Nixie tube clock gets a revisit… after a few years

It’s been a long time since I’ve written anything about… anything. As always, I’m happy if anyone finds the information here useful, but I don’t expect anyone to read it really or give a toss. Really use it as my open lab notes which I refer back to when I take these long breaks to remind myself just what in the hell I was doing!

I have a fairly good memory. To the surprise of some I can recall conversations verbatim from twenty years ago… but I know the instant I say “oh I won’t bother noting that, I’ll remember for sure…” is the moment that said information is forever wiped from existence, never to be seen again. One such, happened when I revisited my Nixie Clock project which I’ve been working on since… 2012 probably?

Looking at the Nixie Clock again

I had a memory of how much fun it was to mess with electronics so I thought I’d dust off my bench and give it a go again. My obsessions work in cycles, and I’ll abandon things for a long time then pick them up again and work in a flurry, getting a huge chunk done, then putting it down again.

I pulled out my partially completed circuit. I remembered that I had the power supply working just fine – the 12V, 5V logic, and 170V nixie supply – as well as the quartz crystal oscillator that serves as my timebase.

What I didn’t remember

I see the board is undamaged and ready to go so I power it up, plugging the pcb power connector into the board. What followed was a load of … nothing. Perplexed I started checking around with my multimeter, following Dave Jones’ troubleshooting mantra “Thou shalt check voltages“ and found even more surprises. The 12V rail was now ~15V… as if the LM7812 wasn’t even there, and the 5V rail was now a 3V rail. Then I got my final surprise – I touched the LM7805 and nearly burned my finger.

Crikey! What’s that all about then?

I unplugged it immediately worried to death I had fried all of the components, particularly the delicate CMOS frequency divider (CD4521). I was so puzzled as to what could be wrong. I looked close at the board and found the culprit.

Polarity matters

For some long-forgotten reason, the red and black wires on the pcb connector were reversed. Well there’s my problem! I was backfeeding the circuit. Small miracle the caps didn’t blow up in my face (which can, has and will happen). I must have assembled it wrong, or grabbed the wrong pre-made connector. Mortified, I correct the polarity issue (permanently this time) and plug it back in, going through the motions to find out what is dead or alive.

To my delight, it seemed to me the only dead bits were the two voltage regulators. So far anyway. I thought – this will be no problem, I have tons of spares! Famous last words. I reach over to my parts bins and rustle around. Have a spare 7805 CHECK, but where’s another 7812? I thought I bought a bunch of these damn things. It seems I have every other regulator possible – except the one I need of course.

Recycling saves the day

Many years ago, when I was living with a clever bloke named Mike, we used to hang out, drink beer, and make an awful stench as we desoldered junked PCBs for parts. We must have pulled out hundreds of active and passive components and probably melting them in the process as we clumsily and very poorly desoldered them. I had chucked all those parts in a plastic box and forgot about them for years. I always bought new parts since they are inexpensive and at stand a better chance of actually working.

In the present, I found this box of lose parts and hoped. Otherwise, I’d have to make a run downtown just to buy a voltage regulator that isn’t even worth a dollar, or wait for an online order. Neither option I particularly liked. I thrashed through the bin, picking out any TO-220 package I could find. Finally, I found a rather worn and crusty 7812. I could have leapt for joy but I’d spill parts everywhere.

I hooked it up quick to a power supply to check it and yes – indeed it works! I desoldered my melted regulators and replaced them with working spares. To my delight, on power up, everything worked just fine. All the voltage rails were correct and the clock was running beautifully.

Thank goodness. I didn’t want to lose my inspiration and momentum just as I was getting it back. One of the things I hate about being an electronics hobbyist is having to constantly run out to get more parts or order more parts… never enough parts!

Where to go from here

So once again I have a working circuit and in peeking at the schematic I realized I had a pretty clear road ahead toward finishing the circuitry at least. I looked it over very carefully and reviewed my past notes here to be sure what everything did. All looked solid or at least reasonably likely to work reliably.

