Designing a converter to drive Geiger-Muller tubes

[knapik]

I'm not extremely knowledgeable, but thought it would be a fun hobby project to try and make some sort of Geiger counter. I would like to have it run on batteries, so I could carry it around. I have been thinking about it and researching it a little, however I'm quite unsure of how I would convert say a 3-5V source from a battery into 400V for the tubes. What would be a good way to achieve this and is there anything that would help for me to research?

[Buriedcode]

Common converters are flyback (basically a boost converter, but with the transformer/coupled inductor turns ratio adding a multiplier for higher out/in difference), and 'royer oscillator' which is a self oscillating converter, which has an almost sinusoidal output (less radiated noise, easy to filter).  Both are isolated converters, although there's not much point in isolation as the tubes detector circuit is hooked to it.

Since the current required by tubes is quite small, common implementations tend to use either the above or a blocking oscillator to get to ~80-120V followed by a cockroft-walton multiplier - just diodes and caps - to get to 300-400V at a few mA.

For battery operation, you'll of course want as high efficiency as possible.  Both flyback and royer converters are pretty efficient.

I have seen DIY implementations using standard boost converters, just off-the-shelf inductors rather than a transformers, but I doubt these will be particularly efficient.  5V CCFL supplies tend to be royer converters, and can easily kick out 400V+.  They also tend to be quite efficient, and if driven by a buck converter, the output can be pseudo-regulated (another google term - current fed royer)

Forgot to add one more idea - disposable camera flash circuits are miniature flyback converters meant to run off of one or two AA's.  These charge a large cap to 330V then turn off - you could use that for your power supply, but the cap can be lethal.  They aren't designed to run continuously but replacing the cap with a much smaller 400v one, and letting it run might work.

I was going to spam you with links, but I'm sure you can google.  Here's one that seems straightforward:
http://4.bp.blogspot.com/-r3W3v1cIgtY/TbiLELsyuLI/AAAAAAAAAEI/VI-Xbo4Wsz8/s1600/500V+HV+inverter+for+Geiger+Counter.PNG

[DaJMasta]

+1 to what's been mentioned, it seems like relatively simple topologies designed for very low currents are going to be most efficient, just make sure that you have a tube in mind before you design because their requirements can vary quite a bit - while many will work at 300-400V, some will require 600V or even 1kV depending on the types of particles they can sense, the technology they're made with, and their sensitivity.  Then just make sure the final design has a way to keep fingers away from high voltage and has parts and traces spaced far enough apart to avoid risk of arcing.

[knapik]

Well, I've been spending a lot of time researching what has been mentioned so far, and its surprising how little I know when the time comes to wanting to actually design something.

Please feel free to correct me, I don't want to learn the wrong things, but from what I understand, in using a flyback converter, a mosfet being used as a switch to create a square wave which can be multiplied through a transformer and then is filtered on the output to create DC. From what has been said, it is also possible to use it like a boost converter which lowers the multiplication needs of the transformer. For the switching waveform, would it be alright to use something simple like a 555 timer, or should I look into using a specialised PWM controller?

With the royer oscillator, its converting the DC input to an AC sinusoid which is then multiplied through a transformer. That would then have to be filtered by a full bridge rectifier with a capacitor, right? I've been wondering about the royer oscillator, as its a DC to AC to DC conversion, would it be less efficient than the flyback converter? Also, from what I see, the royer oscillator requires a multiple input transformer, while the flyback requires a single input, it might be preferable to use a flyback not only to simplify finding a transformer, but also as a learning experience.

I've been learning and playing around with LTspice and its made me realise that I don't really know how to design a flyback converter at all. Are there any resources that can help in specifically designing one?

In regards to the tubes, I have been thinking of just getting one of the sbm-20 tubes off of ebay. I'm not too worried about alpha particles, and it might be a good idea to keep it fully enclosed so that I don't have to worry about getting shocked. I'm not too fussed about accuracy either.

