookup the collector-to-emitter current capacity of the BC547.
* Found in 1987 Motorola databook that BC547 has a continuous collector current capacity of 1ADC.
* The base-to-emitter current needn't be more than 1% of this, due to the current gain being 110 to 800, so limiting current to about 10mA will be sufficient protection.
* Voltage drop of BJT base-to-emitter is always about 0.6v when current flowing; collector-to-emitter when flowing is only about 0.2v, though.
* Then calculate the upper bound voltage across the resistors, e.g. V_R2 = 3.3v - 2*0.6v = 2.1v or V_R3 = 5v - 0.2v - 0.6v = 4.2v.
* Apply Ohm's Law to get your resistances, e.g. R2 = 2.1v/(10mA) = (100*2.1)(v/A) = 210Ω or R3 = 4.2v/(10mA) = (100*4.2)(v/A) = 420Ω.
> Progress is shown in snippets on my YouTube channel (British Tech Guru).
Sounds like the kind of channel I used to watch a lot of -- but if only Google would not keep killing my access to YouTube.
> Today's short videos on the Geiger counter will be online on the 16th and 17th. As it's now 6pm on Sunday evening, I won't be doing more on the project until next weekend. My work day starts with a 4AM alarm and I'm at work from 5:30am until 4:45pm every day. I do get an extended break in the middle from 10am to 1:15pm, most of the time. Last week I had Wednesday free because of a storm. We don't work in high winds or when snow is forecast.
Do you work outdoors? Electricity distribution? Then again, they tend to have more work in storms.
> Right now I'm just hoping the random interruptions in the almost straight line on the 'scope represent background radiation. I would expect background radiation to be in the range of 20-30cps. I didn't open the smoke alarm because my crystal is in a rectangular plastic block which I hope at $20 is real and not just a cute lump of Perspex; thus the crystal would never receive alpha rays from americium-241.
Oh, right, so you are trying to detect background radiation. I didn't think of that possibility. In that case, you are relying on being so sensitive as to be able to detect individual events, which would be great, but would also be tricky.
Have you got a 1nF capacitor that you could try across the crystal instead of the 10nF one in the schematic above? Can you raise the voltage across your crystal any more without damaging it?
> The big thick leads in the photo go to the scope/ (I'll include a photo just in case it doesn't get scrubbed) https://yt3.ggpht.com/N5oH38hpOLriH41kEZjayIbiYw65nYwIMqWoLIOaCWdUpa9pAD4HZbiW3NYEVf3sRpn-OgLh27rgn7I=s640-c-fcrop64=1,20000000dfffffff-rw-nd-v1
That's a slightly cropped version of the 2nd photo you sent me off-list. It does seem to have worked. Is that from a YouTube server? I notice the "yt3" subdomain, and "ggpht" looks familiar in this context. That might explain why you've seen it get scrubbed, not Mailman, but YouTube/Google breaking the link.
> The yellow/green pair go to the photodiode.
Sorry, yes, I forgot that you don't connect to the crystal directly. What is the voltage rating of your photodiode? If it is anything like an LED, then I'd guess it would be 5vDC max reverse bias voltage. But given that you have 2 forward voltage drops below it, that means that you could supply it with upto 6.2v, if it can tolerate 5v across it, because you'd reliably lose 1.2v across the BE-junctions of the transistors, 0.6v each.
Do you have another photodiode that you can experiment with separately from the crystal? Might be useful to see how sensitive you can make it to light in a dark room.
> Yellow is the emitter.
I can just about remember that diodes point from anode ("in-node" where conventional current goes in, usually) to cathode (which is crossed on the circuit symbol like the "t" is crossed in "cathode")... Emitter confuses me. Are photodiodes labelled differently? Is it the same as a diode's cathode? I guess it is by analogy with BJTs, but where the light activates the base. But in this circuit I would expect the cathode to be connected to the positive supply, and current coming out of the anode (like with a Zener when it reaches breakdown voltage). Hmm, I guess it works by analogy with PNP, because the emitter in that case is emitting holes, but the diode of the P-N junction on the emitter side is such that its anode is the same piece of silicon as the PNP's emitter, rather than the cathode in the NPN case.
