/diy/ - Do It Yourself

>>1666900
okay, not exactly trivial, but once you get it you might facepalm
first, when a non-linear gain element is supplied with negative feedback, the circuit gain is reduced by some factor and the gain element's error is reduced by approximately the same factor
second, an opamp's function is to amplify the difference between its two inputs by a large factor, usually about 100k. thus, the opamp will drive its output Vmid to whatever voltage the ??? box requires to drive Vout in turn to a voltage that makes Vsum ~= Vin. in fact, there is a tiny error voltage between the inputs, about (Vout-Vmid)/100k times the closed-loop gain iirc, which is required in order to push Vmid to that point. therefore, opamps can compensate out a lot of crap in post-output circuitry, as long as said circuitry's transfer function is monotonic and non-inverting, Pic related. likewise for feedback circuitry, used to great effect in filter and log amplifier circuits among others
now, with no opamps (and the pullup resistor you didn't show), the transistor will be fully on. with both opamps and diodes in place, one of the opamp outputs will be lower than the other, and its diode will be forward-biased, while the other is reverse-biased and effectively out of circuit. under normal, voltage-limited operation, the voltage opamp output will be the lower of the two, allowing just enough current to pass to maintain the set voltage (but no less than zero), while the current opamp output will call for more power, output near the + rail and effectively disconnected by its diode. in current-limited operation, the current opamp's output goes more negative than the voltage opamp's output and limits the amount of transistor current to approximately the set current, regardless of the voltage opamp's calls for more power. the ultimate result is that the transistor which is normally fully on is throttled back when *either* output measurement exceeds its set point