I2S 24 bit and 16 bit, the Signal to noise ratio was same

[OPQ]

Hi,

I'm evaluating a wireless microphone scenario, using 52's I2S and external ADC chip to sample sound signal, and than wirelessly transmitting it to another 52 for receiving, and the receiver uses I2S to play on the external DAC chip.

Now almost has been completed, but when I compared and tested the audio performance, I found that using 24 bit I2S and 16bit I2S, their SNR performance are the same (normally, the 24bit SNR should be better than 16bit).

During the test, audio AD and DA IC, I confirmed that they have been configured to work in 16bit or 24bit mode, because the actual working effect of the wireless microphone function is normal.

In addition, I have actually grabbed the LRCLK and BCLK pins of I2S. Indeed, there are 24 BCLKs in each LRCLK.

I debug and watch the easyDMA data of nrf_drv_i2s, and it seems that 4 bytes of data will be generated for each acquisition. Although the I2S 24-bit mode is used, I am not sure if there is a problem here?

I am not sure what caused it the I2S 24 bit and 16 bit, the Signal to noise ratio was same ,could you please give me some suggestions,Thanks!

[haakonsh]

I debug and watch the easyDMA data of nrf_drv_i2s, and it seems that 4 bytes of data will be generated for each acquisition. Although the I2S 24-bit mode is used, I am not sure if there is a problem here?

This is a feature of the I2S spec, a sign extension of 8 bits is added when used in 24-bit mode.
See Width, alignment and format, EasyDMA, and Memory mapping for 24 bit stereo. CONFIG.SWIDTH = 24Bit, CONFIG.CHANNELS = Stereo. 

The I2S devices needs to have the same alignment configuration otherwise you'll probably loose 8 lsb/msb depending on config.

[OPQ]

Hi haakonsh,

Thanks for reply.

config.format = NRF_I2S_FORMAT_I2S;
config.alignment = NRF_I2S_ALIGN_LEFT;

Do you mean these two configurations？

[haakonsh]

Yeah I think so.