I have been making some progress this past month, mostly on the USB module. I'm trying to verify the current hardware before doing another revision of the PCBs. So I also got the SAM-BA bootloader from Microchip working on one of the modules.

For the USB stack I ended up using the TinyUSB library because it has both audio and MIDI device drivers ready. It took a while to get it working even with the library, but it is now mostly working in device mode. I have some issues with the audio streaming, the left and right channels are constantly switching and the sample rate is too fast. The channel switching is probably a firmware issue, it is quite annoying as it degrades the audio quality. I have tried using separate DMA channels for each audio channel and checking the word select signal, to no avail. The fast sample rate is caused by an inaccuracy of the internal oscillator, so that's another reason to add an external crystal.

I also made a fork of the library to attempt implementing host mode on the SAMD21 controller. Luckily the library handles all the high-level USB stuff, and "only" a set of functions must be defined to let it interface with the hardware. I got it partially working with the host hid_example. When the PS4 controller is directly connected to the MCU it works, but if it is connected via a USB hub the device will stall forever. The code also gets stuck in a wait function occasionally because the frame number does not increment when no device is connected, but it can recover from this state. My fork is currently limited to 8 endpoints, but the SAMD21 supports virtual pipes to increase this. I will probably implement this later. The host stack is apparently a work in progress, so I will wait a bit to see if it improves.

This was the peak debugging setup. Picoscope to probe the physical I2S signals, In-circuit debugger to read internal data. A loopback connection from DAC to ADC and a speaker on one of the channels.

Next, I will probably improve the firmware of the CV module.