VFD filament resistance when hot/cold?

[Jasper C.]

I'm working on a project using some IV-17 tubes, and I cam across something I need advice on.  The datasheets I've found say that the filament current is 47 mA at 2.4 Vac, so an effective resistance of about 51 Ohms.  But on 3 tubes I've measured, they're all about 20 Ohms.  Does the filament resistance change when hot?  Because that seems like a huge difference...

[gregebert]

Absolutely!  The difference between operating and inrush-current is a leading cause of burnout.

If you recall, incandescent room lights generally fail when you turn them on. It's the same mechanism.

If you dont plan on power-cycling the filaments very often, you can just leave it as is and replace tubes every few years. You will probably have phosphor degradation before the filaments burn-out.

If you like doing reliable designs, you can limit the inrush current with a series resistor, and driving from a higher voltage.

Or you could use a triac to control the filament transformer, and gradually increase the proportion of the AC-cycle to soft-start the filaments.

Here's a table of filament resistance vs current, once temperature had stabilized, from a BA0000P31 NIMO tube:

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I faced a similar issue when designing my NIMO -tube clock. NIMO tubes are very rare, so every means must be incorporated to protect them. I chose to drive the filaments from a higher supply voltage and some series resistance. In my case, I was able to reduce the peak current from 400mA to 250mA, while keeping the operating current at 180mA. The datasheet specifies 200mA. In addition, I have a 250mA fuse for each tube, and hardware+software to check each filament, fuse, and dropping resistor.

The last part is what algorithm should I use for shutting-down the filaments ? Leaving them on continuously will eventually lead to burnout, but turning them on/off too frequently will also cause wearout. Right now I'm using a 30-hour timeout on my IR motion sensor, so if nobody enters the room for just over a day, they shut down. Furthermore, it takes a bit of extra movement to turn them on so if you tiptoe into the room, they stay off.

[ZY]

For my VFD filaments, I've settled on using a power supply module with a configurable soft-start time. I've set it to something huge, like 10 seconds, to try to limit the inrush current. 

[Mark Moulding]

There's a large difference - 3-fold or often much more - between cold resistance and hot in, for example, a nichrome filament.  (It was this effect, in fact, that Mssrs Hewlett and Packard used to stabilize the output level of the Wein bridge in the their first product.)  In my projects that involve vacuum tubes applied to modern products (Numitron clocks and digital dashes, and guitar effects units), I take great care to limit the inrush current.

The simplest solution is just a big resistor in series with the filament supply, which is at least somewhat effective.  A better choice is a current-limited supply.  For example, for a vacuum tube filament rated at 300 mA, I used a supply limited to that value.  The tube takes somewhat longer to heat up, but its lifespan is greatly increased - the life-limiting factor becomes cathode emissions, rather than filament failure.

In a different application driving Numitron display tubes, my firmware keeps the filaments heated to just below the point of visibility when off, bringing them up to the desired current when they're to be visible.  This is all done with PWM, because that's easy to do in both software and hardware.  This also follows RCA's recommendations for lighting up Numitron tubes (although, considering the era, they recommend using a pull-down resistor on TTL outputs, rather than PWM).

I should note that an old clock I built using very rare DTF-104B tubes, with no special handling at all for reduced thermal shock, is still going strong after 10 years as my bedroom clock; for tube filaments, The Numitron products from RCA are very robust, and it's not clear that limiting the inrush current is of great importance.

For your project, a LM317-based current limiter would be an easy thing to implement, and that's certainly a possibility for the filament current supply.  If you need to come up with a separate supply anyway for the 2.4-volt filament, one option is to make it current-limited as well.  The easiest way though, assuming that you're using multiple tubes, is to just hang a resistor in series with each, and drive them off of the 5-volt supply.

