Even if you don't know the name, Nixie Tubes are an iconic piece of display technology that most anyone can recognize. I've always found them quite fascinating and having just completed my Microcontroller Applications course, I wanted to challenge myself and try to make something that uses an MCU, without using the entire dev board.
A Nixie tube is a kind of vacuum display tube that uses neon gas and physics magic to display Arabic Numberal digits. When a high voltage is applied to the tube (around 180VDC) electrons travel though the vacuum (not actually a complete vacuum, its very low pressure inert gas inside) and move from the cathode filaments to the anode. This discharge reacts with the neon gas inside and causes the connected filament to glow a bright orange color. In order to get visible numbers, the filaments are simply in the shape of each of the ten digits. The effect gives off this very retro and steampunk vibe that I really enjoy!
I know already that I'm looking for the final design to meet the following criteria:
I want to understand everything going on in the circuit.
This sounds kind of strange, but there's some really complicated power supply circuitry out there that is beyond my current understanding. I want something that I can point to each component and with confidence say This does this, and this does this, so a basic linear power supply it is.
Be as small as possible, and looking proportional to the size of the tubes.
This is completely subjective. Basically I want it to look good.
The PCB must be entirely designed by me, with no MCU dev boards.
Again, this is so I can learn how to design a PCB that uses an MCU without using the whole dev board. I want to figure out what exactly is needed to get the MCU to run. Some challenges this could cause are providing stable power, programming/debugging, and missing jumpers/connections that are needed for the MCU to behave as I expect. This will require digging through a lot of documentation.
Gotta count seconds
More numbers = more pretty
After digging through some Nixie tube forums online, I found that the HV5122PG-G would be appropriate for controlling the tubes with an MCU. This IC is just a 32-bit shift register with the outputs connected to the gates of high-voltage MOSFETs. This way, I can just serialize the time from the RTC module of an MCU. For the MCU itself I opted for the STM32G031K8T6. I've gotten very comfortable with STM32 microcontrollers as it's what we used in class and the dev board is less than $10 so I can afford to blow a few up if (when) I make mistakes.
I decided for a user interface a nice single-button design would be the best. It's easy to use and easy to design around, though this will be slightly more challenging to program.
For the tubes themselves I decided to go with the IN-8-2 which would later become the IN-8 for reasons explained later.
With all that set in place I started designing the schematic in KiCAD
Above is the initial electrical design. Let's break it down.
Perhaps the most complicated part of this design is just powering everything properly. There's basically 2 main voltage rails: 180VDC for driving the tubes, and 3.3VDC for powering the MCU. This design takes the 120VAC mains and rectifies it to ~210VDC. I expected it to output 170VDC because the transformer is supposed to have a turn ratio of 1-to-1 but it seems that the number of turns on the secondary is higher than the primary. At this point in time I haven't tested another transformer so it's not clear if this is just a single unit issue or not. It was after seeing this output voltage that I selected the 180VDC zener diode.
In addition to driving the tubes, the MCU also needs ~3 volts so I selected a component that can convert mains to that level. Yes this does sort of violate design requirement 1 but hey, I don't have all year to make this work, just gotta make compromises sometimes.
Testing the setup so far, I am able to drive and shift through each of the digits on a single tube as seen below.
If you look closely, you'll notice the tube color is slightly dark, and some of the segments are partially illuminated (2 and 9), this is because this test was done without a linear regulator. These tubes are sensitive to voltage drops and without a regulator in place, the voltage drops to about 140, which is enough to keep some of the smaller segments fully illuminated, but causes issues on larger segments with area that's further away from the center. After including the linear regulator, this issue is resolved.
This is mostly where the project currently stands, as of now. Because of tariffs on foreign imports, PCB prototyping and specialized components have become very expensive and are making it difficult to work on this project economically. Once I can afford a second round of PCBs, more components and new tubes, I will post an update.