Perhaps I'm mistaken, but I noticed in your schematic (and on your PCB) that the microcontroller's analog voltage supply pin (VDDANA) isn't connected to VCC (+3.3V).
In order to get your point of views about my new PCB release, I just joined here both PCB and Schematic I updated based on your advices.
Action I made :
=> I added a voltage supply for VDDANA
=> I checked SWDIO and SWCLK on Pin Header and there are not shorted anymore
=> I moved bypass capacitors (C7,C8,C9) closer to power Pins
Considering these updates, do you think this new release should work ?
I'm uncomfortable with the way the USB signals and Crystal signals are so close together.
Is your crystal really that tiny?
In general, I like to design my PCBs more conservatively, with traces further apart, and vias not too close to other things, unless it's really needed. See https://www.instructables.com/id/Make-hobbyist-PCBs-with-professional-CAD-tools-by-/ (although that's an extreme example, since it's aimed at home fabrication, but the principles are similar even with professionally fabbed boards.)
I also use the Abracon ABS07 on my SAMD21 custom boards, it's a nice compact crystal, together with 22pF load capacitors, located as close the to the crystal as is reasonably possible.
It's probably also worth looking at the SAMD21 datasheet, section 42 "Schematic Checklist". It details the recommended external circuitry and layout considerations for the microcontroller.
I'd also recommend placing an capacitor on your voltage regulator input, the MCP1703 suggest a 1uF ceramic capacitor. Personally, I'd move the regulator and its capacitors closer to the microcontroller.
Another minor consideration, is that you're driving the LED with around 3mA of current. This exceeds the SAMD21's default maximum pin current of 2mA. To drive the LED it's necessary to first set the pin's driver strength bit in software. Setting this bit allows the SAMD21's pin to source a maximum of 7mA.
The fact that your microcontroller is recognising your Atmel ICE is a significant step forward. Before starting on your next board iteration, I'd suggest that this is an opportunity to "play" with your existing board and attempt to get everthing working. As with any engineering project, large or small, any findings can then be used to improve the next version.
I'd carefully inspect the board to check that all the solder joints are good and that there are no dry joints or solder bridging between pins.
I'd also use a multimeter to check that the supplies 5V and 3.3v are correct. If the microcontroller's internal voltage regulator is working, you should be seeing around 1.8V on the VDDCORE pin. Is the RESET pin pulled high?
If you've got access to an oscilloscope, are you getting a 32.768kHz, 500mV, sine wave oscillation at the crystal? (That's what I'm seeing on my custom board using the ABS07 crystal).
Have you tried removing your Atmel ICE and then powering your board direct from the USB? It's also worth double tapping the reset button (two presses in quick succession), to try to place the microcontroller into bootloader mode and see if the board connects.
Thank you so much for all your coments. I succeed to implement all components on board and now it is working.
However, I am now working on a other MCU, the ATSAMD21G18A.
And I get the same problem my device is not recognized by the Atmel Ice and I cannot flash the MCU. I followed the documentation from Microchip, but I am now blocked ...
So, Here I share my current PCB + Schematics concerning this new board.
Please can you have a look and confirm me that my board look good ?