Closed dkonigsberg closed 2 months ago
Given that the ability of MLCCs to filter high frequency noise has more to do with the package size, than the capacitance value, and given that we'd rather not put 0402 capacitors on the main board, this issue may not be worth implementing. Chances are that any promising preliminary results from kludging these onto an existing board were not actually conclusive, and other things had a larger impact.
This is hopefully taken care of, well enough, by some changes to power input filtering and minor bypass capacitor changes around the USB hardware. Considering the details of how the EMI issues have been manifesting, if they still exist on Rev E hardware then its probably easier to just make sure the software is robust enough to recover from them.
The USB2422 IC is especially susceptible to conducted EMI. The evaluation board datasheet suggests optionally adding 1nF capacitors in a variety of places to help with this. Lab tests have shown that adding these definitely helps with the problem. Additionally, adding a 1nF capacitor to the MCU's VDDA pin (which drives the internal PLL, and hence the clock line that runs to the USB2422), also seems to help.
Thus, this task is to add 1nF bypass capacitors (in addition to the ones already there) to the following places:
This is related to #56