Help with Nixie tube power supply -- common ground, hot mosfet

Thread Starter


Joined Nov 13, 2018

I've been working on a circuit for a nixie tube clock but I've run into a couple issues I'd like some advice on. I've attached the schematic for a working prototype I'm using. The actual final circuit will look more like the one here. However, for now, this is what I'm working with.

The transformer is wound by hand, and the n-channel MOSFET is driven by a 25kHz arduino pulse with a 50% duty cycle. The power supply has a 2.5 A limit, above which it will shut off. With 9V in, the rectifier/doubler on the secondary side produces an adequate voltage to ignite the tube and provide the correct amount of current, about 3 mA.

One issue I'm having is with excessive heating of the MOSFET. Its datasheet is here. I have a small heatsink attached, and a flyback diode was added, which all seemed to help, but it's still a little much. One solution I'd like to try is reducing the duty cycle. The original circuit I tested used two 0.01 uF caps, but I'm going to see how the 86 uF cap does. I believe this should allow me to include rest periods of even a few 100 ms where the MOSFET is completely off.

The main reason I'm writing, however, is probably more serious. I have no idea what to do about the ground on the high voltage side! The nixie tube driver is an IC, datasheet here. It serves as a BCD decoder, which grounds the cathode corresponding to the digit encoded by 4 binary inputs (A, B, C, D). It takes a 5V supply voltage, which I intended to use my arduino to provide. Right now, my arduino's ground and the primary ground need to be the same to allow the MOSFET to work properly. What I've read (and seen) before has given me serious misgivings about connecting my arduino (or anything, really) to the secondary's ground, but in order for the driver to work, doesn't it need to share ground with the arduino (the source of its input)? Can I count on the primary ground being close enough to the secondary that I can just connect the cathodes to the driver? Maybe just include a large-sized conductor on the secondary and leave it floating?

Another, related issue, is that I am worried I don't know how to use my oscilloscope properly. It seemed to be able to measure the high voltage signal just fine, but do I risk damaging it by probing it like that? I seem to remember the seller telling me that ground was necessary, that it couldn't just float like a multimeter, but I thought he meant just that it had to be plugged into a 3 prong, outlet. It seems fine, but I've gotten some occasionally weird things happen, for instance, when the ground on the probe slipped off I saw could see the trace jiggling a lot (the readout too), and a cube amp I had connected to the same power strip started to pick up some audible interference. The scope is a gwinstek GOS-6112.

Finally, you may assume I have very little clue what I'm doing and give whatever other helpful tips you may have. I know how to stop MYSELF from getting shocked, but I may be a little ignorant in terms of how to protect my equipment.



Joined Jun 4, 2014
This is probably because you do not have a big enough voltage on the gate to fully turn on the MOSFET. Its on resistance is specified at 10V and that shoud be your target for the gate signal.


Joined Oct 22, 2014
Hi There.... you might like to check out the link to an engineering .pdf about "Snubbers", R-C Snubbers, and go down to page 5. flyback_snubber_design.pdf

At the top of page 5 there is a snubber set-up that might work for you, and calculations for values can be worked out, but do take some math. I see you have a cap across the 25khz switched primary, but I might assume we need a bit more for primary kickback currents during FET off-times. As for the scope ground, I always use an adapter that keeps me from hard grounding, and usually run my scope thru an isolation transformer, 1:1 as I am used to working on HV hot-grounded switchers. Floating my scope on HV equipment can make me a current path if I touch the wrong spots, but that is HV and RF usually. For specific values of switching snubbers and reduce math-time I sometimes find common integrated switching IC datasheets like the TOP224 series and others that often show optimum snubber values/considerations in the notes.... for frequencies as many will switch around the 22k-55khz range.
If you get a small isolation transformer to plug the Arduino and any other power adapters into, then you could tie grounds and not worry much about scope issue's as well. What is max power out at 1:5.8 then half-wave rectified I might see +35 to the Nixie Tube (9v switched into 1:5.8 then half-waved[Vsec x .7]). ? Full-wave rectifying provides clean power and higher Vsec voltages (Vsec x 1.4 filtered) would be 78 max, then at 50% duty-cycle we divide those number in half to about +17dc(half-waved) and +38dc(full-waved) :)
I might also consider protecting the driving Arduino signal with a damping resistor (22 to 200 ohm)and/or gate-drive speed-up diode in parallel with the gate damping resistor which helps faster turn-on/turn-off signals at the gate, and on N-Channel drives we usually want a pull-down to ground on that gate as well of a higher value (10k).
Have fun, good luck, enjoy