rejuvenation of neon bulbs and nixie tubes

[FrodeM]

It is reasonably well known that there's a fair amount of things that can cause the striking and discharge voltage of neon bulbs/nixie tubes to rise, including storage. It is also generally agreed that this can be remediated at least to some degree by the process of giving the bulb an extended period of burn-in, typically at a little over its maximum rated current. However, I have not found any decisive number about just how far you can pull the operating voltage back down doing this.

Now, I have two B5991 nixie-tubes that have risen some 30-40V over their typical values, and they are giving me a hard time lighting fully, even at 15% above their maximum current-rating (above that it tends to illuminate the lead going to the digit and not the digit itself). Would these be possible to recover, or are they completely done for?

[factory]

Are they working in the machine they are fitted in?
If they are I would advise to avoid disturbing them, as these Burroughs tubes are often very tight in the socket and damage can happen from removing them, usually resulting in a crack starting from one of the pins.

David

[FrodeM]

Are they working in the machine they are fitted in?
If they are I would advise to avoid disturbing them, as these Burroughs tubes are often very tight in the socket and damage can happen from removing them, usually resulting in a crack starting from one of the pins.

David

Most digits on these two bulbs only light up about 20-30% of the surface when used in the original machine they were used with, with the original anode-resistor. For instance, like the upper curve of the nr 3, or the foot of the nr 2. That being said, I got these with some spare-part modules so they have probably not actually been in active use for a good while. With a lower resistor value and the current mentioned I can make about 50-70% of the surface of most digits light.

The part that lights is normal brightness I would say, but the rest of the digit stays dark. 5 other nixie tubes from the lot work just fine with the original current limiting. When measuring a good tube, they have a discharge voltage drop of around 135-140V, not 165-175V as the flaky ones.

[Gyro]

What you're seeing is known as Cathode poisoning. It is caused by contamination of the emmissive surface  of the digit cathodes normally from sputtering of material from other, more used, digits. It is almost guaranteed for digits that are almost always blank.

Often poisoning is an indication of end-of-life. Even if the tube is an old NOS one, they still have a gas leakage rate past the pin seals. Atmospheric Helium is a particular problem (for vacuum tubes too) because it cannot be captured by the getters. I'm not sure if Helium leak rate is responsible for increased operating voltage.

Experimental restoration has varied success and methods. The gentlest is to keep all cathodes cycling at normal current, it takes a long time and is the recommended method of retaining normal operation. running at elevated current can help, Burroughs recommend a higher supply rather than reducing the anode resistor, this promotes easier striking.

The brave method is to run each digit in turn using a significantly higher current. This sputters off material from the surface and reveals clean material. This needs to be done with caution, running each digit until it is fully illuminated and then move onto the next one. There is a danger of the sputtered material settling onto the other digit cathodes and poisoning them so sequencing them is important. The other danger is that, if you go too far, material can be sputtered onto the glass and the supporting insulators between the cathode element, causing blackening or leakage. [EDIT: Of course, you would return the Nixie to normal circuit operation after treatment. Borroughs recommended using the lowest acceptable brightness (current) for longest life - by high HT voltage, 175V or so, and higher value anode resistor].

I don't know if you have found the Borroughs Nixie databook, you can find it here if not... https://frank.pocnet.net/other/Burroughs/616.pdf


P.S. Some Soviet nixies, eg. the IN-1, omitted the trace mercury content, this lead to very short life with cathodes literally disintegrating and shorting. This won't be an issue with Borroughs displays though.

[eliocor]

search for "solving nixie cathode poisoning":
https://www.eevblog.com/forum/repair/rejuvenating-nixies-some-comments-photos/
https://surfncircuits.com/2019/04/06/eliminating-nixie-tube-cathode-poisoning-bi-quinary-digit-ghosting-and-heavily-oxidized-leads/

[FrodeM]

What you're seeing is known as Cathode poisoning. It is caused by contamination of the emmissive surface  of the digit cathodes normally from sputtering of material from other, more used, digits. It is almost guaranteed for digits that are almost always blank.

