Nixie tubes

[Balazs]

I'd like to build my own nixie clock, but I'm not sure where to buy the tubes for it... The problem is that they are a bit expensive and I'm nut sure if the "used but working" short legged ebay stuffs whould do the business.
Any hints where I can buy some nice nixi tubes at a reasonable price?

[nowlan]

Ive only seen the QS30 module from china. And i get the feeling they were made in the 80s.

Id like to build a nixie clock myself, but put of a little by the high voltage. Appear to be 2 type of driver chips available, one russian, something more modern.

The kits on ebay are kind of expensive. I realise someone is making a living off it.

[TimFox]

Try "Frag' Jan Zuerst" in Germany.  Lots of Nixies and some sockets, as well as kits.
http://www.fragjanzuerst.de/

[daqq]

Search ebay for new-old stock (NOS) - those are generally in their original boxes. Look for soviet nixie tubes.

[smbaker]

I've been buying my tubes on ebay. Personally, I'm a fan of the IN-12 tube. They're pretty damn cheap and you can build your own socket into your PCB using Harwin H3161 from digikey. The IN-12 is not as 'tube-looking' as the IN-14, but it's sure easier to work with.

For driver chips, I use K155ID1, also found on ebay. That only drives the digits -- if you have a nixie with a decimal point, then you'll need another transistor to drive that. You can multiplex the tubes to save on driver chips, but personally I like to go with one driver chip per tube.

[marshallh]

Multiplexing is fine as long as you do not increase drive current above what you'd normally run at. Sputtering is based on peak current, not average current. So it's not quite like an IR led you bang on (unless you don't mind your cathodes disintegrating)
You can also run at the lowest current possible while still maintaining readability to increase life.


The IN-1 lacks a mercury additive unlike most tubes so lifetime will be very short. Less than 1000 hours. Most other soviet tubes do have additives to last a decent while.


For sockets I used Millmax PC pin press-fit inserts. They are great, but also for 16 * 6 it comes to $20. 

I have got used russian tubes and they work fine. When cleaning them be careful not to wipe off the mfg mark (or OTK stamp)

[Macbeth]

I've used a Russian seller http://tubes-store.com/ in the past. All new old stock and very cheap as these things go. IN-12's are only $2 each. 

IN-9 nixie bargraphs for only $1.50 each!

[Macbeth]

The IN-1 lacks a mercury additive unlike most tubes so lifetime will be very short. Less than 1000 hours. Most other soviet tubes do have additives to last a decent while.

That would be why they are cheap then I suppose. Where did you find this info? I hadn't seen mention of mercury. Not that I was looking for it.

Is the IN-1 the tube in that nixie watch Steve Wozniak wears?

ETA: Nope, looks like a Burroughs B5780

[barry14]

Ramsey Electronics sells Nixie tube clock kits. The price range is from about $230 to $300.  They also sell individual tubes but only to purchasers of their clocks.

[FlyingHacker]

I also like IN-12 tubes because of their price and availability. Get some sockets as well.

You can drive them with MPSA42 transistors and lower voltage counter chips if you don't want to spend the money in the 74141 or equivalents.

Some more info: http://nixietube.info/NixieTransistors.pdf

[Balazs]

I've used a Russian seller http://tubes-store.com/ in the past. All new old stock and very cheap as these things go. IN-12's are only $2 each. 

IN-9 nixie bargraphs for only $1.50 each!

Thanks, this store is just what I need!   

Are the nixie driver ICs a must, or I can use some serial to parallel conversion IC + transistors (because of the high voltage level)?
Any tips to prevent cathode poisoning?

[cncjerry]

If you want to build a clock, I highly recommend tubeclock.com.  I built one years ago, sits on my desk in an open frame.  I've run it as long as 9 months and it gained less than a second.  I don't know why this clock is so accurate though the temperature in my town varies very little season to season, always around 65 degrees.

I set it with my GPSDO and tweaked the internal timing constants over a period of months when I assembled it.  He sent me the software as well since it uses a PIC chip.  Nice guy to deal with, very nice, accurate non-disciplined clock.

[FlyingHacker]

I've used a Russian seller http://tubes-store.com/ in the past. All new old stock and very cheap as these things go. IN-12's are only $2 each. 

IN-9 nixie bargraphs for only $1.50 each!

Thanks, this store is just what I need!   

Are the nixie driver ICs a must, or I can use some serial to parallel conversion IC + transistors (because of the high voltage level)?
Any tips to prevent cathode poisoning?

See the PDF link I posted above.

[Balazs]

I read the pdf and the methods presented are described nicely, but didn't find anything about cathode poisoning.

[deephaven]

This guy http://www.daliborfarny.com/ is making beautiful large Nixie tubes and does sell them.

However, I have to use this saying, "If you have to ask the price, you can't afford them", they ain't cheap.

