I2S interface noise when compressor used

Hello Eason,

I think the compressor is being a red herring for you. I think it is causing another problem to be more audible. 

Thanks for attaching your project. I see you are only operating the serial port as a clock slave. Where are the clocks coming from? Is it from something like a BlueTooth receiver? 

Let me jump right to why I am asking. The Master clock coming in on the MCLKIN pin MUST be synchronous with the clocks coming into the I2S port. the part will count the number of master clock cycles per sample period. It needs to match up with the PLL setting of 256x fs? or 512xfs? The PLL Mode pins have to be setup correctly and the actual clock has to agree. If the number of cycles is not correct it will mute the part. 

So you may be hearing that and the compressor is making it more obvious? You have the compressor curve set to be a brick wall so I am guessing you can hear these mutes more easily. I have experimented with this on the bench and found that if the clocks are close to correct it can be difficult to hear the muting. It is easier with a low frequency tone so a brick wall compressor will accentuate the low end. 

Let me know your clocking details.

Thanks,

Dave T

Hello Dave,

Appreciate for your quick response.

For the clocking issue, i used a I2S transfer module, which receives stereo audio from the laptop and transfer to I2S signal then send it to ADAU1701. the module has the MCLK, LRCK,BCK. the module is the master clock with 24bit/48K, 256Fs. Besides, the CRYSTAL frequency is 12.288MHz.

Since the master clock of I2S module had connected to the MCLKIN pin of ADAU1701, Do you mean there is still clock synchronous issue?

Waiting for your further help!

Thanks,

Eason

Hello Eason,

I am confused. You mentioned a crystal and you mentioned the MCLK out of the I2S module. Which are you using to drive the ADAU1701 MCLKIN pin?

When using the I2S module you must use the MCLK out of the I2S module. You cannot use a crystal. 

Show us the schematic.

Thanks,

Dave T

Hello Dave,

Sorry for the confusion. 

The cystal is used for the A/D converter(PCM1808) of the I2S module, shown as below,

The MCLK is actually connected to ADAU1701 to drive the MCLKIN pin.

Thanks,

Eason

Hello Eason,

Thanks. So there will be no issue with the clocking and you have it setup correctly. 

There is nothing in the compressor that is different from using the ADCs. The only thing I can think of is that the signal is either much hotter (louder) from  the I2S ADC so the compressor is turning down the level (you have it set as a brick wall at a fairly low level) Then you turn up the output and you are hearing the noise floor. The other possible issue is a grounding or power issue with the connections to the I2S module causing the audio going into that I2S ADC to pick up the noise and you then hear it. This makes it seem like it is the compressor because when you remove it the audio is louder and so the noise floor is lower. 

Look for those sorts of issues. I assure you the compressor is not the cause of the noise. It is just making it easier to hear. You can always right clock on the window for the compressor curve and set it to flat and it should behave like when it is removed. 

This I2S module has a USB connector. USB power and ground connections are often very poorly done. I have experienced this myself many times. I would get a ground loop if I used the same computer for audio as I used for the USB connection and when I used two different computers it was fine. I have seen power supplies for computers be a huge problem adding noise. I have seen it with my MAC I use for analog audio on my desk. If it is plugged into the power supply it will inject a lot of digital noise. If I remove the power and let the laptop run off a battery it is fine. 

You could record a short clip of the noise if you like. I am guessing it is what I would describe as a digital noise. 

Dave T

Hello Dave,

So glad to get the confirmation that the project is setup correctly.

According to your suggestion, i changed the compressor curve to flat and test again, yes the behavor is just like the compressor removed. So there is nothing wrong with the compressor.

And thanks for sharing your experience on the USB power and ground issues. The I2S module is just used for learning and test, I will verify the power or ground issue with our new hardware platform later. 

Really appreciate for your detailed explanation.

Thanks,

Eason