all my Nixie tubes are dying :(

[Richard M.]

I bought the pv electronics SPECTRUM 18 Nixie Clock Kit, which, I think, is of high quality and does not do multiplexing.

The electronic is working fine.

But I have an other big problem: all my IN-18 Nixie tubes are dying, one after the other.

I already bought 10 tubes from two different sellers on eBay (which look reliable)

All tubes were working fine, but after only several weeks, dark spots begin to appear, until the Nixie becomes totally dark.

I checked the supplied power and everything looks fine: (is it?)

170V without a tube connected

with tube: 130V drawing 4.7mA

power is clean and without any noise as to my oscilloscope

Slot machine effect is running every 10 Minutes. Clock is off for 6 hours every day.

[Nicholas Stock]

Richard, that's odd. The clock pictures look like an old version of the Spectrum LTC and not the newer version (the older one is multiplexed, the latest version is direct drive..)....saying that though, I have one LTC that has been on for over 3 years and only one tube has died (rather odd failure as well...different story)....are you sure the tubes you purchased were NOS and not used (but de-poisoned by an unscrupulous seller...)?

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+unsubscribe@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/943a03c5-c975-4f86-853c-be5b5a9709d7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Richard M.]

I bought the kit in April 2016, so it's definitely direct driven.

See a photo of the pcb attached.

That's one of the sellers:

http://www.ebay.com/itm/222522723935

that's the other (the auction is no longer accessible on eBay)

https://feedback.ebay.com/ws/eBayISAPI.dll?ViewFeedback2&userid=makdal2013

I developed a "Nixie Tube de-poisoning revive machine" :)

(the one on the picture in my first post)

I can adjust voltage and current and set up the routine from the browser, which digits to revive (how long, how many loops, etc.)

This helped, but didn't last long.

Am Dienstag, 3. Oktober 2017 16:38:13 UTC+2 schrieb Pramanicin:

Richard, that's odd. The clock pictures look like an old version of the Spectrum LTC and not the newer version (the older one is multiplexed, the latest version is direct drive..)....saying that though, I have one LTC that has been on for over 3 years and only one tube has died (rather odd failure as well...different story)....are you sure the tubes you purchased were NOS and not used (but de-poisoned by an unscrupulous seller...)?

On Tue, Oct 3, 2017 at 7:31 AM, Richard M. <vis...@gmail.com> wrote:

I bought the pv electronics SPECTRUM 18 Nixie Clock Kit, which, I think, is of high quality and does not do multiplexing.

The electronic is working fine.

But I have an other big problem: all my IN-18 Nixie tubes are dying, one after the other.

I already bought 10 tubes from two different sellers on eBay (which look reliable)

All tubes were working fine, but after only several weeks, dark spots begin to appear, until the Nixie becomes totally dark.

I checked the supplied power and everything looks fine: (is it?)

170V without a tube connected

with tube: 130V drawing 4.7mA

power is clean and without any noise as to my oscilloscope

Slot machine effect is running every 10 Minutes. Clock is off for 6 hours every day.

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

[Nicholas Stock]

Cool. I've bought tubes from the first seller also (not IN18's) and they were good....that's a pretty fancy de-poisoner! You should consider putting that out there...I'd be interested...;-)

As for the continual failures, not sure at the moment what to suggest, certainly appears to be more than bad-luck....I've got a few of the recent Spectrums and haven't seen that behaviour...have you contacted Pete at all? He's usually more than happy to help troubleshoot...it's almost as if the tubes are being under-driven...

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+unsubscribe@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/d922f115-36f6-46ab-8fec-92eb6e42bff5%40googlegroups.com.

[Terry S]

Hard to tell from your pictures, but it looks like cathode poisoning..... I would try to manually de-poison a few digits individually, see what results, you have nothing to lose. Then I would run the clock at max brightness if it is adjustable, and see if the poisoning returns in short order. If the high voltage is adjustable you might want to tweak it up to 180v. 

I did have one tube where the digits faded over a period of weeks, it was not poisoning but rather a gas pressure leak I presume. I replaced that tube and it has not reoccurred.  

I don't believe the slot machine de-poisoning effects are useful, unless the current is somehow increased dramatically during the effect. The "on" time of the unused digits is just too short.

Terry

[Richard M.]

>Cool. I've bought tubes from the first seller also (not IN18's) and they were good....that's a pretty fancy de-poisoner! You should consider putting that out there...I'd be interested...;-)

Yes, I put a lot of work in it.

You can switch it between manual and automatic mode.

In manual mode, you can set voltage, current and digit directly on the device.

In automatic mode, the digits are controlled by an Arduino compatible ESP8266 WiFi "Wemos D1 mini pro".

I can set, e.g. run digits 2,3,5 and 8 for 90 sec. each, loop 10 times and then send me a push notification to my mobile phone.

After the routine, I can set it to turn off or cycle through all digits 0-9 infinitely with a 10s interval.

