STUDY PATH/11 — FREQUENCY DOMAIN
CHAPTER 11 / FREQUENCY DOMAIN

Why −180° matters: Nyquist and margins

Negative feedback can become reinforcing feedback when the returning signal is inverted and strong enough.

01 / THE INTUITION

First, picture it

The summing junction subtracts the returning signal. If the loop adds another −180° phase shift, subtraction of an inverted signal becomes addition. If its magnitude is also 1, the return can sustain itself. This is why the critical Nyquist point is −1, not simply any point at −180°.

02 / THE IDEA

What the model says

For unity negative feedback, the characteristic equation is 1+L(s)=0. On the imaginary axis this can happen at L(jω)=−1: magnitude 1 and phase −180°. Phase margin is the extra phase lag from the gain-crossover phase to −180°: PM=180°+∠L(jωgc) using the appropriate unwrapped branch. Gain margin is 1/|L(jωpc)| at phase crossover, often reported as −20 log₁₀|L(jωpc)| dB. More margin often means more tolerance to uncertainty, but not a universal safety guarantee. With open-loop right-half-plane poles, multiple crossovers, delays or unusual loops, a single margin number is insufficient; use the full Nyquist encirclement criterion and a stated loop definition.

RELATION / 관계식1+L(jω)=0 ⇔ L(jω)=−1; PM=180°+∠L(jωgc)

At −180° alone, a loop can still be safely below unity gain. The combined magnitude-and-phase condition matters.

INTERACTIVE / L(jω) AT ONE FREQUENCYDrag the loop value toward −1
−1ImRe

Distance to −1: 0.77. This is a geometric illustration at one frequency, not a stability test for an entire loop.

03 / IN PRACTICE

Make it concrete

At 0 dB crossover, a phase of −135° gives a 45° phase margin under the standard single-crossover convention. An additional 45° of lag would reach the critical direction.

BE CAREFUL / 주의

'−180° means unstable' is false. The gain must also be considered, and Nyquist encirclements account for unstable open-loop poles.

Check your understanding+

At phase −180°, is a loop with |L|=0.1 necessarily at the critical point?

No. It is at −0.1, not −1. The standard gain margin there is 10×, or 20 dB, if the usual assumptions apply.