The Xmega samples arrived (two of them: XMega32E5 and XMega128A4U). I had a breakout board all ready for the E5, but will need to wait for a 44 pin breakout for the A4U.

I went ahead and got the chip soldered onto the breakout board, fired up AVRDude (an older version due to an incompatibility between the LUFA implementation of AVRISPMkII and AVRDude version 6.1)... and was met with the error message 'AVR Part "x32e5" not found'. Boo! So it turns out that support for this chip was not added until version 6, which breaks compatibility with my only PDI programmer.

(The fix is pretty simple, of course: add an .avrduderc file which includes the part definition for this chip, and run it against version 5. I will proceed with that later today.)

However, I am still debating with myself the merits of using one MCU per channel vs. using a single master. There are pros and cons to both: most notably, using one chip per channel will give me the ability to have more than 4 channels and will allow for high frequency changes to the voltage (i.e. output an A/C waveform), but will complicate firmware upgrades (you need to flash each MCU for a total of N times), plus it makes it impossible to upgrade the firmware over USB. By using a single master chip, I would have to use some sort of external DAC (whether that be a separate IC communicating via SPI or I2C, or whether it be a RC network to do a low pass filter on PWM output). Whatever the method, it would probably not be fast enough to output a smooth A/C waveform without jitters.

Anyway, the final decision won't be made until I can evaluate the XMega chips, most importantly the quality / speed of the DACs.

Cheers