Six Digit Nixie Tube Clock

Hey kids, that's 250000 volts! (Careful!)

Six Digit Nixie Tube Clock Project

You can reach me at

That's delicate! Maybe I shouldn't use the torch...


Nixie tube Clock

Nixie Tubes?
I remember watching great SciFi films from the late 50s and early 60s as a kid. They all featured stone-age Bendix G-15 computers with equally stone-age register displays (those blinking lights). Later on, as prop technology advanced, the display technology used on the fake "high-tech computers" advanced as well. Eventually, the ubiquitous blinking lights were replaced with those more realistic decimal display tubes invented by Burroughs, named the Nixie tube. These nice little tubes have one common anode with 10 separate cathodes, each shaped like an arabic numeral. These numerals are stacked one on top of each other in a sandwich fashion, such that when any one is grounded, the resultant orange glow completely covers the cathodes. Thus, with only 170 or so volts and less than 3mA, a bright, legible decimal display is created.

Several movies had them as key elements of the computers, due to their extremely high-tech appearance (in the early 60s). A Google search doesn't yield any examples, but I'm sure I could dig some up if I tried. I seem to remember seeing them in the communications exhibit in the basement of the Museum of American History at the Smithsonian, and at an exhibit at the Exploratorium. Either way, this site isn't intended to chronicle the history and development of the nixie tube, rather the continued use of these tubes. Several web sites have detailed the use of these tubes in clock projects. Mike's Electric Stuff website has an extensive gallery of clocks built by enthusiasts.

Since I was so drawn to the amber glow of late 50's high-tech, I had to do something with the nixies and nixit-tube driver ICs (74141s) burning the proverbial hole in my junkbox. So, I began reading designs on the web and pursuing my own design.

Nixies require about 170 volts, and the ones I happened to have (NL-8422) require 175 volts, and light well with less than 3mA. This is a chief problem with driving nixie tubes with 5-volt TTL circuits: they fry with the 60 or so volt drop induced turning on the nixie tubes. Luckily, I had some 74141 driver chips hanging out in my junkbox. I also had 6 tubes and associated sockets. So, I set about designing the logic and pic program needed to drive the nixies. I decided that the pic would generate a 1-Hz interrupt and divide the time appropriately into seconds, minutes, and hours. From there, a "display" routine would put the numbers out on port B in pairs, 4 bits per digit. The 8 bits per digit pair would be stored in an external latch, and the latched data would go to the 74141s directly.

The next hurdle any nixie clock designer faces is the design of an appropriate 170-180 volt power supply. I originally intended to have two transformers work back-to-back, but since Maxim still ships samples of thier simple boost-topology switching supply controller chip (MAX771), I went with a switching supply. Plus, I had all the parts in my junkbox.

I also scavenged some National DM74ALS573BN octal latches from a closing Cabletron office dumpster. The PIC came from samples at Microchip. The only thing I had to buy in the end was the protoboard from Radio Shack. A custom PCB was going to cost too much.

Building the Nixie Tube Clock

I decided to use point-to-point wiring for this prototype. That was difficult. I wouldn't suggest it for a relatively complex digital design, such as this. I also fit the clock into a wooden shadow-box. I think it is quite pleasing to see the wires and the parts on the breadboard. While this road-warrior chic may not appeal to everyone, it is quite to my liking.

Here is my Pic Micro .asm Assembly Code for the clock. It uses Roman Black's modified Bresenham "error-free" 1-second algorithm. Basically, it loads a pseudo-24 bit register with 2 million, and divides it down to 1-Hz with the internal prescale and subtraction. Each time the timer overflows, 256 is decremented from the large register. When it overflows, the remainder is kept, and 2 million is added to the result, and the 1-second flag is pulled. Thus, error from one second to the next is carried, resulting in an overall stable 1-second interval after the first. This idea was taken from another pic project, but I can't seem to find it now. Since there has been interest, I have included the .hex file as well.
It seems that there are a few errors with the .asm. It is possible to work around the A-D swapping by simply exchanging wires, but the watchdog/timer prescale problem has to be changed in firmware. So, I have included two other sets of firmware .asm files and the compiled .hex files that should help.
Oops - found another small error. Richard Ingram found a 1/2 speed problem that a few others had mentioned. Essentially, the code needs to have a 1x multiplier on the clock crystal oscillator. That means that the oscillator configuration register needs to be set to the WDT timer, which requires the OPTION register to be set to 0x88;.

Here's a prescale of 1.asm. If you want the .hex file, I can recompile it for you.

There are 5 pages of schematics. Everything is copyrighted, but nothing is particularly special about the design. Feel free to use aspects of it, or the whole thing. Just don't copy it and then sell it...

This is the processor page. The PIC resides here, with the hour and second set switches. Note that the entire 8-bit bus is terminated. The crystal has the rather un-necessary 22pF caps to ground for stability.

This is the first of three latch/display sheets

The next two are very similar:
Page 3
Page 4
This is the power supply page. The switcher is almost straight from the app note. The 7805 is boilerplate.

I've been catching a lot of flack recently because I didn't post high-res enough images of the schematics. So, here's another copy of them at full resolution. Unfortunately, the cad program (expresspcb) exports only to .bmp files, and gives you a menu of "pixels per inch." So here's the five pages, in full, 300pixel/inch glory. :)
Page 1, Page 2, Page 3, Page 4, Page 5

Here's the Bill of Materials for this design. I suggest not using the exact parts in the schematics, because I updated the BOM after the breadboard design was complete. Also, my components probably don't match yours, as everything came from my junkbox. I suggest playing with the anode resistors to get the current just right. 33k worked perfectly for me. Good luck!

Picture gallery

This is the layout of the components on the protoboard.

This is the clock operational after all the soldering. The clock reads 02:00:38, but the nixies are out of order.

Click for larger. This is the usual nixie clock working shot. Not too glamorous after the first night.

More pictures to follow...
-Fred Niell

Design, text, programming copyright 1/2008 Fred Niell