GM Signals - anode vs cathode?

[Joe Jarski]

I've been playing around with some geiger counter circuits and have come across different versions where the signal is taken from the anode side and some where the signal is taken from the cathode side. Are there any adantages to using one over the other? Either way you get a pulse and can process it and I've seen the same people use each version, but I'm not sure why or when you would use one over the other.

[Starfire]

Joe - it does not matter which source of signal is used but there is advantage to use a signal near ground rather than from a high voltage source. ie the anode ( signal source ) is grounded and the cathode made HV neg - - the voltage is relevant ( to something) in this case near ground potential

Better to think of potential difference rather than voltage - the difference between two points means that to the anode the other voltage is negative but to the cathode the other voltage is positive - it depends on your point of view which you ground

[Doug Coulter]

Most people take it off the anode for one obvious reason. That's so the cathode can be grounded, and that's the outside of the tube and also the noise shield. It's no big disadvantage to have to come up with a tiny coupling capacitor -- here I use from 30 to 270 pf, usually 47 pf, which are dirt cheap even with 1kv rating. I use 1 meg in series with the positive anode supply, you can use more, but not much less, as you never want the anode to see enough current to get to glow discharge -- ruins the tube quick. Which is also why one wants the smallest cap you can get away with and still drive your counting stuff -- so it doesn't store enough energy to damage the tube (or the counter input) if something goes wrong. The tubes do care about polarity, more often than not. The anode is usually a thin wire, and the high field near it helps the avalanche get going by accelerating incoming electrons that will then ionize more gas.

An interesting side note. Working on the standard counter, and BillF was here today. To make it practical to run long times on a battery is a design goal. I'm using one of those CCFL's with a volt doubler to make the voltage. Well, that makes it need about 1.7v input for the nominal tube voltage for those Russian tubes. But - for grins, I just hooked the thing across a 3.5v li cell and charged up the output filter cap (a rather excessive .47 uf polypropylene). That was at about noon. It's 5:50 now and the thing is still counting from the remaining charge in that cap from a one second charge at noon! Thus, in the standard counter, I'll have the uP simply run the power supply for maybe 1 second per hour, and the uP go to sleep except for the watch crystal oscillator between outputting counts every 10 seconds. The thing should run a year on a tiny Li cell if I also only power the rs-232 converter on demand! The extra couple switching P fets I'd need to do that are well worth it.

[chrismb]

John has it right. All the GM tube does is act as a momentary 'low resistor', so you can use it to 'pull-up' or 'pull-down' according to what you want. With the serial resistor between the tube and the ground, when the tube 'shorts' it will pull that side of the tube up. So you will get a positive-going pulse that only gets a few volts above ground - you don't even need a DC isolation capacitor!The down-side is that you will either be taking the signal off of the 'ground' of the tube (which may itself be subject to larger levels of interference, as it is usually the bulk[-ier] of the tube) - unless you chose to connect the tube the wrong way around and you've made the tube shell high +ve, in which case I think most designs would not be very efficient.

You can even connect a GM tube across a dividing resistor. Let's say you have a 1000V source but you want to run the tube at 500V. Just string two 5Mohm resistors across the source, and the GM tube in parallel with one of them. The GM tube will then take 500V, but if it 'fires', then the resistor in parallel with it is much larger than the conducting resistance of the tube, and the mid-point of those resistors will then get pulled up, or down.

[Joe Jarski]

Thanks for the info. So far I've been taking it off of the anode but thought that there might be some deeper reasoning to doing it one way or the other. I'll have to try pulling the signal off of the cathode too.

Doug, it's funny that you mention the capacitor running the GM tube for a while with no power. I'm using a CCFL inverter for the HV running on a 9V battery and everything else is running on a 3V Li cell so the first few times that I unplugged the HVPS it would surprise me when the thing would keep clicking away - not for hours like yours, but for several minutes at least.

