Properly power managing VFD displays for better longevity

[akasaka]

I like to use VFD displays in various builds of mine. While most gear is synth or test equipment that won't stay on for long or frequently, I also build clocks which are supposed to run for lengthy amounts of time.

To go around the phosphor burn-in, I've made the decision to include PIR motion sensors in most of my builds, so that the firmware can put the display to sleep when there is nobody in the room potentially watching it. I also try to run the display at a 50% (or even 12.5% for NOS units) brightness setting as the phosphor is usually pretty bright for how much light I have in my room.

For turning the display off I normally just use the vendor-supplied "Power Saver" command which turns off the AC inverter of the heater circuit. Leave it alone for 10 minutes or so, and the display switches off.

However recently I've acquired an enormously oversized VFD (over 50cm in length!) and noticed that whenever the display starts to run, the filament strings expand so much that their tensioner springs leap to the side of the assembly by around a whole centimeter.

While this insight was fun to learn, I presume the constant stretching and contraction would not be kind to the thin heater filaments over time, thus a much longer timeout is in order — at the potential cost of the phosphor burn-in.

From experience of looking at other devices, I've seen some just have the heaters constantly running. For example, my CD player has a 6V3 tap on the transformer that goes directly into the display, bypassing any switches and power circuitry. Seems to even look fine, but I think it wasn't used for much before it landed in my hands.

If I'm not mistaken, running the heater constantly reduces the mechanical stress from stretching the filaments, but then something called sputtering (?) will happen and there will be visible stripes on the display segments. On the other hand, turning off the heater every time causes the metal to stress and eventually snap due to fatigue (albeit I have yet to see a VFD with a damaged heater that wasn't due to damaging the tube itself or overvoltage).

For now I am considering a two-stage power management strategy — if there is no user action for e.g. 10 minutes, blank the display in software, but keep heaters running. If there is no action in the following 110 minutes, shut down the heaters too. But I'm not sure if this will be for better due to less stress to the filaments, or for worse due to sputtering.

Does anybody have experience with designing such devices and picking a "sweet spot" for usability vs. longevity?

[coromonadalix]

i  thought of lowering the ac filament voltage to maybe half of the voltage   if it is possible ??  somekind of pre-heat ??

some slow  lowering and rising ...  to avoid any brutal voltage augmentation ?

[Zipdox]

Half a meter VFD? Holy shit I've got to see this, post a picture!

[akasaka]

coromonadalix
Sadly all of the ITRON VFDs I'm working with are modules, thus there is no schematic to what is what and this would require modification on a hardware level — the datasheet shows no command for controlling heater voltage (unless the brightness command does exactly that).
I presume though, if the heater is driven with AC then the sputtering should be less fatal?

Thus I have for now implemented a second timer in the firmware and set my smaller VFD clock to blank after 10 minutes, but stop the heater only after 12 hrs of inactivity.

Zipdox
I didn't yet figure out how to drive it myself, but I found a way to enter the test mode on the original RS425/RS232 interface board, so you can have even a video, in which I also conveniently forgot a 50cm ruler in the shot :-)



I'll have to bother a friend of mine who has a 16ch analyzer (mine is an 8ch knockoff from years ago) to capture the signals and map them out to any of the known ITRON protocols, and then I'll make another video with PIS-OS running on it once that's a thing.

[daisizhou]

I got a GU256X32-302 data sheet, I hope it can help you

[akasaka]

Thanks, yes, that seems in line with the analyzer capture for the most part. However this one has two more address lines which I'm not sure the purpose of. I've sent an email to Noritake ITRON for the time being but got no reply yet.

However I wrote a preliminary driver here: https://github.com/vladkorotnev/plasma-clock/blob/develop/src/display/gu312.cpp



Since autoincrement is not locked to the display area, but for some reason goes across the whole RAM, I have to write the bottom row 7 times for the counter to overflow into the next column. Seems like it isn't an issue for performance and still allows a stable 62FPS, but I'm waiting for the PCB to arrive from JLC to get any conclusive results.

[daisizhou]

This is my actual picture.

Did you know that there is an OLED display that can perfectly replace it? Noritake said that GU256X32-302 has been discontinued and can only be replaced with GU256X32-800. It is estimated that GU256X32-800 will also stop selling soon.
So I'm looking for an OLED type display to replace it, preferably a perfectly compatible replacement

[akasaka]

This seems to be an embedded module, so I don't think you'd be able to find something directly compatible. Your best bet is probably to parse the protocol and translate it using a microcontroller, if latency isn't a problem.

But also if the VFD is intact I wouldn't bother, because they can last for decades.

Be that as it may, you're probably better off asking in a separate topic on the forum :-)