In the third part of this series, I want to talk about the PCB design and the custom case for the electronics. I’ll also revisit the transistor array, which I didn’t finish in part 1 of this series and I test the completed project and show it in action.
The transistor array
Earlier in this series, I showed you this image of the completed control-circuit:
The following image shows the simplified array from fig. 1 in more detail:
As you can see, each digital output of the decoder is connected to the base of an npn-transistor via a current limiting resistor. That’s all, really simple.
Just make sure, that the transistors you use can handle a voltage of 170V and a current of 25mA.
To figure out, what your base-resistor value has to be, use the calculator linked at the end of this article in “Sources”.
The PCB design
I wanted to combine everything (except for the step-up circuit) on one PCB, which I think turned out pretty well:
A main goal of mine was to keep the PCB-size as small as possible, but still provide some space, where it could be mounted to the case. I also wanted to use SMD-components, so that I could improve my soldering-technique and they would also help keep the PCB thin so that the custom case doesn’t have to be large and bulky.
Due to the usage of SMD components, most of the connections had to be made on the component-side. I tried to use as few vias as possible. The bottom-layer really only has the GND, AGND, VCC and +170V lines and some connections that had to be made between different pins of the same IC. That’s also the reason why I used the two DIP-16 ICs instead of their SMD variants.
You can download the PCB design files and EAGLE schematics here.
Ordering the PCB
Because this is a tiny design with very small tolerances and traces it was important to find a good manufacturer for the PCBs so that they would turn out nice and work properly.
I decided to order them at PCBWay and I can’t be more satisfied with the product they sent me:
Prices start as low as 5$ for 5 PCBs with a super-fast build time of one to two days. And the best thing is: New customers get a 5$ coupon! So basically your first order might be free! (Depends on the size of your PCB and other factors) So it was practically a no-brainer to get my PCBs here.
Plus another really cool thing (that other manufacturers usually charge for extra): You can choose from green, white, red, blue and black as the solder mask color without additional cost.
You can get an instant quote for your prototypes online without the need to register.
If you decide to order: They also have this handy online-converter which will convert EAGLE files to the correct gerber format. Even though EAGLE has a converter too, I really like online converters from manufacturers, because this way you can be 100% sure, that there won’t be any compatibility issues with the gerber version.
Another huge plus point is the customer support. I can’t stress this enough: A good customer support is essential (at least for me). The site offers a live-chat and somebody will look into your design manually before production and they will get back to you when something critical is wrong with your design. That can save you some money, time and stress!
When first testing my freshly soldered PCB, nothing worked. The tubes would either display nothing at all (decoders reached a value > 9) or random numbers would either stay on constantly or flash on and off, which looked nice but was undesirable in this case.
At first, I blamed the software. So I came up with this Nixie tester for Arduino:
This script allows you to input a number of a GPIO pin (0-8) you want to change the state of. It then asks for the state. When entering pin number 9, the latches are reset.
So I continued my testing and made a truth-table with all the possible inputs for A, B, C and D. I noticed, that the numbers 4, 5, 6 and 7 could not be displayed with either of the two tubes. Additionally, they would react differently to the same combination of inputs.
I figured, that there must be an electrical problem too. I couldn’t find any technical problems in the design, but then I thought about something I’ve learned a long time ago (but never really had a problem with since then): Flux can be conductive. This might not be an issue for usual digital and low-voltage applications, but it seems like it was an issue here. So I cleaned the board with alcohol and afterward it behaved properly.
Kind of. Another thing I noticed: The part that I used in EAGLE when creating my PCB layout was incorrect (at least for my tubes). My tubes seem to have a different pinout. After further testing with the tool, mentioned above, I eventually created the following truth table for my tubes (might differ from yours):
Just some things to keep in mind when your circuit does not work right away.
After everything else was sorted out, I wanted to build a nice case to house my circuit. Luckily I had a lot of wood left from my word clock project, which I wanted to use for building a grid on the inside (I used foam-board instead of wood in the final build):
I built the case using the following measurements:
|6||40 x 125 x 5||Bottom, top, front and back side|
|2||40 x 70 x 5||Small side pieces|
|2||10 x 70 x 10||Structural pieces on the inside (See fig. 8).|
|2||10 x 70 x 5||Structural pieces on the lid (See fig. 11).|
After cutting the pieces, I laid them out together to create the following box:
The following image shows the case from a different angle:
Using the measurements from above, the PCB fits perfectly into the case:
The top of the case was not displayed yet, but it is exactly the same as the bottom, just without the walls and with less high structural parts. It acts as a lid and can be taken off to service the components on the inside. The PCB will be mounted to the lid with the two tubes sticking out of the case:
After I was satisfied with how everything fits together, I simply glued all the parts together and let it dry for some hours.
You might be wondering, how I fixed the PCB to the lid when there are no screws visible on the top. I simply drilled a hole for the screw into the structural part of the lid and then made a countersink for the screw’s head to go in:
Finishing the build
After the main PCB was mounted to the lid, all the other components simply had to be placed in the case, which looked like this for me:
As you can see, I tried to organize the cables as good as I could and I think it turned out rather good. Everything fits into the case nicely:
You can also see that I added a DC-Jack to the case (and went a bit crazy with the hot glue there). But this way you can power the thermometer with any generic phone charger and a fitting cable. However, you could also add a 5V battery, if you wanted to.
Parts used in this build
|1||DHT-11||4,19€||Got it from an expensive store. You can get these for less than 1$ in from China.|
You’ll also need some kind of microcontroller. I used an Arduino Pro Micro.
* For details see part 1 of this series.
|1 bottle||Wood glue||1,29€|
|1 can||Wood paint||5,79€|
I’m really happy with the outcome of this build. For once I managed to cut the wood pieces precisely and also didn’t forget about mounting holes for the PCB. And it actually looks magnificent too:
Besides that, it was interesting to work with tubes and high voltages in general and there are a few things to consider when doing so.
In conclusion, I’d say that it’s good, that we have more convenient ways of displaying numbers today but on the other hand there is nothing comparable to the glow and overall appearance of nixie tubes, which I really do enjoy looking at, especially, when it’s dark:
Hope you liked the build, make sure to leave a comment below if you tried building one yourself!
Table of contents
This article contains paid promotions.