Poisoning already?

[dzing3rParticipant]

NCS314 2.2 8C, been working about a week and a half and already one of the digit tubes is showing possible poisoning on numbers 5,6, and 7. Only one though, the rest are fine including the 19A. Anyone seen this in a short timeframe?

[whatthetubeParticipant]

Same here. I bought 2x 6 tube clocks for which I built casings during xmas holidays. Gave the other one away as a gift and kept the second.
Clocks are set to shutdown the tubes at night, for 6 hours. They run 10s anti-poisoning once an hour.
1. The gift was used only few weeks after one of the tubes totally. Pretty much pitch black, only a slight glow at spots occasionally. I suspect the gas leaked out.
2. The one I kept has been used only for about a week and one tube has ‘4’ starting to fade out.

I did some basic voltage & current checking from one tube socket. I don’t have a scope and not sure if HV57708 can really be measured with a simple multimeter but current seemed quite low, only ~1.5mA.
Voltages seemed ok(ish), 177V without the tube and ~137V with the tube.

[whatthetubeParticipant]

Checked and the actual anode resistor on my NCS314-6 v3.4 board is 27k, so my measured values seem correct. In schema the resistor is even higher value, 39k and in some thread here I saw GRA&ASCH telling they should be 24k.
Still, as the rated max current of IN-14 is 2.5 mA, isn’t 1.5mA a bit too low and doesn’t too low current speed up the poisoning?

[dzing3rParticipant]

Since I posted that I have lost another tube. I too believe the gas leaked out at the bottom. Difficult to tell date code on these, since that has all been rubbed off. Mine were definitely used tubes. I’m pretty sure that the problem isn’t the resistor on the board, since I have just had two random tube failures and the resistor is the same on both of those tubes.

I bought NOS tubes and replaced them and have not have problems since.

[whatthetubeParticipant]

Yuphhh, quite disappointing. My dates have been fully wiped away too. I did ask gra&afch about this but they just offered new tubes for the normal price (25$ + shipping) i.e. no shits were given. I can get new tubes at half price elsewhere.

I like the HW design and the shared code (although it’s quite horrible mess) & 3d models but still, at these prices one would hope to get some warranty for these.

• This reply was modified 3 years, 8 months ago by  whatthetube.
• This reply was modified 3 years, 8 months ago by  whatthetube.

[whatthetubeParticipant]

About the resistor: if it is too high it can still speed up the poisoning (afaik). Like with proper current these could last decades vs with wrong one only few years so would be nice to get comments on that too.

• This reply was modified 3 years, 8 months ago by  whatthetube.

[dzing3rParticipant]

I didn’t even get a response when I inquired. I just went around them and bought new. Not worth my time fighting with them about it. I guess for the money I expected a bit more, but that is fine.

You are correct. The proper current must be maintained and I suppose it’s worth doing the math to calculate and verify. I have considered doing my own firmware fork of their messy codebase, but have not had time yet.

[whatthetubeParticipant]

Here’s my fork. Mostly just quick hacks but some fixes and refactoring too. Most of the functional changes listed in readme.
https://github.com/willmackey2nd/NixieClock

[dzing3rParticipant]

Looks good. Bonus points if you fork the 8C version and add your changes to that codebase. I don’t use GH anymore but use GL instead.

I have been meaning to add things like DST and comment what I can figure out as I dig through it but have not yet had time.

[whatthetubeParticipant]

If the points can be turned into vouchers to Laphroaig store, I may consider it 🙂
8C might need different masking and bit mapping (mapping is most likely same tho) but shouldn’t be too hard to copy and mod the changes.

The tube control bit lottery didn’t make much sense without reading the chip datasheet and seeing there are 4x16bit shift registers. I verified the schema by measuring few pins and made this matrix and simplified the bit assignment loop a bit. 8C version probably uses the same order so the same loop etc. most likely works.
Then there was some other weird masks e.g. for the dots that didn’t really do anything but confuse, at least in 6C version, so just removed some of them.

• This reply was modified 3 years, 7 months ago by  whatthetube.
• This reply was modified 3 years, 7 months ago by  whatthetube.

[dzing3rParticipant]

Ha that is fair . Maybe that could be arranged…

Would be neat to see someone make a wiki or information repository compiled somewhere about these things. Definitely have to go back through everything if you want to make changes as the methodology of design isn’t really easily seen just in the source. It would have been nice to get really detailed comments in the code but I suppose it’s at least good they gave a few pointers here and there.

[Author]

Posts