If this page looks like the formatting is wrong, the video doesen't work, and the images are all fuzzy, please stop using internet explorer. If it's all good, carry on.
What is it?
A clock is a device that tells time, duh...
Oh, you meant the nixie tube part... Well, a Nixie tube is a cold cathode display tube. When 180V is applied to the anode, and one of the numeral shaped cathodes is grounded, the voltage difference becomes enough to excite pull the electrons from the cathode to the anode. This excites the low pressure neon gas in the tube, and the cathode glows a mesmerising orange with hints of blue. If you're curious about a more detailed explanation, google How does a nixie tube work?
How it began
Back in June of 2010, a friend gave me an electronic thingy and neither of us knew what it was. Upon googling the part number I found that it was a nixie tube driver. Having no idea what that was, I googled "nixie tube". I suddenly saw glowing numbers surrounded by shuny chrome cases, and became obsessed. I wanted one. A $150 is a lot to a 16 year old kid, so buying one was out of the question. My first thought was, "I wish I could build one of those..." Then I thought, "the heck with it, I AM going to build one of those!"
Actually building it...
I had never done anything with digital logic before, and didn't really know what a transistor was... so you could say that I was a bit in the dark. Since it was summer, I spent all my free time researching designs and schematics, trying to figure out what the heck was going on inside one of these. Then, one day, when I woke up at 6AM and spent the entire day on the internet until 9PM, it all clicked. What really helped was THIS site. From there, I set out to build one.
I decided to base my design off of the design on the website I just linked to, because it was so simple and easy to understand. It is essentially 5 simple modules, 3 of which are almost identical. There is...
The Signal Generator
The signal generator is quite standard for timekeeping devices. It is a 32.768KHz crystal oscillator (signal generation created by a vibrating quartz crystal). This signal is divided by a binary divider, the 4060 chip. This divides the signal over and over until all that's left is a 2Hz signal. That signal is then fed into a 4013 flip flop chip, which divides the signal by 2 to reach a final 1Hz signal. the capacitors on the power pins are called decoupling caps and exist to smooth any irregularities created by the power supply or the other circuitry.
The division logic
This circuit takes the 1Hz signal, and counts it out 60 times before sending out the next pulse to the minutes circuit. The 4017 is a decade counter, and each time it recieves a high signal on the input, it changes it's output pin, until the tenth pin is high, then it puts a signal out and rolls over back to the first pin. The signal it puts out when it rolls over is fed into the input of the next 4017, whish is wired so that when the 6 is supposed to light up, it triggers the reset pin, which resets the chip to the first pin and puts a signal out to the minutes. This creates the 16 seperate seconds pins needed for the 10 cathodes of the first tube and the 6 needed for the second tube. The diodes are there to keep the resetting of the second chip from resetting the other chips, as well as keeping the output pin from going high when the reset is triggered, which could damage the chip. We'll get to the reset in a bit.
As I said, the 3 logic sections are almost identical, with The minutes being almost exactly identical, and the only difference with the hours being that it has a bit more logic on it.
As you can see, theres some extra stuff there. The two transistors are wired as an AND gate, which resets everything only when both the 1 on one chip and the 2 on the other chip are high. Also, the electroltic capacitor is there to pull the reset high for a few moments when power is first applied, which makes sure all the chips are set to zero to start with.
Switching the cathodes
The final section is the tubes themselves and a way to ground the cathodes. The nixie tube will have a substantial voltage drop, but it is still usually around 60V, depending on the tube you use. This means that it will fry most chips (not all). So, the MPS-A42 transistor is used. It can handle voltages of up to 90V, so it can be used to ground the cathodes. Once you make a schematic of the logic chips and the transistors and the tubes together, it looks like this-
The resistors between the outputs are all 33K Ohm, and are there to limit the current to the transistor, so the chip and transistor don't fry. The resistors on the anode of the tube are there to limit the current to the tube, preventing the cathodes from overheating. The value needed depends on the tube you use and the voltage of the power supply, I used 27K.
There is another section to provide the 180V to the tube anodes, but at this point I was using a pre-made nixie power supply unit.
