As per title, I'm trying to send a simple 'READ' (0x03) command to the device and get the contents of the CANCTRL register (0x0F). I even moved past our project board and bought Microchip's development board for the MCP25625 to rule out any weird HW design issue. In both cases I've measured the CLKOUT pin using my oscilloscope and it's correctly reading 2.5MHz (20MHz crystal / 8 default prescaler). So assuming the chip is getting the correct clock signal in and should be started correctly.
The only odd thing about both my setups is that I'm powering VDD from 3V3 and Vdda from 5V.
SPI MISO is pulled high using a 10K resistor to 3V3. SPI is running in mode 0,0 with data valid on the clock's leading edge, as per the picture attached. I tried everything so far: 1. Disconnected the MISO line from the MCU. Tried both pulling it up and leaving it floating. Same result (0xFF when it's pulled up, 0x00 when it's left floating). 2. Tried all SPI modes.
I'm genuinely out of ideas, the SPI data looks correct, but the chip isn't replying at all.
LE: Have discovered that, as soon as SPI is enabled and configured on the processor (Kinetis MKW41Z), the high voltage of all SPI lines controlled by the peripheral drops to 1.9V. I'm using 10K 3V3 pull-ups on all SPI lines. Currently trying to figure out why that would happen.