Troubleshooting a mux'd clock

[Terry S]

Been asked to troubleshoot a clock, 6 tubes, multiplexed 2 x 3, that is 3 anode drivers for 6 tubes.Two of the tubes that share an anode driver are not lighting.

Seems pretty straightforward, repair the suspect anode driver. Standard high side NPN/PNP pair.

But when I put a scope on it, I do see the +170 seemingly switching. The waveform does look subtly different on the bad driver.

I'm scratching my head a little as to why I see switching HV at all on the bad tube anodes. Can anyone shove me off the plate and towards first base -- explain what I'm seeing? I feel a little stupid -- been troubleshooting over 30 years....

Terry

[gregebert]

Are you measuring the anode signal at the anode-pin of the tube, or at the driver ? Always possible something went awry on the board, etc (failed anode resistor, tin-whisker shorting between traces).

Is the anode supply adjustable ? Try boosting it a bit to see if the tubes are aging and need more voltage to ionize. Seems unlikely, though, because I would expect 1, not both, tubes on the driver to fail at the same time.

Are the tubes socketed ? If so, number them, remove all of them, then use a good tube to check each socket one-by-one. It will provide another clue.

I was thinking that one or both of the tubes that doesn't illuminate has an internal short; I've seen this with IN-1's though it was always cathode-to-cathode, which resulted in 2 numerals turning-on at the same time.

If the tubes are soldered, I would try disconnecting the anodes of both suspect tubes, then reconnecting them individually to see if one of them has an internal short.

[Terry S]

I measured at both points, the traces look the same. Anode resistors measure OK, as do the voltage divider and pull downs, etc. 

Tubes are not socketed, and I have not tried adjusting the anode supply. 

I suspect the PNP base is open, although it measures the same as the other PNPs in the other two anode drivers. If the transistor is unable to turn on, I expected to never see 170 at the anode. But it is there. It looks like it decays at a faster rate than the other drivers. And I never see the voltage drop as one would expect when the tube ionizes, which I do see on the other drivers. Tells me no current is flowing, hence no lit digits of course. My scope is rather primitive, so not as helpful as it could be.

The board is delicate, so I'm reserving my rework attempts for a well reasoned diagnosis and fix. 

Tonight I'll attempt to get some pictures of the scope waveforms.

Terry

[gregebert]

What might be happening in your clock is the PNP driver is damaged, and only allowing a tiny amount of current through. That can still result in ~170V across the tube with no visible glow. I wonder if the PNP's Vceo rating is less than 170V, and it suffered from long-term exposure.

--------

I've observed erratic behavior on 7971 nixies when the current is limited to a few uA. I was attempting to measure the ionization voltage, but the behavior was so erratic I cant get consistent readings. But I did notice the voltage was significantly higher in this pre-ionization phase, versus normal operation. For example, I would measure around 140V across the nixie at 100uA (dim segment); during pre-ionization I'd see 160-180V and faint erratic glowing.

I'm sure other nixies show similar behavior.

BTW, I used my nixie bench supply to plot I-V data in Excel for my 7971 nixies (7tubes * 15 segments * 7 currents = lots of data) and came across this behavior.

[Terry S]

Thank you for the suggestion... That is essentially what I was thinking as well, simply leakage current C-E. I've ordered 300V transistor pairs from Digikey.

Good news is the wife brought home a wonderful vintage Tek scope today, with a cart, -- digital, 150 MHz, early but with lots of capabilities for its' day. Model 2340A in perfect working order.  $5 as it was being dumpsterized soon... I may well end up with a few more. I think I can safely retire my 15 MHz analog Hitachi scope now.

Terry

[Dekatron42]

You might already have tried the ideas below, or already know about them but I mention them anyway as I have come across those faults.

You should make sure that you load the driver properly, otherwise it might not show its fault as small leakage currents will be present and show you the wrong voltage levels. If possible you should disconnect the two Nixies, you can do this by unsoldering the end of the anode resistor closest to the nixies and hook up a known good nixie to the anode resistor, wiring one of the cathodes to the same cathode on the nixies. This way you will know that you have a known good nixie there and if it still fails it must be some component in the driver chain, all the way back to the controller driving the NPN transistor.

You can also unsolder the resistor driving the base of the NPN transistor, the end closest to the controller and then hook it up to any of the other anode drivers, if the driver works then it is the controller.

