Nixie tubes have two things going for them: they’re awesome, and they’re out of production. If you’re building a clock – by far the most popular Nixie application, you’re probably wondering what the lifespan of these tubes are. Datasheets from the manufacturers sometimes claim a lifetime as low as 1000 hours, or a month and a half if you’re using a tube for a clock. Obviously some experimentation is in order to determine the true lifetime of these tubes.
Finding an empirical value for the lifetime of Nixies means setting up an experiment and waiting a very, very long time. Luckily, the folks over at SALTechips already have a year’s worth of data.
Their experimental setup consists of an IN-13 bargraph display driven with a constant current sink. The light given off by this Nixie goes to a precision photometer to log the visual output. Logging takes place once a week, and the experiment has been running for 57 weeks so far.
All the data from this experiment is available on the project page, along with a video stream of the time elapsed and current voltage. So far, there’s nothing to report yet, but we suppose that’s a good thing.
Although the folks at SALTechips have done a great job, the IN-13 is not a Nixie tube. First of all, Nixie is a trademark of Burroughs. Second, a “nixie type” tube displays the digits 0-9. Third, the life expectancy ratings of Soviet era tubes are EXTREMELY conservative. Tubes rated for 5000 hours have run for YEARS in well designed clocks with no signs of degradation or failure. I suspect that running this experiment with an IN-12 digit display tube would take a decade or more before darkening became an issue.
Wouldn’t it make more sense, and be more useful if they were changing what it is displaying constantly, or at least if they were testing a nixie tube, every minute. Turning things on and off often seems to shorten the run life of most things.
Industrial multi-phase electric motors have fast and slow start, I’m guessing for a reason. For Nixie tubes the equivalent of slow start may be a circuit with very little hysteresis characteristics, or similar refinement.
This is especially true of luminous tubes. For example, a fluorescent light with an instant start ballast will hardly last a month if turned on and off multiple times a day from sputtering in the tube.
The experiment specifically aims to measure the darkening of the glass envelope of the IN-13 tube and how that darkening can dictate the end of the usable life of the tube. The effect is accumulative; once metal has been deposited on the glass, it will stay there as a thin film. As such turning the tube on and off will have little effect on the darkening process. For sure there are transients in the very detailed physics of the process when it starts, but these are insignificant compared to the material being deposited under steady state.
Nixie tubes are quite different to fluorescent lamps as the glow is not initiated by thermionic emission. In a fluorescent tube the filaments are heated momentarily forcing them to emit electrons (traditional ballast). A high voltage pulse then accelerates the electrons causing emission the electrons collide with the atoms inside the tube. In a nixie tube a voltage is applied to the terminals and you wait, patiently, for cosmic radiation to strike the tube ionising a couple of atoms. The free electrons are then accelerated causing more collisions and emission, similarly to the fluorescent tube. The life of the filaments in a traditional fluorescent lamp is greatly shortened as they are heated, much like an over-driven incandescent lamp.
Forgive my ignorance, but your’e saying a nixie tube requires “cosmic radiation” to work? Please clarify if possible.
They need something from outside if the voltage inside isn’t high enough. They can fail to light at low intensity if the room is dark. Often ambient light is good enough to do the job.
You’re kidding me, right?
You’re complaining they used a non-Burroughs (and thus not, by name, a Nixie) tube? Holy hell, the pedantry with this one is amazing.
Both the IN-13 and official, trademarked Nixies have an anode, cathode, and a low pressure gas inside. That’s it. That’s a Nixie tube. They are testing cold cathode tubes in this experiment. They are using cold cathode tubes. Have you ever heard of a genericized trademark?
Second, since they’re only testing the output of one cathode, wouldn’t it make sense to use a tube with only one cathode? Numeral nixies don’t have that. The IN-13 does.
Third, your complaint that russian/soviet datasheets are conservative is exactly what they’re testing. You wrote an entire paragraph that says nothing. Perhaps you should get into speech writing or public relations.
