3.5.42 · D5 · HinglishGuidance, Navigation & Control (GNC)
Question bank — Gain margin, phase margin — stability margins
3.5.42 · D5· Physics › Guidance, Navigation & Control (GNC) › Gain margin, phase margin — stability margins
Traps se pehle, hum vocabulary fix kar lete hain taaki is page ka koi bhi symbol mystery na rahe.
Recall Critical point ek sentence mein
Ek feedback loop tab khatam hoti hai jab signal wapas aata hai bilkul ulta (phase ) aur utna hi strong (magnitude 1) — yahi complex plane par point hai. Saari margins sirf tak ki distances hain, do alag tareekon se measure ki gayi.
Neeche do annotated pictures sab kuch anchor karti hain. Pehle Bode view — jahan dono margins vertical gaps ke roop mein padhte hain:
Phir Nyquist view — wohi dono margins curve se point tak distances ke roop mein:
Traps solve karte waqt dono pictures ko dhyan mein rakho.
True or false — justify
True or false: Bade gain margin wala system hamesha stable hota hai.
False. Tumhe dono chahiye — GM dB aur PM ; ek loop mein gain ka bahut room ho sakta hai lekin phir bhi cross karte waqt gain unity se upar ho sakti hai, jisse phase margin negative ho jaata hai.
True or false: Phase margin, ki phase ke barabar hoti hai.
False. PM tak ka gap hai: . Raw phase (jaise ) margin nahi hai — margin woh room hai jo bacha hua hai.
True or false: Loop gain badhane se dono stability margins improve hoti hain.
False. badhane se poori magnitude curve upar uthti hai, isliye , 1 tak ek zyada frequency par pahunchti hai jahan phase aur zyada gir chuki hoti hai — isse GM aur usually PM bhi shrink hoti hai. Gain speed mein help karta hai, margins ko hurt karta hai.
True or false: Agar Nyquist curve kabhi ko touch nahi karti, toh closed loop stable hai.
Generally False. Sirf miss karna kaafi nahi; full Nyquist stability criterion ke around encirclements ko open-loop right-half-plane poles ke relative count karta hai. Point miss karna lekin usse encircle karna unstable hai.
True or false: Pure time delay loop ki gain badalta hai lekin phase nahi.
False. Delay magnitude ko untouched chhod deta hai () lekin phase lag add karta hai jo frequency ke saath badhta hai — yeh phase margin khata hai. Dekho Time delay and Padé approximation.
True or false: Gain margin aur phase margin ek hi frequency par measure hoti hain.
False. GM phase crossover par rehta hai (); PM gain crossover par rehta hai (). Har margin wahan measure hoti hai jahan dusri quantity apni critical value par pahunch chuki hoti hai.
True or false: ka phase margin se har tarike se behtar hota hai.
False. Bahut zyada PM matlab bahut heavy damping — sluggish, over-cautious response. Sweet spot roughly – hai; zyada margin utna hi design smell hai jitna kam. Dekho Damping ratio and overshoot.
True or false: Negative gain margin (dB mein) matlab loop already unstable hai.
True. Negative dB matlab hai phase par, isliye curve pehle se ke beyond negative-real axis par baith chuki hai — loop khud ko feed karta hai aur badhta hai.
True or false: Har stable loop mein exactly ek gain crossover aur ek phase crossover hoti hai.
False. Curves dB ya ko kai baar cross kar sakti hain (conditionally stable systems). Tab ek single margin number ambiguous hoti hai aur tumhe poora Nyquist plot inspect karna padta hai.
True or false: Rule kisi bhi feedback loop ke liye kaam karta hai.
False. Yeh sirf ek approximation hai jo second-order, dominant-pole loop ke liye valid hai aur sirf roughly tak ke PM ke liye. Usse bahar (higher-order, extra poles/zeros, bada PM) yeh kaafi galat ho sakta hai — isse sanity-check samjho, law nahi. Dekho Damping ratio and overshoot.
