Nixie tubes too low brightness

Hello everyone,

I am working on an Arduino project involving nixie tubes and I have been stuck with a tube brightness problem for about 10 days. I would greatly appreciate any input on how to solve this issue. I apologize in advance in case I made some dumb mistake, I have some experience with Arduino projects and soldering circuits but I am no expert with electronics.

As i was saying, I am building a nixie clock using eight IN 18 nixie tubes and an Arduino Mega. The tubes are directly driven (no multiplexing, at least for now), each with a K155ID1 driver.

A 12 V power supply is used for both powering an Arduino Mega and as input for a high voltage DC-DC transformer (NCH6100HV). The voltage supply for the anodes ( U_pot ) can be modified through a potentiometer placed on the transformer. In my case I set 170 V before connecting the tubes. A 330 uF capacitor was placed between the 12 V and the ground between the power supply and the transformer. The anodes of the eight tubes are connected in parallel to the high voltage supply, each of the anodes connected through a 10 kohm resistor ( R_an ). The choice of supply voltage and resistors was made based on examples of similar clocks [1][2].

After some tests on a breadboard, I have soldered the electronic circuit using a few stripboards. Pretty much all the connections have been soldered and the tubes have been tested and observed to be working correctly: all digits are correctly displayed, the anode current is within the datasheet range [3] and the tube brightness is appropriate).

These tests were all carried out for a single tube to verify I had correctly soldered a tube socket. However, when I inserted all the eight tubes in the sockets to verify if the clock was working properly, I noticed that the digits were not completely lighted and that the lighted parts were not as bright as when the tubes had been individually lighted. I measured (see Table 1 in the attachments) the current draw out of the transformer ( i_down ) while inserting more tubes in the sockets and noticed that the current per tube ( i_down/N ) decreases from 4.9 mA for one tube to 1 mA for eight tubes. The current upstream the transformer ( i_up ) changes very little.

I tried changing the supply voltage U_pot (see Table 2 in the attachments). However, voltage downstream of the transformer seems to drop at « running » conditions, when current flows through the tubes, to a similar value ( U_run ) no matter the set value of U_pot. Both U_pot and U_run are measured at the HV and GND pins of the transformer.

I read a bit regarding nixie tubes’ physics and, if I understood correctly, this is normal. As long as the voltage is sufficient to lighten the tubes, current is what matters for brightness and voltage will depend on the tube voltage drop and selected resistors.

Since changing the supply voltage was not helpful, I tried changing the resistors R_an. In the beginning 15 komh resistors were used. Decreasing to 10 kohm only slightly changed the current through the tubes, with i_down being 7.8 mA instead of 7.4 mA with eight tubes.

I have read [4] that undervolting nixies can lead to long delays for the tubes to light up but I do not think this is the case. I tried waiting around ten minutes with four tubes in the sockets and the measured current i_down seems to remain the same.

I would like to have a current between 4 mA and 6 mA in each tube to comply with the datasheet indications but I am out of ideas. Further decreasing the resistance below 10 kohm may help but I do not see why it is not already working when the parameters I am using now correspond to what has been done by other people.

Once again, I would appreciate any help you can provide.

[1] http://www.mindspring.com/~tomtitor/IN-18.html
[2] https://gra-afch.com/content/uploads/Schematic_Nixie_Clock_Tube_NCT318-v1.1.png
[3] http://www.tube-tester.com/sites/nixie/dat_arch/IN-18_03.pdf
[4] neonixie-l@groups.io | Undervolting Nixie Tubes

Tables.pdf (23.8 KB)

Hello Zurent,

Please will you post a schematic, ideally one annotated to show were you measured what voltage and current. Maybe someone else will be able to make sense of your description, but it just hurt my head trying to understand your circuit based on text.

Thank you.

What is the voltage output now from your "170 V" DC-DC converter? It would appear it is not working, most likely because your 12 V power supply is not able to provide sufficient current (more than half an Amp). So check your 12 V voltage.

Thank you for your replies.

I have drawn a schematic of a case with 4 tubes to help understand how the circuit works and where the mentioned quantities were measured.

Paul__B:
What is the voltage output now from your “170 V” DC-DC converter?

As you can see from Table 2, I have tested a few different voltage values. Most of the tests were made for 191 V. Since modifying this voltage wasn’t very helpful, I am now back to 170 V.

Paul__B:
It would appear it is not working, most likely because your 12 V power supply is not able to provide sufficient current (more than half an Amp). So check your 12 V voltage.

That is something I thought as well when I saw that the current i_up remains pretty much the same for any number of tubes. However, the power supply should be able to deliver enough current, being rated for 2 A. I attached a picture of the power supply specifications for clarity.

12V_Supply.pdf (237 KB)

NixieCircuitSchematic.pdf (112 KB)

I know quite well how the circuit works. :grinning:

If your “Urun” is in fact the voltage measured at the output of the DC-DC converter, the common point of all the anode resistors and reduces when you connect more tubes, it is indicating that your DC-DC converter is simply not capable of powering the Nixie tubes, no more no less. And the fact that iup does not increase significantly with more tubes confirms this. Clearly the power supply current should be directly proportional to the number of tubes connected, that is to say twice as much for two tubes as one, three times as much for three tubes and so on.

