-
I was thinking more of the method that varies the rail voltage cleanly and efficiently whilst allowing sufficient current. Need to do more reading!
Not sure a PC PSU is clean enough for audio without lots of RFI screening.
Dramatic_Hammer-
I was thinking more of the method that varies the rail voltage cleanly and efficiently
Should be easily achievable with a SMPS, you're effectively just looking for a 0-50V DC bench power supply, they can be had pretty cheap from China.
Not sure a PC PSU is clean enough for audio
Everything is powered by SMPSs now, obviously to drive just the logic in a class D amp you're looking at something more like the internal PSU of a phone rather than a 250W gaming PC PSU. The RFI issue is greatest at the high power end, and you have that problem from your class D output stage regardless of what you do with the logic PSU.
The alternative to messing with the rail voltage would be to mess with the pulse width, but I don't know whether you can get pulses down to narrow enough for the quiet passages if you also want the volume down to -40dB relative to full power. If your pulses get elongated due to finite rise time, you effectively have a compressor because the short pulses (low amplitude) have more excess area under the graph than long ones.
gbj_tester
Not exactly, because it wouldn't use the input signal to control the rail voltage. If the volume knob controls the rail voltage, you don't need to apply any attenuation in the digital domain, so you don't lose resolution.
I was looking at some Class D modules which will run on anything from 12V to 56V, the lower limit presumably being the minimum required to operate the logic chips. If you separate the logic power supply from the HV rails powering the output MOSFETs, you can just use a normal small low power computer PSU to power the logic and you can set the output stage rail voltage to a more or less arbitrarily low value to make it quiet.