Strange Strobe Behavior on my Nixie Clock

[Dman777]

I have this strange behavior with my Nixie clock. The right column marker strobes like there is some kind of short. This clock has a timer function where I can turn off the tubes at a certain time. When the tubes go off, that column marker still strobes on. And also, when the 2 tubes to the right of that column marker are slightly strobing on and off with the column marker. 

What exactly is causing this? And does it mean that the circuitry may not last as long as the life of the tubes? 

Video 1:

Strobing column with tubes on:

http://youtu.be/HeQjneaF7AA

Strobing column and tubes when tubes are off:

http://youtu.be/KKrKTdNek6I

Thanks,

-Darin

[Nick]

Looks like a leaky driver... or could be a dry joint around it somewhere... or some surface contamination on the board... etc.

Nick 

[GastonP]

Or a glitchy output in the microcontroller pin that drives the erm... driver.
If I were to risk an answer... that clock seems to be multiplexed in a 3 x 2 way (3 anode drivers, 2 tubes per driver). I'd check the components around that positive (from B+) driver. It could be some  faulty transistor, or a wrong value resistor...
Of course an oscilloscope could be of great help, but beware of the high voltage.

Gaston

[gregebert]

I have seen multiplexed designs use a bias resistor to hold anodes at a voltage below ionization when they are not selected; in fact one of my Sperry / Beckman datasheets makes specific mention that this type of biasing is necessary.

Do you have a schematic ?

I've seen similar behavior with NE-2 bulbs years back; a faint erratic glow is visible when one lead is connected to the AC line (in my case, 120v /60 Hz) and the other is unconnected. I dont see this behavior with my nixie clocks (they are all direct-drive) because all tubes have exactly 1 cathode driven at all times.

Now that you've raised my curiosity, I'm going to poke around with some tubes on my bench to see if I can get a faint glow with only 1 terminal connected.

[Charles MacDonald]

On 14-07-24 07:57 PM, gregebert wrote:

> Now that you've raised my curiosity, I'm going to poke around with some
> tubes on my bench to see if I can get a faint glow with only 1 terminal
> connected.

Don't forget the old Science lab demonstration where the demonstrator
holds a 4 foot florescent tube near a Tesla coil, or even a van-de-grafe
generator, and it lights up in her hand. no wires needed.


--
Charles MacDonald Stittsville Ontario
cm...@zeusprune.ca Just Beyond the Fringe
http://Charles.MacDonald.org/tubes
No Microsoft Products were used in sending this e-mail.

[Dman777]

I contacted the seller and he said he would replace or refund the Nixie clock. The  only problem is that I have to mail it to Ukraine. Does anyone know what that might cost from the USA? 

I'm not a hardware person so what ya'll are mentioning is pretty foreign to me. Is this bad...what you all have mentioned? Is it likely the clock might die out in a few years? Or start a fire?

Thanks,

-Darin

[gregebert]

If it was my clock, I would just enjoy it as it is. From your earlier postings, it has a lot of nice features. Mine are just clocks; no alarm, no remote-control, and in some cases no battery-backup so I envy what you have. There isn't any fire or safety risk from the leaky digits in my opinion. Hard to guess if it will adversely affect the life of the tube.

I did do a few quick experiments on a few tubes with a single connection, and the glow is definitely related to the leakage-path. When I used a NON-isolated 325VAC supply, I could get a dim glow. When I made the supply isolated, it got even dimmer. Changing to 5,000 vac didn't make a huge difference. Maybe it's a corona discharge as contrasted to a full-on discharge when a lot more current is available.

[gregebert]

Just to clarify, from the 2 videos, the first one looks like it was intended to "blend" the seconds-digits as they change. It's not leakage because the digits are almost fully illuminated.

My guess is the designer actually went to some extra effort to make the digits behave this way; it's far easier to just make them "jump" to the next count.

The second video is probably some combination of leakage and short glitches where the anodes get pulsed on for a short time, perhaps a few microseconds.

[petehand]

Looks like a design problem to me. The proper way to blank a nixie tube is to remove the anode voltage, either by killing the HV power supply or disabling the anode drivers. If you leave the power on and the anodes multiplexing with no cathode grounded, the anode has some capacitance to ground, and the multiplex signal is effectively AC, so some small anode current flows and the gas glows. I had a cathode go open circuit on one of my clocks - it was the '8' of minutes - and when that digit was selected, the whole inside of the tube lit up with a dim diffuse glow. Being as it stayed like that for a whole minute it was quite noticeable. Since in this case it's entire digits lighting up, probably the cathode driver is breaking down. Does it use one of the Russian 74141 equivalents? They're only rated for about 70V and if there is no cathode grounded to pull the anode voltage down, considerably more than 70V can appear at the driver.

