An update: I modified the design for the Zedboard to use the "hard" I2C controller in the PS instead of the FPGA implementation. This frees up a bunch of resources, and reduces power too. After I got this up and running, the zero samples problem mysteriously disappeared on my test board. Probably just a change in timing in the DMA periferal (which also changed interrupt vectors which kinda took me by surprise).
↧