I am working on a nixie tube clock design using IN8 tubes and a combination of MPSA42 and MPSA92 transistors to implement a multiplexed design of four tubes. The issue I'm running into now is that when I power the system on and drive one of the tubes' numerals, I get a slight glow out of the other tubes on the same numeral. My anode circuit consists of a mpsa42 connected to the base of a mpsa92 to enable my microcontroller to multiplex between tubes (see anode circuit attachment). The cathode side of the tube has each of the corresponding numeral cathode pins connected to an mpsa42 which is also controlled by the same microcontroller. My plan is to multiplex through the anode of each tube while simultaneously enabling the correct numeral with the cathode side mpsa42.

This all seems to work fine for a single tube, but once I install the other tubes I see that even though I'm driving the anode selects for the unused tubes to zero (IE driving only one of the anode's mpsa42), I'm measuring a voltage across the active numeral of the theoretically undriven tubes. I'm struggling to understand how this is happening, as from what I can tell when probing around, the PNP should not be on so no current should be flowing through it...

I have noticed that when I apply a voltage to a tube's anode (with a series resistor) and ground the cathode that all of the other cathode pins appear to have a voltage on them even though they clearly should be floating. Could it be possible that when I have multiple tubes in parallel and apply a voltage to one of their anodes that the common cathode pins are actually allowing enough voltage to flow back from an inactive cathode to the tubes anode and then back through the "active" cathode (I've tried to diagram this possible current path in the last attachment)? Is this possible with the fundamental nature of nixie tubes? I'm admittedly a bit stumped here as I believe I have seen other circuits of a similar design to mine for controlling nixie tubes, and when modeled in SPICE the circuit works fine (with the caveat that the simulation uses a purely resistive load).

Any help would be appreciated, or any feedback on my circuits since clearly I'm doing something (or understanding something) wrong!

Anode Circuit Cathode Circuit Annode-Cathod Loop?

  • \$\begingroup\$ If you think that path is possible, could you consider inserting a diode appropriately in order to test the idea? You should be able to block that ghosting path, if you are right. \$\endgroup\$ – jonk Oct 12 '16 at 7:11
  • \$\begingroup\$ I had thought about that, but I have this built up on a custom PCB and thought I'd try figure out of this was even possible before I cut up my board more than it already is (I've already found a few other problems in other areas...) \$\endgroup\$ – Wildwoodashes Oct 12 '16 at 7:40
  • \$\begingroup\$ How about adding a resistor from anode to ground (high enough value and wattage) to provide an alternate path? Just as a test? \$\endgroup\$ – jonk Oct 12 '16 at 7:43
  • \$\begingroup\$ @jonk The goasting path is not difficult to explain. Afterall the nixie tubes are neons. See my answer. Therefore with nixies the cathodes should NOT be combined. \$\endgroup\$ – Decapod Oct 12 '16 at 10:04
  • \$\begingroup\$ @Decapod Thanks. Appreciated reading the info. I knew these were neons and that they don't differentiate current direction. And this makes me a bit uncomfortable with the idea of multiplexing them, at all. I'd expect spattering to lower the lifetime ... perhaps a lot. I don't think I'd even try muxing them. Especially at the prices charged for some of these. \$\endgroup\$ – jonk Oct 12 '16 at 23:00

enter image description here

Your approach is based on the model to drive led matrix or 7 segment displays with all the cathodes combined. However a nixie is in principle a neon lamp. As soon as the voltage over the elements (single or in series) becomes high enough the digit(s) will be activated. This makes separating the tubes from each other absolute necessary.

There is a more simple and straight forward way by driving each nixie with a 74141. Then the multiplexing is done at a TTL level. This way is a proven methode and should work without any problem.

  • \$\begingroup\$ I believe you might be onto something. As a test I've disconnected the PNP transistor from one of the tubes that was showing the unactivated glow such that it should "theoretically" have no source for current, but it's still glowing faintly on the selected numeral (cathode). Based on this, the only possible source of the current IS from the cathode of a different numeral, back to the anode, and then back through the cathode of the selected pin (as attempted to be shown in my third image). Does this sound possible? Is my only option to separate the different tubes cathodes? \$\endgroup\$ – Wildwoodashes Oct 12 '16 at 17:33
  • \$\begingroup\$ Yes I believe so. You could do a very simple and more ridgid test by connecting two nixie cathodes of the same tube to your source. Dont forget the serial resistor. If they glow then you do not have to look furter and you have to change your system. \$\endgroup\$ – Decapod Oct 12 '16 at 18:40
  • \$\begingroup\$ I'll give that a try. Thanks for the idea. Do you know if the 74141 (or Russian K155ID1) uses a standard 16pin PDIP package? I'm evaluating if it'll fit on my board, but having trouble finding a good source of information about the footprint. \$\endgroup\$ – Wildwoodashes Oct 12 '16 at 19:56
  • \$\begingroup\$ Standard 16 pin dil package. There is plenty of info in the internet. Also look at the picture in my answer. \$\endgroup\$ – Decapod Oct 12 '16 at 21:08
  • \$\begingroup\$ @Wildwoodashes Please also upvote my answer \$\endgroup\$ – Decapod Oct 13 '16 at 4:47

Multiplexing nixie is not impossible neither difficult . As for previous comments the nixie accumulate energy like a capacitor . When the cathode is switched off anything is connected to it in addiction to the open collector is receiving this energy . The strategy is really simple . The multiplexer scanning never should be over than 150 Hz . Too low freuqencies lead in the 'flickering ' in my case appear around 50 Hz . The scanning pulse must be not 100% of duty cycle but less than 70% . You must leave the time to the nixie to discharge its energy before the nex is switched on . A resistor of 200- 300 Kohm before the series anode resistor (mean on the anode driver collector) connected to ground will help the nixie to speed up the discharge . If you look with a scope the anode pulses one channel nixie anode 1 second channel nixie anode 2 you should see a positive pulse dropping before the next will rise . This gap clear completely any ghost number . A light intensity drop is to be considered after with this application but you can check the voltage drop on the anode series resistor and reduce the value in order to increase the anode peak current at the limit allowed . It will be very short as will drop very fast . To drive the cathode I use the SN75469 and to the pin 9 is connected a 100V 1W zener diode in order to protect the the darlingtons . Not only the zener improve the performances as using the classic MPSA42 in open collector configuration the ghost effect increase .


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.