So I guess yellow is the diode's anode, with the diode always being used in reverse bias mode like a Zener.
> The other yellow lead is +3v. The blue lead on the left goes from batt -ve to circuit GND.
>
> We really need to set up some place where we can post photos and videos. I don't do social media other than Mastodon. I do have a Peer Tube account but it takes way too long for videos to be approved. Mastodon only does still images on the instance I use.
Are you having trouble displaying my ASCII-art schematics? They will only look right in a monospace font, and given that we email in plaintext, there is no way to ensure that the font that they are displayed in is monospace, other than to assume that every email client has sensible defaults for displaying plaintext as monospace, which is unfortunately not the case, as a lot of email clients do not default to monospace for plaintext. Some email clients, especially on "smartphones", do a lot of other stupid things like mangling whitespace, so even if you set your default font to monospace, it can still look wrong. But I keep checking that the emails that I receive back from the mailing list are not mangled at all, they are exactly as the schematics are meant to look. So if they appear mangled, they are only mangled at the very last stage and so you can see the correct diagram by trying some different software or settings or mode (such as "view source", or copy/paste into a programmer's text editor).
I find it quicker and easier to use the keyboard than to use graphical software, but if it helps, I can try sending screenshots of my ASCII-art schematic diagrams.
I don't do social media either, but I have a VPS running -- might be able to help here somehow. Also, there's also BitTorrent magnet URLs which loads of Linux torrent downloaders support, including Aria2c, Transmission, Deluge, etc.. You could try posting magnet URLs of anything too big for the Mailman file-size limit.
PeerTube is great too -- I've dabbled with it a few years ago on NewPipe before I stopped using smartphones. It worked fine, but nothing much on it interested me. Interesting that you tried to upload to it and found a bottleneck there. That's sad. A moderation bottleneck probably heightened by all the legal threats of running such a service.
But on a mailing list that is already moderated in its own way, I don't see the harm in posting magnet links that don't require any moderation. This works fine for images, collections of images, and even videos.
Hey, have you heard of Tribler? I dabbled with it about a decade ago. I'm not sure whether it is still going but it seemed quite promising back then. It was a torrent client designed specially for video.
Before that, I used to use Miro as well, but I don't think it still works anymore. I haven't tried it in years. It was really good though back when it did work. I was a big fan of Miro.
Video isn't everything, though. I find still images quicker to process, and text even quicker, especially seeing as I can search back through text very easily just in my emails, whereas searching images and videos relies on technologies that give me the creeps, and I don't use them. So then I can only search images and videos that I have metadata for, such as subtitles, or chapters, or tags, but often they come without any of that, and it is time-consuming to add that myself -- I don't even have a system for it, other than to search for the email that linked to a particular video or image, by which time, said links are probably long since broken.
That's another nice thing about torrent magnet links is that they only break if everyone stops seeding, and anyone can keep them alive, or fix them by enabling seeding again. The cryptographic hash gives pretty decent assurance that if someone else provides the missing file, that it is the verbatim original as intended when the email was sent. Therefore they are easier to archive without breaking links, and easier to fix those links if missing files are later found or supplied by someone else.
In theory, I could set-up Aria2c on my server and automatically mirror magnet links. But I'm not exactly sure how I would do this. It'd probably take quite a bit of programming to get it to work properly, but it is something that I have been meaning to do for a while, because I keep running into this problem, and it sounds like you are too.
Kind regards,
James.
--
Wealth doesn't bring happiness, but poverty brings sadness.
Sent from Debian with Claws Mail, using email subaddressing as an alternative to error-prone heuristical spam filtering.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.lug.org.uk/pipermail/swlug/attachments/20250310/b3c6b79c/attachment.sig>
More information about the Swlug
mailing list