The clock probably won't die from this problem for years, but why take a chance? If I were you, I wouldn't use the timer function to turn off the tubes, since the tubes run continuously will last longer than a TTL chip being broken down by HV pulses on a regular basis.

[Darin Hensley]

Here is the diagram I tried to capture with my phone. With this helps any I'm trying to see what is causing it?

Sent From Samsung Note 2

[gregebert]

Well, it's multiplexed and there isn't any biasing for the anodes when they are off, so that explains the dim glowing digits.

[Dman777]

I was reading http://en.wikipedia.org/wiki/Biasing

and it states "A hot bias can lower the tube life span, but a "cool" bias can induce crossover distortion."...

Would this be considered hot biasing that could lower the tube life span? 

[gregebert]

For multiplexed nixie tube designs, the idea of biasing is to keep the voltage of the anodes for tubes that are off below the sustaining voltage. Most nixies will ionize around 170V, and sustain around 135V. So if you bias them at 85V (half the supply voltage if you run at 170V), then you are assured it will be off.

In the context of the wikipedia article you mentioned, the bias they refer to is the static current thru a vacuum tube,  and like nixie tubes, they will degrade faster at higher currents. I doubt biasing a multiplexed nixie tube to keep it off will affect it's life. Crossover distortion is irrelevant for nixie tubes because they are not used as amplifiers.

One reason I dont use multiplexed drivers in my designs is because the tubes are not on continuously, so I believe that a multiplexed nixie would be dimmer than a direct-drive nixie running at the same current. If additional current is driven into the tube to make it brighter (to compensate for the loss of brightness due to multiplexing), I would be concerned about reduced lifetime. I dont believe tube lifetime is a linear function of current; doubling the recommended operating current would probably cut the lifetime much more than 50%.

Multiplexing uses fewer transistors (which saves on component cost and PCB area), but that doesn't bother me because I can get transistors and PC boards very cheap; billions are made every year. Not true for nixies; I think Burroughs stopped making them about 40 years ago (I apologize for being such a zealot about brand-loyalty....) so the supply of nixies is finite and dwindling. Another reason I dont multiplex them is the datasheet advised against using long-life tubes in pulsed-mode, which implies not to multiplex them. Other tubes might be OK with multiplexing.

[Arne Rossius]

Hi,

Darin Hensley wrote:
> Here is the diagram I tried to capture with my phone. With this helps any
> I'm trying to see what is causing it?

The PC817 optocouplers are only specified for a collector-emitter
voltage of 35V, so it is very likely one of them is breaking down when
its tube is turned off. It might be a good idea to replace them with
optocouplers that can withstand a higher voltage.

Biasing to 85V won't work with these optocouplers, because
(175V - 85V) = 90V > 35V.

If the pin numbers in the schematic are correct, the optocoupler
transistors are also connected in reverse (pin 3 = emitter to +HV), in
which case the specified voltage is only 6V. However the position of the
pins makes me think that very likely, the pin numbers are incorrect and
the transistor is connected correctly (usually, pin numbers are counted
in counter-clockwise direction around the package, however here it is
1-2-4-3).


Best Regards,
Arne

[Dekatron42]

Try using TLP627 from Toshiba, they are usually used by these designers but they are more expensive so some use the cheaper and inadequate PC817 and hope that it will work. However if the designer has switched the base and emitter that is a design error. You can look at this document that has been around for ages on driving Nixies: http://web.jfet.org/nixie-1/NixieTransistors.pdf

/Martin

[Dman777]

I think I might of caused some confusion....when I say tubes are turned off....I mean.....there is a dimmer setting and you can dim them all the way to not being on. I see that weird flickering in the back even when they are turned on their brightest....but it's barely noticeable because the tubes are on full bright. Does this change anything in the diagnostics? Sorry, I'm pretty daft to this stuff. 

[Dman777]

I really like this clock because of the remote control for the LED lights.  Instead of sending this one back, which would be expensive...I was thinking about asking the seller if he could sell me another one at a discount. But, I would only want to buy one that was in  correct working condition. What are the chances the next one would be ok? Is this problem from design or a anomaly from a flaw in one of the parts?

