Jeff Duntemann's Contrapositive Diary Rotating Header Image

Steampunk Geiger Counter, Part 4


Sparks. Let’s talk about sparks. Last night I finally got things lashed up sufficiently to see whether I could translate three volts–a pair of C cells–to the neighborhood of 600 volts, using an old 25,000 ohm : 3.2 ohm output transformer and a spark gap. Got sparks. Didn’t get 600V. (Got about 350 at best) Drained the batteries pretty quickly.

Nonetheless, it was a fascinating experiment, in a technological backwater I’ve never really messed with before. In summary: Put a pulse of current through the low-impedence winding of an output transformer, and a pulse of high voltage (compared to the input voltage) will appear across the transformer’s high-impedence winding. Rectify the pulses, and you can accumulate voltage in a good, high-value low-leakage capacitor.

One way to rectify the pulses is to send them through a spark gap. The air gap breaks down against sufficiently high voltage and current passes one way across the gap. Put a cap in series with the spark gap and it will store a certain amount of charge each time the spark jumps.

At least, that’s how it works in theory. In practice, with a very high resistance voltmeter across the capacitor, I saw two phenomena I wasn’t expecting:

  1. About half the time, I get sparks on both a make pulse and a break pulse. (Ordinarily you only expect a spark on the break pulse.) If both make and break generate a spark, a pulse jumps the gap in the opposite direction as the pulse that preceded it. This means that the charge placed across the capacitor is then of the opposite polarity, which drains the cap by about as much energy as the previous pulse placed in it. Tinkering with the gap spacing didn’t help, though the effect happened more often with a higher voltage (>6VDC) into the transformer.
  2. Eventually, the spark refuses to jump. It looks to me like accumulating a certain voltage on the cap bucks the spark gap and makes it harder to jump with the same pulse from the output transformer. And of course, once the spark ceases to jump, voltage on the capacitor ceases to rise.

With my lashup, once voltage got to about 320, there were no more sparks, with about .003″ across the gap. Putting a stronger current source across the input didn’t help. I was eventually pulsing 12.6VDC from my 30 amp linear bench supply, which heated up the poor transformer pretty badly but didn’t give me any more voltage across the cap. Now, 320V may be enough to get conduction through a Geiger tube (I’ll find out shortly) but the articles I’ve read suggest 600-900V, and seem to think that this can be had from a couple of C cells and a spark gap.

I did better placing a husky 1000 PIV 1N5408 silicon rectifier diode across the spark gap. The charge went up and only up (because current reliably passes only one way through a rectifier diode) but it still topped out at about 350V. I suspect that that limit may be inherent in the relatively small output transformer I’m using, and when time allows I’m going to troll the collection for the largest one I have and swap it in.

Now, a steampunk mad scientist never runs out of #40 copper wire and thinks nothing of winding his own transformers, so if that’s the secret, a steampunk Geiger counter remains a possibility. However, I’m beginning to wonder how well I can achieve the steampunk ideal (no active devices) with only what I have lying around. Winding my own step-up transformer is just on the other side of what I’m willing to do.

The next step is making sure my two Geiger tubes are good by lashing them up to my 525V DC supply and exposing them to a (mildly) radioactive gas rectifier tube. Don’t know yet when I’ll be able to do that (real work has been piling up this week with the two of us trying to recuperate) but I’ll continue the series here as time permits.


  1. Tom R. says:

    When things don’t work as expected THAT”S when you learn.

    I don’t know when the Steam Punk era that you are trying to match ended, but I think the old induction coils went back pretty far. I used to have one as a kid with two metal handles that would give a good shock, but was probably a wimp compared to bigger ones. Magnetos may also go back that far, and a geared crank whipping some magnets past a magneto might produce a few volts.

    One thing I seem to remember about induction coils was that the MAGNETIC circuit wasn’t as good as they are now. They didn’t have “E” cores like are in most transformers and that may have contributed to a much faster magnetic field collapse, and if I remember the voltage is a function of the rate of change of the magnetic flux.

    It would be interesting to see what was generating the reverse polarity H.V. on the MAKE of your switch. I would think it might also be an effect of the “tight” magnetic circuits in modern transformers.

    Sorry I can’t be more specific, but I can only remember so much from Physics and EE courses about 40 years ago.

    Keep going this is REALLY interesting!

  2. Jack, K8ZOA says:


    A quick SPICE simulation with typical transformer values (3.2 ohm winding estimated at 2mH inductance and 0.5 ohms series resistance; 25K winding 15.6H, series R 50 ohms) shows a peak voltage of around 270V. I’ve arbitrarily (but not without some reason) set the rise/fall time on the exciting waveform as 1us. In fact, the waveform probably has multiple makes/breaks due to contact arcing at your hand cranked switch.

