> Any help with explaining what each stage would be appreciated.
Transistors amplify. More transistors amplify more.
C2 R10 provide a pole at ~~4.2MHz.
Be realistic. Writing a "complete" essay on that plan would take time, and at $0.02/word(*) should be worth some money.
And what do you already know, specific or general? I'd hate to toss-off 1,000 words about basics if you really wanna know pole-splitting.
It makes more sense for you to post 300 words "How I think it works". Doing that WILL force you to think, and start to understand some bits. The parts you get wrong, someone here WILL jump-on and correct; that's just human nature. It is easier to correct a detail than to explain the whole thing. And nobody should be snooty about it; nobody is born knowing this stuff, we all been where you are.
> See attached if you dont have the schematic!
That schematic is wrong. If Q7 Base connected there, the output would be stuck up near the + supply rail. If you understand basic circuits, another more-likely connection may be "obvious"; but IIRC there is another more subtle mistake in this drawing and I don't feel like hunting for it again.
Read and digest Doug Self's pages on "design with discrete transistors":
http://www.dself.dsl.pipex.com/ampins/discrete/discrete.htm
Also his discourse on Power Amps (most PAs are giant opamps):
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
Take a break:
http://www.dself.dsl.pipex.com/ampins/canteen.htm
The 2520 amp is an electronic puzzle. Jensen's JE-990 is another very-good opamp, and Deane wrote it up in great detail. Must Read! Go to this page and request a copy by paper-mail:
http://www.jensen-transformers.com/apps_wp.html
Jensen analyzes frequency effects on a generic (though 990-like) opamp:
http://www.jensen-transformers.com/an/an001.pdf
For a quick-look, Hardy's JH-990 is nearly the same thing and schematic (but without heavy technical analysis):
http://www.johnhardyco.com/pdf/990-2007.pdf
---------------------------------------------------
(*) Actually technical writing should be paid more, even $0.20/word, but that rarely happens.