For my next project I would like to do a big VFD display using Russian IV-25 tubes. This has lead me to read a lot about these tubes and their notorious filaments. Long story short: they are very fiddly and require a low AC voltage which is quite hard to obtain without specific transformers.
Another way to power it is through straight DC; but the problem is that one side of the display ends up being more luminous than the other side.
The third way is to feed it a high frequency square wave floating slightly above the ground level. Ideally it should be ~25Khz or above to avoid the human hearing range. This is the way I have been exploring and I believe I have managed to do quite well! Here’s the method:
- Generate a 1V DC signal.
- Generate a 5V square wave (using whatever method you prefer. A 555 timer if you’re old school. I use a Attiny13 because they’re dirt cheap and require no external components).
- Pass these 2 signals in a non-inverting summing amplifier.
- Buffer the resulting signal using a BJT power transistor.
Very simple isn’t it? Here’s the schematics:

I built a quick prototype of this using perf board, SOIC to DIP adapters for both the ATTINY13 and the op-amp in quad package form:

Note: I use a simple N23904 BJT NPN transistor here instead of a power transistor. A puny BJT can source enough current (less than 100mA) for a single VFD tube so it is good enough for a prototype.
Originally, I used a LM2902DR2G quad Op-Amp from On Semiconductor, a proven design of the LM290 family of operational amplifiers. Here’s the resulting signal:

At barely 5Khz, the square wave already doesn’t look very square… Clearly the LM290 won’t cut it for this application. It’s good enough to light up the tube though –with a very annoying buzzing electricity sound:
This power supply is looking good! In the next post I’ll discuss op-amp selection!