    This is about what would be expected from the winding step-up ratio, sqrt (25000/3.2) = 88. Driving the primary with 3V would therefore generate around 270V.

    The ability of the magnetic field to quickly collapse and thereby generate extraordinarily high voltage isn’t seen in the simulation. For faster field collapse, you need to break the circuit cleanly and fast. It may also need some capacitance across the switch points. (Remember the old automobile ignition system – same concept.)

    Given your hand crank system, I think you would be much better off to use higher voltage and a current limiting resistor. Try your 12V supply, but with a series resistor of a value to limit the current (when combined with the resistance of the 3.2 ohm winding) to a couple hundred mA. Something in the range 50 to 75 ohms. Driving the transformer with greater current can be counter-productive as you are going to drive the iron into magnetic saturation which reduces the inductance big time.

    In addition, not all standard silicon diodes are going to be usable in your circuit. Most standard Si power diodes have a slow turn-off period (reverse recovery time). Some diodes will stay in conduction for several microseconds. If the waveform has oscillation (ringing) at a few hundred KHz range, your diode may not have sufficient recovery time. I’ve written about this at

    The normal solution for this is a fast recovery diode, such as a Schottky. Most Schottky diodes have a reverse voltage rating too low for your application, however.

    You may also wind up with arcing in the secondary, depending on how well the transformer is insulated.

    Jack K8ZOA

    1. A larger transformer (too large, actually; see Part 5) got the voltage up to 620V or so very quickly, and with 3 MFD now available to hold charge, the voltage stayed in a useful range for about 30 seconds, which was my target.

      I haven’t yet tried a cap to suppress arcing at the pushbutton switch, but it’s easy enough to do, and in the rotary interrupter will probably be essential to avoid destroying the contacts, which were not meant for anything like that kind of service.

      By the way, abundant thanks for taking the time to reply here in such detail!

  3. Jack, K8ZOA says:


    I can report some success on the bench. I couldn’t find a tube-type audio transformer in my junk box, so I used a small 60 Hz power transformer, around 1.5 inches on a side. Only has house numbers on it, but the voltage ratio is 1:11, so it’s probably a 10V out, 120V in transformer after giving allowance for the IR drop in the windings when a load is put on the transformer. The presumably 120V winding has a DC resistance of 2K ohm and the presumably 10V winding has a DC resistance of 28 ohms. At 3V DC, therefore, the short circuit current would be around 100mA, which is way too much to avoid saturating the small core.

    Your 3.2R – 25K transformer has a voltagte step up ratio of 88:1, so all else being equal you would see 8x the voltage I did today.

    To simulate your hand cranked switch, I used a small relay with the coil driven from an HP3312A function generator at 10 Hz. The low turns winding of the transformer is powered by an HP6612C DC power supply at 3.0V output. The power supply connects to the relay and a 100 ohm series resistor and thence to the low turns winding on the transformer.

    Output side of the transformer connects to a 1N4007 1KV standard silicon power diode and a 0.005uF/2.5KV disc ceramic cap that must be 50 years old or more.

    A Tektronix 10:1 oscilloscope probe on the output and a Tek 2246 100 MHz analog scope completed the setup.

    The output is a saw tooth waveform, as the 10 Mohm scope probe discharges the filter capacitor, but the peak voltage I measured is 155 volts. At the lowest level of the sag, it’s still 68V.

    Since your transformer has an 8x advantage in turns ratio, you should easily see > 1KV out of it.

    A couple of things I noted … one is that the waveform out of the transformer is asymmetrical – the voltage spike from the applied field is much smaller than the spike from the collapsing field. This isn’t too surprising since the applied field has a slow time constant with the 100 ohm series resistance. The decaying field, on the other hand, can drop quite rapidly and generate a substantial voltage. The waveform asymmetry means you have to properly polarize the diode – it must be connected so that it rectifies the decaying field voltage spike.

    The second thing I noted is that you can feel 150V – too much time tinkering with solid state electronics has caused me to forget the rules about keeping fingers out of working circuits with more than 24 volts.

    So, you are on the right track, I think.

    Jack K8ZOA

    1. Wow. You’re ahead of me here, but again, thanks for writing it all up.

      I haven’t taken a shock off this rig yet–I build with tubes most of the time these days, and my tube habits are still with me–but I took the first (small) second-degree burn off a soldering iron in probably 25 years. The last time I waved it front of my nose to see if it was still on–and missed. That’s a hard thing to explain to your friends.

      1. Tom R. says:

        In my younger days I was able to catch a falling soldering iron that I knocked off the workbench before it hit the floor.

        Fortunately I am older, less agile, and after that experience decades ago MUCH more inclined to just let the irons fall where they may.

        Great Project Jeff.

Leave a Reply

Your email address will not be published. Required fields are marked *