There’s something about Nixie Tube Clocks that keeps drawing hackers to build their own iterations, even if its been done a gazillion times before. Their depleting supply, and the high voltage drivers to control them, makes it all the more interesting. [Pete Mills], a veteran of several interesting projects, many of which we have featured here, is no exception and decided to build his own version of a Nixie Tube Clock, but with several nifty features.
To put it in a nut shell, his Clock uses Nixie tubes for display, has USB serial communication, temperature measurement, AC frequency measurement, time and date keeping with a software based RTC, software driven boost converter for the 175V DC nixie tube supply and a windows app for clock configuration.
The software based time keeping is pretty interesting. It is essentially a method to calibrate the crystal to more closely match real time, and some code to keep track of the time and date. This obviously leads to a reduction in components and the spin-offs that comes with that; increased reliability, cost reduction, real estate savings. The RTC code can easily be ported to other clock projects irrespective of the display used. Besides keeping track of time and date, it can also account for leap years, and report the day of the week. A zero-crossing detector connected to the low-voltage transformer supply that powers the clock can also be used as an alternative way of keeping time.
When connected to a serial console over UART, the clock can report back many variables depending on the queries it receives. The high voltage DC needed to drive the Nixie tubes is generated using a simple boost converter controlled by the micro controller. An important “gotcha” that [Pete] deduced after blowing off several fuses, was to disconnect the micro controller port connected to the PWM timer and explicitly set it to output low via software. There’s a couple of other issues that he ran into – such as board layout, power supply, incorrect pullups – that make for interesting reading. The clock enclosure is still work in progress, but [Pete] hopes to get it done sometime soon.
He also wrote a Windows application – Nixie Clock Communicator – to help with time setting and calibration. Finally, he describes in detail the process of calibrating the clock’s software based RTC. Based on his calculations, the clock will drift by about 48 seconds over an 8 month period. Since he will be adjusting for DST much sooner than that, his clock ought to be off from correct time by not more than a minute at any given time. Not bad for a clock that does not use a dedicated RTC chip. [Pete] still has some of the prototype boards to give away if someone is interested. If you’d rather build it yourself from scratch, [Pete] has posted the software code, schematics and PCB, and a BoM.
19 thoughts on “Nixie Clock Uses Ingenious Software RTC”
Why not run the 175V boost converter from the unregulated post-rectifier 18VDC, instead of from the regulated 5V? (Looking at the schematic is proving more difficult than I’d hope)
Because he is! and because I misread “I had underestimated how much current the HV driver ICs would consume – they doubled the circuit total.”
No, he isn’t! He “solved” it by using a buck converter – now the linear regulator has 6.5V instead of 18V on input…
I have always liked the “ambiance” of Nixie tubes and may make one someday. I had a thought on an alternative way to make something close. Yes I know it wouldn’t be the same, but here it is. Capture images of the numerals and display it on small LCD displays. While some programming would be required, the materials cost could easily be less than an actual clock using Nixies.
I thought this was an interesting alternative:
It has separate hardware for each digit similar to a nixie tube, and you can customize the look of the digits.
Wow, tell that to people in the 1960s, after explaining what an LCD is and how we have tiny computers, and they’d be amazed. Cheaper to use a hi-res full colour display to show a picture of a tube full of gas and wires, than to use one.
It’s the analogue, high-voltage glowing charm of the Nixies that makes people bother with them, an image just wouldn’t be the same. Same feeling you get from forcing bits through a Z80 in an old computer, rather than the more convenient method of emulating it.
Agreed on both counts, but especially the first. It’s pretty amazing how often “put a computer in it” is the *cheap* way to solve a problem.
I also like the idea of not having to have a local hardware RTC in order to be a decent short- and long-term accurate digital clock.
the way I solved it was to have periodic beacons be sent over wireless (xbee, not wifi/ip) and each time you get one, you simply set the local clock to that value. send the beacons as often as you want or as little as you want. a few times a day or a few times per minute. but the arduino (for example) software rtc is good enough that you can miss lots of beacons and still be within a second or two before it finally gets one and updates its clock to that value. it does mean you have to have a good time source -sending- those beacons, but that’s not hard; any old linux box and a usb/xbee dongle will do the sending. once you do that, you can have many receivers, all in simple transparent mode and they all receive the same time beacon updates. it works well, actually.
(I also send out ‘unread email counts’ on those beacon channels, so any listener can display time and also alert me if I have new mail) ;)
clockless clocks. as long as you have some wireless transport, you don’t need local clock hardware anymore. just ‘message receivers’ and short term software rtc’s.
I want t o know why there are so many surplus Soviet countdown displays floating around out there. Did they have something in mind for all of those?
So everyone could have a nuclear countdown clock for everyone’s nuke! Communism! XP
The Soviet Union made a crapload of the things, then collapsed. Much of what they made was lost (or “lost”) and much of what was left got sold off to raise money.
Why not just send it internet time over usb every day? Or use a WiFi module and do it that way.
Rather than, say, a simple counter. People used to make electronic clocks using Johnson counter ICs, one per digit. Maybe just a flip-flop or two for the ten-hours digit and AM/PM. Then some random logic to drive the display. Or maybe use a binary counter and an LED digit encoder chip. This wasn’t that long ago! Of course people made clepsydras using jugs of water back in the old day. Moving to a microcontroller makes the whole thing simpler in hardware and more versatile, makes sense. But now they should have Wifi!?!? For a clock?
And they probably will. If LED clock radios, that are an alarm clock and a radio and nothing else, still exist in the near future, they probably will have Wifi. Maybe they’ll have no buttons too. Or a touch pad. Or you set them through your phone using an app.
It’s just getting ridiculous, the amount of computing power, transistors, sophistication, thrown at the simplest ideas and uses. WTF, seriously! A clock with Wifi. Or a clock with Wifi even if you don’t want Wifi because it was cheaper to use a Wifi/CPU chip than a dedicated clock chip. And a 1080p display cos it was cheaper than LEDs.
Zero crossing mains rtcs are nothing new … My 80s digital clock uses that xP … But I much prefer to use hardware rtcs … More reliable and recovery from power loss … Not to mention hardware handling of leap year and daylight savings (for the 5 people who still take advantage of that x3 )
The hybrid approach, using a computer as the clock, but keeping track of the mains, is a good idea. Even though almost nobody uses mains powered clocks anymore, the mains is still kept very accurate for that purpose, if they have a few slow cycles (eg from heavy demand on the grid) they make it up with a few faster ones later. Overall it’s probably much more reliable long-term than any quartz crystal.
You could even do a clever method, of comparing the clock’s own time with the time calculated from the mains, keep track of both. That would give you a factor of how off the internal clock is, and you could use that to correct the time keeping in software, adding or removing the odd millisecond here and there. That way it would calibrate itself.
The writing on the PCB in Russian “Русалка труба часы” appears to be a literal translation of English “Nixie Tube Clock”. But the way it translated it actually means something like “Mermaid Pipe Clock”. For the next PCB revision, the proper translation would be “Часы на газоразрядных индикаторах”.
Actually a Nixie is a mythical water-spirit creature in English. Originally from German apparently. And a pipe is a kind of tube. So I can understand where the seemingly odd translation came from. So what’s your own suggested translation, transliterated into the Latin alphabet, and translated literally, word by word, into English? Just curious.
What did the Russians call Nixie tubes originally, in parts catalogues etc? Or maybe in advertisements or documents when used in clocks etc? Again a transliteration and literal word for word translation is always interesting to pedants and miscellanists.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)