Help request for custom Nixie clock

[Florian van der Dussen]

I’m a 21-year-old Mechanical Engineering student and have always wanted to own a Nixie clock. Since buying a fully assembled one is out of my budget, I ignorantly figured, “How hard can it be to build one myself?”

Well… turns out, a bit harder than I expected.

This is one of my first real electronics projects, and I’m designing a custom PCB despite having almost no electrical engineering background. I’ve completed my first version of the board, but I’m honestly nervous about ordering it. The idea of running 170V through something I designed feels like a recipe for frying every single thing I have on my desk.

I’m keeping it as simple as possible: just a hh:mm:ss display. I did add a light sensor to dim the display based on ambient light (saw someone do this in a video and thought it was neat). I used KiCAD and autorouting for most of the tracks except the 170v net. I’m trying to make the casing compact and clean, aiming (or better called dreaming in this case) towards something close to the Puri Nixie Clock.

Am I missing anything critical? Are there any other sources that I can look at? I am expecting a lot to be wrong, so any help and tips are welcome! 

Current PCB

Current schematic:

[Florian van der Dussen]

My apologies, the images seem a bit compressed. Here are the PCB and schematic as pdf aswell.

Op maandag 22 september 2025 om 20:43:56 UTC+2 schreef Florian van der Dussen:

[Mac Doktor]

On Sep 22, 2025, at 2:43 PM, Florian van der Dussen <florianva...@gmail.com> wrote:

Current schematic:

The schematic is too small to read. Please share a larger copy.

Terry Bowman, KA4HJH
"The Mac Doctor"

https://www.astarcloseup.com

"If only you could see what I've seen with your eyes."—Roy Batty, Blade Runner

[Adrian Godwin]

Thanks for the pdf. Yes, the schematic wasn't usable.

The first problem I've found is that SCLK and SDATA are connected to 3v3 and then have resistors in line. I'm pretty sure you had the right idea but made an error on the wiring, but if it's not obvious the resistors should be between 3v3 and the clk/data lines.

You will likely need a pullup resistor on the light sensor., unless the Pi has one internally that can be enabled on analog inputs. They normally just measure voltages and you want to measure resistance.

Overall, it's pretty good for a first try !

 

--
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 view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/33D10011-0F75-49D6-81B7-8049459EDE45%40gmail.com.

[Adrian Godwin]

Also, put 100n capacitors between supply (5v or 3v3) and gnd near each and every IC.

[newxito]

Have you checked if the traces leaving from the  barrel jack are wide enough for the expected current?

[Nicholas Stock]

You'll need some mounting holes in the PCB too... :)

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/58c67414-cd96-4730-8aad-0d07666f4023n%40googlegroups.com.

[Leroy Jones]

For a 21 year old guy with no previous electronics experience, in my humble opinion, you are jumping in far far deeper

than your present skill level is capable of producing any meaningful results.

First of all, you have NEVER even fired up a nixie tube yet.   Nor have you studied and worked with logic gates.

If you want success, the first thing to do is get a nixie tube and a 170 volt DC power supply and a 15k anode resistor, then start

experimenting lighting the tube digits.    Next, get a 74141 or a 7441 nixie driver IC, and connect it up to an appropriate 4-bit switch of come sort

so that you can feed it binary bits at the 5 volt level, and light the tube digits.   Next thing then is to switch those bits using a counter such as 74LS160.

As for a nixie clock, I would strongly suggest making one that uses no processor of any sort.   Use TTL or CMOS logic to run counters that drive

the 74141 nixie driver ICs.    A very nice nixie clock can be made using about 16 to 20 ICs.

Learning electronics and learning to use digital ICs and nixie tubes requires many, many practical tests and experiments.

Jumping right in cold, with no prior experience right away thinking that a complex PCB can be designed and a clock made to operate

without doing any experiments to prove the fundamental concepts is THE classic recipe for failure.

I can help with book recommendations, parts lists to experiment with, and experiments to do.

Take it slow.    Stay down at the level of reality.

-Chuck

[David Pye]

While I agree with some of this, I don't necessarily agree entirely.

There are lots of things about the PCB that could be improved (eg using ground planes/fills) and simple starting experiments - eg a trial single-digit-nixie is a good place to start, to work out where the gaps in your understanding are, and you can then incorporate what you learn into the clock.

