a lot of the class D amps also have a high output impedance because of the lack of a feedback loop. it is rather difficult to provide feedback in a class D amp (it can be done but it isn't as simple as with class A and B amps). i have seen tube amps with some feedback, which does reduce the output impedance below the impedance of the transformer, but usually not by much. an interesting tube design is the "circlotron" which uses two floating power supplies, and drives the output transformer from the cathodes. there was an electro-voice amplifier that used this method, and actually had a control in the feedback loop that could adjust the output impedance down to about an ohm or so. solid state amps, because of their very large open loop gain and very effective feedback, often have output impedances of about a milliohm. some Crown PA amps have output impedances of a few tenths of a milliohm (Damping Factors of anywhere from 5,000 to 10,000). of course, adding the resistance of the speaker wires increases the output impedance to the amp plus the speaker wires.
for anybody that thinks AudioGuru and i just took the discussion "into the ozone" (pardon the pun), the output impedance of an amplifier determines how the amplifier and the speaker interact with each other. speakers are actually a linear magnetic motor. magnetic motors produce a "back EMF" when the coil moves through the magnetic field. to see this effect with a speaker, connect a voltmeter to the speaker terminals and tap on the speaker cone. you will see the meter read some voltage. this is the back EMF a speaker produces. when you play music, the cone moves. the cone has mass, so it has inertia. when the amplifier changes the current through a speaker, the cone moves to duplicate the change in the form of motion. since the cone has mass, the cone doesn't follow the current instantaneously, and the inertia produces an opposing voltage at the terminals of the speaker (this is an oversimplified description, there's really a LOT going on here....). this back EMF can actually cause the speaker to "do it's own thing" and not follow the amplifier as intended.... now back to a physical demonstration for a second. while you are tapping on the speaker cone, place a short across the terminals of the driver itself. you hear the motion dampen (this demonstration works best with a subwoofer with very loose suspension) and the sound of your tapping changes. this is because the back EMF is driving a short circuit and produces a braking effect in the magnetic field around the voice coil (the same effect happens with a DC motor when you short the motor wires and try to turn the shaft). this is what a low output impedance in an amplifier does, and in effect allows the feedback loop in the amplifier to exert a certain amount of control over the cone motion. the lower the output impedance, the more accurately the cone follows the amplifier. how the feedback lowers the output impedance, is that any voltage that shows up on the output terminals, that is not present at the input (whether it's distortion produced by the amplifier or back EMF from the speaker) gets fed back to the input and compared against the input signal. if it's different, the amplifier corrects it. this feedback effect can by this correcting action lower the output impedance far below the actual output resistance of the amplifier. in other words, (and i've actually done this experiment) if we took the speaker output terminal, and placed a 10 ohm resistor in series with the output, that will increase the output impedance by 10 ohms. if we then move the feedback "tap" from the speaker line of the amp board to the speaker side of the 10 ohm resistor, the output impedance will drop back down to far below 10 ohms. the amplifier will do whatever it must to keep the feedback and the input signal equal. of course with the 10 ohms "internal" resistance, the overall gain of the amp remains the same, but the level that the amp clips at is reduced, because the amplifier is working harder to produce the same voltage at the feedback "tap". the actual output of the amplifier is still being driven rail to rail, but the voltage drops across the "internal" resistor as it forms a voltage divider with the load.