But the time to build this thing is nothing someone would pay me. :)

And for mass production there are too less people interested in such a thing.

There are already similar devices on the market at a reasonable price:

http://www.ebay.com/itm/262483292894

>have you contacted Pete at all?

Yes, I already contacted Pete, that's what he replied:

This is classic cathode poisoning. IN-18 tubes suffer more than most tubes in this respect. The problem is that some batches are perfect and run for years without issue, and others degrade very quickly.

I just wonder why this happens to nearly all of my tubes.

>it's almost as if the tubes are being under-driven

As mentioned, they only draw about 4.5mA, which might be too less, as the minimal current, according the the datasheet is 4mA, nominal is 6mA and maximum 8mA.

What I notices is, that soon after I purchased my kit, the anode resistor was changed to a lower value, from 10kOhm to 8.2kOhm.

>a gas pressure leak

That's what I also thought. But, again: all of them?

>you might want to tweak it up to 180v

Are you sure? As what I read, you should never drive the IN-18 Nixies above the specified 170V.

[gregebert]

Are all numerals inside the tube dying, or just pieces of some numerals. Cathode poisoning should only affect seldom-used numerals.

If all of the numerals in a specific tube are failing, I can think of 2 causes. #1 would be a leak. Check for cracks around the pin base. Are the tubes snug/difficult to insert/remove from the socket ? If so, there could be pin-stress that's breaking the glass-seal. #2 would be a power-supply issue. Can you bench-test your tubes at a higher voltage ? I recall my IN-18's were glowing nicely around 140V; the ionization voltage is a bit difficult to measure, and I was measuring around 165-175 volts. It's a long-shot, but you may have some tubes that need a few more volts to ionize as they age.

My IN-18 clock has 14 tubes to display date and time (MM.DD.YYYY   HH:MM:SS format) . The tubes on the left are basically static, which is a bad thing to do. Every night, I run a depoisoning routine for 1 hour to display all numerals on the static tubes and I see no signs of poisoning on any tube after almost 2 years of usage. The display is on for about 16 hours per day.

[Richard M.]

At first, the unused digits begin to show poisoning effect (e.g. digits 6-9) of the first minute tube.

Sure, that's nothing strange as they never light up.

Then the other digits get dark spots with gets worse and worse until all digits are black.

I will now change the anode resistors to values that the tubes draw ~6mA instead of only 4.5mA.

Because, as already mentioned here, it's not necessarily good for the tubes to be driven with very low current.

[Terry S]

You can apply a higher voltage -- the datasheet only specifies the guaranteed ionization voltage. 170 is "guaranteed" to be the maximum required voltage for ionization. Maximum applied voltage is 200 according to the datasheet.

Jeff Thomas modified the Nixichrons to run at 5 ma when some clocks showed poisoning.  But 6 to 7 should be safe as well.  Tube life will be shortened, but these tubes will still last decades. 

Terry

[gregebert]

Richard - What are the date-codes on your tubes ? I recall hearing problems with IN-18 tubes manufactured in 1982 (82xx date-code).

I have one from 1977, another from 1986, and the rest are from 1989/1991

[Paul Andrews]

Make sure your wall wart is pumping out a steady voltage close to that required - you would need an oscilloscope. I have heard that this can be a problem for these clocks.

I’m also not a big fan of running tubes near the bottom range of their current specs. It seems to lead to more cathode poisoning. There are various threads where this has been discussed in great detail - with evidence.

Apologies if this all sounds like hearsay, but it is based on the conclusions of similar threads.

[Dan Hollis]

Is the way the tubes are being driven possibly a factor?

I have a tubehobby kit and my IN-18 have been constantly running for
almost a decade and the digits still look basically brand new.

It multiplexes them at a relatively low frequency so there's an audible
whine from the tubes, the only downside to the kit.

-Dan

[Richard M.]

The tubes are from 1983 and 1986. From both batches, ones working fine and others gone totally dead.

I already checked the voltage with the oscilloscope and it's stable and totally clean, without any noise.

[Jeff Walton]

There was another thread about the IN-18 in a Spectrum clock where the wall wart was not outputting sufficient power and replacing it with a 2A supply took care of a persistent poisoning problem.  Running a nixie with insufficient drive will definitely poison all of the cathodes because when they run at spec, the cathodes are able to drive off deposits as part of normal operation.   

Driving the IN-18 at 6 mA is not unreasonable. 

Try changing your wall wart supply first and do the anode resistors second.  If changing the supply shows ANY difference in brightness, then you may have solved the problem.

I have a couple Spectrum clocks with IN-18 and have 3 years on the oldest clock with no tube issues. 

Jeff 

-------- Original message --------

From: Paul Andrews <pa...@nixies.us>

Date: 10/3/17 1:28 PM (GMT-06:00)

To: neonixie-l <neoni...@googlegroups.com>

Subject: [neonixie-l] Re: all my Nixie tubes are dying :(

Make sure your wall wart is pumping out a steady voltage close to that required - you would need an oscilloscope. I have heard that this can be a problem for these clocks.

