search

nxp-mcuxpresso / mcux-sdk

MCUXpresso SDK
BSD 3-Clause "New" or "Revised" License
301 stars 136 forks source link

wm6890.c: Suspicious bit rate calculation. #68

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#L103

This line looks strange, since it has the ' 2U' term in both numerator and denominator. In fact that function creates a bit rate which is half of what is should be. While recording from a microphone with a sample rate of 16000 and 16 bit width, codec set to master mode, the bit rate was 512kHz. It should have been 1024 kHz. Dropping the ' 2U' in the numerator fixed it. Maybe the calculation is right for playing and wrong for recording only, but then that must be handled.

mcuxcc commented 2 years ago

Hi @robert-hh

Thank you for your feedback.

I am sorry for late reply.

Let us discuss the issue you report,

The 2U added in the numerator is used to expand the bit clock divider to an integer, as the calculated bit clock divider maybe a fractional value image So 2U is added for non-float calculation.

And the below code convert the expanded divider to the register value image

robert-hh commented 2 years ago

I understand the intention. However, using that code I got a wrong bitclock frequency in master mode when recording from a microphone. The frequency was half of what it should be. So maybe the problem is somewhere else.

mcuxcc commented 2 years ago

is the issue come from the *2 of denominator? since there is only one channel MIC?

robert-hh commented 2 years ago

Even in Mono mode, both channels are present. For Mono, the signal is taken from the left channel. And, there is a second mic input at the headphone connector of the Dev boards.

If the divisor is smaller, the resulting frequency is higher.

robert-hh commented 2 years ago

I compared that with the timing in slave mode, and it seems fine.