Not the best of the designs…

I wanted to have a go at building my own re-usable pcb with a microcontroller, similar to a Shitduino but without buying one, and with ample functionality.

My resources were more than limited and I didn’t want to waste time with a non-feasible project, so I had a look at ATSAMD21G18, the same microcontroller the Arduino Zero has, which has very good software support (Atmel Studio/Visual Studio and Atmel Framework), is a more or less a recent microcontroller, and, in case I’m too dumb for Atmel Studio, I could always go with Arduino IDE.

I sampled the cpu from Atmel, that is, Microchip, and I’d really like to thank them, because it was one in numerous samples which they could’ve rejected altogether, and because they actually sample1. I also had access to my school’s pcb laboratory, composed of a drill machine, milling machine, UV lamps and all the accessories to develop and etch a board (HCl + sodium perborate). This(img) is from a failed miniboard I trashed while I was learning how to use Eagle. Limited to one layer; bare plastic with exposed copper doesn’t look very attractive.

So, according to the datasheet,

datash1.jpg

I’ve tried my best at copying this, which resulted in:

scheme.pngscheme2.png

It does have errors, which I would like to comment on:

-the microcontroller is designed for various layers, so using one plane made me commit a lot of errors;
-the pins are in mixed-order (signal separation?) on the controller but the schematic symbol numbers them sequentially, so that’s what’s with all those ugly pin blocks;
-there’s an unnecessary trail of capacitors since they all follow the same paths on a single plane;
-I wanted to be able to make use of it both with an external oscillator and without, so I thought I could free the pins by placing a jumper near the filter capacitors but what I’ve should’ve done was place at least two of those to join and sever the track ties going to the oscillator. As it is now, that jumper will always stay there;
-I’ve actually labeled the pins into the copper layer because there’s no silkscreen, assuming they might come out well after developing.

>Why wouldn’t the manufacturer allow groups of pins that close to each other?

Well, I have no idea regarding EMI compliance but,

cinduced.jpg

that’s my phone’s GSM antenna inducing a current in the board, lighting up the blue LED, just by placing it near a centimeter of distance of the LED’s exposed tracks.

I’ve made no pictures of the mounting and soldering process, which would have been futile and difficult to do, but it ended looking like this:

mcrpr.jpg

,after quite a lot of attempts,

nefu.jpg

The farthest to the right is the good one, and, yes, as you can observe, I drilled the holes at eyesight precision because I wasn’t trusting Eagle with the drill map2.

All those are anonymous FR-4 boards from various suppliers, I don’t know any of them, I don’t know the pcb’s composition, thickness, nothing. But the only thing I can complain about is that the resin simply wouldn’t come out of the copper tracks so I got very ugly solder joints with burn marks from the resin.

I also had to use three types of solder, because lead-free offered me the comfort to solder almost everything else and it wouldn’t melt with the air from the heat gun, because I would use a lower temperature for the tin-bismuth (that’s how the microcontroller got soldered and ironically I have burn marks from when I’ve tried to heat the ground pour because the shitty copper plane wouldn’t tin itself no matter how much flux I fed it3). Some joints were redone with leaded solder.

I couldn’t prevent the degradation of the paint from the laser marks, and I still can’t believe it came out by wiping some flux out with ethanol on a cotton bud.

atslogo.jpg< before | after >atsboard.jpg
(for comparison, a working 18F65(img) that nearly melted while I glued it with SnCu, those cuts are the tweezers’ marks)

It does work and that shall come up in a follow up article.

Footnotes

  • ^1 which isn’t the case with, for example, Texas, or STM (rejects without reason), or NXP.

    this being said, I’ve bought capacitors from LCSC and they were all way off the specs with their tolerances. honestly, I’ve bought better from AliExpress.

  • ^2 but the real reason was the drill machine, made in the 80-90’s, which can’t properly handle <1mm drill bits (it breaks them in the process because the spinning head isn't fully vertically aligned, so the tension is distributed unevenly on the bit when it tries to drill down). I would even be ashamed to tell what's the normal use of that machine.

    this is why under the led there's a grainy plastic glued onto the legs, which is cyanoacrylate; I had to stick it somehow into place because the led's legs were way too thin for the 0,8 mm hole and the solder between the leg and the pad would get fractures.

  • ^3 I own no preheater, unless an aluminum case onto which the board was placed counts by having absorbed some of the heat gun’s heat.