What I don't necessarily agree with is starting by a clock using discrete ICs and not a microcontroller.   *IF* you come from the generation where arduino/rPi Pico are more familiar to you than large numbers of chained 74-series ICs (like me), then you might well find it easier to use an MCU to drive your clock.  If you're 'old school' and MCUs are also new to you, then sticking to that should flatten out the learning curve a bit.

David

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/022b559a-12fd-4b3c-ac21-73224c815c0dn%40googlegroups.com.

[Adrian Godwin]

I agree with David : it depends where you're starting from.

In addition, there are many pitfalls in building with TTL logic - such as proper decoupling and ground impedance - which will not be encountered when the more complex logic is embedded in a ready-made processor board such as a Pico.

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/CAOQ6x0F0SW2VBfrn_faayVgjODgXE%2BHNaOBrXhF_vB4xH79GVg%40mail.gmail.com.

[Leroy Jones]

David Pye:   You are missing the point of what I am trying to convey here.   My point is that trying to design and build a microproceeor controlled nixie tube

clock from scratch is an ambitious project even for experienced people.   Ambitious for an experienced man.   Impossible for a beginner.  That is my point.

It is about the same as someone who has never flown an airplane nor been involved with aviation in any way, suddenly thinks they can design and build a working

helicopter from scratch and have it work flawlessly the first time without having any prior experience of any kind.

The only way to do it is to learn each small thing in small steps.    Building a working nixie clock from scratch is a massive undertaking.  No way around that.

I am attempting to give guidance to this young man so that he might learn.

Got to crawl before walking.   Got to walk before running.   Folks today seem to want everything to have a processor in it.

And made on a PCB.    Lots of luck with that as a first-off effort!

Get a solderless breadboard.   Get some #24 gauge solid tinned jumper wires.  Get a nixie tube.  Get a 170 volt DC power supply.

Get a 5 volt DC power supply.   Connect both supplies so they share a common ground.  Use a 15k anode resistor for the nixie tube.

Get a 74141 nixie driver IC.   Get a 74LS160 decade counter.    Learn how to operate the counter using a 555 timer IC running in astable mode.

Connect the counter to the nixie driver IC.    Turn it on and watch it count.    For detailed explanation and study, get these two books:

1) The TTL Cookbook by Don Lancaster.

2) The CMOS Cookbook by Don Lancaster.

How the hell can anyone expect a newbie to understand what the processor is actually accomplishing without having him

first deal personally with each binary bit?    

Or are we now in such an age where fundamental understanding has been relegated to the dustbin, because it is "Too Hard, and Too Much Work"  ?

Flatten out the learning curve?   Nope sorry.   Not buying it.  ("flatten the curve".......hmmm......where have we heard THIS before?)

[Leroy Jones]

Adrian,  I am so glad that you brought up the complexities of digital design.   Decoupling caps.   Hefty grounds.   

These things should not ever be overlooked in ANY well designed circuit.   I am not ready to give this young man the go-ahead

for the processor operated clock until he builds at least a few from discrete logic first!   Of course it is up to him how he wishes to spend his time.

How many of you guys on here have designed and built nixie clocks that run on discrete logic and function properly for 25+ years without any trouble?

Because I have done so.   Many different versions.   I speak from experience.   But maybe that is of no value now.  We just grab a raspberry pi and act cool

about it and hope it works.

[Peter Virica]

Sorry if this has been already covered, but there don't seem to be any buttons to control it? set time, configure etc. ? But I don't do 'Pi' so maybe there is something I am missing. 

Pete

[David Pye]

My point remains. If you already know to program MCUs already, there's no need to go down the route of 74 series counters to build a clock, unless you want to.  You can fuse that knowledge with the fundamentals of driving nixies.

I certainly didn't build a clock based on because I already knew how to use microcontrollers.

Following your logic to it's endpoint,  you'd be telling him to learn how to build a clockwork grandfather clock before daring to build a digital clock.

David 

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/f9f50dd1-abf1-4eed-810a-95a7f50db3d1n%40googlegroups.com.

[Leroy Jones]

David Pye,

That is an oversimplification.    Operating nixie tubes comes with a whole universe of unique considerations.

Knowing how to program a micro cannot, and will not prepare a newcomer properly for building a nixie tube clock.

I am done here.     Not going to waste my valuable time screaming at a brick wall.   The new guy either learns or he does not.

That is up to him now.   I have laid down a few of the basics for him.   (Hint:  A PCB should not be made until the circuit has been tested and

proven out.).................PCBs are for mass production.    One off experiment is much better and faster on a solderless white board.

Much easier to change and to learn from.    

