Has anyone switched from Class AB/A to Class D? Was it better? Was it worse?

@atmasphere "Actually there are a few class D amps that have a better first Watt than a good number of class A amps.” - list of “few” and “good” ones will be helpful! 

Analog output filter question: how output filter is designed, order, phase v frequency., loading considerations (reactive speaker x-over + spkr-dr complex loading), R/L/C components tolerance, frequency dependency, aging / magnetic core inductors nonlinearity? 

@westcoastaudiophile You might try the AGD Audion, one of the Orchard Audio amps or one of ours.

WRT to the analog filter, its job is to remove the switching artifacts that occur at the switching frequency. What is left will be a sine wave called the "residual". This sine wave is at the switching frequency so might be 500KHz. Its amplitude is so low that its of no concern to any loudspeaker driver, whose inductance will block it.

The filter is designed according to normal filter theory; typically second order. Usually its designed with a target load in mind, which might be 4 or 8 Ohms. Increasing the load (speaker) impedance results in a higher Q (Quality) of the filter effect. The load will shift the frequency a bit but usually its well outside of the audio band so that’s not a problem.

The characteristics of the capacitors and inductors involved is critical, since stray inductance (which might be caused by the leads of the capacitors, which extend not just from the circuit board but well into the part itself) can cause parasitics as well as the design of the choke.

It is for this reason its usually not a good idea to replace the caps or inductor in a class D amp in an effort to get better sound since reducing RFI is part of the filter’s task. Parasitic oscillation may be introduced! We found we had to be quite careful about the inductor in our amp, which uses a ferrite core. It was so critical we had to build our own jig to wind the chokes so we could control the inductance without strays. The vendors we tried to have the part custom wound simply didn’t seem to have the ability to get it right. To make the choke work properly we use a Litz wire. 

Of course the filter introduces phase shift as any filter. However, in our case the amplifier is using quite a lot of feedback and so is able to correct phase at high frequencies.

@atmasphere "target load in mind, which might be 4 or 8 Ohms” - in my 40y + sound works I haven’t seen such speaker, with purely resistive load (2/4/6/8/12/16)! In order to design output filter properly, we need to take complex model of loading, including variety of speaker cables, DC protection relay ckt, speaker crossover, spkr driver/s complex, often resonating impedances, which is tremendous engineering task! I’ve used class D amps in PA systems with active x-overs, where spkr was driven directly by the amp with short cable and w/o any x-overs in the path. For home use, I would definitely look for class D amps if I need power efficiency, that’s it. 

I like and have kept both. I’ve got a vintage Class A Pass Aleph3 (driven direct from a Wyred DAC2DSD) and ICE-based Wyred SX500 mono’s paired with a highly modified Rogue Metis (running Tung Sol VT99’s via adapters). The Aleph3 has an intimate continuity in image depth.. magical midrange while the Wyred’s (paired w my specific pre) have finesse and LF control while painting a detailed but natural cinematic stereo image. One is great for colder months and the other for summer and fall. I have tried the SX500’s driven direct from the DAC2DSD and thought it was nearly unlistenable, however. The right tube pre can have a way of ’sorting out’ older (but w well implemented output stage) class D topology into something that can still work now. The Rogue Metis Magnum is fairly dynamic (and a tube-roller’s dream/nightmare as an unbuffered tube design), but definitely a bit rough in stock form with stock tubes. I’ve got a good dozen other boxes in another room of other good-value separates; tube, Class D, and Class A, Class A/B, but these are the two setups I still use.. and I still go to shows. If I had money burning a hole in my pocket I’d like to try a Riviera Levante, Linnenberg or Kinki Mono’s.. a Nagra Classic amp.. perhaps some AGD Duets. I’d also like to try a Perla S50 integrated some day.. which I’ve heard with two way monitors before.

it’s more system synergy and implementation of the topology than topology, imho.

in my 40y + sound works I haven’t seen such speaker, with purely resistive load (2/4/6/8/12/16)! In order to design output filter properly, we need to take complex model of loading, including variety of speaker cables, DC protection relay ckt, speaker crossover, spkr driver/s complex, often resonating impedances, which is tremendous engineering task!

@westcoastaudiophile Actually its far simpler than that. A good class D amp can double power into a 4 Ohm load like any good class A or AB amp. There's no concern if the load isn't resistive! This is a no worries situation WRT the output filter.

In our case the amp is self oscillating, which means it also employs a lot of feedback. This prevents problems with capacitive or other odd loads causing resonance or the like- the amp is very well behaved with such loads as is any self-oscillating class D and so will be load invariant.