Understanding SAI1/2 and internal sampling rates, and speaker output

No one?

This is what I'm thinking:

Is there any reason not to use the Speaker outputs with 30-40mm 8ohm drivers?

Yes, I would advise against it using a speaker output as you will have an elevated noise floor with the speaker output. The full scale output from the headphone amplifier is 1 VRMS with a Dynamic Range of 101dB when driven from the DAC. The speaker’s maximum output is 2VRMS with an output noise of 53uVRMS. We can estimate the noise output of the headphone amplifier in VRMS using the dynamic range. In the datasheet the dynamic range measurement is defined in note 4 as the EIAJ dynamic range. This estimate will be a little different due to the measured methodology but will still be a good estimate. Using the maximum input level and scaling it by the SNR of 101 we get an estimated noise level of 9uVRMS or a 15.5dB lower noise floor. The recommended option would to be use the line output and send the audio to two separate headphone amplifiers. I am recommending some headphone amplifiers that have a output power vs load resistance plots that are defined at 8ohms so you can select one at your leisure. The following headphone amplifiers would be good for your application the MAX9820, MAX13330/1 & MAX9791/2 not all of them have analog input.

My question here is can we run DAI1 as a target to the radio so it takes in 16 or 8k digital audio, but still run the internal sampling rates higher? 32 to 48kHz?

The part has a sample rate converter than can be used to match a higher internal rate (see page 88 in the datasheet).

 I'd love to also do USB audio from STM but it doesn't look like DAI2 has audio out other than looping DAI1's in?

The only restrictions on the DAI is that you have 1 record path. However, you can have two separate DAI inputs. Your headphone outputs are separate from your record path’s routings.

Thank you Phenix. Understood on the output drivers.

One more question. If I wire my STM32 to SAI1 and run it at 48kHz, along with the ADCs, DACs, and everything else, and I wire my Bluetooth controller to DAI2 at 8/16k:

1) Do the clocks (BCLK/LRCLK) need to be synced with DAI1?

2) Can I just feed the input path down through everything running at 48kHz? will the DACs and filters and things just use 3 or 6 samples of the same data or will this just not work at all?

I can get around not having SDOUT to the radio because I have the audio at the STM32, so I can down sample and pass it to the radio from the STM32.

That limitation means I can't run the Bluetooth SDIN2 down and through the Sample Rate Converter because it would end up getting looped back into the Bluetooth from the STM32.

PhenixNunlee Any thoughts on what happens if we run the SAI2 in at a lower frequency than everything else?

Thanks

1) Do the clocks (BCLK/LRCLK) need to be synced with DAI1? yes, they would need to be synced by turning on the sample rate converter.

2) Can I just feed the input path down through everything running at 48kHz? will the DACs and filters and things just use 3 or 6 samples of the same data or will this just not work at all? See previous answer.

1) Do the clocks (BCLK/LRCLK) need to be synced with DAI1? yes, they would need to be synced by turning on the sample rate converter.

2) Can I just feed the input path down through everything running at 48kHz? will the DACs and filters and things just use 3 or 6 samples of the same data or will this just not work at all? See previous answer.