But I am sure someone will want to get the last word in and certainly not allow any talk of discrete logic circuits to go unchallenged.

[Adrian Godwin]

If he wanted to fly a helicopter without training, I would stop him too. 

But he has respect for the HV supply which is the only real danger. He's aware of the need for anode resistors. He might burn out a few ICs by shorting some wires together but I don't see any real harm. One learns from one's mistakes. Likely he'd do that with TTL too.

Sure, an education in digital logic would be valuable, as would a lightweight introduction to nixies.  But I don't see any reason why you can't learn on a microprocessor just as well. Thousands of people's first encounter of electronics has been with Arduinos and have found them a great start. The Pico is more complex but actually easier to use.

You're not wrong, Leroy, but I think a course of action that interests and enthuses him is the most important part. Personally, I'm quite impressed by his efforts. There are problems, sure, but he's come asking for help knowing that he doesn't know enough. What's the harm in encouraging him ? 

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/7a1ae0a3-ecf3-4b3c-802a-137cee4dc38dn%40googlegroups.com.

[Florian van der Dussen]

Dear all,

Thank you for the valuable information and help you’ve shared. I never expected so many replies. I really appreciate it!

It seems I’ve sparked an interesting discussion here. 

First of all, David, thank you for your encouraging words. Your outlook aligns closely with mine: my focus is on creating a product, not just accumulating theoretical knowledge. What excites me most is working toward a tangible end result, and the learning that happens along the way is a byproduct I greatly value. Creating the logic gates for tiny clocks will devalue that goal.

Leroy, thank you as well for your critical points. I do take your more traditional approach into account, and I understand its importance. Your 25 years of experience are invaluable. However, I don’t have the luxury of unlimited time or resources to master every tiny step in isolation. I realize that using a microcontroller is overkill in this case, but since I already have experience with MCUs, and even a stack of Picos ready to use. In that case it makes sense for me. That said, I will go through the books you suggested to strengthen my foundation and fill in the gaps I’m skipping. In my humble opinion as an inexperience student, never use the word "impossible", it's a word that shouldn't exist in an learning forum. It limits yourself and the ones around you for the unimportant argument of "realism".

A few additional notes that I didn't mention in my first message: I’m still waiting on some nixie tubes I ordered, though they’ve proven either very expensive or difficult to get in the Netherlands. In the meantime, I’m proceeding in parallel. I have always planned with a breadboard version to identify shortcomings before moving further. I never expected this much feedback, but I did hope to learn from the wisdom of those who have already gone through this phase in their own projects to hopefully ease mine.

I’m a young student pursuing this entirely outside of my formal studies, with very limited resources. From the start I knew this wouldn’t be a “flat” learning curve, but to me, that’s exactly what being a student should mean. If I avoided every challenge that seemed too steep, this extracurricular journey would have ended the moment I opened my first book/page. I understand I am cutting corners at times, but I believe that diving into the deep end allows me to learn more, faster, even if that means frying a few IC's along the way. And that is an attitude I hope to carry with me throughout my life.

You cannot be an expert about everything in life, but I hope that I can still make some cool projects along the way.

P.S. I might just take you guys up on the challenge about building and flying that helicopter :)

I will keep you guys posted of the progress,

Florian

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/1ElHfdtIMOo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/CALiMYrs%3Ds6P1YKRtwOwr1SKbMC89iVcPv9XoWEHcysi7u2-_gw%40mail.gmail.com.

[newxito]

I'm not really qualified to give advice, but I'll do it anyway :-)

If using a barrel jack, consider putting some kind of reverse polarity protection. A simple schottky diode or better a mosfet with low Rds(on) should do the job.  If you want to be on the safe side, you can add also an overvoltage protection but that's a bit more complicated. I use a specialized IC for that. You have also room on your pcb for an appropiate (resettable) fuse. 

I don't know if feeding your HV module through thin traces and a via is a good idea. You may consider using thicker 5V and GND direct traces instead. That's an easy change on your board.

Have fun with your project.

[Adrian Godwin]

Good point. You might also try a 'filled zone' which can connect all the gnd points together using all the remaining copper on the board. Unfortunately the autorouter has sliced it up thoroughly which may leave it in pieces but if it manages a complete margin around the edge of the board it will help. You can also fill both top and bottom surfaces with gnd, then link the two together with vias in a grid.

To view this discussion, visit https://groups.google.com/d/msgid/neonixie-l/94d5b9d6-ddd2-48a0-af96-cc5f9dd76f4bn%40googlegroups.com.