Atmel SAMC breakout

Aside from Microchip apparently deciding to rename all the peripheral registers, going from "running" to "blinking the LED" went pretty quickly.  Running at 48MHz, 2 wait states, and outputting the clock on its respective IO pin.  Woo hoo!

int main(void) {
    // Clock at reset is set to the internal 48MHz clock, divided by 12.
    NVMCTRL_REGS->NVMCTRL_CTRLB |= NVMCTRL_CTRLB_RWS(2); // 2 wait states for 48MHz
    OSCCTRL_REGS->OSCCTRL_OSC48MDIV = 0;  // no division
    GCLK_REGS->GCLK_GENCTRL[0] |= GCLK_GENCTRL_OE_Msk; // output the clock on a pin
    PORT_REGS->GROUP[0].PORT_DIRSET = GCLK0_OUT;
    PORT_REGS->GROUP[0].PORT_PMUX[GCLK0_PIN/2] |= 7;  // PMUX H is clkO
    PORT_REGS->GROUP[0].PORT_PINCFG[GCLK0_PIN] |= PORT_PINCFG_PMUXEN_Msk;
    PORT_REGS->GROUP[0].PORT_DIRSET = LED;
    
    while (1) {
        PORT_REGS->GROUP[0].PORT_OUTTGL = LED;
        mydelay();
    }
    return;
}

After fiddling with tools a bit (it looks like neither AS7 nor MPLABX 5.1 is quite up to using a PICKit4 with SAMC21), I have the board responding, have uploaded an empty C program, and can single-step through the C runtime startup code.

(Using (upgraded) MPLABX 5.25 and the PICKit4, on MacOS...)

So that's pretty significant!  Next up, Blink!

(A couple of people have recently "liked" this project, so I thought I should post an update...)

I'm ready to call the existing PCB design a failure.

Aside from the non-presence of USB in the chip, the PCB was designed with the "sweet spot price" prototype PCB of 50x50mm.  Since then, most of the cheap vendors have increased the max size of low-cost boards to 100x100mm, which would easily fit a much more "comfortable" design (Arduino Shield compatible, for instance.)

I haven't gotten around to actually trying to build one of the boards (scary QFN pinouts!), and now I'm not even sure it's worth trying :-(

So it turns out that i didn't read the datasheet carefully enough. A "quick" look showed that the SAMC21Gxxx was exactly the same as the SAMD21Gxxx, except for the 5V-ness. All the pins were in the same place, and etc.

But it turns out that none of the SAMC parts have a USB controller! SAMC21 has CAN on the same pins that SAMD21 has USB. I guess I can't blame anyone but myself, but I wish Atmel had used more distinguishing part numbers. Sigh.
(Found while looking at SW. I guess the boards are still usable.)

I sent the design out to OSHPark, and have received real boards that look pretty good (the slots for the USB connectors didn't work, but everything else looks pretty good.)

Gee, those pads look awfully tiny "In Real Life", compared to how big they looked blown up on my monitor when I was working on them in EAGLE! I'm reviewing tutorials on QFN soldering without fancy equipment, and have practiced with some simpler SMT components and the hotplate that I own; that seemed to go "ok", but I don't really have any "practice" QFN chips (I've got some SSOPs) or boards (though I did find some SO-sized scrap boards...)

And it seems that a toothpick is probably "too big" a tool for applying solder paste...