Output Devices

To figure out if I could use the LED by itself without the strip, I desoldered one and soldered jumper wires to the VDD, VSS, and DIN pins.

It was a challenge to do this, and part of the packaging got melted, but it still worked, albeit a bit unreliably. With the hunch that I could get this to work without resistors and capacitors, I decided I would make digit components that could be connected in series, just like a NeoPixel strip but for digits.

Happy about the result I asked Henk on Mattermost if these LEDs would be available in the lab, but he told me they might not be in stock. So, to ensure I had something to work on, I decided to also make schematics for NeoPixels cut from the strip (because desoldering is too error prone).

I created a new symbol in the KiCad Symbol Editor.

Then I created a footprint in the KiCad Footprint Editor.

Note: pin 1 (VSS) and pin 3 (VDD) are physically connected (I checked with a multimeter), so I drew them as one big pin.

I recreated the schematics using the new symbol.

And recreated the PCB from the new schematics.

The next day at the Waag, I discovered there were still some WS2812B LEDs in stock, so that was an interesting lesson in making symbols and footprints.

Milling my first design went badly. The traces were too close together, and Mods fused them. I did not notice at first, but as I was showing off to Henk, he noticed it immediately.

Henk also reminded me that this had happened to me before (did I miss this documenting?). So, I decided to pull up the PCB Design Rules.

Apparently, setting the Minimum Clearance to 0.4 with a 0.4 mill in Mods is asking for trouble. I changed the Minimum Clearance to 0.6, adjusted the PCB design to pass the rules and reran Mods. The preview look good and the milled PCB came out perfect.

After soldering the first LED (and the pin header), I tested it, and it worked. Then I soldered the second LED, but the second LED would not light up. I added the capacitors as shown in the datasheet (I was expecting I could do without).

But that did not work. I removed the second LED; the first still worked. Henk resoldered the second LED; it still did not work. I replaced the second LED with a new one; nope.

Because the PCB was looking worse with every iteration, I tried to solder the LEDs without a PCB to see if it would even work.

Same problem.

I even added the capacitors:

No show. Stunned, stressed, hungry, and irritated.

Then (local instructor) Leo entered the lab. He gave some advice, none of which was fruitful, but then I realized the code I wrote to test the first LED did not activate the second LED..

F**K!

I changed the code, and the second LED on the butchered PCB lit up.

Less stunned, still stressed and hungry, but with regained drive, I started milling another board because the traces starting to come off the first PCB after all the soldering and desoldering. While that was running, I cleaned up the first PCB as best as I could and soldered the other 5 LEDs on.

Soldering ws2812b components was annoying because solder doesn't really want to stick to the pins. I used soldering flux (S-39) to pre-tin the pins before putting them on the board. That made it a lot easier to solder them down, but it was a bit harder to get them flush on the board. I had so much practice soldering this day that making a second board was a breeze.

Running the code from the earlier strand test.

Now let's make a two-digit counter!

import machine, neopixel, time

A=0
B=1
C=2
D=3
E=4
F=5
G=6

DIGITS=[[A, B, C, D, E, F],
        [B, C],
        [A, B, G, E, D],
        [A, B, G, C, D],
        [F, B, G, C],
        [A, F, G, C, D],
        [A, F, G, E, C, D],
        [A, B, C],
        [A, B, C, D, E, F, G],
        [A, F, B, G, C, D]]

C = (100, 100, 100)

def digit(np, arr, offset=0):
    for i in range(7):
        np[offset + i] = (0, 0, 0)
    for i in arr:
        np[offset + i] = C

n= 14
np = neopixel.NeoPixel(machine.Pin(0), n)

for i in range(100):
    digit(np, DIGITS[i // 10], 0)
    digit(np, DIGITS[i % 10], 7)
    np.write()
    time.sleep(.5)

It works!