My first power supply!

[Paul Andrews]

Hi,

I've been beavering away on a nixie tube power supply using an off-the-shelf flyback transformer (I wanted to be able to use it in battery powered clocks, but be able to use the same basic design in larger clocks with a 12V supply). I've written the whole process up on hackaday.io if you are interested. It works fine. It isn't in the same league as (for example) the John Taylor power supply, but it is small, produces enough power for most purposes (easily producing 20mA/200V with 12V in for example) and is pretty feature rich - you can enable/disable it and vary the output voltage from 0V to 200V either with a resistor or a PWM signal.

My intention was really to have a design that I could integrate directly into a clock PCB, rather than it be a standalone module, and one that I could tailor to specific clocks - so for example I have a smaller version that uses a smaller 1:20 transformer that works fine for four tubes being driven from a LiPo or USB - but I have actually produced some boards over at OSHPark.

I hope you all don't think it is too bad an effort - I am no electrical engineer - and if you are so inclined, feel free to use the design however you like.

- Paul

[Paul Andrews]

I put together a graph showing max output current for Vin and Vout:

And a picture of it driving some IN-12 tubes (my test load of choice) at 177V  from a LiPo battery:

[Nick]

Nice job! - looks really well done. Might be interesting to see if you can push the efficiency up a bit.

Are you sure there's no core saturation going on, and what about output ripple?

Nick

[Paul Andrews]

There is no ripple that I can detect with my 'scope. As for saturation. I don't think so. I don't see any collapse in output voltage or surge in input current (would I? I don't know). The transformer is pretty capable, even though it is a little on the small side. I have got some potentially beefier transformers to try out.

I went for the regular 1oz, 2 layer board from OSHPark. I was wondering about the 2oz copper, but not sure if it would make any difference. I couldn't avoid a small ground loop with the 2-layer option, not sure if that makes any difference, but I wanted to stay away from 4 layers because ultimately I want to integrate it into a clock PCB and wanted to avoid the expense of 4 layers. I also avoided laying a ground plane under the transformer, but I wonder about leaving the ground plane there so I could use it as a thermal sink, but slap a bunch of vias in it to reduce (remove?) the eddy currents.

I'm pretty glad I went for surface mount everything. Way better selection of suitably spec'd components and of course it keeps it small :-)