User Tools

Site Tools

PiGI Hardware

Schematic V1.1

Layout V1.1

Prototype PCB's V1.0

Download Layout & Schematics (Eagle 6.2.x):

Circuit Details

The Pi-GI circuit is divided into two parts:

  • Kickback High-Voltage Switching Power supply
  • Impulse Inverter

Inductive Kickback Switch Mode Power Supply

The basic idea was to build a cheap but efficient Kickback High-Voltage Generator to convert 5V up to 450V - 600V in order to reliably feed almost any available Geiger-Mueller tube for maximum flexibility & hackability. This circuit design was inspired by Tom Napier's article in Nuts & Volts, Jan 2004 (rebuplished in issue #184 of Circuit Cellar Nov 2005) which has been used successfully in multiple DIY Geiger-Counter projects worldwide for example:

Since the original circuit incorporated some expensive and hard to get Through-Hole parts (like the STX13005 & UF4007), it was redesigned with modern and available SMT components. This also led to an increased tolerance of the high voltage components up to 1kV instead of 600V. The great thing about this design is the really low BOM count and the fact that it regulates the HV really well WITHOUT the need to have feedback from the HV rail.

How does it convert 5V up to 500V and more?

The transistor (T1) is turned on and current flows into the inductor (L1). When the transistor is turned off, the input current that formed and maintained the inductor's core magnetic field becomes zero. The magnetic field collapses causing a voltage reversal to occur in the inductor and induces sufficiently high voltage (known as inductive kickback voltage) into the diode (D1). If you want to learn more about it and use a java-enabled browser you can use this inductive kickback simulator. Toggle the switch at the bottom (simulating T1) and watch the voltage graphs to see the effect.

For the HV generator only two components seemed to be critical:

Component Function & Parameters Selected Part
D1 Fast recovery diode (75ns) for up to 1kV BYG23M
T1 NPN transistor with a >=1kV collector-emitter breakdown voltage STN0214

The inductor is a high Q wound dust core choke with shielding to minimize EMI. The output voltage is set by the maximum current, which is controlled by the adjustable trimmer (R10) in the emitter lead of the STN0214. A lower resistor value (turning CCW) will result in higher output voltage. Alternatively, when only a certain type of GMT is going to be used with a PiGI module, R10 can be replaced with R10a to set the specific required voltage for the GMT. Please add working values for R10(a) with your particular tube to the Common Table of Geiger-Müller tubes.

Based on the CMOS 555 The 555 oscillates at about 3.2 kHz Pulse on time is controlled by the inductor and emitter resistor (which sets the maximum current), which in turn sets the high voltage value.

Impulse Inverter

Each impulse will pull the selected (J1/J2) GPIO pin to ground via T3 so that it's possible to generate an interrupt which can be counted. This ensures the safety of the PI and also helps to prevent getting false events generated by external EMI.

Prototype & projected production costs

Volume PCB Parts Soldering method Risk margin Final product
7 V1.0 Prototypes EUR 7,23 EUR 8,67 Hand (EUR 0) EUR 15,90
< 100 V1.1 EUR 4,97 EUR 9,70 Hand (EUR 0) 20% TBD
>= 100 V1.1 EUR 2,79 EUR 7,05 Outsourcing (TBD) 15% TBD
>= 1000 V1.1 EUR 0,70 EUR 3,95 Outsourcing (TBD) 10% TBD

As we can see, it scales very well with numbers and a non-profit oriented production run could bring many modules to people for less than 20 EUR. If it were sold for more than 25 EUR (fully assembled), someone would really be ripping people off.

Assembly Instructions


Check the Mouser project page for specific lists and datasheets. You can also directly order all parts with one click from there:


# V1.0 V1.1 Part/Value Package
1 D2 MMSD4148
2 C4 1nF 0805
3 R5 330 ohms 0805
4 R1 R4 100k 0805
5 R6 220k 0805
6 R4 R1 1k 0805
7 R9 27k 0805
8 IC1 TLC555QDRQ1 - Case mark facing towards R6 SOIC-8
9 T2 MMBT4401 SOT-23

If you're not stacking two modules use either a 0-ohm resistor/solder-bridge on J1 or a capacitor to connect the GPIO Pin of the PI to the inverter's collector. Don't use a capacitor when connecting to a uC with internal pull-up (like a Raspberry Pi).

