ZM1040 nixie Clock

This nixie clock has ZM1040 nixie tubes. These are nice and large: 28 mm in diameter, 60 mm tall, having a digit height of 30 mm. They have a B13B 13-pin all glass base so they need a socket to mount them. I have 5 identical tubes with the vacuum pip at the base, and one tube with the vacuum pip on the top. The 5 identical nixies had their orange coating already removed.

This clock has been designed by Thomas and Claus from nixieclocks.de. It is a straightforward and practical design. Like a lot of the clocks made by nixie enthousiasts today, it has a microcontroller for the timekeeping and control functions. The entire clock, including the nixies, is built on a single PCB. All the components are "wired", i.e. no SMD components.

Home made nixie tube sockets

To mount the nixies on the PCB, the designers have intended the PCB for the use of so-called "Mouser Pins". These seem to be hollow pins, receptacles actually, that will each hold one pin of the tube. They are soldered into the PCB individually. But these Mouser Pins are not sold widely. I loathed the idea of having to order these Mouser pins, pay shipping and ordering costs on top of a high price for low volume sales, so I tried to think of alternatives. The idea came more or less by accident. At work, during a somewhat boring discussion, my thoughts drifted away while I was staring at the back of my colleague's PC. Musing on uselessness of the parallel printer port and how ridiculous this legacy of the 1980 IBM PC was on an office workstation in 2007, I suddenly realised that the connectors could be removed and put to use. The pins of the sub-D connectors used for printer ports, serial ports etcetera are almost 1 mm in diameter so it should be possible to use the contacts of a female sub-D connector for the 1 mm pins of the nixies. Of course, removing parts from office PCs, however superfluous they are, would lead to awkward situations at work, but this was unneccesary as I knew I had plenty of old equipment at home.

At home, I tried to insert one pin of a nixie into a female sub-D connector and noticed this went well. I disassembled an old female 25-pin sub-D connector and harvested the individual contacts. The photographs below show the process.

Then I slid 13 of these contacts around the pins of a nixie tube. I inserted the contacts into the PCB and soldered them in place. The contacts fitted into the board loosely, almost stuck at the place where thy are thickest, so they seem to be slightly thinner than the intended Mouser pins. The abuse of a nixie tube was necessary to align the contacts (by the way: in order to prevent damaging my precious ZM1040's, I used a ZM1041 for which I have no use at present, a nixie with "+" and "-" symbols, ). It was easy to pull the tube and insert a ZM1040 in the newly formed socket.

So my idea worked: I could make free nixie sockets from old computer connectors!

Building the clock

I soldered resistors, capacitors, transistors and IC sockets for all the ICs onto the PCB. The hardest component was the inductor for the HV power supply. The circuit diagram specifies 100 µH/1 A. After some digging in my pile of inductors I found a 150 μH coil with amply thick wire. I removed some windings to get 100 μH and mounted the coil. It was a bit largish, but I decided this would have to do.

I had a minor improvement for the electronic design: I added two 15 pF capacitors to the clock oscillator pins of the the PIC microcontroller. The manufacturer recommends these in order to make the oscillator more stable.

Starting up the clock

After I finished the clock, it was late and I went to bed. I left it standing on my workbench for a few days, as it was Christmas time and I had dinners to cook and relatives to visit. Then I turned to test the clock and start it up.

First I tested the low voltage power supply. I put a piece of wire in the NE555's socket to short the gate of the MOSFET to ground, to prevent it from accidentally switching on. I connected a 10 V AC wall wart to the circuit board and switched on. The voltages were all well within range.

Then I tested the HV power supply and checked the anode drivers. I inserted the NE555 that is used here as a switch mode power supply controller, I put one of the 74141's into its socket and turned the voltage control in center position. I inserted two nixies (position 2 and 5). Then I switched on again. The voltage on the buffer capacitor of the LV power had slightly dropped, which is only logical as the 555 is now drawing current. The HV was about 170 V. The NE555 was switching at about 18kHz with a low duty cycle. The anode drivers all looked fine as I measured the voltages on the bases and collectors of the MPSA92 transistors.

Finally I inserted the remaining IC's and switched on. The nixies started to glow. I pushed the "set" button a few times and the clock started to run. The HV had dropped to 160  V so I turned it up slightly. The NE555 was now running at about 32 kHz and 50% duty cycle, so it was working hard to power the nixies. There was a 50Hz ripple on the HV and I noticed that the voltage on the buffer capacitor had dropped to less than 8 V. So perhaps I should look for a slightly more powerful wall wart. Checking the multiplex signals on the anode drivers I found they looked just fine. The clock signal on the crystal was OK, too.

This concludes the first phase of my ZM1040 clock. The next step will be to craft some kind of case.