The configuration of the cathodes and the concerns with darkening are completely different in numeric Nixies. When I think “nixie tube”, I certainly don’t think bargraph tube either, but still — it doesn’t make sense to test the lifetime of “nixie tubes” with an IN-13. It makes sense to test the lifetime of an IN-13.
What makes far more sense is to accelerate their testing with a variety of tubes and driving them higher than specs. Have a few that you drive progressively harder to see how nonlinear the accelerated failure is. Have fun, extrapolate, get data in weeks, not years. I can make a neon lamp get as black as it would in 20 years in about ten minutes.
If only it was so straightforward. We know that sputtering is highly non-linear and the best empirical approximation we have is that the sputtering rate is proportional to the quotient of current over gas pressure raised to the power of 2.5. But that only applies to a limited range of currents and pressures. Overdriving a tube will certainly increase the sputtering rate to a point where measurable darkening will take place over a short period of time. However, you can’t extrapolate down to normal operating currents as the equation developed for increased currents will no longer be valid. This is especially true for a tube such as the IN-13 that is operated at low currents in the ‘normal glow’ region unlike numerical nixies that are operated in the beginning of the ‘abnormal glow’ region. What is more the gas pressure is indirectly related to current through the gas temperature.
You can see that it is very difficult to ‘predict’ the lifespan from extrapolation. What we have opted for instead is a precision brightness measurement under normal operating conditions and then use those data points to fit a curve that we know is valid as it is purely the end result of the sputtering process. Have a look at our working and bibliography for a deeper understanding.
At last someone who knows what a Nixie tube is and where the name came from. Thank you.
It’s an interesting experiment, but really proves nothing, except the lifespan of a single sample of an IN-13. The lifespan is determined by several factors, the primary two being the current and voltage the elements are driven at, and how good the seal around the pins are. Other factors are vibration, bond-out wires from the pins to the cathodes, and purity of the original gas.
As mjrippe says, Russian tubes are very conservatively rated, and there are clocks (and other displays) out there that have been operating for decades with no issues.
There has been some speculation that driving nixie tubes by multiplexing MAY shorten the life span, particularly if you can hear the tubes “singing”. That means they’re vibrating, and if they’re vibrating, they’re being mechanically stressed. I’m sort of a purist in that respect, and prefer tubes to be directly driven.
The experiment is focused on the loss of brightness as the reason to discard a tube. Other factors are briefly mentioned in the description. As you mentioned unfortunately specifications in datasheets, when actually published, are too conservative and/or the end of life criteria not stated. For the first time we can point people at real data with a transparent measurement technique and end of life criteria. That’s a big deal as nixies are used so frequently and good homebrew tubes have just appeared. Just imaging if we had such data for plasma TVs (made public, I mean) in the days they were selling by the 1000s!
For SALTechips such data is crucial as we design-in a tube in a product and know with confidence that it won’t go dim. Customers often query about the lifespan of these tubes precisely because no substantiated answer has been given.
I have a Cherry Nixie display (single piece 14 segments x 20 digits
multiplexed) made in 1980 from the date code. It was probably used in a
pinball machine for a few years before sold to surplus.
I found the tube is still very bright at 160V instead of the 180V as
marked on the PCB. Other than some “burnt-in” and some chipping of the
glass around the edges, the display and electronics was working when I
fired it up a few years ago.
So multiplexing, burnt-in, 30+ years of gas leakages doesn’t seem to
kill off the brightness of my display.
The display has been unpowered and sitting in my junk/treasure pile. I am still looking
for a good use for it as it is a waste to make a clock/twitter/RSS “ART”.
Thanks for the article Brian!