Spot the error
"." — kya galat hai?
Yeh reciprocal hona chahiye: . GM woh factor hai jisse tum gain multiply kar sakte ho jab tak 1 tak pahunche, isliye agar hai toh 6 ka factor hai, ka nahi.
"." — kya galat hai?
Sign ulta hai; yeh hai . ke saath yeh deta hai (sahi), jabki galat form deta hai.
"Kyunki GM dB mein hai aur PM degrees mein, unhe compare nahi kar sakte, isliye bas GM bada karo." — kya kharabi hai?
Kharabi yeh hai ki unhe interchangeable priorities ki tarah treat kiya ja raha hai. Ye alag failure directions measure karte hain (extra gain vs extra phase lag). Design mein dono ek saath satisfy karne chahiye; ek ko inflate karne ke liye dusre ko trade nahi kar sakte.
"Critical point hai kyunki wahan sabse bada hota hai." — kya error hai?
Critical point hai, jahan hota hai (denominator vanish hota hai, poles blow up karte hain). denominator ko banata hai — bilkul benign, dangerous nahi.
"Ek minimum-phase system jisme PM aur GM dB ho, phir bhi secretly unstable ho sakta hai." — sahi ya galat?
Minimum-phase, single-crossover, open-loop-stable loop ke liye galat — wahan positive margins stability guarantee karte hain. Yeh caution non-minimum-phase, conditionally-stable, ya open-loop-unstable systems par apply hoti hai, is case par nahi.
"Phase margin add karne ke liye, bas gain badhao taaki crossover higher ho jaye." — kya error hai?
Gain badhane se gain crossover higher frequency par move hota hai jahan phase aur zyada gir chuki hoti hai, isliye PM typically decrease hoti hai. Phase margin add karne ke liye tum phase shape karte ho (jaise lead compensator), raw gain nahi badhate — dekho Loop shaping.
" exactly hai, isliye PM matlab ." — kya error hai?
Yeh relation dominant second-order loops ke liye ek approximation hai jo PM ~ tak valid hai; tak extrapolate karna uski valid range se bahar hai aur ek misleading number deta hai. Actual damping puri pole–zero layout par depend karti hai.
Why questions
Critical point kyun hai, ya kyun nahi?
Kyunki closed-loop poles ke roots hote hain, yaani . par, value matlab returning signal unit-strength aur perfectly inverted hai — ek sinusoid jo har cycle mein khud ko reinforce karti hai.
Gain margin ke liye hum ki jagah reciprocal kyun lete hain?
Kyunki hum woh multiplier chahte hain jo ko exactly 1 tak scale kare. Agar current magnitude hai, toh woh multiplier hai — instability se pehle ka headroom.
Phase margin damping aur overshoot se kyun relate karta hai?
Dominant second-order loop ke liye, low PM matlab crossover ke paas hai, isliye closed loop mein lightly-damped poles hain → ringing aur overshoot. Approximate rule (sirf ~ tak valid) is trend ko capture karta hai; zyada PM ≈ zyada damping. Dekho Damping ratio and overshoot.
Time delay ek "well-designed" loop ko bhi kyun unstable kar deta hai agar tum kaafi wait karo?
Delay phase lag add karta hai jo ke proportional hai. Jab phase margin (radians mein) se zyada ho jaata hai, gain crossover par phase se guzar jaati hai aur PM negative ho jaati hai. Max safe delay hai.
Hum specifically negative-real axis ke along GM kyun measure karte hain?
Phase crossover par ek negative real number hai, us axis par lie karta hai. Isliye tak pahunchne ka ek hi tarika hai — magnitude scale karke real axis ke along slide karna — GM exactly woh scaling distance measure karta hai.
Do loops jinka identical gain margin ho woh time domain mein itne alag kyun behave kar sakte hain?