I’m also not a big fan of running tubes near the bottom range of their current specs. It seems to lead to more cathode poisoning. There are various threads where this has been discussed in great detail - with evidence.

Apologies if this all sounds like hearsay, but it is based on the conclusions of similar threads.

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

To post to this group, send an email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/7244521f-6f56-40c3-942a-91dd71147d4e%40googlegroups.com.

[Dan Hollis]

On Tue, 3 Oct 2017, gregebert wrote:
> If all of the numerals in a specific tube are failing, I can think of 2
> causes. #1 would be a leak. Check for cracks around the pin base. Are the
> tubes snug/difficult to insert/remove from the socket ? If so, there could
> be pin-stress that's breaking the glass-seal.

The only tube failure i've had was from broken seal when I used to rotate
tubes, due to stress on pins. I don't rotate them any more.

I was thinking of making an IN-18 adapter board so I can rotate the tubes
without stressing the tube pins.

-Dan

[gregebert]

Next-step is to bench-test one of your tubes that has failed to see if it can be depoisoned, or determine if it's truly dead. Do you have a supply that goes up to around 200V ?

What is the multiplexing frequency ? 

Years ago, my neighbor who was an EE told me about an ultrasonic cleaner he was developing back in the 1960's. It worked great, but after a few months it would fail. They found out that the bondwires on the transistor reached mechanical resonance at the same frequency as the ultrasonic cleaner and failed from mechanical fatigue. They literally had to have the transistor manufacturer create a new transistor with different mechanical properties. End of problem. Could the multiplexing frequency be causing micro-vibrations inside the tube, leading to glass-seal leaks ? Who knows. One thing I've learned is never to rule out ridiculous-sounding causes of failure, because they do happen.

[Nicholas Stock]

Richard, it sounds like your 'de-poisoner' is a lot more capable than the one you listed (I have one already, but as has been discussed before, it has it's limitations...).

I'd up the voltage on the clock (I think the max is somewhere around 184V for the Spectrum 18) and see if that helps...hope you get it sorted, that's a real pain in the proverbial....

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+unsubscribe@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/1806bdf6-2208-43a9-a87e-e1e81c4b47fe%40googlegroups.com.

[Richard M.]

You could buy this one:

http://www.ebay.com/itm/262488260835

so you would already have a housing, a 125-180V high voltage supply and an IN-18 socket.

Just add an Wemos D1 mini pro (https://wiki.wemos.cc/products:d1:d1_mini_pro)

And voila, you've got your own, "app controlled" Nixie de-poisoner. :)

I can give you my source code, if you like.

I just checked my power supply. It's working fine and is rated up to 1.25 Amps. (and delivers!)

The whole Nixie clocks draws 725mA max, including an ESP8266 acting as an GPS emulator. (gets the time and date from NTP and outputs an GPS NMEA sentence the Spectrum clock syncs to)

I changed the anode resistors from 10k to 8k2.

As expected, the digits are brighter now.

Hope, this helps.

I don't wont to give it up. It's just too beautiful and nostalgic, this pretty glowing clock. :)

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.

[Tomasz Kowalczyk]

There was once a thread on this group, where this topic was discussed - 1. will undercurrenting tubes actually decrease their life and 2. does multiplexing affect tube life and how. As far as I remember, the poisoning problem of undercurrented tubes was mentioned there.

I'd like to get the code, too. Not only I'd like to have a nixie healer, I also just want to start using a Wemos board I purchased some time ago, but never had the guts to start actually playing with it.
BTW. The clock will draw this current on average, but it will do so in a sawtooth form due to boost converter. You'd need a very large capacitance on the board to smoothen it.

However, you have measured the voltage and current of the nixies, right? If these two parameters are correct, then it means the boost converter is working properly.

[Manuel Azevedo]

Hi,

Like Jeff and Tomasz have said, having an underpowered PSU can cause this.
I had this issue with a PV IN-18. At the same time it didn't help that I had two tubes that degassed at the same time, so troubleshooting took more time.
If you haven't yet, try to buy a good power supply (don't order on eBay, order from Amazon, Digikey or Mouser) with 2A minimum... And see if it makes any difference.

[Richard M.]

A 2A power supply is definitely not needed.

0.008A (per Tube) x 130V x 6 (Tubes) = 6.24W 

6.24W / 12V = 0.52A

+ power for the rest of the electronics, a 1A power supply fits best.

I'm using this one (rated 12V 1.25A) https://www.pvelectronics.co.uk/index.php?main_page=product_info&cPath=10&products_id=161

I'll clean up the code a bit and post the Arduino source code of my Nixie Tester as well as my GPS time emulator the next days!

[threeneurons]

 

What kind of sockets are used ?

I don't like using those socket pins that soldered directly into the PCB, with glass based tubes. Okay for older style octal based tubes. I like the pins to float. Especially for pricier tubes, like IN-18s. Could be losing gas.