When you want to stack two modules, connect J1 on the lower and J2 on the upper module.


# V1.0 V1.1 Part/Value Package
1 C5 330pF 0805
2 R7 1.5k 0805
3 R8 100k (default - see signals for other values) 0805
4 T3 MMBT4401 SOT-23
5 C1 220uF/10V LowESR Aluminium Electrolytic Capacitor see BRD
6 C2 Vishay MLCC 10nF/1kV 1206
7 C3 Vishay MLCC 10nF/1kV 1206
8 R2 KOA Thick Film 1M 0.25W 0.5% 1206
9 R3 KOA Thick Film 4.7M 0.25W 1% 1206
10 D1 Vishay BYGM23 Fast Recovery Diode 1.5A/1000V/75ns DO-214AC (SMA)
11 T1 STN0214 Bipolar NPN 1k2V SOT-223
12 L1 Murata Shielded Inductor 15mH see BRD
13 R10 Bourns Trimmer 100R see BRD
* R10a optional, with a fixed value instead of R10 0805
14 H1 Stacking Raspberry Pi Header 2×13 2.54mm

Test Points

High Voltage

The kickback power supply is very efficient (in order to draw less power) and the whole setup is tuned very much to accommodate a geiger-mueller tube as a power consumer. Introducing another consumer (a multimeter with 10 Megaohm for example) draws too much current from the circuit.

The supply drops down and we can only see significantly lower voltage values. To mitigate the effect a high value resistor (1 Gigaohm) was used to obtain a somewhat more reliable measurement. The perils, pitfalls and possible solutions to measuring the high-voltage (with some accuracy) and alternative methods to adjust the high-voltage without measurement are described here.


This test was run on a V1.0 prototype board, plugged into a Pi with running counterd. A radioactive test source (thorium) was placed in front of a FHZ-76 tube. The high-voltage was increased (turning R10 counter-clockwise) until the count-rate reported by counterd didn't increase anymore. The following table shows the resulting measurement of the set voltage using different setups:

Resistor Setup Read Voltage Calculated Voltage
10M (Fluke-87V internal resistor) 230V 230V
10M + 42.3M (9x 4.7MOhm in series) 63.7V 333V
10M + 1G (20kV 1 GigaOhm single resistor) 4.56V 460V

Measured at: [+] Cathode of D1 (pin facing towards R2) [-] Cathode of GM Tube (HV- Pad)



V_actual = V_read * {{ R_meter + R_probe } / R_meter}


63.7V * {{10 MOhm + 42.3 MOhm} / {10 MOhm}} \approx 333V

4.56V * {{10 MOhm + 1000 MOhm} / {10 MOhm}} \approx 460V


PiGI in long-term test harness on the Bench (all the wires are just for testing:)

555 output: Via next to T1


It seems, that in practice (industrial/military), cathode connection is always supposed to be used with GM tubes if possible, mainly because it's less likely to affect the characteristics of the tube and reduces undesirable effects. The extra capacitance of the output circuitry added to the cathode is considerably less significant than when added to the relatively small anode. This was taken into consideration while designing the PiGI.

For optimal tube operation R3 (anode resistor) and the equivalent of R7/R8 (cathode resistor) should have a 45:1 Ratio, at least that seems to have become “industry standard”. A short test with the LND712 GMT showed that indeed the signal flanks became a bit more precise when R8 is 220k instead of 100k in order to match the 10M for LND 712. However, in many other tests the circuit has shown a very high resilience against “suboptimal” operating conditions.

Worst Scenario (Undervoltage & very high anode/cathode resistor ratio mismatch)

Recommended operating voltage & 45:1 anode/cathode resistor ratio

GPIO output

This is how the final impulse signal looks like to the counting IO Pin. It also leaves a question: For the LND 712 the dead-time is supposed to be 90us, but when you count the divs in the scopeshot above it is about 140us. Shouldn't in this case 140us be used as a base value for the dead-time algorithm?

DSO Trace of the falling edge on the GPIO Line

Radioactive Test Sources

Since everything is over-regulated these days, up to the point where only multi-national-corporations can obtain the resources needed to develop something easily, it's quite a challenge to find radioactive test sources to probe a Geiger Counter with more than just the local dose rate. So instead of having simple access to quality test-radiators which can be handled and stored in a safe manner, we have to improvise and hack something out of whatever we can find. When we consider that the regulation's original intent has to protect of people from harm, it backfired pretty well.

