3.5.39 · D3 · HinglishGuidance, Navigation & Control (GNC)
Worked examples — PID tuning — Ziegler-Nichols, loop shaping
3.5.39 · D3· Physics › Guidance, Navigation & Control (GNC) › PID tuning — Ziegler-Nichols, loop shaping
Kisi bhi symbol se pehle, aao un characters ko pin down karte hain jo hum har jagah use karte hain.
The scenario matrix
| Cell | Case class | Kya special hai | Example |
|---|---|---|---|
| A | Closed-loop ZN, well-behaved plant | se full PID | Ex 1 |
| B | Open-loop ZN (reaction curve) | dead-time , slope | Ex 2 |
| C | Loop-shaping, positive phase deficit | lead chahiye (D) | Ex 3 |
| D | Loop-shaping, zero / negative phase deficit | plant ke paas already enough PM hai → lead nahi, ya lag chahiye | Ex 4 |
| E | Degenerate: pure integrator plant | , ZN closed-loop fail ho jaata hai | Ex 5 |
| F | Limiting value: aur | P-only mein reduce ho jaata hai, formulas ki sanity | Ex 6 |
| G | Real-world word problem | temperature oven, units, detune | Ex 7 |
| H | Exam twist: table ka wrong form | vs trap | Ex 8 |
| I | Pathological plant: non-minimum-phase zero / unstable pole | RHP zero bandwidth cap karta hai; unstable pole ko minimum gain chahiye | Ex 9 |
Har cell A–I neeche work ki gayi hai. Forecast padho aur scroll karne se pehle guess karo.
Ex 1 — Cell A: Clean plant pe Closed-loop ZN
Step 1 — Phase condition likho. Yeh step kyun? Stability ke edge pe fed-back signal inverted aur equal hokar wapas aata hai, yaani total phase . Yeh definition hai ultimate frequency ki (dekho Nyquist stability criterion).
Step 2 — Numerically solve karo. Kyun? Teen alag poles wali clean trick nahi dete, toh hum root-find karte hain. Solution hai rad/s (yeh hai; Verify dekho).
Neeche figure — yeh kya dikhata hai: purple curve plant ki phase hai jo frequency badhne ke saath girती hai; coral dashed line instability edge hai; mint dot wahan mark hai jahan dono cross karte hain, jo hai . Note karo ki curve sirf ek jagah ko touch karti hai — woh single crossing whi number hai jo ZN ka closed-loop test physically sustained-oscillation frequency ke roop mein measure karta hai.
Step 3 — Magnitude se milta hai. Kyun? Edge pe humein chahiye, toh .
=\frac{2}{\sqrt{12}\sqrt{15}\sqrt{20}}=\frac{2}{60}=\frac{1}{30}.$$ Toh $K_u=30$. **Step 4 — Period. Kyun?** $T_u$ ek full oscillation hai, rate $\omega_u$ pe $2\pi$ radians: $$T_u=\frac{2\pi}{\omega_u}=\frac{2\pi}{3.3166}=1.895\text{ s}.$$ **Step 5 — ZN PID table.** $K_p=0.6K_u=18$, $T_i=T_u/2=0.9474$, $T_d=T_u/8=0.2368$. Convert karo: $K_i=K_p/T_i=19.0$, $K_d=K_p T_d=4.263$. > **Verify:** $\omega_u^2=11$ toh $\omega_u^2+1,+4,+9 = 12,15,20$; product $=3600$, $\sqrt{}=60$, > magnitude $2/60=1/30$ ✓. Forecast check: $\omega_u=3.32>\sqrt3=1.73$ — spread-out, faster > poles critical frequency ko **upar** push karte hain. Units: $K_i$ [1/s]·gain hai, $K_d$ [s]·gain hai — dimensionally > $C(s)=K_p+K_i/s+K_d s$ ke saath consistent ✓. --- ## Ex 2 — Cell B: Open-loop ZN reaction curve > [!example] Statement > Ek plant pe step test se S-shaped response milti hai: apparent **dead time** $L=0.5$ s, aur > sabse steep slope reaction rate $R=0.8$ per second correspond karti hai (output ka slope ÷ step height, > per second). ZN open-loop PID gains do. > > **Forecast:** dead time chota hai — kya tumhe *aggressive* (bada $K_p$) ya *timid* controller ki umeed hai? **Step 1 — Reaction-curve table yaad karo. Kyun?** Humne plant ko unstable nahi drive kiya; balki humne ek first-order-plus-dead-time model fit kiya aur delay $L$ aur slope $R$ use karte hain (parent §3). $$K_p=\frac{1.2}{R L},\quad T_i=2L,\quad T_d=0.5L.$$ **Step 2 — Plug in karo.** $$K_p=\frac{1.2}{0.8\times0.5}=\frac{1.2}{0.4}=3.0.