In the system I'm developing using the ADAU1702, I'm getting a broadband noise floor out of the DACs that is ~10dB higher than the eval board (ADAU1701MINIZ). I'm using an AP to do an FFT of both systems looking at a DAC output pin in order to see this difference. The DACs are muted (but not powered down).
- If I power down the DAC I'm monitoring, the noise floor drops dramatically.
- If I power down the analog reference only, the noise floor drops as well, but not as far as when I power down the DAC.
- Noise is broadband; not any particular frequency that's pokes out for me to point where the noise might be coming from.
- CM, FILTA and FILTD pins look clean; cleaner than DAC output.
- Audio band noise on 3.3V looks equivalent between the two boards.
- Clock source on my board is from separate MCU. I connected my clock source to the demo board's (and swapped position of S1) and found no change in noise floor.
- DAC outputs have no loading currently (beyond the AP's 100K input impedance). Loading them down does not change the noise floor.
- Bulk caps on my board are electrolytic; bulk caps on demo board are MLCC.
- Decoupling caps (100nF) are as close to pins as on demo board.
- Analog and digital grounds are not isolated, as per datasheet. One giant ground plane, 4-layer board.
Where should I go next in order to determine where this noise is coming from or how to resolve it?