The mightyohm blog has compiled a list of things you can try to use:


2013/08/04 00:07

“Since the original circuit incorporated some expensive and hard to get Through-Hole parts (like the STX13005 & UF4007), it was redesigned with modern and available SMT components. This also led to an increased tolerance of the high voltage components up to 1kV instead of 600V.”

For the sake of accuracy the STX13005 / UF4007 provide a maximum HV of ~970V in my kit, not 600V. That's not to say the STN0214 isn't an improvement. Using that I get ~1125V max. It is also much easier to source. John

2013/08/04 06:38

Well, for the sake of accuracy, I was talking about the tolerance of the components specified by the manufacturer and not the resulting output voltage :) So I gather you've already tested the STN0214 in your setup?

2013/08/04 20:14

Well there's accuracy and reality ;-). Since the duty cycle is so short the HV transistor can go well above it's ratings. A moot point though.

Yes, I did test it with the DIY Geiger Kit, and wrote up how to add the STN0214 to existing kits at the bottom of this page:

So thanks for finding it, and good luck with your project. John

2015/04/13 12:20

It would have been nice if you would have added a link to our site instead of just doing a grab and run, to give people the whole picture where you got that information. After all you weren't stingy to throw links to your site at every opportunity here (aside from the links provided in the wiki already), no?

2015/04/13 11:50


In the picture under the Signals section I see a resistor right on the anode. Is that an anode resistor I have to add (depending on the tube type) or was that some temporary fix?

2015/04/13 11:54

You mean the red one connecting the anode of the tube with the PiGI board? Indeed that is an additional anode resistor. R3 is the on-board resistor equipped with 4.7M. Some tubes need more and since some solder connection was needed to get the anode of the tube connected to the pad of the board, I used another 4.7M resistor to come closer to 10M and have my tube connected at the same time. If you need more anode resistance, you could do it just like that or replace R3 with a higher value and use a plain wire to connect the tube. Keep in mind though, supposedly, the anode resistor should be as close to the anode as possible (so I've heard).

2015/04/13 19:32

I see. The 4.7M should be fine for my STS-5 tube, although adding another few hundred k couldn't hurt either.

I have a question though. I have an assembled board now, but with a PZTA44 transistor (400V) rather than the STN0214. That could be OK at low voltages (or…?), but the problem is when I adjust the trim pot, the voltage goes from 0 to 420V (measured with a 10M multimeter between HV- and HV+, so it's probably 800V+ in reality) in one step; there are no intermediate voltages. Knowing most GM tubes' operating voltage is lower, I guess this is not normal. Any ideas?

2015/04/13 19:51

Unfortunately, I currently know nothing about this particular transistor but it sounds weird that the poti is acting like a switch ;) According to the notes above, the voltage measurements points were:

Measured at: [+] Cathode of D1 (pin facing towards R2) and [-] Cathode of GM Tube (HV- Pad) and NOT the HV+ pad after R2 & R3. I don't really know exactly anymore, I just can hope that the notes are correct, partly the reason to take these notes in the first place :)

Maybe you can try to re-measure with these points to get a result closer to the original reference measurements above. In the first tests the voltage was set way too high and the voltage was actually at 5V. After turning the poti down the whole way and coming back up slowly showed the voltage build-up. If you have good ears you can hear a feep sound (with the STN0214, others may vary) which increases with the voltage level. As a basic rule, when we could hear it we weren't far off. Having a test radiator in front of the tube while bringing the voltage up and using something to count the ticks also helps finding the best plateau voltage of the tube.

2015/04/14 12:19

Yes, but why measure there? This is something I don't understand. What matters ultimately is the voltage the GM tube sees. That will decide if the tube will work at all or you will fry it.

2016/03/01 14:56

Hi, where did you order the PCBs for 7,23€ per board? I only find services with prices about 12€ per board (10 boards order). Or do you have a PCB for me? If not I will order 10x PCB and 10x parts from mouser (~12€ per package).

2016/03/01 18:41

IIRC, the original prototype boards were ordered on It's already been some time ago, so prices may have changed. Have you tried to get a wider range of quotes?