[ealex]

take a look at this : http://cds.linear.com/docs/en/application-note/AN118fb.pdf

for a quick experiment i used a transformer recovered from a newer CCFL supply with an added feedback winding - it worked after i got the phasing right.

this one also works - i've used it with a SBM-20 tube : http://www.electroschematics.com/wp-content/uploads/2015/09/Figure-8-Front-end-circuit-of-a-diy-Geiger-Counter-kit-.png
you can replace the active components with something else.

another thing that you need to remember is to add a series resistor, otherwise the tube will not extinguish after the initial discharge.

also, when measuring converter voltage after the limiting resistor you have to take into account the input resistance of your meter:
looking at the 555 timer based schematic and assuming you are measuring at the tube-side of R2 with a meter that has 10Meg internal resistance on the 1000V scale you form a voltage divider where the actual voltage will be ((4.7Meg + 10Meg ) / 10Meg) * measured voltage.
hopefully i got the math right

[Gyro]

another thing that you need to remember is to add a series resistor, otherwise the tube will not extinguish after the initial discharge.

That series resistor is absolutely critical. It is vital that the tube is fed from as high an impedance as possible - that means minimising parasitic capacitace of the resistor and tube connections too. This has a huge effect on the life of the tube - ideally it should only be discharging its own internal capacitance, while you detect the current pulse through the cathode connection as it recharges.

[schmitt trigger]

If you are willing to build a kit instead of designing it yourself, electronic goldmine has many interesting Geiger detector kits.

http://www.goldmine-elec-products.com/products.asp?dept=1468

[knapik]

I've been looking through the other forum topics and trying to search for flyback transformers and its all just so confusing, I think I'll try simulating that mightyohm power supply a bit later on to see how it works. I also want to keep the project mainly as a learning experience, so I'm not too interested in kits.

With the series resistor, I see many sources using a 4.7M resistor, would it be preferable to increase the value to use as "high an impedance possible", or is it fine?

[David Hess]

Since the required output current is so low, use a switching boost converter to drive a diode-capacitor charge pump.  Many controllers are available which will run on 1 or 2 AA cells.

[dexters_lab]

I'm not extremely knowledgeable, but thought it would be a fun hobby project to try and make some sort of Geiger counter. I would like to have it run on batteries, so I could carry it around. I have been thinking about it and researching it a little, however I'm quite unsure of how I would convert say a 3-5V source from a battery into 400V for the tubes. What would be a good way to achieve this and is there anything that would help for me to research?

i was going to suggest looking through the mighty ohm counter schematics for some ideas but it looks like your well on your way now anyway...

i started with the mighty ohm kit and now i have 5 counters, it can be addictive!, two of them run SBM-20 tubes, another runs a massive Mullard tube and the other has a 3" scintillator+PMT probe and a small mica windowed tube, so you might want to design it with a wide operating voltage, say 250-1200v so you can use both GM tubes and photomultipliers for when you want to experiment with other types of detector

[Gyro]

With the series resistor, I see many sources using a 4.7M resistor, would it be preferable to increase the value to use as "high an impedance possible", or is it fine?

The value of the series resistor has three effects:
 
- It prevents the tube from going into uncontrolled avalanche (needs to be high enough in value to allow the tube voltage to collapse sufficiently)
- It determines how quickly the tube can recharge after a count (one element of the 'dead time' - the tube has its own dead time too) and so, the maximum count rate.
- It affects the size of the current pulse that your detection circuit in the cathode circuit has to work with (the higher the value, the lower and slower the recharge current waveform).

Therefore it's a compromise. It depends quite a bit on the tube, how big it is, its internal capacitance etc. The one I uses is a thin end-window type that runs at 1400-1600V, I use a 30M resistor for that. The tube datasheet may give some clues (eg. capacitance, dead time) or may actually recommend a value. If not, then around 4M7 sounds about right for a small 400V tube.

As I said, circuit capacitance is a key factor, you don't want the tube having to discharge your parasitic circuit capacitance. Making the [Edit: resistor capacitor] out of two 2.4M resistors in series is something that will help (and also help with resistor voltage rating). No need to get paranoid about it but with the cost of GM tubes, it's something that is worth the effort to get right. Be prepared to experiment a little though.

One thing on the power supply, you don't need to overdo it size-wise, at 400V even a short circuit through a 4.7M resistor will draw less than 100uA, in practice it will be an order of magnitude lower than that.

P.S. Yes, make to output voltage adjustable (you need to be able to set it to the middle of the plateau voltage anyway).

[Buriedcode]

Please feel free to correct me, I don't want to learn the wrong things, but from what I understand, in using a flyback converter, a mosfet being used as a switch to create a square wave which can be multiplied through a transformer and then is filtered on the output to create DC. .....

