3.5.42
Ek feedback loop tab unstable hoti hai jab woh "apni hi tail ko chase" karti hai — signal wapas aata hai bilkul out of phase aur utna hi strong (ya zyada) jitna tune bheja tha. Stability margins measure karte hain tum us catastrophe se kitni door ho : kitna extra gain add kar sakte ho pehle ki system self-destruct kare (gain margin), aur kitni extra phase lag bardaasht ho sakti hai (phase margin).
Ek closed loop consider karo jiska open-loop transfer function L ( s ) = G ( s ) H ( s )
T ( s ) = 1 + L ( s ) G ( s )
Intuition Instability kahan rehti haiDenominator 1 + L ( s ) L ( s ) = − 1 − 1 − 18 0 ∘ L ( j ω )
Toh stability ka matlab hai curve L ( j ω ) − 1 . Gain aur phase margins do rulers hain jo us doori ko do specific directions mein measure karte hain.
Definition Gain crossover & phase crossover
Gain crossover frequency ω g c ∣ L ( j ω ) ∣ = 1 0 Phase crossover frequency ω p c ∠ L ( j ω ) = − 18 0 ∘
Kyunki point − 1 = 1∠ ( − 18 0 ∘ ) dono — magnitude 1 aur phase − 18 0 ∘
Jis frequency par phase already − 18 0 ∘ → Gain Margin (GM) Jis frequency par gain already 1 hai, wahan phase − 18 0 ∘ → Phase Margin (PM)
Loop ko phase crossover ω p c ∠ L = − 18 0 ∘ L ( j ω p c ) negative real number − ∣ L ( j ω p c ) ∣ − 1
Extra multiplicative gain k − 1 k ⋅ ∣ L ( j ω p c ) ∣ = 1 ⇒ k = ∣ L ( j ω p c ) ∣ 1
Gain crossover ω g c ∣ L ∣ = 1 − 18 0 ∘ extra phase lag hai jo hum tolerate kar sakte hain usse pehle.
ω g c ∠ L ( j ω g c ) − 13 5 ∘ − 18 0 ∘
PM = ∠ L ( j ω g c ) − ( − 18 0 ∘ ) = 18 0 ∘ + ∠ L ( j ω g c )
Nyquist view: L ( j ω ) − 1 − 18 0 ∘ Bode view: Do stacked plots (magnitude dB, phase deg). Ek vertical line kheencho jahan phase = − 18 0 ∘ 0 dB tak ki distance hai. Ek vertical line kheencho jahan magnitude = 0 − 18 0 ∘
Worked example Worked Example 1 —L ( s ) = s ( s + 1 ) ( s + 2 ) K K = 1 Phase crossover find karo. ∠ L = − 9 0 ∘ − arctan ω − arctan ( ω /2 ) = − 18 0 ∘ Yeh step kyun? Hume woh frequency chahiye jahan phase danger angle hit kare, taaki GM compute kar sakein.
Set karo arctan ω + arctan ( ω /2 ) = 9 0 ∘ tan → ∞ 1 − 2 ω 2 = 0 ω p c = 2 Wahan magnitude: ∣ L ∣ = ω ω 2 + 1 ω 2 + 4 1 ω = 2 = 2 ⋅ 3 ⋅ 6 1 = 36 1 = 6 1 Kyun? ω p c ∣ L ∣ GM = 1/∣ L ∣ = 6 = 20 log 10 6 ≈ 15.6
Worked example Worked Example 2 — Phase margin ofL ( s ) = s ( s + 1 ) 1 Gain crossover: ∣ L ∣ = ω ω 2 + 1 1 = 1 ω 2 ( ω 2 + 1 ) = 1 ω 4 + ω 2 − 1 = 0 u = ω 2 u = 2 − 1 + 5 ≈ 0.618 ω g c ≈ 0.786 Yeh step kyun? PM wahan measure hoti hai jahan gain unity ho.
