How do Digital Amps Mfrs. compare in sound?

Excellent points Stereo. I would like to add that at least 2 manufacturers (Bel Canto and JRDG) have started to add pre power supply current rectification in some models. This means that the power supply is essentially fed DC from a current reservoir of capacitors. The idea is the DC current is expected to:

1. largely eliminate 2 way AC line noise in and out of the amp.
2. Keep internal capacitors more optimally charged.
3. Be able to draw power for large transient from the DC reservoire regardless of the current state of AC 50/60 Hz phase.
4. Increase overall power conversion efficiency.

Some of these amps do sound amazingly musical to my ears. . . if and when more devices adopt various form of rectification prior to power supply, we may eventually be able to determine if there truly exists a correlation between input current rectification and sound quality.

I probably won't be adding much to the discussion by simply suggesting (as another poster did) that it depends on the manufacturer. For the past 2 weeks I've been listening to an ARC 150.2 which I swapped into my system in place of a tubed ARC D90B. The 150.2 is a tripath class T amp, which is really just another name for class D. I'm feeding it with a tubed ARC SP-8, and it sounds as good as, if not better than, the tubed amp it replaced. I also have the TAS issue which compared a number of class D amps, and if you read the review of the ARC 300.2 (the bigger brother of my amp) you see two opposite opinions - the assigned reviewer loved it (as I understand, he actually bought one after reviewing it) but Jonathan Valin, another TAS reviewer, definitely didn't like it. It seems to me that when you have two professional reviewers for the same publication disagreeing, it certainly points up that there is no "easy answer" to the question of class D sound.

SONICS
"Immediately good/refreshing sound" (how Class D is commonly described... at first) can be a very superficial thing. How many times was the sound of some gear (or music) you are long term happy with NOT necessarily so good at first ?

B. NEW TEAM - SUSPICIOUS NEWBISM
"Good" (sic) Class D is a massive Analog/audio/RF/Digital engineering tangle problem, & analog engineers are sage enough to know to leave such a vipers' nest alone. Seriously clever people and the biggest corporations have all thrown "brains and bucks" at, and into Class D, since the early 1960s. Over 40 years of trying! With still very few results on the high end stage, and, very late results all round.
People playing with overly-mathemeticised plans of reality may need to write out 1000 lines :
"Digital knowledge is not power amplifier engineering." !
Not a single master of analogue power amplifier engineering has turned to Class D. Shouldn't one know this, and, ask 'Why?’
By their nature, electronic engineers are usually eager to "migrate into the future technology".

C. Because the people who make Class D amplifiers are somewhat "ignorant", let alone their advertising copy-writers, they forget that the ACTUAL efficiency of an amp has to be multiplied by the power supply's efficiency, to arrive at the nett, realistic sum. This down-plays their rather spurious/overstated claims of high efficiency, once the comparison is made on an apples/apples basis.
In other words, taking best case practical round figures, an analog amp (70% efficient) is only a tad less efficient than a Class D amplifier (90%), whether they both use a mains frequency power supply of nominally 70% efficiency (70% x 70% = 49%; 70% x 90% = 63%); Here, Class D's 63% is only "28% ahead" of 49%.
Or, if both amps use a high-frequency (switching) power supply, then 70% x 90%, and 90% x 90%, are 63% & 81%. Again, 'D' is only some "28%" better. Do you wreck high-end sound for that? But, even the average 63% (of both) is FAR more than 200% more efficient than the most efficient car or indeed, power station.
If we dare include the power station efficiency (circa 28%), then any eco benefit from just one final part of the total energy loop, acting efficiently in Class D, is further reduced!

D. To anyone with an understanding of what makes existing good high-end hi-fi amplifiers, it is hard to see WHAT Class D achieves, that is useful. No one has ever announced it or written it down. It is not a logical next step. ONLY IF manufacturing convenience, cost, and material usage were put at the head of the list, before sonic quality.
This is from Ben Duncan , prominent audio designer.

Stanwal -

"Digital knowledge is not power amplifier engineering." !

Class D amps are NOT digital. There is no limit to resolution. Time (duty cycle) is as analog as voltage. Analog modulator used in class D Icepower is pretty much same thing as sigma-delta DAC without filtering. Also SACD is class D (you like it or not) as well as DSD recording.

"Not a single master of analogue power amplifier engineering has turned to Class D."

Not true - Jeff Rowland for instance not only turned to class D but also stopped making anything else. This alone should suggest something.

As for efficiency - It depends what you compare to. Efficiency of class AB is not 70% - it is around 40% or less. Class A is complete disaster with 7/8 of power wasted. Just imagine true class A 1kW amp. By definition it dissipates about 15kW. Now lets look at 1kW Icepowers . They have total efficiency of 79% (including power supply). In addition music power is only a few percent of average power.

If you don't understand what class D achieves - I will try to explain. In traditional class AB nonlinear characteristic of output transistors is corrected by negative feedback. Bandwidth is increased and THD as well as IMD reduced. Unfortunately TIM distortions are introduced causing not only unpleasant sound but fatigue as well. When feedback is not fast enough to respond amplifier goes momentarily into saturation and charge is trapped on semiconductor junction of output transistors making them non responsive. Small TIM sound pretty bad with sharp sound and expanded odd harmonics but more of TIM might not be so audible. Small gaps in sound are created and our brain fills missing pieces - causing fatigue after even short listening. In Class D there is no TIM since time and not the voltage is an analog quantity.

Convenience, cost, material usage have nothing to do with class D. It just sounds great. I know - I have one.

Audio research, Jeff Roland, Nuforce, Spectron, Channel Island, Kharma, Bel Canto, Pioneer, Rotel, Panasonic, Acoustic Reality, Flying mole, Digital Amp Company, Sonic Impact, Peachtree and Carver are all companies that quickly come to mind that now make a type of Class D amp.

Another benefit is Class D dissipates much less heat.