....I've been learning and playing around with LTspice and its made me realise that I don't really know how to design a flyback converter at all. Are there any resources that can help in specifically designing one?

http://www.dos4ever.com/flyback/flyback.html  - I often drop this link in when people mention flybacks, boost converters, or high voltage.  It covers it well.  I just deleted a very lengthy explanation.

You can think of a flyback simply as a boost converter, with an extra winding on the inductor.  The key difference between this and simply considering it a 'transformer' is, in a flyback, the "transformer" stores energy, and current doesn't flow in the primary and the secondary (or secondaries if there's more than one) at the same time.  This is why technically, its a coupled inductor.  A true transformer shouldn't store energy, and simply transfers it from primary to secondary, at the same time, via its magnetic current.   

So, say you had a boost converter that generated 30v from 5v.  Now, if you added another winding on that inductor, with say, twice as many turns as the primary, and looked at the waveform on a scope you would see its voltage jump to 60V during the primary off-time (the primary will jump to 30).  So, even though the ratio is only 1:2, you get 60V from 5V.  The reason qutie a few ~300V flyback supplies have a 1:10 ratio is that it keeps the maximu primary voltage at 300 / 10 = 30V.  This is the largest voltage the switch will have to cope with when it is off, and allows one to use very low on resistance MOSFETs (with a Vds of 40 max).

From what has been said, it is also possible to use it like a boost converter which lowers the multiplication needs of the transformer. For the switching waveform, would it be alright to use something simple like a 555 timer, or should I look into using a specialised PWM controller?

555 timers can be used, but... they were designed to simply create a delay or generate a square wave.  Proper converters generally sense current in the primary, for current mode converters, and have many other features for the task (voltage reference for feedback, dead time etc..).  Using the 555 for a boost converter can work, you just have to set the maximum on time based on the inductance and the peak current, and set the off-time to something sensible. Whilst its rather cool to use a cheap 555 - and the fact they are everywhere - these days, if you can, I would use a boost converter chip, still not easy, but less hassle trying to tweak with a 555.
 

With the royer oscillator, its converting the DC input to an AC sinusoid which is then multiplied through a transformer. That would then have to be filtered by a full bridge rectifier with a capacitor, right? I've been wondering about the royer oscillator, as its a DC to AC to DC conversion, would it be less efficient than the flyback converter? Also, from what I see, the royer oscillator requires a multiple input transformer, while the flyback requires a single input, it might be preferable to use a flyback not only to simplify finding a transformer, but also as a learning experience.

Again, more complicated. Sorry to resort to wiki... but its explanation isn't half bad. https://en.wikipedia.org/wiki/Royer_oscillator  It looks deceptively simple.

Here it really is a transformer, with two primaries.. or rather a single, centre tapped primary, and a feedback winding that drives the switching transistors.  The positive feedback from this winding is what causes it to oscillate.  It generally outputs a square wave (a high for one primary switching, and a low for the other) but the addition of a cap can create a form of resonance that greatly reduces the harmonics of the switching waveform.  Its quite involved!  In practical terms - its all about the transformer, the number of windings, the core material etc.. making it quite difficult to design from scratch. (its not just about turns ratio).

[Lee Leduc]

Maxim has an App Note for a geiger power supply. I've used this and it works well. It's appnote AN3757. See attached file.

[KMoffett]

Here's a simple circuit for an 800VDC unregulated boost converted. Just drop the last two stages of the multiplier for 400v. And yes, I have tested it.

Ken

[Someone]

So, say you had a boost converter that generated 30v from 5v.  Now, if you added another winding on that inductor, with say, twice as many turns as the primary, and looked at the waveform on a scope you would see its voltage jump to 60V during the primary off-time (the primary will jump to 30).  So, even though the ratio is only 1:2, you get 60V from 5V.  The reason qutie a few ~300V flyback supplies have a 1:10 ratio is that it keeps the maximu primary voltage at 300 / 10 = 30V.  This is the largest voltage the switch will have to cope with when it is off, and allows one to use very low on resistance MOSFETs (with a Vds of 40 max).

From what has been said, it is also possible to use it like a boost converter which lowers the multiplication needs of the transformer. For the switching waveform, would it be alright to use something simple like a 555 timer, or should I look into using a specialised PWM controller?

