Decoupling of input ports

RMSacoustics Question

Mar 21, 2016 00:00

Hi there,

I am struggling with noise that might be generated inside the ADAU1772.

When working with the evaluation board I get a lower output noise with the same settings as with my final PCB design

A possible cause of this could be in the way of connecting the common mode port with the reference inputs and its decoupling.

According to the ADAU datasheet these should be connected and decoupled to ground via 10 uF.

However, on the evaluation board CM is not connected to the input references but all ports incl CM are separately decoupled to ground.

The datasheet tells that the decoupling of CM is necessary as it prevents crosstalk between de DACs and ADCs.

Now my question is whether this also influences noise and what the best way is to connect these ports.

Further I wonder if it is necessary to connect unused input ports to CM.

Thanks, Rob

adau1772

Integrated/Special Purpose A/D Converters

Audio CODECs

Report

ADIApproved Reply

Mar 22, 2016 00:00

Hello Rob,

Here is a THD+N graph of the evaluation board.

The red line is the response of the evaluation board with no modifications (input references and CM separately coupled to ground). The blue line is the response of the evaluation board with the input reference tied directly to common mode, and common mode coupled to ground (per datasheet recommendations).

As you can see, there is not much consequential difference between the two configurations, so the way that you have the input reference connected is correct. I suspect that the noise is introduced elsewhere in the design.

Could you provide a little more detail about the type and level of noise you're seeing, as well as your PCB layout, stackup, and schematic?

Thanks,

Tyler

581dc592b85819a90149ccc45cc8512b.png-620x235.png

Report

RMSacoustics Reply

Mar 24, 2016 00:00

Hi Tyler,

Thanks for the reply. Your graph is very useful.

I could solve the noise by meticulously comparing signals from the same situation on the evaluation board and the new design and it appeared to be caused by a failing ADAU 1772 input AD converter, probably due to a previous accidental error in the power supply (wrong polarity before the main regulator). After repair of theregulator all seemed to work fine and we forgot about the accidental error, but yesterday we remembered something from long ago that bipolar transistors can increase noise after inversion of the supply voltage so we made a second board with new parts. This second board showed -10 dB less noise. So problem solved.

But anyway thanks for the additional info.

Best regards,

Rob

Report