Unfortunately there are no more spare PCBs from the first batch left, otherwise I would have send you one. There is still plenty of traffic by people looking for PiGIs so if you do a new batch, I'm pretty sure there will be more people who would like to chip in and share the costs, to get one as well.

Nice astronomy page you got there, love the C/2014 Q2 Lovejoy pic, especially considering that it was done in a light polluted area.

2016/03/01 20:16

Thanks! Unfortunally is only for corporate clients. gives me some cheap manufactures in China but I didn't create or order any PCBs yet. So I have no idea what's the important options. I opened the PiGI Layout in eagle and clicked on the PCB service button. There I got an pre filled form on that looks ok. An order of 20 pieces would cost about 135€ (~6,75€ per board) and would be affordable for me. I can resell them for the same price (6,75€ + shipping cost from Germany). If more people are interested I can also order a bigger contingent.

Thanks! C/2014 Q2 Lovejoy was just a “fast shot” before the clouds arrived :) I was surprised that I could see the comet with my naked eyes from Dresden.

2016/03/01 20:33

Pah, don't let yourself be discouraged by that :) Hackers in a hackerspace/hackbase are in no way corporate clients and yet there is a lot of development and low/medium scale fabbing going on. As long as they get their money, most of them don't even ask. I guess in most cases it's just to avoid some stupid law forcing them to have to comply to certain ways of pricing and taxing services and products, if they officially cater to “enduser-consumers”.

Chinese suppliers are also a great option, it has become incredibly easy these days to order stuff directly in Asia and (with the usual few exceptions) the quality is also quite good. You can look in electronic boards or search for reviews and pictures of PCBs produced by a certain manufacturer that comes up in your quoting evaluation to get a better feeling for their products.

(OT) Btw, maybe this is something for you too :

2016/03/21 21:11

So I got a package of 20 PCBs for ~4,70€ each at eurocircuits :) I soldered the first PCB but C5 is missing in the mouser list. It seems to work without C5, an Arduino Nano and an FHZ76V but I have no real radioactive material here to test the rate. I will buy the C5 capacitor at Conrad the next days.

If anyone wants to buy a PCB just send me an email → info[at]

2016/03/21 22:38

Awesome, thanks for sharing and the time and effort to do it. As for C5, it's no absolutely vital component, I guess the circuit should work just as well - but I've never simulated or tested the board without it. Is it not in the BOM at all anymore or were none available at mouser? Because the BOM used to be complete at the time… hmm, need to check.

2016/03/22 13:44

Hey you made the PCB layout, I just ordered it :) C5 isn't on the BOM. I ordered this capacitor at Conrad (I don't have to pay shipping costs because we have a store in our city) I hope it's the right one?!

2016/12/19 19:30

Hallo, Have got some basic questions: Are there one PCB left? Where do I buy the actual GM Count Tube? How much is it? How accurate is everything? Is the bill list still on the latest revision?

I saw this project in german tv i think :)

2016/12/19 20:33

I got some more requests in the last days. At the moment there are 5 boards left but 2-3 of them are reserved. So I can offer you a board for 4,70€ (PCB) + 12€ (parts from mouser list, already soldered) + shipping (0,90€-5€ from Germany). At the moment I'm waiting for parts for the last 5 boards because some parts weren't available. Just send me a mail to info[at] I attached a SBM-20 tube to my PiGI. It's about 15-20€ on eBay: I tested with a gas mantle (thorium nitrate) and some uranium glass.

2016/12/19 21:40

@Fehlfarbe: If it's getting too much and you'd like to get your email-address removed from the comments and replaced by a new mailing-list just give me a ping.

@Dani: Like fehlfarbe said, you can buy tubes at many places online, new and used ones, there is a big surplus of some older cold-war era SBM-20 available, even on ebay kleinanzeigen:

I don't have the time right now to do the search for you, but I'd recommend to spend a little time figuring out what you want to achieve in terms of detection, i.e. gamma only, beta + gamma or alpha + beta + gamma. Once you know what you want you can start to search for specific tubes (and their common names/ids) to get good search results and offers from Germany/EU. Prices vary very much depending on those parameters, can go from 10 EUR like the SBM-20 above or up to 100 EUR when you want a brand new abg capable LND712.

Again, this came up before, accuracy is tricky in terms of ionizing radiation metrics without very expensive gamma spectroscopy gear, which can not only tell us how much radiation there is but also what kind of source (isotope) it is.

