Here's another idea for the single-ended output stage. This is the beauty of audio (or electronics) design, there are so many variations of all the basic building blocks that hundreds of different circuits can be made with all the different combinations.
Studying for one of my exams, (not a course in electronics), I came up with this idea of including a P-channel device in the current sink for lower output impedance. I always seem to come up with ideas like this when I’m supposed to do something entirely different. I simulated it in Electronic Workbench using BJTs and indeed it seems to work. I used the BJTs shown simply because they are readily available when making circuits in EWB, MOSFETs are what I'd use in real life.
This illustrates the typical single-ended output with all NPN:
Then I include a PNP in the sinker section of the stage:
The output impedance should be half as much as the circuit using n-channel devices only, assuming the n- and p-devices are complementary.
(The small valued resistors running from output to ground were used to load the circuit during simulation, to get a guesstimate of the output impedance and are not part of the circuit in practice.)
Even though there’s a P-channel device in there, it is not a push-pull circuit, it’ll only source current like a normal all N-channel circuit would, dependent on idle current, (as well as output impedance).
So it’s home to the breadboard, and there goes the grades again!
PS. It didn't take long till I found out this is a common circuit in ICs. Darn! Everytime I think I come up with a unique idea, I get shot down by the fact that it's all been around for ages.
The day after: Looking a little closer at this, I think the extra p-device will make a serious miller capacitance problem. However that's something that can usually be worked out. Whether it's a good idea or not I'm not sure yet. Unfortunately I am so short on time at the moment that I'll have to get back to this later. It'd be smartest of me to remove this from here in case I find out it's a real bad idea, but what the hell, I learn best from those circuits that do not meet expectations, perhaps someone else out there will too. Anyone out there know if this will work properly?