Thanks,

-Darin

[Quixotic Nixotic]

Ask the brain is a very lifelike animatronic machine. Slip him a dollar bill and he will give you a minute of his wisdom.

John S

[Niek]

Like some others already mentioned, it looks like an issue with the design. So, ordering another one probably wouldn't fix the underlying problem. A "quick" thing you could try is replace the PC817 optocouplers with TLP627's (as mentioned above). Just buy them on aliexpress.com or something - you'll need 6 of them. (try these, you'll get 10 for about 5$ including shipping http://www.aliexpress.com/item/Free-Shipping-10PCS-LOT-Optoisolator-TLP627-1-TLP627-DIP-4/1406962252.html ) You'll have to do some soldering, but it may be well worth the effort! Consider it a fun project. I'd be extremely frustrated having bought a clock, to see that flickering all the time. 

Look, the problem with those PC817's is the max. 35V collector-emitter voltage, see this datasheet of the part: http://mkpochtoi.narod.ru/pc817_ds.pdf

You can see under "Absolute Maximum Ratings", the Vceo rating of 35V. Now, look in your clock's schematic, and you will see HV (175V) connected to lead #3 (which should actually be #4, but that could be an error in the schematic, hopefully) of the PC817. When the nixie is on, the voltage across 3 and 4 will be close to 0, but when it's off (it's off at least 5/6th of the time: your clock works by multiplexing; only one tube is on at a time, but because this happens really quickly, your eye will perceive them to be all on), it will be higher than 35V, meaning that the 'switch' (a transistor is essentially a switch) won't stay closed well anymore. If it's not closed, the nixie will light up. This is what you see as flickering. The TLP627 works the same as the PC817, but it has a max. Vceo rating of 300V, much more than the 175V you need. (see http://media.digikey.com/pdf/Data%20Sheets/Toshiba%20PDFs/TLP627,-2,-4.pdf)

Replacing those PC817's will be a good idea regardless, but it's possible that it's not the cause of your issue (although it would be the main suspect right now). There may be other issues, for instance, you may have microstrip crosstalk, caused by long stretches of high voltage (175V) traces on your PCB running alongside low voltage traces that lead to your PC817's #1 leads. If you post some high resolution photo's of your clock's PCB (the printed circuit board), we may be able to see this (as well as see if leads #3 and #4 are placed correctly or not), but you really need a scope to figure this stuff out. If this is the case, you could potentially fix that by bypassing some of these traces with wires. There may also be a problem with the software that runs on the PIC microcontroller, which you couldn't easily fix without the source, or writing it yourself from scratch. E.g. there may be issues with the order and slight delays in which the anode's and cathodes are enabled.

If I were you, i'd either send the clock back and ask money back, or consider it an interesting learning project and try some of the above suggestions.

Good luck!

 Niek

[jf...@my-deja.com]

 

One more suggestion. I had ghosting in some multiplexed B-7971s. I could affect the degree by varying the high voltage, and I eliminated it by increasing the anode voltage by about 15V above its original value. Try adjusting your +12V. I would feel pretty safe in the 10-15V range. 

[Dekatron42]

You could try to remove the two neon bulbs and see what happens with the Nixies, if the Nixies stop flickering it could be a software problem in the PIC-processor. You can also replace the 74141 (or K155ID1 - K155ИД1 ) and see if the problem disappears, sometimes they are a bit flaky and won't work correctly.

or, ask the seller if it is a hardware or software problem and if you can correct it yourself!

/Martin

[Dman777]

Actually, I suspected this also because they seem to be the source/the strongest point of the strobbing. If I clip a leg off of each neon bulb, will the LED's and Tubes still work as normal? I can't tell from that diagram if there is some kind of serial thing going on or not. 

Thanks,

-Darin

[Alex]

Looking at that schematic there are a few dubious things about this design, but one which strikes me immediately is the 16f678 proc running from 4MHz with no apparent other method of keeping time. Software RTC with a 4MHz primary clock source. Hows that working out accuracy wise?

[Alex]

Meant 16F628, sorry. That additional piggy back on the multiplexing to drive the neons AND the LED indicator through one transistor seems a bit dubious also, granted since the transistor is NPN and neither is powered at the same time there should be no issue but still seems a bit unconventional...