$$ $$T_i=2(0.5)=1.0\text{ s},\qquad T_d=0.5(0.5)=0.25\text{ s}.$$ **Step 3 — $K_i,K_d$ mein convert karo. Kyun?** Taaki controller $K_p+K_i/s+K_d s$ ke roop mein code ho sake. $$K_i=K_p/T_i=3.0/1.0=3.0,\qquad K_d=K_p T_d=3.0\times0.25=0.75.$$ > **Verify:** chota $L$ → fraction $1.2/(RL)$ bada → aggressive controller, forecast se match karta hai > (short delay matlab hum hard push kar sakte hain). Units: $R$ [1/s] hai, $L$ [s] hai, toh $RL$ > dimensionless hai aur $K_p$ pure gain hai ✓; $T_i,T_d$ seconds carry karte hain ✓. --- ## Ex 3 — Cell C: Loop shaping **positive** phase deficit ke saath (lead chahiye) > [!example] Statement > Plant $G(s)=\dfrac{4}{s(s+2)}$. Ek PD controller $C(s)=K_p(1+T_d s)$ design karo taaki gain crossover > $\omega_c=3$ rad/s aur phase margin $\text{PM}=50°$ mile. > > **Forecast:** plant mein ek integrator ($1/s$) hai jo $-90°$ contribute karta hai plus ek pole lag — kya > required lead positive hoga (D add karo) ya lag chahiye hoga? **Step 1 — $\omega_c$ pe plant phase. Kyun?** Controller ko jo bhi phase plant crossover frequency pe miss kar raha hai woh top up karni hogi (dekho [[Bode plot & frequency response]]). $$\angle G(j3)=-90°-\arctan\tfrac{3}{2}=-90°-56.31°=-146.31°.$$ **Step 2 — Required loop phase. Kyun?** $\text{PM}=180°+\angle L(j\omega_c)$, toh $\text{PM}=50°$ ke liye humein $\angle L=-130°$ chahiye. **Phase deficit** (upar define kiya) hai $\phi_\text{deficit}=-130°-(-146.31°)=+16.31°$ — positive, toh yeh Cell-C, "need lead" problem hai. **Step 3 — Lead needed = target − plant phase.** $$\phi_\text{lead}=-130°-(-146.31°)=+16.31°.$$ Positive → derivative zero ise supply karta hai, forecast confirm karta hai. **Step 4 — $T_d$ solve karo. Kyun?** Ek PD zero $\arctan(T_d\omega_c)$ phase contribute karta hai. $$\arctan(3T_d)=16.31°\Rightarrow 3T_d=\tan 16.31°=0.2926\Rightarrow T_d=0.09753.$$ **Step 5 — Magnitude $=1$ force karo. Kyun?** Definition se $\omega_c$ wahan hai jahan $|L|=1$. $$|G(j3)|=\frac{4}{3\sqrt{3^2+2^2}}=\frac{4}{3\sqrt{13}}=0.3698.$$ $$|1+T_d j3|=\sqrt{1+0.2926^2}=1.0420.$$ $$K_p=\frac{1}{0.3698\times1.0420}=2.595.$$ **Neeche figure — yeh kya dikhata hai:** purple curve *plant ki* phase hai; mint curve woh loop phase hai jo derivative lead add karne ke *baad* milti hai. $\omega_c=3$ (coral dashed vertical) dekho: mint curve exactly $-130°$ butter dotted line pe baithti hai, yaani $+16.31°$ arrow phase deficit ko precisely wahan close karta hai jahan humein chahiye. $\omega_c$ se door lead kam karta hai — isliye hum zero ko crossover pe exactly bite karne ke liye place karte hain. ![[deepdives/dd-physics-3.5.39-d3-s02.png]] > **Verify:** $\omega_c=3$ pe, $\angle L=-146.31°+16.31°=-130°$ → $\text{PM}=180-130=50°$ ✓. > $|L(j3)|=K_p\cdot0.3698\cdot1.0420=2.595\times0.3853=1.000$ ✓. --- ## Ex 4 — Cell D: **Zero/negative** phase deficit (lead nahi; lag shayad chahiye) > [!example] Statement > Plant $G(s)=\dfrac{10}{s+1}$. Hum $\omega_c=3$ rad/s aur $\text{PM}=60°$ chahte hain. Dikhao ki ek *lead* > controller **zaroorat nahi** hai, aur balki ek pure gain (ya low-frequency accuracy ke liye gain + lag) > kaafi hai. > > **Forecast:** ek single lag pole sirf $-90°$ tak phase de sakta hai — kya PM kabhi *problem* ho sakta hai yahan? **Step 1 — $\omega_c$ pe plant phase.** $$\angle G(j3)=-\arctan 3=-71.57°.$$ **Step 2 — Pure gain $C=K_p$ ke saath loop phase. Kyun?** Ek constant $K_p$ $0°$ phase add karta hai, toh $\angle L=\angle G=-71.57°$, giving $\text{PM}=180-71.57=108.4°$. **Step 3 — Interpret karo. Kyun yeh matter karta hai:** $\text{PM}=60°$ ke liye target loop phase $-120°$ hai, toh phase deficit hai $\phi_\text{deficit}=-120°-(-71.57°)=-48.4°$ — **negative**. Hamare paas surplus phase hai. Derivative lead add karna bekar hoga; agar kuch add karna hai toh **lag** compensator steady-state accuracy ke liye low-frequency gain boost karne ke liye lagao bina ample margin disturb kiye. **Step 4 — Sirf crossover ke liye gain set karo. Kyun?