The chassis is grounded and signal ground is coupled via 47nF capacitor to the chassis. For all measurements, signal source has been Topping D10s DAC that is connected through USB-ISO isolator to kill the loop, ADC is an USB soundcard. The signal from the amp output goes to the resistive voltage divider and then to the soundcard RCA input. This path completely wired with RG-59 cables. Look, I am very good in killing interferences and interference shield currents
. The reasons is that I was designing wideband measuring instruments that were asked to work with small signals in the hostile environment of High Voltage testing labs during lightning and surge impulse tests and in high power testing labs. This is unbelievably similar to small signal audio issues.
I will post some ultrasound spectra resulting from these SMPS operations. They start at about 70kHz, nothing in audio band. Still less issues than with class D switching that masks SMPS residuals.
If you are interested in the 2nd half of the story - the amp with SMPSs is supplied from 1:1 isolation transformer from the 230V/50Hz room mains. This is how the SMPS leakage currents are killed almost totally. But even without this transformer, the 50Hz leakage is completely acceptable, you will see it somewhere at -110dBr in those plots.