Phase wahan: ∠ L = − 9 0 ∘ − arctan ( 0.786 ) = − 9 0 ∘ − 38. 2 ∘ = − 128. 2 ∘ PM = 18 0 ∘ − 128. 2 ∘ = 51. 8 ∘ 3 0 ∘ 6 0 ∘
Common mistake Classic errors ko steel-man karna1. "Zyada gain hamesha safer / faster hota hai, toh K
Kyun sahi lagta hai: zyada gain se faster response aur chhota steady-state error milta hai.
Fix: K upar shift ho jaata hai, GM shrink ho jaati hai. GM cross karte hi loop unstable ho jaata hai. Speed aur stability mein trade-off hai.
2. "Positive GM akela matlab stable hai."
Kyun sahi lagta hai: GM ko literally stability margin kaha jaata hai.
Fix: Tumhe dono chahiye — GM > 0 aur PM > 0
3. Kaun si frequency kaun se margin ke saath hai — confuse karna.
Kyun sahi lagta hai: dono "crossings" hain.
Fix: GM phase crossover par rehta hai (∠ = − 18 0 ∘ PM gain crossover par rehta hai (∣ L ∣ = 1 margin wahan measure hoti hai jahan DOOSRI quantity apni critical value par ho.
4. "Phase margin ∠ L
Fix: PM = 18 0 ∘ + ∠ L ( j ω g c ) − 18 0 ∘ gap hai, angle khud nahin.
Recall Feynman: 12-saal ke bache ko samjhaoSocho tum ek bache ko swing par dhakka de rahe ho. Agar bilkul sahi rhythm mein dhakka do, toh chhote chhote dhakkon se badi swing ban jaati hai — yeh hai loop jo khud ko feed kar raha hai. Ek control system yeh accidentally kar sakta hai: uski apni correction galat waqt par aur galat size mein wapas aati hai aur cheezein aur aur kharaab ho jaati hain. Phase margin poochta hai "meri timing kitni off ho sakti hai pehle ki dhakke add hone lagein?" Gain margin poochta hai "main kitna zyada zor se dhakka de sakta hoon pehle ki dhakke add hone lagein?" Bade margins = tum kahin bhi near nahi swing ko crazy karne ke. Chhote margins = ek nudge aur sab out of control.
"GaP": G ain margin P hase crossover par; aur elimination se P hase margin G ain crossover par.
"180 + angle = phase safety." Unity gain par jo bhi phase ho, usme 180 add karo.
1 + L ( s ) − 1 − 18 0 ∘
Phase crossover frequency ω p c Woh frequency jahan ∠ L ( j ω ) = − 18 0 ∘
Gain crossover frequency ω g c Woh frequency jahan ∣ L ( j ω ) ∣ = 1
Gain margin ka formula? GM = 1/∣ L ( j ω p c ) ∣ − 20 log 10 ∣ L ( j ω p c ) ∣
Phase margin ka formula? PM = 18 0 ∘ + ∠ L ( j ω g c )
Gain margin kis frequency par evaluate hoti hai, aur kyun? Phase crossover par (∠ = − 18 0 ∘ − 1
Phase margin kis frequency par evaluate hoti hai? Gain crossover par (∣ L ∣ = 1 − 18 0 ∘
GM aur PM ke typical healthy design values? GM ≳ 6 ≈ 3 0 ∘ 6 0 ∘
PM aur damping ratio ke beech approximate relation? ζ ≈ PM ∘ /100
Instability se pehle max tolerable time delay? τ ma x = PM (radians) / ω g c
Loop gain K Yeh magnitude curve ko upar shift karta hai, phase crossover par ∣ L ∣
Kya sirf positive gain margin stability ke liye kaafi hai? Nahin — dono chahiye: GM > 0 > 0
Nyquist stability criterion — parent theorem; − 1 Bode plots — jahan GM aur PM graphically read hoti hain.Root locus — gain badhne par poles kaise move karte hain stability limit ki taraf.Time delay and Padé approximation — delay phase margin ko kaise khaata hai.Damping ratio and overshoot — PM ↔ transient quality link.PID controller tuning — GNC loops mein margins as design targets.Loop shaping — L ( j ω )
Phase crossover wpc, angle=-180
Gain crossover wgc, mag=1
Stability distance from -1
