Hi there, how are you?
I recently developed a car audio signal processor using two ADAU1701s. Now I am working on the testing jig, which among the many tests it performs, it is supposed to measure the noise and distortion. As you can see bellow, it has an ADAU1442 + AD1938 codec. Because noise figures of AD1938 are better than ADAU1701's, I assumed that it would suffice to check if the noise floor of the device under test is equal or lower certain threshold, say -85 dB for example. Audioprecision measured -93 dB SNR.
However, even with everything disconnected, the noise floor is pretty mediocre (see second image). I really don't know what could possibly be causing it. Digital and Analog rails are separated and each one has a LM317 regulator, that alone should allow for 80dB PSRR, plus LC filters prior to the regulators. Board is 4 layer, which one is a solid ground plane. Nothing is connected to the board besides USBi. Other than that, the DAC of AD1938 puts out a beautifully clean sinewave, so I suppose system is healthy. Is there any chance that the DAC is working fine but the ADC may not be working so well?
It is running at 48Khz FS at the moment, but I will increase to 192Khz later, once I develop the functions to initialize the AD1938.
I had similar noise in a mixer design, which turned out to have several causes as described at https://ez.analog.com/message/169919#169919
Perhaps reading through this experience may…
Perhaps reading through this experience may help spark an idea to run additional tests and find the issue.