Few things, if any, are certain in life and the divide caused by the use of the word ‘nixie’ is one of them! For sure, one of the first things one finds out when reading about ‘nixies’ is the original use of the word. Since then and especially following the recent ‘rediscovery’ of the technology the word is used for almost everything indicator-like using neon. Numerical indicator tubes, bargraph tubes, thyratrons, dekatrons even neon-based voltage regulators have been called nixie tubes! Personally I’m surprised how neon signs have escaped this so far. Imagine that! The widespread use of the word is a side-effect of the popularisation of the technology. Purists need not be too alarmed as the correct origin and strict meaning of the word has gone down in history. Instead be pleased by the huge following the technology enjoys! Ok, maybe if neon signs start being called nixie tubes then it might be time for action :-)
I can’t tell whether I can’t read, or whether they don’t actually have a graph for 57 weeks. All I see is a graph for 3 weeks, and their ustream claims ~10kMin, or about a week.
Yes, I was curious about this as well – I’d really like to see if there’s any decrease over that period of time.
Over-driving the tubes to accelerate aging is possible, but it’s an extremely nonlinear process, so if you drove it at say 100mA, how are you going to know what it does at 10mA.
I can confirm that the experiment has been running for just over a week. There are two data points, one at the beginning of the experiment and one yesterday (1-week sampling period). I think Brian got confused with the x-axis on the graph and the fact that we have been using the IN-13 tube in the thermNeon for more than a year without any visible decrease of brightness (using our eyes!). No matter as it is clear in the description text.
You hit the nail on the head with the sputtering being non-linear. The non-linearity is not the real issue though, the real issue is that the non-linear equations that can be used to describe darkening are only valid for a limited range of currents. So you can’t accurately extrapolate.
Great work with the vacuum experiments.
Please come back and post a followup in a year! I’d love to see whatever data you have when there is any. :)
Many years ago, at the end of the Christmas season, I had one of those decorative neon candle bulbs. The ones with a flame shaped electrodes and would flicker all over.
I was curious about how long one of those would last. I decided to leave it plugged in and turned on non stop to see. That was 2005. Other than a very mild darkening, it’s still functional after mostly non stop power save for a rare power outage.
I made a clock from IN-12B nixies about 5 years ago and it’s been running continuously ever since. There are six display tubes. I bought four spares but have not needed to use any of them to date. The spec sheet for them says “work capacity 7500 hours” but if that means lifespan then they have certainly performed well. I’ve not noticed any dimming either. I’m not sure when they were manufactured, but the original packaging looked 60s ish.
Everything the soviets made has dates on it. You can pull a tube and look. There’s roman numerals for the month, and two normal digits for the year.
You can make a 3-500 tube and other bigger transmitting tubes lives a bit longer by reducing the filament voltage slightly. I wonder if that applies to nixie tubes, but also lighting them up using some kind of PWM to reduce the duty cycle?
Later models will run for years.
Early models that didn’t have added mercury have a pretty short lifespan, some months. (Soviet models IN-4, IN-1, maybe others, are like this. short life).
If the tubes have a sort of faint blue fringing, it has the Hg added.
I was under the impression that was Argon. But Alex is right about the aging tests – glow tubes degrade much quicker at higher currents, I think it relates to the plasma temp but I’m not sure.
I’m not very knowledgeable about tubes..
I like the color, and use orange and yellow L.E.D.s as nightlights meter/dial lights, and to ‘fill in’ the spectrum of white L.E.D.s. My bench light uses 2 dozen orange and assorted other colors (mostly white) I’m age testing them, but only by visual inspection.. I’m not sure but I think ‘amber’ and orange are the same thing. (what they now call orange they at first called amber)
I heard a lot of bickering, I just wanted to see the average lifespan and maybe what people have seen in the while. Grow up folks.
I m using Russian IN-12 tubes in a clock, multiplexed and with “soft digit change” (PWMming, simulating brightness as one digit fades and the next brightens which especially affects the seconds, of course). It has been running 24/7/365 for at least six years with no signs whatsoever of dimming or metal deposition. Pretty sure that these are “mercury” tubes. OTOH, I have seen the plug compatible US equivalents (used) with incredible amounts of “silvering” – no way to know what their operating conditions were.
I’ll bet that the Russian who thought that saving money on not needing a “5” die on the IN-12 got a nice apartment in Moscow for his/her ingenious money-saving idea :( WHY????