Often poisoning is an indication of end-of-life. Even if the tube is an old NOS one, they still have a gas leakage rate past the pin seals. Atmospheric Helium is a particular problem (for vacuum tubes too) because it cannot be captured by the getters. I'm not sure if Helium leak rate is responsible for increased operating voltage.

I would expect this to be more of the second effect described than purely cathode poisoning, since all digits are equally bad.

Experimental restoration has varied success and methods. The gentlest is to keep all cathodes cycling at normal current, it takes a long time and is the recommended method of retaining normal operation. running at elevated current can help, Burroughs recommend a higher supply rather than reducing the anode resistor, this promotes easier striking.

These are in some spare HP 5212L decade modules I recently got, and I have them installed in a HP counter with the correct slots for these when I power them. That should bring the striking-voltage potential to a maximum of 300V for the active digit.

I did run the math when adjusting the load resistance, to account for the higher discharge voltage drop with these. To get the best visual results, I landed on around 3.5mA, or about 15% over the original max current rating from the datasheet. The problem is that while it gets a little better with more current, I can't raise it further cause then I get discharge on the leads going to the digits and not on the digits themself.

Of course I will adjust the current as I go on, if I can observe that the discharge voltage drop decreases. I have only been running* them for 8 hours this far, under observation, but as mentioned I know there is a chance this may take days or weeks, if fixable at all in this case. It's of course a chance these may be gassy, but the issue is that when I try to look up how to identify gassy bulbs in a search engine almost all hits I get are clickbait slop along the lines of "Neon bulbs are bulbs that do indeed contain neon gas".

(* cycling betwen the digits of course, it's in a counter after all so I just have that set to count input pulses from a signal generator.)

[FrodeM]

For reference, here is the two nixies in question at the 3.5mA:
https://youtu.be/m_fIS6m8Mek

The two modules are not original to this exact counter, but it's just the modular display-attachement on the front that differs.

[Gyro]

I think you are doing what you can, without pulling the nixie tubes, apart from a waiting!

Looking at the video, I can see the discharge slowly creeping into the dark areas of the cathodes, so they are at least 'trying'. It's hard to tell on the faster cycling digits but there looks to be better cathode coverage, I don't know whether this is down to more exercise or chance, or maybe they have been illuminated and cycled more in previous operation. Depending on what you observe over time, it might be worth swapping around the individual decade modules.

Yes, trying to increase the current further is likely to cause discharge between the internal cathode connections, due to leakage in the cathode driver ICs or transistors (use caution not to exceed their breakdown voltage!). You would need to pull the tube and use clip leads to manually energize each cathode until you get consistent illumination. This is a last ditch treatment if what you are already doing fails to give acceptable results.

[FrodeM]

Yes, trying to increase the current further is likely to cause discharge between the internal cathode connections, due to leakage in the cathode driver ICs or transistors (use caution not to exceed their breakdown voltage!).

Little to worry about here, HP used a photoresist-based opto-isolator decoder to drive these, so there's no active components at all in direct path of the nixies.

[factory]

Those in the video don't look too bad, with patience & cycling through the digits at a slowish speed you should get them working again. By the way the 3734A counter in the video was the second product designed at the HP factory in Bedford England, I'm guessing you bought one of the two that appeared for sale recently.

David

[tooki]

Are they working in the machine they are fitted in?
If they are I would advise to avoid disturbing them, as these Burroughs tubes are often very tight in the socket and damage can happen from removing them, usually resulting in a crack starting from one of the pins.

I expect you could apply the same trick I use when removing fluorescent tubes from sockets they’ve been in for years: spray some contact cleaner/lubricant onto the socket and pins first. The amount of force required to remove the tubes drops dramatically.

[FrodeM]

By the way the 3734A counter in the video was the second product designed at the HP factory in Bedford England, I'm guessing you bought one of the two that appeared for sale recently.

David

Ah, nice to know.

No, actual, this one was dumpster-dived at the local university by the student hackerspace. My own counter is a 5245L, but it's awaiting some new capacitors so I can't use that one for this kind of stuff just yet.