I decided on a course of component layout to minimize the number of connections needed between the two boards which I was pleased with, and also noted some areas of concern which needed looking into.

No Backup?

I did notice something absent from it – I had designed in no battery backup. Anyone who had (or even still has) a plug-in clock knows the annoyance of having to reset the damn thing every time one loses power or unplugs it to move it to another location. I know it’s one of those little things that I had previously thought not to bother with but it was nagging at me now. I looked into options to supplement my circuit by adding in a simple battery backup.

Ideally, I would have loved to use a coin cell in a nice PCB mount to keep the time when I unplug it… but I quickly ruled that out. CR2032s are great, put out 3V and last an age… but I need at least 5V to drive my circuit, more even to overcome the dropout voltage of the LM7805. I looked into various options, boost converters really, and it all seemed way more complicated than it needed to be.

Haz Backup

I finally decided on just using a regular old 9V battery and two diodes to do the business. I already had all the parts on hand. I replaced the wire from the DC input to the 7805 with a 1N4002 power diode to make a one way flow from that direction, and used another diode to connect to the positive lead of the 9V battery clip, then grounded that. Simple done! The diodes ensure both that the battery doesn’t dump power backwards in the circuit to the wall wort, nor can the DCin try to charge the non-rechargeable battery, which would have been destructive, as well as amusing.

Power up test confirms it works. I can switch from DCin to battery without skipping a beat, the clock keeps on ticking. The higher voltage of the DC in (about 14V) prevents the battery from being drained while it’s plugged in. No need for a fancy latch circuit or anything. Good enough and a modest improvement to my design.

Next Steps

With these modifications, I realize a few things will change, but not much so it’s a pretty straight road ahead from here on out. One concern are the blue decorative LEDs that will sit under each nixie. In my schematic, I have it feeding off the 5V rail… that’s going to have to change as I don’t want it draining the battery when unplugged. That whole section with the darlington transistor-pulsing flasher thingy will have to move to the 12V rail. Somehow I have to remember how I biased the thing and recalculate how I do the resistors for the LEDs given the higher voltage. I’ll either adapt the lot for 12V or use a 5V1 zener regulator just to power that, though that seems a bit silly (never stopped me before though).

I also whacked in a simple red LED to give me some feedback as to whether the thing is powered on and at least the 12V rail is working. I tried to add a green one to the 5V rail, but sadly it seemed to stop the clock operation when under battery power. That’s a very minor disappointment. I’m hoping the battery can handle driving all the CMOS chips to keep the time at least! It should do, they are very low power devices. I probably could whack in a cap or something to keep the power drain of the LED from affecting the CMOS circuit but that’s… why even bother? I ripped out the LED and resistor, just keeping the red one for “power on” indication.

So next stop is redesigning that darlington LED pulser to work off 12v, install the first 2 of 6 decade counters (the seconds, and tens of seconds). Both of those parts should finish off the first board and I will consider that a victory if all the parts remain functioning and I don’t make any more silly mistakes. Then it’s off to the second board, the remaining decade counters for minutes and hours, then all the transistor drivers for the nixie tubes (about a thousand of them it seems).

Other notes

This post is pretty dry in terms of visual content, meaning there are no diagrams and pictures, I do apologize for that. One reason is my circuitlab subscription expired and now I have to wait until I have the funds again to restore it. Great service for doing my schematics and circuit simulation. I’ll have to do it the old fashioned way until then.

The other is that there isn’t much to show at this stage. Apart from a few component additions and a big battery clip hanging off it, there isn’t much to show that wasn’t there last time. More updates to follow in a subsequent post.

I also haven’t forgotten about my other projects either. The dummy load is another one which has reached design completion and merely needs assembly. I think I have all the parts for that already so I’ll be seeing to that when I get the clock either done or get bored of the clock again (or run into an inevitable snag). I’ll try to get at least one done!

Comments are closed.