/Martin

[Terry S]

I did trace the logic level signal from the CPU to the anode driver -- it's good all the way up to the NPN and past that thru the voltage divider (although harder to see) to the base of the PNP. So while I am *fairly* certain the PNP is the culprit, I haven't ruled out a tube yet. Resistors ohm out good.

Terry

[Terry S]

Transistors arrived today --- I swapped in a new PNP and still no dice. Since the tubes have individual anode resistors, I remain at a  loss as to what the problem is. Waveforms did not change.

On the working tubes, I can see the voltage drop across the anode resistor as the tube lights. No such drop across the anode resistors on the non working tubes that share the HV anode driver.

Could I really have 2 bad tubes, that just happen to share a driver? Would all the digits fail?

I'm tempted to apply my neon sign tester to the tubes to see if they ionize, but I'm a bit fearful about damaging the clock.

Any suggestions appreciated.

Terry

[gregebert]

OK, all of the obvious culprits have been ruled out so now we have to consider weird things.

Wish you had a schematic, but it's not mandatory.

1. With your scope, do you see similar on/off times at the base of each NPN 'predriver' ? It should be in the milisecond range. Trying to rule-out the possibility of the control logic 'skipping-over' the tubes that aren't lit.

2. This might be risky, but you could try momentarily connecting the anode of the bad tubes to the anode of a good tube. You may want to play it safe and use a resistor of a few K-ohms instead of a short. If all 4 tubes are off, there's a low-impedance path somewhere.

3. Connect a 10K resistor to +170V supply. This will not damage the PNP, and assuming it takes 150V to illuminate the tube, the 10K resistor will limit the current to a safe value of 2mA (1 mA per tube). It wont be a strong glow, but you will see it if the tubes are alive.

Connect a voltmeter to the other side of the resistor, and GND.

Next, momentarily touch the resistor (voltmeter side, not +170V side!)  to the dead tube's anode. You should see the 2 dead tubes glow, and see around +150V. If you see significantly less voltage, or zero, there is a short or low-impedance path somewhere. It could be the voltage divider, or a bad tube. You may need to unsolder one or both anodes to confirm

If the voltmeter still reads ~170 when you touch the tube, and there's no glow, there could be an open somewhere. You'll need to buzz-out the connections between the tube's anode pin, anode resistor, and PNP. If that still fails, then I see no other culprit besides the tubes.

[Paolo Cravero]

Hello.

Far from being an expert and I never troubleshoot these circuits. Yet.

Could I really have 2 bad tubes, that just happen to share a driver? Would all the digits fail?

I'm tempted to apply my neon sign tester to the tubes to see if they ionize, but I'm a bit fearful about damaging the clock.

So you have not tested the tubes if they're outgassed. Couldn't it be that they need a higher voltage to ionize? It was the case with a B5094 (symbol Nixie) that doesn't spring to life unless it sees to 210V, but it glows at the HVAC test.

If you disconnect the anode resistor and the whole board is left "floating" there should not be induced current and/or a closed circuit to be damaged. Unless your tester is very powerful: with mine I have to touch the glass to get the glow.

What is less risky? Higher voltage or the wireless neon sign tester?

Paolo

[alb.001 alb.001]

I have a nixie tester that I bought at a Goodwill store. There  are several shopping channel type sites which sell something called a derma-wand. It is AC powered and has a glass tip filled with neon which women use on their faces to improve their complexion.  It generates high frequency high voltage at a fairly low current.  I have safely used it on the terminals end of a nixie to ionize all the gas inside which glows orange everywhere and instantly tells if the tube is outgassed. I paid about $8 for it and use it every time I get a new nixie. It comes in a carrying pouch and I have never destroyed a nixie using it.  You should keep an eye out for one. 

If your wife has one you should take it away from her and never allow her to use it again. It generates ozone which can destroy delicate facial skin and it should be banned from being advertised this way.

--
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 email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CABj2VaaYKPVGNoz4R_yxT0A6FcsdQ%3DmPZ4C2ZyNkvYob3uzcog%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

 

[Terry S]

Excellent suggestions. Here is an update.

I do have the schematic -- however I do not have permission from the designer to publish it. I have ohm'd out the entire circuit, anode and cathode sides.

I did check the timing -- as you said, it was in the millisecond range, and no tubes are being skipped.

I finally did test the tubes with my sign tester. The working tubes lit up with a nice orange glow. The two non-working tubes lit just slightly, with a purplish glow. This told me there was something different about those tubes.

