DIY ES9018 Hi-end USB DAC

I'll try to answer myself. LM49710 max output current is 25mA as opposed to AD797 suggested by ESS which has typical 50mA.

Not sure why LM ever worked for me. Maybe it was marginal. It might have been also different sampling freq which loads the regulator less...
 
Although a colleague from Germany had not looked into the topic for a long time, I decided to point to the source of the power problems that caused him to hum in the sound. I figured out the board of this Chinese manufacturer. He has a typical error in power calculations. The transformer actually produces 7 volts of voltage. However, diodes with a very large drop in the forward direction are used in the rectifier. Therefore, the output of the first stabilizer remains a component variable. It goes further to the analog power of the DAC and then to the output.
The problem is solved simply by replacing the diodes with Schottky diodes with a very low voltage drop. This is verified by me in practice.
Probably the correct diodes were used in the prototype. But with a series in China, it is customary to reduce the cost by using cheaper components. In this case, it played a fatal role with sound quality.