Nixie Tube Clock update! :)

[Mighty Burger]

Hi! I was wondering if you all could give some feedback on my circuit, and what can be done to improve it.

Here's some background info. A while back I chose to make a Nixie Clock as my high school senior project. I posted here asking for help and many people gave help generously. You can find that post here if you're curious. https://www.eevblog.com/forum/beginners/32-768khz-pierce-crystal-oscillator-for-nixie-clock/ The school cancelled the project due to Covid so it became my own personal project. I learned a ton working on this, largely thanks to you all. It was my first time trying to lay out a circuit board and I certainly made many beginner mistakes!

School started up and I got busy, and put off the project. beforehand I had one prototype up and running but there were lots of flaws with it. I made a second revision and almost finished soldering up the second board, but I tested it before soldering the tubes + power supply and I found problems. The fixes were pretty bodgy and I kinda put it off. Overall I rushed the development process and didn't catch design flaws.

After a year of college and a couple months working an EE-related job (that's a whole other cool story) I've learned a lot and wanted to revisit this project! Particularly, I took classes in C and C++ and I wanted to give embedded programming a try. (My first design just had discrete logic to drive the tubes.)

So!! In my off time this past week I drew up a couple schematics for the new design. Boy did that go faster than last time!!! Here's an overview.

- The circuit is built on two boards. The "bottom board" has the power supply and the microcontroller. The "top board" houses the Nixie Tubes and the shift registers used to drive them. A 2x5 header connects them, and they will also be mounted together via standoffs. I plan on laying out each board with basically all the components on one side, leaving the other side free for a nice big ground plane, with hopefully minimal interruption from signal traces. Heavily populating both sides was a headache with the last design, and I don't think that was good design practice anyways.

- I plan on picking up Sparkfun's AVR programmer doohicky and using that to program the chip. I thought it would be a good step up from using Arduino, which I'm already familiar with. I'd learn a lot more this way!

- I'm using shift registers for the microcontroller to output to, to directly drive the tubes. I was watching Dave's series on YouTube and he had success using something similar to the TPIC6B596. They're open-drain shift registers with 50V output clamping, so they should do the trick!

- I wanted to be able to turn off the digits. With clamped driving circuitry it's not a good idea to blank Nixies by turning off all the cathodes - it's necessary to drive the anodes. I'm using opto-isolators to drive the Nixie anodes. By shifting data to the shift register that controls these opto-isolators, I can turn on and off individual tubes. I can then send a PWM signal to the enable pin of that shift register to control the brightness of the whole clock.

- I did a little bit of learning on how switch-mode power supplies work. The theory isn't so bad, Dave had a really helpful video, but it seems like tweaking one to work well is tedious and difficult! Thankfully someone has gone through and done that tweaking already, and I am using what he found works well. It's called the "Pile O' Poo": https://threeneurons.wordpress.com/nixie-power-supply/
The design for his kit seems to be more up-to-date, so I based my design more off of that: https://threeneurons.wordpress.com/nixie-power-supply/hv-supply-kit/

Here are the schematics for both boards.

Bottom Board - Power Supply and Microcontroller:


Top Board - Nixie Tubes and Drivers (Apologies for the strange layout, I made it thinking about how the board will be laid out so it's backwards):

The dotted line there indicates that, if you want to only have a four digit clock, you can chop off the end of the board!

I had a couple specific questions I would like to ask:

- Is my use of the TLP388 appropriate for driving Nixie tubes? I don't know if I've seen another design that uses opto-isolators to drive the anodes and I'm worried there's a reason.

- I had some serious trouble decoding the pinout for the ATtiny84 regarding their use in ISP programming. Did I connect the right microcontroller pins to the ISP header pins?

Thanks! And sorry for being so verbose.

[james_s]

Why are you driving both the anodes and the cathodes? Anode switching is needed for multiplexing but shouldn't be necessary when you are direct driving the cathodes. I built a bunch of nixie clocks years ago and the only time I ever switched the anodes was in one that used bi-quinary tubes that have two anodes. The TLP388 ought to work, I just question the necessity of having it.

[Mighty Burger]

Why are you driving both the anodes and the cathodes? Anode switching is needed for multiplexing but shouldn't be necessary when you are direct driving the cathodes. I built a bunch of nixie clocks years ago and the only time I ever switched the anodes was in one that used bi-quinary tubes that have two anodes. The TLP388 ought to work, I just question the necessity of having it.

I did some testing with parts I had on hand before designing this circuit and found it to be necessary.

For my test setup I used parts left over from the last revision. Namely an IN-14 Nixie and the ULN2803 plus a 47V zener on the common pin. I used a 22k resistor on the anode. This setup should be nearly identical to the 50V clamping of the TPIC6B596 registers. I kept the anode attached to +170V but did not drive any cathode, and instead of the tube being blank like I'd expect, all the digits glowed a little and it looked like an ugly mess. I dropped the supply voltage to 165V, and still got that awful ghosting whenever no cathodes were driven. It wasn't until I dropped it to 160V when that ghosting stopped.

I believe this is caused entirely by that Zener clamp. A little bit of current flows through each cathode and it's enough to cause ghosting when no cathode is being driven.

I'm curious about how your circuits worked? Did you rely on zener clamping, or did you use drivers capable of handling the full voltage of the Nixie? If it's the latter it would make sense why you wouldn't see any ghosting.

[james_s]

Most of mine used MPSA42 transistors which are able to handle the full voltage. Some others used 8 channel driver arrays, I remember having that same issue when I tried using a clamping zener due to leakage through the zener. Untimately I omitted the zener and left the pin floating in a couple of them and have not had any failures in the ~20 years the clocks have been operating. In others I just never blank the tubes so it's never an issue. In the bi-quinary clock I used a MPSA42/MPSA92 pair to drive each anode. 

[Mighty Burger]

Update on my design! Parts arrived and I've done some prototyping on the breadboard. This time, BEFORE I order any circuit boards 

Good thing I did! Here are the mistakes I've made. I'm posting this in case anyone wants to use my design. I haven't gotten to test everything yet, mind, there could still be issues.

- The biggest flaw: the PNP transistor in the 170V power supply is backwards (Q1, bottom board). Oops! Before switching it, the circuit drew half an amp, and that transistor heated up a ton and started smelling bad.
After switching it, the only part that heated up was the 22k resistor I put on the output as a test load. I forgot power scales with the square of voltage, and poor guy was drawing nearly a watt and a half. The supply seems to work fine now though! Thanks to ThreeNeurons' for the design.
I have yet to test with Nixies.

- The RESET pin of the microcontroller needs a pullup resistor. (There's an internal pullup, but the datasheet says it isn't sufficient in noisy environments.)

- I plan on having a "sleep" or "off" mode for the microcontroller, where you can turn off the clock. It would be best to allow the microcontroller to cut power to the 170V power supply in this state. I plan on using some combination of a MOSFET and NPN transistor combo for this, just in series with the input to the supply. Have yet to design it. With this change, it would also probably be good to throw in a 10uF cap right where the 12V supply comes in.

- I tested the supply without the 10nF cap on the end (C10, bottom board), it seems to work OK. I don't think Nixies need super good regulation so you can probably omit this part.

One more thing. The bypass caps for the shift registers are really important. I breadboarded without them and the dang things wouldn't work. So, don't omit those parts from the schematic!

In the meantime I've been learning how to program AVR and it's been lots of fun!