555 timers can be used, but... they were designed to simply create a delay or generate a square wave.  Proper converters generally sense current in the primary, for current mode converters, and have many other features for the task (voltage reference for feedback, dead time etc..).  Using the 555 for a boost converter can work, you just have to set the maximum on time based on the inductance and the peak current, and set the off-time to something sensible. Whilst its rather cool to use a cheap 555 - and the fact they are everywhere - these days, if you can, I would use a boost converter chip, still not easy, but less hassle trying to tweak with a 555.

Its not complicated to use the 555 as a regulator (see below) but a dedicated chip is usually easier. I'll disagree about the desire to use low voltage and low resistance MOSFETs, there are many commodity MOSFETs with both 400-600V ratings and avalanche specified that have adequately low resistance for these low power converters.

Here's a simple circuit for an 800VDC unregulated boost converted. Just drop the last two stages of the multiplier for 400v. And yes, I have tested it.

Can we please get away from the unregulated 555 supplies at these low frequencies. Its not hard to go with a regulated and compact design with modern parts: https://www.eevblog.com/forum/projects/555-timer-boost-converter-(and-buck-converter)-switching-power-regulator/

[Gyro]

The supply doesn't need to be 'regulated' as such. The current demands of the gm tube are so low that it will be very difficult to prevent the feedback resistors from loading the output way more than the tube. What it does need to be is stable and adjustable at nearly zero load, probably by regulating the input supply voltage.

[vindoline]

I've got a couple of suggestions and a "trap for young players..."

1) if your main interest is in getting a working Geiger counter so that you can look at background radiation, hunt for radioactive Fiestaware, experiment with low level sources, etc. I would recommend starting with a kit. This is a great one, and the site has tons of info: https://sites.google.com/site/diygeigercounter/

2) If your main interest is learning about the types of high voltage supplies that can be used for Geiger counters, there have been some good suggestions and starting points in this thread already, but I would recommend checking out Charles Wenzel's site: http://www.techlib.com/science/geiger.html

Several of the circuits there have been optimized for battery life and they work great.

Finally, these HV supplies are generally very safe, because they are very high impedance. They are only capable of sourcing several microAmps (not milliamperes!). That means you can poke about with your fingers and not get hurt. It also means that you can't use your multimeter to measure the real voltage level that is being generated. The standard 10 Meg input impedance of a meter will load the circuit down too much.

Have fun!

[MagicSmoker]

So, say you had a boost converter that generated 30v from 5v.  Now, if you added another winding on that inductor, with say, twice as many turns as the primary, and looked at the waveform on a scope you would see its voltage jump to 60V during the primary off-time (the primary will jump to 30).  So, even though the ratio is only 1:2, you get 60V from 5V.  The reason qutie a few ~300V flyback supplies have a 1:10 ratio is that it keeps the maximu primary voltage at 300 / 10 = 30V.  This is the largest voltage the switch will have to cope with when it is off, and allows one to use very low on resistance MOSFETs (with a Vds of 40 max).

The voltage the primary switch must withstand during the offtime is the reflected output voltage (Vout / n) plus the input voltage.*

Also note that the maximum voltage the secondary diode must withstand is ((Vin * n) + Vout); that is, the reflected input voltage plus Vout.



* - plus, of course, any leakage spikes (though this, along with ringing, should be addressed with RCD clamps and/or RC dampers).

[ChristofferB]

Not exactly "home-made", but I've had reasonable success driving a STS-5 (sorta sbm20 equivalent) tube with an electric fly-swatter  tapping off the multiplier at as close to 400V as possible, then add 400V worth of zener diodes and a high-ohm series resistor.

It does work, but I don't think it sits too well with the swatter circuit to be run over longer periods of time..

[knapik]

Wow, there's a lot that has been mentioned lately, I'm going to need some time to properly read through it all. I borrowed a few books from my library dealing with switch mode power supplies and spice simulations of them which I was having a lot of trouble with. I think I'm going to take the time to read over everything carefully and try to simulate some of them to see what works before buying actual components.

[jbb]

This probably is a good time to have a go with SPICE.
Some tips:

• Don't try to simulate the whole circuit in one go. Just simulate the basic blocks.
• Start off with a pure voltage source (no resistance or inductance) for the input, and perhaps a second one for the output. That way you can isolate the fundamental switching behaviour (remember, divide and conquer).
• leakage inductance matters a lot. Instead of a coupling factor k = 1 (perfect), try k = 0.99 (more realistic).[\li]
  
  • instead of putting in a control chip, try a pulse voltage source (I.e. fixed duty cycle control). It's very rough, but can get you started.[\li]