[quote author="featherpillow"]I've finally gotten to the point where I think I've figured some of this out--thanks so much for the explanation, PRR. It took me a couple of months of reading it to get a better understanding of it.
So, for the sake of simplicity, when we're driving into the Miller C, is it the collector current (as Leach says) that's the determining current here? That seems to explain the reason why, as Self points out, a current mirror on the collector would roughly double slew without the problems typically associated with emitter degeneration.
However, I was reviewing some stuff that Luns (UC Berklee?) wrote, that seems to suggest that it's actually TOTAL current in the differential that we want to consider here. Am I unnecessarily confused by this?
Also, for the life of me, I can't seem to find the formula to determine the value of our compensation C. I know the items involved here that determine this--1/Gm, bandwidth, emitter resistance, etc. I would guess it would be some type of 1/distance thing involving bandwidth, right? Something tells me there was a 5534 datasheet of yore that had that forumula in it.
I also can't seem to get my head around the manner used to determine where to split the poles--everything I read seems to discuss it as some vague notional thing (IIRC the exact value wanders a bit anyway, right?) but I've never read any specifics. Or maybe I am reading specifics, but it's not gelling in my tiny mind :roll:[/quote]
The current mirror also doubles the small-signal gain I believe, as well as adding its own phase shifts and noise. It does have the advantages that come with inherent balance at low-moderate frequencies compared to a simple load resistor, so that the precise value of the emitters' current is unimportant.
The slew rate, assuming the standard diff input + mirror circuit with second-stage feedback integrator, is just I/C where I is the total emitter current. We are hard-switching the input stage and steering all the emitter current or its inversion through the mirror to the integrator input, which responds with a voltage ramp at the output.
There are many discussions in greater detail, including more quantitative ones, of compensation.
An old reference by Burr-Br*wn folks and their AZ pals at the time has the basics. Feucht has a brisk and much more general discussion in his Handbook of Analog Circuit Design and its CD-ROM expanded/revised version.
Like most things, the simple first-order approaches will get you a long way but the subtleties can fill volumes. There is no simple single formula though---you must know what the parasitics are, like the output stage behavior as outlined by PRR, before you can figure out what you need. You also need to know what the total system configuration is---feedback network, output load, etc.
No matter how well the open-loop gain of an op amp follows a single-pole response, it is always possible to get it to oscillate given the "right" feedback network :razz: Conversly, you can usually take an amp with a nonideal open-loop response and have a stable circuit with the proper choice of external feedback components. Feucht's material is particularly good on the latter.