Closed pico-coder closed 2 years ago
In the rev2 release, data parsing for digital only configs greater than 4 channels was rewritten to be clearer but also to be optimized to handle 1B,2B and 4B widths of DMA storage. And along with that this bug was fixed was well.
By inspection found a bug in how the DMA data is interpreted and sent to memory. In the send_slices_function there is a second_nibble and related code which supposed to use the upper 4 bits of an 8 bit double sample. But it's broken so we always send the lower value twice. Workarounds: Either: -Use 1-4 channels without any analog which will use the D4 optimized mode. -If analog channels are enabled, use 0 channels or 5 or more digital channels. Fix will be to simplify the storage of sample captures in these modes.