[linux-works]

I'm getting very close to finishing this design and making a pcb for it.  if you want to build mine (or be a beta test for it when the pcb is ready) PM me.



its a little unique that its 'IoT' based (so to speak); it uses packet radio modules to relay time from a single source (in my case, gps based) to anything in the house, such as this clock.  it stays accurate with periodic 'beacons' and since its all message-based (and open source) you can expand it so that you can display any numeric on the tubes and it could be the number of unread emails in your inbox or it could be a count-down kitchen timer or anything else you want.

the tubes are in16 iirc, since they have the better looking numeral 2 and 5 (instead of the in14 that uses the same shape but just inverted, which looks quite dumb imho).

it will be arduino based and like I said, source will be provided so you can change things.

(currently, I'm thinking of breaking the board into 2 parts; a display board and an 'engine' board; the engine would be the same but you could pick a display that is led or lcd or nixie or oled and the functionality would still be the same).

[FlyingHacker]

I read the pdf and the methods presented are described nicely, but didn't find anything about cathode poisoning.

No. Sorry, was answering the first question only.

This was about the only thing I ever found on cathode poisoning... More about how to reverse it. But to prevent it you just need to use all the digits regularly. Naturally a clock doesn't lend itself to that unless you do some kind of rollover effect' or Matrix effect or something to get the other numbers used regularly. (E.g. 6 through 9 in the minutes column.)

http://www.tube-tester.com/sites/nixie/different/cathode%20poisoning/cathode-poisoning.htm

[linux-works]

here's what cath-poisoning looks like:



I used that clock (another of my nixie builds) as a regular clock with digits not showing 7,8,9 on the 10's place.  when I updated the firmware so it could show other numbers, you can see that some of the digit will not 'light up' properly.

I replaced that tube with another one but it also started to show the same problem and that's WITH cp logic enabled!   I have not found a definitive algorithm for what truly keeps the tubes working for all digits, but I tried cycling thru all the digits for about a second each, every 5 minutes that even THAT ws not enough.  if you defer the slot machine to just at night, it won't be enough, I fear.  my 5min routine was not enough.  I will continue to try new ideas and maybe keeping the 7-9 on longer and not even cycling the other ones is a better idea.

if you only use the digits for normal clock use, you don't need to care.  its only the unused number ranges that will look bad; the 0-5 digits will always be fine in clock mode even with CP happening.

[Towger]

As mentioned previously.  Are you better off if you drive each tube separately, rather than multiplex them?

[linux-works]

I didn't mux mine; I directly drove each one.  the current isn't very high and this isn't for battery use, so I didn't see the purpose in muxing them if I didn't have to.

I used 8bit port expanders for groups of 2 nixies.  you write bcd nibbles to them and those go to the usual 74141 or equiv russian chips.

[Macbeth]

Perhaps your nixie display driver needs to keep statistics on each tube and each digit, and when it goes into "screensaver" mode it exercises the unused digits over the others?

[FlyingHacker]

Or maybe, to really complicate things... the clock has a mechanical component to rotate the tubes through the places. So each tube spends a day (or a few hours) in the ones place and then moves on to the next, wrapping around. Perhaps this is done with mirrors. 

On a more serious note maybe the exercise routine needs to apply more current to the lesser used numbers, like the article discussed.

[linux-works]

Perhaps your nixie display driver needs to keep statistics on each tube and each digit, and when it goes into "screensaver" mode it exercises the unused digits over the others?

not too hard to do, actually.  as long as the clock controller has enough ram and code space to support that.  sometimes I fill up the flash with app code and there is not much space left for extras like that.

if you are going to use it as ONLY a clock, you probably don't need to care.  having bad digits that never turn on anyway is not going to affect the good digits.  the russian in16's that I use have been on for about 2 yrs constantly, now, and I see no signs of any problems on the 'clock digits'.  I have not tried the non-clock digits on THAT clock (the 6digit one I posted) but on the 4 digit one (using US tubes, not russian) those get poisoned very quickly and I did use that clock as a generic number display device and even after 2 weeks of 'clock use' I saw badness in the non-clock digits.  I would probably have a nightly routine that leaves the bad digits on for maybe half an hour at a time and cycle between only the baddies; the goodies get enough time during the day to be fine

if your tubes are in a socket, its not so bad.  on my builds, they are all hand wired perf and so replacing a tube is NOT something I look forward to.  if/when I go to a pcb, it won't be a problem.

I've also noticed that the direction you go, matters.  I had a semi-bad 7, iirc, and the top flat part of the 7 would come on if I went from 8 to 7 instead of 6 to 7 (I may have the specifics wrong, though).    I noticed this when I had a volume control (preamp system, DIY) send vol change messages to the clock and if I turned the knob up or down, I could see some of the dim segments come on or NOT come on, depending on what the previous lit thing was.  it is not really related to the numerical higher or lower, but the one that is next to the digit, in terms of metal and glass and gas   but I did see the thing act differently if hit from above vs below, so to speak.  so the lesson in that is, how you do the exercising matters as much as the duration.

if you find actual data on this, I'd love to see it; but I have not found a definitive guide on how to prevent CP on the nixies.  no one wants to say exactly what their exercise routine consists of and what they tried and found didn't work.