search

nxp-mcuxpresso / mcux-sdk

MCUXpresso SDK
BSD 3-Clause "New" or "Revised" License
323 stars 142 forks source link

wm8960.c: Precision loss in calculation. #69

Closed robert-hh closed 2 years ago

robert-hh commented 2 years ago

https://github.com/NXPmicro/mcux-sdk/blob/7b545788e3f6e0c1a8decae21e428ed8d5b5a8e7/components/codec/wm8960/fsl_wm8960.c#L68

The order of arguments in this expression causes wrong values for pllN. For example with iniputMclk of 24MHz and outputClk of 11.2896MHz, the calculated pllN is 6, when it should be 7. If sticking with integer arithmetic, the line must read;

            pllN      = (pllF2 * sysclkDiv) / inputMclk;

I added brackets to make clear, what must happen. The same happens with fewer problems a few lines down when calculating pllR. There, the error is 1000 times smaller.

mcuxcc commented 2 years ago

Hi @robert-hh

Thank you for the feedback.

I am sorry for late the reply.

I am agree with you about the issue, so i will create a Pull request for it.

Thanks!

robert-hh commented 2 years ago

Thanks.

mcuxsusan commented 2 years ago

Update: the improvement patch is available in https://github.com/NXPmicro/mcux-sdk/commit/ce9a8be7b39ac939fba6a4527a079ae98eb045b2

robert-hh commented 2 years ago

Thank you very much. Just for information: I use a version of the driver translated to Python. It is here: https://github.com/robert-hh/wm8960