A proper DC-DC converter with a potentiometer to set the voltage will give the same voltage when you connect your Nixie tubes with their resistors as with nothing connected, and the same with the maximum number as with one. Failing to do so is not in any way “normal regarding Nixie tubes’ physics”; is is simply not working. :astonished:

Hello Zurent,

Thanks for the schematic, much more helpful. The answer Paul__B has given you confirms what I suspected, so there is nothing for me to add.

I'm interested to know more about your project as a Nixie tube clock is something I might get round to building one day.... Some photos would be interesting :slight_smile:

Not sure it is a clock! Eight digits!

(Do you really need to know the date as well?)

Paul__B:
Not sure it is a clock! Eight digits!

(Do you really need to know the date as well?)

My best friend calls me Tempus because I have clocks all over the place and am too concerned (in his opinion) with time :slight_smile:

Surely 8 tubes gives HH:MM:SS

Yes, I have some "colon" tubes here - I think! Possibly a box of them.

I suspect this is not what the OP is proposing.

Zurent:
I am building a nixie clock using eight IN 18 nixie tubes and an Arduino Mega. The tubes are directly driven (no multiplexing, at least for now), each with a 5IDK151 driver.

OK, he says it is a clock, so why eight tubes I do not know. :roll_eyes:

Typical overkill with a Mega 2560, and you do not multiplex Nixie tubes themselves.

Presumably the Russian 7441 clones were supplied with the tubes. :astonished:

As Paul__B pointed out, the transformer was not doing what it was supposed to.

Before giving up and thinking about replacing it, I thought that maybe the transformer was not capable of delivering the expected voltage because of a too small input voltage. Therefore, I changed the 12 V power supply with a 20 V supply. Much to my delight, the transformer is now able to maintain the 170 V with current flowing through the tubes. The datasheet indicates 12 V as a “typical” input but values in the range 10 - 24 V are acceptable. With the current parameters, i_up is about 260 mA.

Thank you for pointing out about the behavior of the transformer, it really helped.

Paul__B:
Not sure it is a clock! Eight digits!

Well, technically it’s a divergence meter, a prop from a japanese animated series from ten years ago [1][2]. Eight tubes are necessary for the prop but six would, of course, be sufficient for a clock. I have the impression that a good part of the attention nixie tubes have got in recent years comes from fans of the show. It certainly was the case for me.

PerryBebbington:
A Nixie tube clock is something I might get round to building one day… Some photos would be interesting :slight_smile:

Sounds like a great idea. I have attached a couple of pictures of my project. I hesitated posting them in the beginning because the clock right now is mostly a big mess of cables. At the bottom floor I placed the Mega 2560, a RTC DS3231 clock and the transformer. Above is the main circuit with the drivers and blue LEDs for illuminating the tubes from below.

Having the possibility to work with PCBs, or simply dedicating a bit more time to the design phase, would probably lead to a cleaner result. I could post some better pictures when the project is a bit more advanced.

[1] Divergence Meter | Steins;Gate Wiki | Fandom
[2] http://www.mindspring.com/~tomtitor/IN-18.html

DivergenceMeter.pdf (812 KB)

Zurent,
Cheers for the photos! Nice piece of work! :slight_smile: ++Karma;

I do have a concern though, in the photo in the pdf document you appear to be using hardboard as an insulator. Wood is not a very good insulator, especially for the voltages needed for nixie tube. See Lichtenberg figures

Also, you don't need a stack of nuts on your bolts like that, you just need a nut at the bottom next to the bolt head then 2 nuts, one either side of the wood, tightened against each other at the end of the bolt.

PerryBebbington:
I do have a concern though, in the photo in the pdf document you appear to be using hardboard as an insulator. Wood is not a very good insulator, especially for the voltages needed for nixie tube.

Thanks for the heads up PerryBebbington. The hardboard comes from my local hardware store. I think they also offer to cut plexiglas, I could try with that.

To be honest, with no calculation, I can't be sure that hardboard will not provide sufficient insulation, or even that plexiglas will be good enough, for that matter. In any case, I don't think I can get my hands on a better insulator than a plastic material, so it will have to do for now.

Thanks again

I can't be sure that hardboard will not provide sufficient insulation

Wood is largely carbon, carbon is a conductor.... (unless it's in the form of diamond).

Plexiglas, also know, I think, as perspex, is fine (also made of carbon! Any chemists here to explain please?).

Perspex - "Plexiglas" - is an excellent insulator, perfect for such an application.

Wood is a porous organic material, absorbs moisture even from the air to a varying degree so has limited and varying insulation. Similar materials are used in the DHT22 humidity sensors often used with Arduinox.

Diamond when pure, is an insulator with the same crystalline structure as silicon, so if it could be fabricated and "doped" with similar elements in the same way as silicon semiconductors, could be used in the same fashion. There will be researchers still working on this potential, graphene is of course, an excellent conductor.