DIY Numpad: More Case Prototyping (and Nixie Clock Case)

I did some more prototyping of the case several weeks ago using a 3D printer (first time I actually used a 3D printer) to test out an opening for the USB port:

Freshly printed.

This was the second time I printed it; the first time I didn't have a brim and one of the edges peeled up. The screw holes are undersized as expected since the plastic will have some squishing outwards and after cooling there will be some internal stresses, but I don't plan on using this right now so it is fine. The plate fit on it just fine.

Since this week is spring break, I was able to go home and use my CNC. I made a few changes: I got a new spindle after the previous one died (and it wasn't worth replacing brushes and bearings), bought a new spindle mount, and finally mounted my e-stop button.

New router (variable speed!) and mount.

E-stop.

Yesterday, I finally got around to making the case for my nixie tube clock. I did a quick test pocket before making the case:

Test pocket. 3/8" straight cutter.

I used a facing operation for the top with and an adaptive clear for the pocket. I made a few changes for the final case: changing the chaining for the facing operation from smooth to linear (so the tool wouldn't suddenly engage more material in the smooth curve outward), leaving a finishing pass for the adaptive, and adding a 2D contour to cut out the box from the stock.

Not bad.

This looked really nice but there were a few problems. The x-axis lost steps in the middle on the operation twice and I had to re-touch off to re-zero my x axis. This was due to some resonance in the x-axis leadscrew, and I had to add some damping to the screw with my hand. I need to make a bearing support from the other side of the screw, or use faster leadscrews (for a future upgrade). Secondly, my finishing pass was too aggressive (0.02" axial and radial stock to leave) and I think the thick double-sided tape had more deflection than that. I tried to pocket the entire depth of the pocket for the finishing in one pass and the final pocket was exactly 0.04" too thin. As a result I had to manually route the pocket a little bigger for the PCB to fit.

Pocket too small. Looks nice though.

I had to sand the top to remove some gouges in the stock.

Gained a case but lost a tube.

In my carelessness of handling my clock a tube dropped onto the ground and broke. A replacement is on the way.

Cracked tube.

 (update: new tube installed!)

Today I just wanted to run a quick job to test out a clear plate on the bottom of my numpad case (last week I got the APA102 RGB LEDs to work with SPI so now there are nice colors to see). I had some 1/4" acrylic to test this out in.

My first attempt led to some chip melting and the edges have poor finish (was at ~10k RPM and 45 IPM). I thought I was going fast enough but I guess now. Also the bore for the screw heads to sit in is undersized likely due to some melting.

Plate.

Sad edge.

I tried to go faster for a second time (60 IPM in program, with 120% feed rate in LinuxCNC, which comes out to 72 IPM) but still got melting. I held a brush up to the bit the whole time and this helped prevent chips from accumulating on the bit. I think the melting is because the acrylic was probably extruded and not cast (I don't know exactly what type it was, but it had clear plastic covers).

I had to use a drill to clear out the holes for the screws, but after that I could test fit it:

Not bad.

The edge finish definitely needs work, but it should be easy to fix by changing acrylic type and putting in a dedicated finishing pass. I still need to make the top part of the enclosure and put in threaded inserts for the screws to grab. After that, if everything looks good, I will make more cases out of nice wood and not MDF.

I don't know how long I plan to stay at home due to the coronavirus outbreak shelter in place policy and school all being moved online now. If I stay I should be able to make more progress on this case and finish it up.

Nixie Clock: Programming

I was able to get some simple code running on the Nixie Clock that displays the time and flashes the separators once every two seconds. I initially tried to write my own code for interfacing with the DS32321, but I kept running into issues. The ATMega328 datasheet (for the I2C setup and status codes), the DS3231 datasheet (for register addresses), and this website were all indispensable when writing the code. In the interest of time, I decided to just use a library. It worked perfectly, and I was able to set and retrieve the time.

I now need to add fancier animations and more modes for showing the date, temperature, and timers.

Nixie Clock: Finally Ready for Programming

After several pass with the toothbrush, isopropyl alcohol, and the occasional use of the soldering iron, I finally got all the issues ironed out. It tool quite a while to clean off all the flux, and every time I had to re-solder a pin, I made sure to clean the flux off.

Here is a video showing all segments operating independently:

Nixie Clock: Debugging

Well, as soon as I began programming, I ran into issues everywhere. Along with all the issue I listed yesterday, I realized that the flux I used, Kester SP-30, (which my dad for me a got for me a long time ago, but I never used it) was plumbing flux. And according to the data sheet, it is "too corrosive for electrical or electronics soldering applications." More searches turn up that not before long this flux will probably corrode all the SMD joints.

I decided to test my hypothesis that the flux residue was conductive and causing the numbers to light up randomly. I wrote up a simple program to cycle between all the digits one by one to see if there were any shorts between and pins. Interestingly, sometimes digits near the one that was actually powered would light up. As the number which was powered got closer to the numbers that weren't supposed to light, the ones no supposed to light up got brighter. There was definitely some sort of connection between multiple digits, and only thing I could think of was the residue.

I took out some isopropyl alcohol and cotton swaps to begin cleaning off the residue, hoping that the lighting up issues would go away and to see if I could minimize any future corrosion from the flux.I didn't have a brush handy so this took quite a long time, and in the end there were cotton fibers caught all over the place on pins and not all the flux in tight spaces were gone.

After cleaning up as much as I could, I tested the clock out again and it improved significantly. The last digit was no longer lighting up, and fewer digits were lighting up out of place. The reset line was now holding at 5V and didn't require the extra wire any more.

I will definitely need a brush to clean out all the flux residue. I hope this will solve all the issues so I can get back to programming.

Nixie Clock: Circuit Complete

I got the PCBs from OSHPark two days ago and finished assembling one board.

The assembled circuit.

Nixie Clock

I haven't done a project in while and decided I wanted to build another clock. I've always liked the look of Nixie tube clocks and decided to build one. This time I wanted to try using surface mount parts to keep the costs down.

For hardware, this clock uses six IN-14 Nixie tubes for the digits and two IN-3 tubes for the digit separators. An ATmega328 (I'm still not sure if I'm going to use an external crystal or not, but there is room for one on the PCB) is the main controller and a DS3231 RTC is used for time keeping.

Nixie tube with high voltage power supply in background.

Nixie tube testing.