Month: August 2020

The first enclosure I’ve ever designed arrived a few days ago (I wonder if it still count as an enclosure if it has an open back, as it’s not fully enclosing anything)!

I did a quick fit test with the board and some standoffs and it looks pretty good. I’m so happy that my cut-outs for the tubes and the PIR sensor worked out well!

I’m very impressed with the material and with the laser cutting service. Everything fit well (if I were to remake this, I would probably shave a quarter mm from the joints as they were a little tricky to put together, but that’s on me) and I’m very happy with how it looks. Here’s some more detail on the material, which has a matte (external) and glossy (internal) side. I chose to put the matte side on the exterior because I don’t want my fingerprints all over the face of the clock, but there’s one face (the triangular face on the left) that’s inverted due to the orientation it was in when it was cut – something to thing about for next time.

I also wrote a quick script to display the seconds portion of the current time on the two tubes I’ve got soldered down, and that”s worked well except for some digits missing due to some questionable solder joints.

The new boards arrived early this week! Here’s a side by side comparison of the old board and the new one (notice the difference in the of the white silkscreen around the tubes – the new silkscreen reflects the actual size of the tube, whereas before I think I just plotted down a circle of arbitrary radius).

I finally bit the bullet yesterday and installed one of the tubes on the new board (after installing the decoder and series resistors and testing that out). It was a bit of a process because it turns out that the alignment of the tube itself to the pins is not completely controlled. So when I put all six of the tubes down on the board (with no solder), they were all at slightly different angles!

I may have to come up with a way to combat this if I want to make clocks for everyone in my family, but for now, I settled with bending the pins and aligning it manually. My girlfriend gave me the idea to align the two white dots using a box and my calipers as right angles, as seen below.

After this I tested each digit manually (by driving each pin with the supply and a series 15kΩ resistor), soldered it down, hooked it all up, and wrote a little program to count up (and then go back to zero). I made a little animation to show you!

Now there’s no excuse. It’s time to get everything else soldered down while I wait for the enclosure to arrive!

It’s a new post in the series and this time there isn’t a new PCB! So there is hope after all…

I haven’t had much of a change to work on the electronics part of the project since the new board hasn’t arrived yet, so I’ve been mainly focusing on the enclosure design aspect. A friend of mine suggested looking more into manufacturing techniques before continuing, so I took his advice and decided I’d give laser cutting a go for my first prototype. I really like how the finished products look with this technique and it seemed cost effective for a larger design like this one.

Since a laser cut enclosure is just a set of 2D shapes of fixed thickness, I could no longer have the small features I’d designed to hold the various boards in place. In order to deal with this, I added screw holes on the front face to hold up the main board and PIR sensor. I’m planning to have the screw heads on the front face, with the screws going through a small spacer, the boards, and then a nut on the other side. I have no plans for securing the RPi and supply boards to the bottom of the enclosure for now, but I might make a ledge or something like that on a future revision. Note the hole on the right side for a button!

I like the look of this “open air” enclosure for a prototype, so I might keep that going forward. My main concern for now is that the whole thing is going to want to tilt forwards as it’ll be top heavy because of the tubes. I may have to increase the size of the base or make it heavier going forward. For now I can always weight it down with something.

Another post, another PCB. Hopefully that’s not par for the course…

I just sent the Rev 1.1 PCBs off to fab. I made the following changes:

• Swapped out the 74HC42 decoders for HEF4028BT parts, to address the issue discussed in the previous post. This involved a fair amount of re-routing as the pinout was quite different.
• Changed the footprint for the nixie tubes and rotated them 180 degrees so that the board can be mounted right-side up on the enclosure
• Added screw terminal headers for connecting the PIR sensor and the switch to the board (they will both be mounted onto the enclosure itself, so they will need to wire into the board)
• Moved the BJTs closer to the nixie tubes to shorten the length of the 40-70V collector traces
• Grew the board’s width by 30mm and re-arranged the nixie tubes so that they’d be spaced a little further apart
• Changed the M3 mounting holes into M4 ones. I had the space and the M3 posts on my enclosure were looking a little fragile.

In parallel, I reworked one tube’s worth of the current board to work with the HEF4028BT to make sure that the interface between than and the RPi was OK (the part’s power pin is specified down to 3V, but the input high level is only spec’d when it’s powered off 5V, and it’s 3.5V) and that it wouldn’t have a problem driving the BJTs. It seems to have worked well. In order to make it work, I had to at least re-route two of the previous outputs to the Raspberry Pi and vice-versa, see below. This would ensure that inputs would still be connected to inputs and outputs would be connected to outputs. However, doing only this leaves the logic table very scrambled! I wrote down the options and “decoded” (pun fully intended) the logic.

I didn’t want to solder a nixie tube onto the board to test it since I’d need to de-solder it later (to put it on the new board) and that wouldn’t have been the most fun without clipping the leads. So I soldered a single digit (the “1”) up to a wire which I then hooked up to the tube with clips. You can see a picture of the full system in action below.

I’ve also scrapped the previous version of the enclosure (which had the board mounted upside down) and started over. Here’s some shots of the new enclosure design with the Raspberry Pi, power supply, and main board shown.