ES9038Q2M Board

However in contrast to quantum world in audio the uncertainty principle has one exception: close-in phase noise. So the lower the close-in phase noise the better it sounds regardless of anything else.
You missed "noise floor modulation" effects, invisible on a FFT, "time variant" effects, etc... almost all reported by a "professional audio designer friend" with alleged no vested interest in spreading FUD.
 
Greetings. Forgive me if this was mentioned somewhere in the depths of this thread but I cannot find it.
Does anyone know the expected frequency content at the AVCC, VDD, DVDD, and AVDD pins?
It's difficult to design regulation without knowing what I'm filtering.

Also which ones take priority in filtering? The AVCC is obviously the most critical but the rest are a bit more obscure in what they do.
I have no idea what the circuits that VDD and DVDD drive actually do and the AVDD is said to be for the oscillator and the gain compensation, however if an external XO is used then it is only for gain compensation.
It is difficult to know which supplies are more important to filter and what frequencies are present at each.
Your wisdom would be appreciated.
 
Filtering? For ES9038Q2M?

IIRC, Vdd uses in internal regulator and the datasheet recommends a particular filter cap. Actual cap is not too critical IME just so long as its good enough.

Regarding VCCA, even if using an external clock there may be internal clock buffering and distribution that benefits to some degree from low noise on the power rail. Just saying I wouldn't assume that rail isn't important without experimental confirmation.

With the exception of AVCC, all the other dac power rails usually work fine with standard decoupling practices (e.g. .1uf X7R and 10uf tantalum close as possible to the load). If optimal performance were sought, I would use a separate regulator for each rail. For lowest cost some designs use a single regulator. In the latter case sometimes ferrite beads are used for additional isolation, a practice I would personally rather avoid if possible (since ferrites can produce hysteresis noise that is hard to make sense of on FFT yet which can in some cases affect SQ; if doubtful I would suggest to try it both ways and see what you think).

Regarding frequency range of noise on each rail, its hard to say exactly. Clock edge noise can extend up to several hundred MHz if not higher. Easy to see with an E-field probe and spectrum analyzer that dac chips produce extremely broadband noise locally.

Regarding AVCC, depending on how well everything else in the dac and the reproduction system are designed, almost any change in AVCC regulation or filtering can be audible. So, pick something you like the sound of if this is going to be for a listening dac. If a measurement dac then just use whatever measures well.