Construction of MK I
For the first model, I added a simple set feature by wiring the seconds counter into the inputs of the minutes and hours 4017's through switches (and a fast set by using a different output of the 4060), as well as a switch to pull the reset high. I then attempted to construct it on radio shack perfboards. It worked at first, but inevitably, something shorted out and something fried and the chips went poof. I did get some photos and a video forst though.(Yes, I know that the video is poor quality, and that the tubes are over currented.)
After determining that a whole bunch of radioshack perfboards wired together were not a good solution, I attempted to do it on one, giant perfbord without copper pads. Construction was essentially the same, though I added some more features, such as slow/fast set, a flashing colon using IN-2 neons, an alarm function settable using rotary switches, and an AM/PM light. At one point during the construction, I overlooked the maximum current of my purchased step up converter, and it overheated and stopped working. At this point I switched over to a 555 based power supply, which can be seen in the attached schematic. I also got around to making some real schematics, which I will post here, click on them to enlarge.
Signal and power
As I said, I attempted to build it on a large piece of perfboard. This resulted in the mess of wires you will see below.
As you can see, it's not complete. That's because when I started on the second of teh two large perfboards, I encountered a problem. I don't know what was, I checked the circuitry more times than I can count, and couldn't find a short. Whatever it was, it made my tube power supply start smoking. At this point I decided I was tired of all the problems arising from point to point wiring, and decided to do this the right way... by making a PCB.
Having given up upon point to point wiring, I began to create a PCB (printed circuit board). Since I had done all my schematics in the program ExpressPCB, I used it to create the first circuit board. All of which had to be routed by hand. Then I realized that ExpressPCB wont let you create gerbers, you have to order it from them. Since I didn't want to pay their $125 for one board, I had to remake the schematic.I re-did the entire schematic on one page, including creating custom parts, layouts, and packages for the tubes, including a layout package that allows for the use of either the IN-14 tube, or the IN-16 tube. Then I realized that the free version of eagle won't let you make a board more than 4*5 inches (or something closeish to that). So then I broke the schematic up into the following sections-
And then I made the correlating boards
I also made a bill of materials
This is where I am now. I have the files ready, and will be ordering the boards and parts sometime. Hopefully I can have the completed product on here in a couple weeks! Then I'll be selling kits!
If you're wondering why it took me a year and a half to do this, keep in mind that all the other projects on this site, and more, were completed during that year and a half. Also, other than the basic beginning of the logic, I've had to learn, design, and test all of the circuitry. That includes everything from what a transistor is, to logic gates, to how to make gerber files for production, and everything in between.
And I had to make all the schematics like 3 times. That takes time.
Unfortunatley the idea of breaking the clock into so many boards did not help the cost as much as I had hoped, and only added to the complexity and overall cost if I wanted to sell it as a kit. Taking this into account, I decided to switch again from eagle to KiCad, an open source development tool with no restrictions. I changed the board layout to have everything on one board, and swithched to only using 4 tubes in order to save on the cost. I redesigned and ordered a set, and unfortunatley after biulding one noticed multiple errors and/or areas for improvement. Most notably, the pad layout for the tube sockets was a mirror image of what it should be. Another significant change that needed made was that the 1N4007 did not have a fast enough recovery time to function in the voltage boost circuit, and I had to switch back to the UF4007. Additionally, there were no copper pads on the back copper layer for the neon indicators. Other changes I decided to make after version 2.0 were to move the power port from the side to the back, and switching the orientation of the tube power switch.
The greenwire prototype version of these changes makes up verion 2.1.
Version 2.2 implemented the changes mentioned above. I sent a sample of this kit to a real kit selling business in texas, apogeekits.com, for evaluation of potentially adding it to their line up, and for feedback. They were kind enough to provide very detailed, helpful feedback on the kit, including the advice to increasse the size of the resistor footprint, and remove part values from the silkscreen, among other things.
I did sell a fully assembled version of the V 2.2 clock on ebay in order to make back some money.
Pictures of V 2.2
Version 2.4 will implement all the changes and information gained in previous versions, and may possibly be the final version of this project.