I extracted one of the non-working tubes from the board, and unlike a known good tube, I could not get the individual cathodes to light with the tester. So I am now convinced the clock has two bad or gassy tubes.

As of yet, I've been unable to install a new tube -- having trouble cleaning the solder out of the PCB holes. Not my clock, so I'm working very hard to not damage the PCB.

Thanks for the help -- I will post updates as I make progress.

Terry

[gregebert]

I'm always paranoid about ESD, so I would suggest removing tubes before testing them with something that generates an e-field. Most, but not all, CMOS devices have limited ESD protection that is intended to protect the device during assembly; certainly not for deliberately-induced charge.

If out-of-circuit testing is not possible, then I would leave the clock energized and try touching against the tube's envelope. I believe that a powered-on device will shunt induced charge better than when unpowered. I think the risk of ESD-induced latchup  in the driver IC is minimal, because you wont be touching the circuit with a charged device, only the tube's envelope.

[Terry S]

Another quick update.

Replacing the 10's hours tube worked fine. So I do indeed have two bad tubes. I've extracted the other bad tube (1's minutes) and will replace it this evening.

Once I got that going -- after tearing out what is left of my hair trying to get all 13 tube wires into the PCB -- I noticed that the decimal point on the 10's seconds tube was not lighting. Closer inspection revealed that not only was it not only not lighting -- it was missing all together. Looks like the connecting tab is there, but no cathode. No glow. Owner is not concerned.

Terry

[David Forbes]

Terry,

I am surprised that a Nixie tube with wire leads would not be socketed. That's
the easiest type of tube to put a socket on!

These tubes often are sold with a plastic alignment wafer that will align the
wires for you, if you hold the wafer against the PCB while pushing the wires
down through it.

On 11/17/2016 12:27 PM, 'Terry S' via neonixie-l wrote:
> Another quick update.
>
> Replacing the 10's hours tube worked fine. So I do indeed have two bad tubes.
> I've extracted the other bad tube (1's minutes) and will replace it this evening.
>
> Once I got that going -- after tearing out what is left of my hair trying to get
> all 13 tube wires into the PCB -- I noticed that the decimal point on the 10's
> seconds tube was not lighting. Closer inspection revealed that not only was it
> not only not lighting -- it was missing all together. Looks like the connecting
> tab is there, but no cathode. No glow. Owner is not concerned.
>

> <https://lh3.googleusercontent.com/-knlOkmTOlNQ/WC4DNOCnFDI/AAAAAAAAAEk/t0Nf0QLDVOAgyyJgft9B8aCPj6WIvE-kgCLcB/s1600/missing%2Bcomma.jpg>
>
>
>
>
> Terry


--
David Forbes, Tucson, AZ

[Terry S]

David,

They are also the easiest to solder without sockets... so I can see the builders' intent.

Some of the replacement tubes (used tubes) had the plastic spacers, but I couldn't leave them on the tubes, as there are components -- LEDs & resistors -- beneath the tubes. In fact 3 different LEDs below each tube. Quite a color show.

The replacement tubes had the wires cut, and of the 4 tubes provided to me, only 2 were usable because of the wire length. One of those had some horrible iron curtain glue all over the wire leads. Almost an hour of work to clean that off. No solvent I had handy would touch it.

Terry

[gregebert]

After you've removed both tubes, I'm curious how both will bench-test. I still cant figure out how 2 would fail simultaneously.

Check with the clock's owner to see if 1 tube failed first, and they used it like that for awhile, and then the second tube failed later. If that's the scenario, there has to be a circuit-wise explanation for the second tube's failure. Perhaps excess (ie, 2X) current ? Maybe a bad batch of tubes (compare date-codes) ?

I've only had 1 nixie actually fail on me, and it was due to a manufacturing defect (internal spot-weld broke). OK I lied, I had a bunch of IN-1's fail, but they dont count.....

[Terry S]

I replaced the 2nd tube last night. It tests bad in the same manner as the 1st -- a small purple glow from the neon sign tester, no cathodes light up. I installed a replacement tube and ran the clock overnight -- almost 24 hours now, and everything seems fine. The printing is gone from the tubes, I cannot compare date codes. The tubes look identical.

If the current was excessive, then the digits should be brighter than the others. They are not. And the anode resistors measure OK.

Curious note, the same owner has sent me another, identical clock to repair.... the saga continues.

Terry