loop gain

If negative feedback is a worker who keeps correcting the output, loop gain measures how strong that worker is. It answers a single question: if you sent a small signal once all the way around the loop — through the amplifier, out, through the feedback path, and back — how much bigger would it come back? A loop gain of 10,000 means the correcting force is 10,000 times stronger than the error it is fighting. That strength is what makes feedback work.

Loop gain is the product of the amplifier's forward gain A and the feedback fraction beta, written T = A times beta. Example: open-loop gain A = 100,000 and a feedback fraction beta = 1/10 give a loop gain T = 10,000. The bigger T is, the more tightly the closed-loop gain hugs the ideal value of 1/beta. In fact the relative error from the ideal is about 1/T: with T = 10,000 the real gain is within roughly 0.01 percent of ideal. Loop gain, not raw gain, is what buys you precision, and it is the quantity to look at when you want to know how good your feedback really is.

Why this matters: loop gain is not a fixed number — it falls as frequency rises, because the op-amp's open-loop gain A drops with frequency. At low frequencies you may have a loop gain of 100,000; by a few hundred kHz it may be down to 1. The single most important frequency in stability is where the loop gain passes through 1 (0 dB), because beyond there feedback can no longer correct anything, and the phase shift at exactly that point decides whether the circuit is calm or oscillates.