Nixie Clock

This page documents my Nixie Clock project. Although the project has been sitting on the shelf for a long time, in an almost working state (the clock was getting behind, and also had a "feature" of adding one to the hours once in a while), I finally got annoyed by it and decided to make a few changes to the firmware, and also take a few pictures and put on this page.

Let's take this the opposite way around, and start with a picture of my finished Nixie-clock, sitting on the shelf below my TV:

Finished Nixie-clock

My Nixie Clock project started after seeing several project like this on various sites on the 'Net. The first thing to do was to find some Nixie-tubes that could be used for the purpose. I asked a friend who did find an old multimeter using nixies, but it was only 3½ digit, so I would only get a 12-hour clock, and no seconds. While I did consider this option for some time, a department at the University was moving, and cleaning out a lot of old equipment. I went there and found a frequency counter consisting of the main signal conditioning and timing circuits in one unit, and mounted on top of that, a display unit with 6 Nixie tubes. I grabbed this unit, which became the basis for my Nixie Clock project.

Opening the unit, I discovered that the entire thing was made without any integrated circuits - everything was made with transistors. The display unit actually consisted of 6 PCBs, each containing a single decade counter with a driver for one Nixie-tube. I decided to keep only the display module, which was contained in its own casing, originally mounted on top of the main unit with signal conditioning and timing circutry.

I also found out that re-using the power-supply or any of the Nixie-driver circuitry would not be viable. I therefore stripped the case of anything except the sockets for the Nixie-tubes.

In the mean time, a friend of mine (Asbjørn) had heard about my project, and found it cool, so he decided to also make a Nixie clock. He actually found a supplier on eBay where he bought a box of 100 Russian Nixie-tubes NOS (New Old Stock). Also, he found a place to buy the Nixie-driver ICs. These are compatible with 4741 or 74141, but are actually Russian ICs. In fact that means that they are more robust than similar 7441's, because the 74-series is for commercial use, while the 54-series is for military use. Back when these drivers were made in Russia, almost all electronics were for military use, so these drivers are only available as military grade.

Before I had my frequency counter stripped, and was ready to hook up the Nixie-tubes, he had actually made a complete Nixie-clock, based on some schematics found on the 'Net, with a few modifications. The design uses a transformer for mains-isolation, and a step-up switching power supply for the Nixie-tubes (since they need around 200V DC). The 50Hz mains-frequency is used for the timing, thus the system can easily run on the internal RC-oscillator of the ATmega8 used.

I kindly asked for a copy of the schematics, but then later realized how much work I would have to put into laying out and routing the PCB. I did, however, require a few changes, since I wanted to have the ability of adding a serial connection. I therefore needed to free the RX and TX-pins, and also use an external crystal for the clock (to get stable UART communication). It was therefore decided that I would re-use his design, but apply my changes by putting the AVR-controller on a small separate piggy-back PCB. This worked out quite nicely, although I still don't have a schematic of the whole thing (I have separate schematics of the piggy-back and the main PCB). I should probably make a combined schematic, as it can be quite handy for reference.

Also, the design my friend made actually consisted of two separate PCBs. One for the main circuit, and one for the Nixie-tubes. This was perfect for my needs, since it allowed me to skip the PCB for the Nixies, and just juse the tubes mounted in sockets already in the enclosure.

Inside of Nixie-clock

The picture above shows a peek inside the Nixie-clock. If you look closely, you can see the piggy-back PCB with the microcontroller. Underneath the two driver ICs are located. As mentioned earlier, these are Russian military-grade chips in ceramic packages. To keep in line with the design of this era, it was decided to use PTH technology in favor of SMD. However, I did cheat a bit on the piggy-back PCB, as you can see: There is an unpopulated footprint for an SMD IC. This is for a MAX232, should serial communication be needed.

Note the 6 striped wires going from the small white connector to each of the tubes. These are the anodes and carry about 200V DC. On the middle of each wire is a 33KΩ current-limiting resistor (this was on the display PCB in my friends design). I made the value of the resistor significantly higher, in order to get the correct display without worrying that the Nixie-tubes would burn out. Of course the output voltage of the high-voltage DC step-up can be changed from software, but it turned out that if the voltage is too low, the segments (cathodes) do not fully light up. Changing the resistor value allowed me to get steady display while keeping the brightness down at a resonable level. On the back of the case are two small push-buttons, used to set the hours and minutes of the clock.

If you find this fascinating, I will be happy to provide the source code for the firmware, as well as the schematics documenting the hardware design. However, at this point I haven't taken the time to prepare these, so if you are interested, drop me a line, and I will try to get it done sooner.

Last updated: 2007.09.26