A GM tube can only accurately count ionizing events triggered by radiation passing the tube. By using masks/windows we can also determine if we're dealing with alpha, beta or gamma radiation (depending on the tube's capabilities). These are our base metrics.

From here on, everything is done via math and magic numbers for each tube, to end up with so called equivalence values like uSv/h. This equivalent dose rate represents a certain radiator like cobalt-60 or Cs-137 and is based on the surface/volume ratio of the human body model, to give us more meaningful data.

Having said that, it's not important to be accurate. As long as the detection behavior doesn't change over time (which it didn't so far) we have a very good instrument to record, share and compare our collected baseline data while no event has happened. Whatever the absolute numbers are in terms of uSv/h, we'll most definitely see an increase in case of an event and the counts per minute values will always be a solid, reliable and comparable base-metric.

Do you happen to remember when/where you saw it on TV?

2016/12/20 08:57

Thank you for your very detailed anwser. It was in zdf mittagsmagazin from yesterday:

2016/12/21 06:45

Ah, funny. Thanks for the update.

2016/05/28 13:51

”[…] it's quite a challenge to find radioactive test sources to probe a Geiger Counter with more than just the local dose rate. […]”

Here in Germany you can buy >130000 CPM (that's >800µSv/h if that'd be calibrated for Cs-137) (measured on a SBM-20) pitchblende for 50€, 1000 µSv/h for 60€-65€ and 1200-1500 µSv/h for 70-80€. Perfectly legal to buy. There are known Ebay sellers, and if you want these rather active stones, send them a private message, because they might not put them on Ebay directly, not that it would be illegal to do so anyway AFAIK, but they are not as cheap. Pretty sure you will get one.

“You can also directly order all parts with one click from there: ”

I don't see the PCB included there..wonder what else is missing..

2016/05/28 14:41

Well, of course you won't get a custom made PCB from Mouser, since they are a component distributor and not a PCB fab house. Please ask Fehlfarbe as mentioned above or build your own. Also, it seems C5 has been dropped out of the original Mouser BOM link, so keep in mind to add a 0805 330pF/50V yourself.

We've seen the pitchblende offerings on ebay but most of them were either to expensive or would have created too much effort to make a sensible/secure storage casing for the minerals.

2016/05/29 21:09

“Well, of course you won't get a custom made PCB from Mouser” I thought there was a way for a small private individual to put its own PCBs there, but I guess I should have known better.

“Also, it seems C5 has been dropped out of the original Mouser BOM link, so keep in mind to add a 0805 330pF/50V yourself. ”

What don't you create a new link, otherwise people are going to order from this link and miss parts.

As for the pitchblendes: Well you pay only once in your lifetime like 65€ and get constant 160kCPM ~ 1mSv/h for that time. That may be worth it if you want to design a PCB and test how stable it performs (design a stable one for those CPM flux values can be a problem). Such hot stones (or even hotter than that: there are 400kCPM ones for several 100€) are more rare and so have of course their price. But I give you that the shielding/storage is indeed a problem.

PS: are you still planing a crowdfunding?

2016/12/19 22:09

Well, we're not spoon feeding here :) It's our own freedom and therefore our responsibility to re-check data we use/get, in this case with the BOM on this page. Anyhow, C5 has been re-added to the mouser BOM.

Part availability is a constantantly moving target, I don't get why mouser would drop a single item of the BOM, but hey, database maintenance is gonna maintain, or not.

Since there has been a surge of new interest, due to the situation in east belgium/western germany and the great work of the FiFF folks of deploying an independent p2p radiation sensing network based on PiGI we may give it a try to restart the crowdfunding campaign. Since basically all relevant public media outlets refused to publish anything about PiGI back in 2013, we didn't get the required critical mass to reach enough people who'd consider to back the project. Maybe this time?

2017/02/22 21:41

Update: I sold the last PCB and assembled PiGIs some days ago. Maybe I will order a new load PCBs in march or april when I have more money.

Comment text

Please type the letters you see on the left into the box on the right:

     __  __  __  _____  _____   __ __
 __ / / / / / / / ___/ / ___/  / //_/
/ // / / /_/ / / /__  / (_ /  / ,<   
\___/  \____/  \___/  \___/  /_/|_|