PYBv4 opinions and suggestions by JohnHind

1. The board is too small to be an effective baseboard for shields/skins like Arduino (particularly if skins are half size as suggested). Better to make it a DIP module like Mbed designed to be plugged into a baseboard.
2. This implies that the connections should be pins on the underside of the board only. However the servo PWM pins could be double ended so they have dual use for servo connectors on top of the board (with the servo power pins on the top-side only).
3. I like the symmetrical IO concept with left and right sides rotated by 180 degrees. But rather than skins on top of the board, use a split baseboard. You lay two baseboards side by side with one rotated 180 degrees with respect to the other and the PyBoard plugs in on top as a connecting bridge. The baseboards can protrude out from under the PyBoard by as much as is required for their specific circuitry.
4. With careful design, the PyBoard could also be able to be mounted on two standard mini-breadboards e.g. Cool Components Breadboard laid side by side in the same manner as split baseboards. It is already just the right width, it is just a matter of avoiding any obstructions on the underside of the board. It would be possible to displace the 3v3 and GND pins so they engage with the breadboard power busses, releasing two extra pins from the 16 for IO. This would also provide better structural support for a split baseboard assembly.
5. Rather than (or as well as) a vbat pin, provide pads/vias for the user to optionally fit a CR2032 battery holder. At first sight, this seems far too big for the board, but a holder such as Adafruit CR2032 Battery Holder could be glued on top of the microcontroller chip and soldered with short wire tails to the board on either side of the chip.
6. The accellerometer should be optional - i.e. make the IO it uses available on pins as well. It can be removed if not wanted, or a version of the board can be produced with it omitted.

Some comments (DG):

1. The aim of the pyboard is to be as small as possible, but still functional. And having small skins is inline with this philosophy. Most skins can be made quite small with SMD parts, certainly fitting on half of the pyboard (even a CC3000 wifi module).

2. Some of the accelerometer's IO is available as GPIO in the form of an I2C bus. There is no extra room to break out the additional IO.

Comments (JH) after release of PYBv4 design:

1. I still think it would be better for the "skins" to underlay the PYB rather than sit on top - this is not precluded by the v4 design as detailed today it is just a matter of how pins are fitted. Underlay boards have a number of advantages: Do not obstruct the buttons and LEDs on the PYB and skins can protrude from under the PYB to mount components requiring access/more height; Skins can make up for the lack of mounting holes on the PYB; In many cases skins will be much larger than the PYB, so this will give a more stable assembly.

2. If the left and right rows of pins extend downward, and the two adjacent rows each side (i.e. the servo power pins and the four pins in mirror position on the left side) then it will be possible to use the PYB on breadboards for prototyping/experimenting/educational use. The left and right single row of pins would be plugged into two separate mini-breadboards. In this case 2x [mini breadboard] (http://www.coolcomponents.co.uk/breadboard-mini-linkable-black.html) will be more suitable. The servo power pins can extend upward only and the servo IO pins can extend both up and down. The assignments to the four pins in the same position on the left may require reviewing so they do not require pins extending downward.

3. There is no MOSFET switch on the USB-OTG power line - so the connector must be fixed as either host or device? Would it be possible to fit a handbag link rather than an optional zero-ohm resistor so function can be switched without soldering?