[Doug Coulter]

I used a volt doubler and lower volt input to the CCFL as they are more efficient that way. Quiescent current goes up with volts, usually about 40 ma at 12v, down to about 10 ma at 3v.
Don't forget you need a high R in series no matter which side you take output off. That's for tube turnoff protection, and one meg is none too much, all the old circuits use 5 meg and up, some even bootstrapping that with a vacuum tube to absolutely ensure the tube won't stay lit like a neon bulb. Remember the extra capacity to ground from the larger outer electrode can also store enough energy to hurt things. No one does it that way anymore, and for good reasons. Of course, you use shielded cable for the signal side, as you can pick up lots of noise across a meg.

Some of the really old, non self quenching tubes used a design like in John Strong's book with a vacuum tube bootstrap to force the geiger tube off, those did use the cathode as part of the active circuit. You can't get one of those these days unless you make it (not hard, BTW).


If you use a volt divider -- make sure the parallel R is still over a meg (for a modern tube). Really, it's just better to mess with that on the HV supply input instead, and keep the HV-side stuff as simple and minimal as you can. More stuff at volts means more electro-statically attracted dust and other issues. Do use a large size (like 1/2 watt) so no way it can arc across the current limit resistor.

I got all those hours because those blue block caps have near to zero leakage, ditto the diodes and coupling cap. The tube was just counting my background (maybe one/second) for that, so it wasn't drawing much either.

[Joe Jarski]

I had to go back and give those caps a try, if it's the ones I'm thinking of and wow, what a lesson in capacitor leakage. It's almost comical how long that thing will run without being plugged in. This whole project has been a good lesson - I'm just playing around with this stuff on a breadboard right now and all of those little antennas sure like to pick up the noise from the CCFL inverter. Shaping the pulse has been an interesting experience too. I'll be sure to make use of the advice when I get around to putting this into a real package.

[Doug Coulter]

Yes, the CCFL's make a lotta noise at a good frequency to get into things. I put the diodes right on the thing, output cap too, and it's not "on the board" with the rest of stuff.
This is also why the "generic detector supply" gets its own box to live in around here. The noise is less the lower the AC volts, so that's also why I'll use a doubler (or more) to run the thing at lower input voltage (which also saves power in this type of application - where there's basically no load). If you wind up with one in the same box with sensitive things, you wind up making a box inside a box around the CCFL to keep the noise inside, and only DC in and out of the sub-box. Lucky a geiger tube makes a big fat signal, less worries there. The thing I'm currently working on shows a very good SNR -- almost no pickup (short signal leads) and a very large signal -- over 100V pulses. If I was doing a phototube or a proportional tube it'd be a different story and much more care needed. For example, in that case I'd also have a filter cap on the CCFL input DC, right at the board, so we wouldn't be drawing pulsed current through long wires to wherever the power for it comes from -- that would make a one turn primary for a noise coupling transformer with everything else in your circuit being little secondary windings.

Standard counter kludge parts

The PCB in this picture is a dev board I did awhile back for working with 28 pin PICs. Kinda handy to start right off with RS 232, power supply, and in-circuit programming all working.

[Starfire]

You can use a 555 as a crude follower and get a good o/p pulse - the i/p signal goes to the trigger i/p on the 555

[Doug Coulter]

Indeed you can, especially if you want them "all the same". I eschew even that -- I just use a big series R (100k or more) right into a CMOS counter input, works great. The reason for the series R is to keep from drawing so much current through the tube via the coupling cap, which makes ~100 volt pulses. No need for a preamp when driving CMOS logic at all. Timer 0 on a PIC is the counter of choice here.

I set it up for 16 bits, have it increment another 16 bit number on overflow via an interrupt, so as to keep a lifetime count (stored in eeprom at power down). Basically, you can just read the combined 32 bit number every second, or every 10 seconds (I find this best) to report out to a computer via RS 232, or if you like analog meters as I do, use the PIC pwm output to drive one of those -- you can use a fixed series resistor to the meter and do any scaling in software. Since even a 555 draws more power than a PIC under many conditions, and almost costs as much, well....

Remember, the actual hardware counter in a PIC is full speed (if programmed not-synchronized) even if the pic itself is in sleep mode or running on a slow clock at the time. Easily much faster than needed to handle the max count rate a geiger tube can produce. Not so true for much faster phototubes, but for geigers, not a problem at all.

Weren't you playing with one of the small PIC processors awhile back? I really like them for jobs like this -- it's a good fit.