GM sirf ek point (phase crossover) ko ek direction ke along sample karta hai. Do loops wahan match kar sakte hain phir bhi phase margin, bandwidth, aur pole locations mein wildly differ kar sakte hain — isliye transient behavior (overshoot, settling) diverge karta hai. Margins rulers hain, complete portraits nahi.
Hum same margins ke liye Bode aur Nyquist dono views kyun prefer karte hain?
Bode GM aur PM ko directly do stacked curves se read karta hai (design tweaks ke liye fast via Bode plots); Nyquist global geometry aur encirclements dikhata hai, conditionally-stable, multi-crossing, aur open-loop-unstable cases ko pakadta hai jinhe Bode margins misreport kar sakti hain.
Edge cases
Agar phase kabhi bhi finite frequency par tak nahi pahunchti toh gain margin kya hai?
GM effectively infinite hai — koi phase crossover nahi hai, isliye koi bhi extra gain curve ko us axis ke along tak nahi pahuncha sakti (jaise ek first-order loop jiska phase par asymptote karta hai).
Agar sabhi ke liye 1 se neeche hai toh phase margin kya hai?
Koi gain crossover nahi hai, isliye PM conventionally infinite li jaati hai ("koi value nahi" nahi): curve kabhi unit circle tak nahi pahunchti, isliye phase direction se tak kabhi nahi pahunch sakti, toh loop robustly stable hai gain terms mein. "Infinite" kaho, "undefined" nahi, warna instability imply ho sakti hai.
GM (yaani dB) physically kya matlab hai?
Loop exactly boundary par baitha hai: phase crossover par hai, isliye precisely. Yeh marginally unstable hai — ek sustained oscillation jo na badhti hai na ghatti hai.
PM physically kya matlab hai?
Gain crossover par phase exactly hai, isliye : wahi marginal-oscillation boundary, is baar gain direction ki bajay phase direction se reach ki gayi.
Gain aur phase margins kaise kaam karti hain jab open-loop plant already unstable ho (right-half-plane poles ho)?
Margins utne hi tarike se compute hoti hain, lekin unki interpretation flip hoti hai: stable closed loop ke liye ab Nyquist curve ko ko sahi baar encircle karna zaroori hai (har RHP open-loop pole ke liye ek baar). Positive GM/PM akele stability guarantee nahi karte, aur aisi loops often conditionally stable hoti hain — full Nyquist stability criterion use karna padta hai.
Open-loop-unstable loop ke liye, bahut kam gain itna khatarnak kyun ho sakta hai jitna bahut zyada?
RHP poles ke saath loop ko stabilize karne wala encirclement produce karne ke liye minimum gain chahiye. Gain girne par curve se neeche shrink ho jaati hai, required encirclement kho deti hai — isliye ek lower stability boundary bhi hai upar wali ke saath. Root locus dono boundaries clearly dikhata hai.
Conditionally-stable system ke liye, gain kam karna kyun unstable kar sakta hai?
Uski Nyquist curve ko sirf gains ki ek band ke andar safely encircle karti hai. kam karne par curve shrink ho sakti hai aur plot ka ek loop ke galat side par slip ho jaata hai, encirclement count flip karta hai — yahan single margin se zyada Root locus / Nyquist par trust karo.
Non-minimum-phase loop (right-half-plane zero) ke liye, healthy-looking margins kyun mislead kar sakti hain?
RHP zeros extra phase lag add karte hain aur fundamental bandwidth limits impose karte hain; ek crossover par local GM/PM theek lag sakta hai jabki actual robustness poor ho. Full Nyquist stability criterion aur careful Loop shaping zaroori hain.
Agar PM degrees mein measure hoti hai, toh max time delay compute karne se pehle radians mein convert kyun karna padta hai?
Kyunki ek angle ko ek angular frequency (rad/s) se divide karta hai; units sirf tab seconds mein cancel hote hain jab angle radians mein ho. Degrees use karne par factor ki wajah se nonsense number milta hai.