Cathode poisoning and tube's current flow

[Luka C]

I've recently performed healing on cathode poisoned IN-18. I used the method of increasing the current (never more than 8mA, datasheet specifies it as max) to the point when all the parts of the digit light up and then letting it "bake" until the digit is evenly lit on nominal current of 4mA. I succeeded on every digit and noticed something in the process. I was measuring the current flowing trough the tube and it seems to increase gradually while the tube is healing. I guess this is perhaps related to the fact that removing the coating from the cathode increases it's "conductive" surface and thus resulting in increasing current. What seemed interesting was also, that unrelated to how large the poisoned segment was on the tube, the rate at which the current was increasing corresponded to the amount of time it took to heal the tube. In my case, digit "3" had almost all of the upper part poisoned, but once setting the current to 6.0 mA, over the next 15min it went to 6.9mA and after that the digit was healed when returning to 4.0mA. On the other side, digit "8" was lightly poisoned on two places (both areas probably 3-4mm each) and the current was really slowly increasing (I had to set it to 7mA to start with) to 7.6mA and it took over 6 hours to regenerate those areas.  So those are my observations, I wonder if anyone else noticed something similar?

[gregebert]

I had no luck trying to depoison 1 or 2 Burroughs 5031 tubes, even after several hours. I considered these tubes expendable and put so much current thru them that they got hot. Darn! I didn't measure it, though.

I'm amazed how quickly you were able to depoison your tubes; I was expecting days or even weeks.

Did you purchase your tubes in this condition, or did it occur while they were in your project? I'm curious about the conditions and timeframe that cause poisoning. My 14-digit clock has several static digits, so I run a 1-hour depoisoning every night; maybe it's overkill ?

[Luka C]

I bought the tube with cathode poisoning so I don't really know how long was it operating and under which conditions. To be honest, I was also surprised to see most of them recover in such shorter periods. I'm not sure if it's of any significance but I digits that took less time to recover were closer to the anode grid, perhaps there is something in that.

From what I read on the web, the reason why your tubes didn't recover may be that tubes have a certain point after which they no longer seem to be repairable because the current that would be needed for healing to effectively take place is way too high and would cause the destruction of the tube. I don't know if your tubes still operated after cooling down, mine seemed to be warm to the touch, but not hot?

Regarding my clock, I have the first version of it running for around half a year now and there are no signs of cathode poisoning, probably both due to the fact that the time was way too short for it to occur and also, I enjoy the slot machine effect that I made for cathode poisoning prevention so I let it spin for ~5 seconds every 1min. So I guess letting them run the routine for 1 hour at night can't be a bad idea, I mean, you're sleeping anyway and the tubes will sure benefit from it (I guess there is no damage from exceeding the "standard" amount of time dedicated to anti-poisoning routines) :)

[GastonP]

Do you mean 1 hour overcurrent each night or simply 1 hour of slot-machine effect?
Slot machine would not damage anything, but overcurrent will kill the nixies sooner than later, IMO.

[gregebert]

I dont boost current at any time; it's regulated at 5mA. 

Tubes that run 0-9 uniformly during operation are turned-off during the depoison session.

Tubes that run 0-5 are 'depoisoned' while running 6-9

When I update the RTL code, I'll swap the date and time for a few minutes each hour and reduce the depoison session accordingly.

[Luka C]

Here is an example of the healed digit on IN-18. Btw, there is no need for overcurrent when you heal the tube on time I believe.