If you think about it, most power supplies expect to provide some DC current continuously. Particularly switchers. Not that a PSU cant be designed to behave well through zero current load; there's many lab grade benchtop units that can do this...
Consider an H bridge driven to make a sine wave into a resistive load, from the perspective of the powersupply. The bridge tips positive and current flows - from nearly zero - following the lobe of the 1st cycle half. Then the bridge tips negative and a similar current flows; the current for a complete cycle looks like rectified DC - two sequential positive going lobes, that touch zero - or darn close to it.
Consider a class A amp. That "cooks" the output devices with lots of current and voltage drop, while the output node moves up and down within the constant current flow; the "quiescent" current burn being much more than that delivered to the speaker. So the powersupply sees - essentially - a constant current load. At least a lot moreso than with the H bridge.
When we consider the sound resolution of very faint levels, we can see the class A is delivering such with a good chunk of constant current load on the powersupply, while other amplifiers... It would certainly be an easy experiment to do; add some DC load to your power supply and see if it sounds better. Class A, of course, not necessary!
Consider a woofer flapping about it's full Xmax - than directed to stop suddenly by the amplifier. Damping factor is one way to describe what that woofer "sees" electrically, looking back into the amplifier output, whose character gives it something to electrically "push against". Where does the emf go, as the output devices conduct the speaker current "somewhere"? Does this not reduce any small bias current the powersupply is providing even more toward zero? Perhaps another reason to preload your powersupply.
A couple of reasonable considerations - never mind any wooey-hooey about electrons preferring to flow continuously one way - or not having to stop.
Where's your powersupply operating when within that first watt?
Consider an H bridge driven to make a sine wave into a resistive load, from the perspective of the powersupply. The bridge tips positive and current flows - from nearly zero - following the lobe of the 1st cycle half. Then the bridge tips negative and a similar current flows; the current for a complete cycle looks like rectified DC - two sequential positive going lobes, that touch zero - or darn close to it.
Consider a class A amp. That "cooks" the output devices with lots of current and voltage drop, while the output node moves up and down within the constant current flow; the "quiescent" current burn being much more than that delivered to the speaker. So the powersupply sees - essentially - a constant current load. At least a lot moreso than with the H bridge.
When we consider the sound resolution of very faint levels, we can see the class A is delivering such with a good chunk of constant current load on the powersupply, while other amplifiers... It would certainly be an easy experiment to do; add some DC load to your power supply and see if it sounds better. Class A, of course, not necessary!
Consider a woofer flapping about it's full Xmax - than directed to stop suddenly by the amplifier. Damping factor is one way to describe what that woofer "sees" electrically, looking back into the amplifier output, whose character gives it something to electrically "push against". Where does the emf go, as the output devices conduct the speaker current "somewhere"? Does this not reduce any small bias current the powersupply is providing even more toward zero? Perhaps another reason to preload your powersupply.
A couple of reasonable considerations - never mind any wooey-hooey about electrons preferring to flow continuously one way - or not having to stop.
Where's your powersupply operating when within that first watt?