VFD display filament driver schematic

[bob808]

Hello,
I gutted my microwave oven to turn it into a SMD oven (it has two IR heater elements on top and a convection system in the back). I got it's VFD display and thought about making some kind of standalone clock (or hack it for the future oven).
I managed to discover it's pinout and before I took it off I measured the filament voltage. It was somewhere around 2.2VAC, and there was a limiting resistor in series of 6.4ohm with 0.8V drop on it.  That's about 125mA.
Now, while figuring out the pinout I used a DC supply but I had visible fading from one part to the other so I know I should use AC for filaments.
The issue is I don't know how to do that.
There was a schematic using a 555 timer to generate 100khz with a duty cycle of 50% but that didn't work out for me. I attached that schematic. I didn't have BD139 so I used MJE340 and TIP50 but the signal degrades really bad at transistor base right after the limiting resistor. And there are some large spikes going into the toroid transformer so I didn't bother with it.
I'd like to be able to use 5V as I will be using an Atmega328 chip maybe, and could install a hefty regulator to power everything.
So, a nice and accesible schematic for AC power of 2.2V/125mA to filaments?
PS. I made the transformer with 40turns on primary and 2x15 turns on secondary. I tied one primary with center from secondary to +5V. Seems too sensible as I couldn't manage make it work.
Display is NEC fip6bm9r, couldn't find anything on it.

[SeanB]

Can you not use the existing transformer in the microwave? It will have the required voltages and secondaries, and an existing circuit to copy to use driving the new display.

[bob808]

Not actually.
The transformer has multiple taps besides the one for the filaments.
2.2VAC with 125mA is about 250mW. I really don't want to add a transformer the size of a small apple for 250mW I guess there's also a 5VAC tap as well but I prefer to derive the needed voltage from 5VDC as I'll be using to power everything.
There was a single IC doing all the work for the VFD and oven UI. It's a 54 pin chip that I could find no info on. It's written S-674GP AEICC9578623 on it.
That's why I'd prefer to generate a square wave of about 3.1Vp with 50% duty cycle to power the VFD filaments. I guess I'll install it back into the oven for temperature measurement and timer or stuff like that.
The 555 timer works good, it outputs a 100KHz square wave of 50% duty-cycle with an amplitude of 4.6V. But I can't power directly as the 555 can't do more than 100mA and it's not good practice. Also the output must float above ground as I know, need to do some more research.

[bktemp]

Beside the 2.2VAC for the filament you need about 20-30V for the segments/grids. The cheaper systems use a negative -30V and apply it to the center tap of the filament winding. The microcontroller can then drive the grids/segments with simple open collector pmos outputs build into the asic.
Larges VFDs use dedicated driver ICs and positive voltages.
If you want to generate all voltages from 5Vdc I would use the following approach:
Use a simple H-bridge driven by signals from a microcontroller to drive the filament from 5Vdc. The duty cycle can be adjusted to achieve 2.2Vrms.
The segments/grids can be driven using 4094 shift registers powered with 18Vdc generated from a small stepup converter. 18V is enough for such a small display with low multiplex ratio.
Apart from special VFD driver ICs this appears to me the best solution using the least number of parts.

[bob808]

Thank you for the info.
I managed to make it work, I had messed up the transformer. I rewound it and it works nice now.
I attached a screenshot of the signal across the filaments. It's not exactly a square wave of 3.1V but it's thereabouts.
The thing is that I used another supply, referenced it's ground to ground of 555 and needed about 11-12V to light up a segment. Sadly, there's ghosting all over. Attached a picture of that as well.
Any chances of it being broken?

[bob808]

Beside the 2.2VAC for the filament you need about 20-30V for the segments/grids. The cheaper systems use a negative -30V and apply it to the center tap of the filament winding. The microcontroller can then drive the grids/segments with simple open collector pmos outputs build into the asic.
Larges VFDs use dedicated driver ICs and positive voltages.
If you want to generate all voltages from 5Vdc I would use the following approach:
Use a simple H-bridge driven by signals from a microcontroller to drive the filament from 5Vdc. The duty cycle can be adjusted to achieve 2.2Vrms.
The segments/grids can be driven using 4094 shift registers powered with 18Vdc generated from a small stepup converter. 18V is enough for such a small display with low multiplex ratio.
Apart from special VFD driver ICs this appears to me the best solution using the least number of parts.

I knew I remembered 4094 from somewhere:
http://spritesmods.com/?art=vfdcontroller&page=2

[bktemp]

Is the center tap of the output winding connected to 5V or GND?

[bob808]

Is the center tap of the output winding connected to 5V or GND?

To 5V.
I forgot to add that this VFD is of the type in the attachment. "Dynamic Drive"
I connected one grid and all of the elements to 12VDC, that's how it's connected in the picture where the ghosting shows.

[bktemp]

Did you connect all other grids to GND or left them floating? They should be pulled to GND to keep them always negative with respect to the filament.

[bob808]

They were floating, duh!
So yeap, that fixed it. Looks nice. I also played with the supply and they get pretty bright at 25V. But I like it best at 11.5V or so, beyond that I get uneven brightness inside the individual segments.
Now to make a circuit that pulls the grids high to 11.5-12V when on and at ground when off.
The segments should also be pulled to ground when off?

[bktemp]

You should not drive them at high voltages for a longer time. This will degrade or even destroy the segments. Only when multiplexed you need to go higher to get the needed brightness.
All segments and grids should be pulled low when driven with open collector outputs. Using something like a 4094 will actively drive the output low so you can omit the pulldown.
You can drive the grids with another 4094: For every new multiplexing cycle clock a 1 into the shift register. On every clock pulse the outputed 1 will jumps to the next output.

[Arnvior]

What is the Pinout of the Display? I can't find any information on it