The Mola Mola is class D, but I have a true digital Lyngdorf pulse width modulated amp as well. No analog conversion. Input is only digital and there is no traditional DAC in it. The PWM bitstream is converted to analog by a passive RC network. I wish I could say it sounds great. It sounds sterile to my ears.
FWIW Dept.: Pulse Width Modulation is an analog encoding scheme that's been around since the tube era. You can convert from digital to PWM (PWM is a very typical encoding scheme for a class D amp); so this sounds like a class D with digital input to me, regardless of what the marketing text says.
Regardless, I don't see why there should be a controversy about the 'sound' of amps except for human nature. The simple fact is that harmonics of fundamental tones sculpt the sound of musical instruments; if an amplifier has distortion (and all do) that distortion further sculpts the sound of those instruments. Too often I see the comment that (paraphrasing) 'this distortion component is at -80dB so is inaudible' as if its somehow a separate signal from the audio signal itself. Its not separate and its a modifier of the audio signal.
Our ears respond to harmonics by assigning tonality (as a result, assigns tonality to all forms of distortion, including IMD and aliasing).
The tricky bit is that measurement has gotten so good that we can now predict the sound of an amplifier from the measurements, if all the measurements were actually made. Both the measurement and subjectivist guys seem to hate this!
******************
Music is processed in the limbic portions of the brain. This is why we often feel a toe-tapping or movement response. But if the brain detects a problem, it unconsciously moves the processing to the cerebral cortex. Things like distortion, the speed of the presentation, FR errors and the like can all trip this tipping point. At that point a lot of the emotional involvement is lost.
Topologies sound different because the kind of non-linearity in the topology is different and so generates a different harmonic spectrum. As an example this is why I've been an advocate of fully differential balanced circuits, since they generate what is mathematically a cubic non-linearity. This is because even ordered harmonics are cancelled throughout the circuit, resulting in a dominant 3rd harmonic, with succeeding harmonics falling off on an exponential curve as the order of the harmonic is increased. That 3rd will be at a level slightly less than the same harmonic is seen in an SET. SETs generate a quadratic non-linearity so the 2nd is dominant. They too can have the harmonics fall off on an exponential curve but one that has a different exponent. Put another way a cubic non-linearity is inherently lower distortion since distortion is not compounded as much from stage to stage throughout the circuit. When you combine both non-linearities, as seen in PP power amps with a single-ended voltage amplifier, you get a prominent 5th as well (see Norman Crowhurst; we've known this for a good 60 years or more...).
If higher ordered harmonics are not masked by lower orders, then they will contribute to harshness and brightness. This has been the sin of solid state amps since their inception and is why tubes are still around. But its an engineering issue, not an inherent failing of semiconductors, to sort that out and get it right.