5 volts - 2.4 volts = 2.6 volts, at 47 mA.  R=E/I, so the necessary resistor would be 2.6/.047, or 55 ohms.  The next standard size is 56 ohms.  Power is V^2 / R, which is 0.12W - a quarter-watt resistor would work easily. So, short answer: hang a 56-ohm resistor in series with each filament, and run them off the 5-volt supply.  (You're supposed to use AC for the filaments for uniform brightness, but in my experience with single-digit tubes like the IV-17, it didn't make a whit of difference.)

~~
Mark Moulding

[Paul Andrews]

What power supply are you using? I seem to collect power supply design almost as much as Nixie tubes.

[threeneurons]

VFDs, since you don't want the filament to visibly glow, vary less than other tubes. 2.5:1 that you're getting is rather mild. On typical "radio" tubes, I get 6:1 routinely. And that's on small signal tubes, such as 12AX7s. There, a filament that's ~85 ohms hot, will measure ~15 ohms, cold, on an ohmmeter.

On my 1 Nimo clock, that I made 2 1/2 years ago, the filament supply is pretty simple. The nominal filament is suppose to draw 200mA at 1.1V, for 5.5 ohms, hot. I use an unregulated "12V" supply, which is really ~11V. That goes to the filament by way of 2 22 ohm 2W resistors. The Nimo filament is in series with those in the middle to get the ~6V bias, too. But 2 1N4007 rectifiers are shunted across the filament. See the drawing. Note an extra 1.5 ohms is in series, then that group is shunted in parallel with the two rectifiers, and that whole bunch is in series with the 2 22 ohm power resistors. Its simple and reliable. Even if the cold, the current will only be ~50mA over rated, for the brief time it takes to warm up. Surge currents are usually several times the running current, during warm up. 

This clock has only been unplugged for a couple of months, last year, while I was in the process of moving.

https://threeneurons.wordpress.com/miscellaneous-projects/

[Jasper C.]

On Friday, 17 July 2020 01:56:47 UTC+8, Paul Andrews wrote:

What power supply are you using? I seem to collect power supply design almost as much as Nixie tubes.

I'm planning to use an LM9022 based supply.  Yes, it's out of production, but apparently it's the same as an LM4871.

And yes, it's overkill for a single IV-17, but I plan to start here, then get one of those multi-digit tubes to play with.  Having one PSU design will cut down on the work.

Thanks for all the replies.  I'd never thought about managing inrush current and burnout.  I'm not sure it's possible to do with the LM9022, or at least not with my level of analog know how.  But it occurs to me that the Vsupply for that chip is 2.0 to 5.5 V.  I could use a resistor-divider-set LDO to supply the LM9022, and use a digital pot + small uC to do a "soft-start" by starting at a low voltage then ramping up.  As a bonus, since I plan to have a uC in the clock anyway, I can tie the digital pot to that in my final design...

[Hannah Mishin]

You must be careful with the filament.  The filament will accept whatever current you give it.  Overtime, constant voltage source on the filament is problematic.  As such - for sure- drive the filament with an AC wave

There is lots of corollary info on this blog regarding the filament.

https://hannahmishin.com/blog/2017/4/30/russian-tri-color-vfd-indicator-clock

I accidentally drove some Itron VFD - overdrove the filament and ult. killed the VFDs.

I am controlling the brightness of my current vfd clock by modulating the filament current. meaning if you underdrive it - it will be dim, if you use the circuitry linked in the blogpost, you really will find it hard to overdrive it.

also ->

http://www.noritake-itron.com/subpages/applicnotese/vfdoperapn.htm

http://hannahmishin.com/

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/5bab3541-effd-44bc-a61a-2e7fd64eb5aco%40googlegroups.com.

[ZY]

The chip I'm using is TPS54328, which has an adjustable soft start. I'm still building my supply though, so I haven't actually tested it in action yet to see if the soft start actually limits the inrush in any way.

[David Pye]

Presumably you could also use the microcontroller to do so if it has a spare pin, by using a resistor in series with the filament power supply, which is then shorted out by a FET after a predetermined time.

Valve amplifiers used to do similar but with a time delay relay.

David

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/096a188d-eb05-4bda-986b-c42f8c80edf9o%40googlegroups.com.