** $K_p$ choose karo taaki $|L(j3)|=1$ ho. $$|G(j3)|=\frac{10}{\sqrt{3^2+1}}=\frac{10}{\sqrt{10}}=3.1623.$$ $$K_p=\frac{1}{3.1623}=0.3162.$$ > **Verify:** $K_p=0.3162$ ke saath, $|L(j3)|=0.3162\times3.1623=1.000$ ✓, aur PM $=108.4°\ge60°$ ✓. > Degenerate-case lesson: ek first-order plant kabhi $-180°$ tak nahi pahunch sakta, toh uska *infinite* > gain margin hota hai — closed-loop ZN yahan kabhi oscillate nahi karega (Ex 5 se link). --- ## Ex 5 — Cell E: Degenerate plant — pure integrator (ZN closed-loop fail ho jaata hai) > [!example] Statement > Plant $G(s)=\dfrac{1}{s}$. Closed-loop ZN method attempt karo. $\omega_u,K_u,T_u$ ka kya hota hai? > Degeneracy explain karo aur ek working alternative do. > > **Forecast:** ek integrator *har* frequency pe exactly $-90°$ deta hai. Kya yeh kabhi $-180°$ hit kar sakta hai? **Step 1 — Har jagah phase. Kyun?** $G(j\omega)=1/(j\omega)$ ka phase $\angle=-90°$ hai **sab** $\omega>0$ ke liye — flat. **Step 2 — ZN condition $\angle G=-180°$ ka koi solution nahi. Kyun fail hota hai:** koi $\omega_u$ nahi hai jahan phase $-180°$ tak pahunche, toh $K_p$ badhana sirf magnitude scale karta hai, phase nahi, aur loop **kabhi** sustained oscillation maintain nahi karta — yeh sab $K_p>0$ ke liye stable hai. Isliye $K_u=\infty$, $\omega_u$ undefined, $T_u$ undefined. **Neeche figure — yeh kya dikhata hai:** purple line integrator ka phase hai, har frequency ke liye dead flat $-90°$ pe; coral dashed line $-180°$ edge hai; dono ke beech butter shaded band woh phase gap hai jo **kabhi close nahi hota**. Kyunki dono lines parallel hain aur kabhi nahi milti, koi crossing frequency nahi hai — visually confirm karta hai $K_u=\infty$ aur kyun closed-loop ZN recipe ke paas measure karne ke liye kuch nahi hai. ![[deepdives/dd-physics-3.5.39-d3-s03.png]] **Step 3 — Working alternative. Kyun?** Kyunki ZN closed-loop timing measure nahi kar sakta, **direct loop shaping** use karo. $C=K_p$ ke liye, $L=K_p/s$, toh $|L(j\omega_c)|=1$ directly $\omega_c=K_p$ deta hai, aur $\angle L=-90°$ har jagah → constant $\text{PM}=90°$. Jo bandwidth chahiye uske liye $K_p$ choose karo, e.g. $K_p=5$ gives $\omega_c=5$ rad/s. > **Verify:** $\angle(1/(j\omega))=-90°$ independent of $\omega$ ✓ (kabhi $-180°$ nahi). $K_p=5$ ke saath, > $|L(j5)|=5/5=1$ ✓ aur PM $=180-90=90°$ ✓. Lesson: **degenerate = flat phase → closed-loop ZN > undefined**; tumhe loop shaping ya open-loop methods pe fall back karna hoga. --- ## Ex 6 — Cell F: Limiting behaviour — PID ko P-only mein collapse karo > [!example] Statement > **Ex 1** ke ZN PID gains lo ($K_p=18, K_i=19.0, K_d=4.263$). Algebraically dikhao ki jab > $K_i\to0$ aur $K_d\to0$ to controller pure proportional reduce ho jaata hai, aur confirm karo ki ZN P-only > value $K_p=0.5K_u$ PID value se *less aggressive* hai. > > **Forecast:** kya P-only gain PID ke $K_p$ se higher hona chahiye ya lower? Kyun? **Step 1 — Limit lo. Yeh step kyun?** Controller $C(s)=K_p+K_i/s+K_d s$ hai. $K_i\to0$ karne se $1/s$ branch remove hoti hai; $K_d\to0$ karne se $s$ branch remove hoti hai: $$\lim_{K_i,K_d\to0}C(s)=K_p.$$ Koi integral lag nahi, koi derivative lead nahi — sirf ek constant gain, jaise expected tha. **Step 2 — ZN rows compare karo. Kyun?** ZN P-only $K_p=0.5K_u=0.5(30)=15$ use karta hai, jabki ZN PID $0.6K_u=18$ use karta hai. **Step 3 — Ordering interpret karo. Kyun P chota hai:** derivative lead ke bina koi *phase margin help* nahi hai, toh gain back off karna padta hai (0.5 vs 0.6) same stability edge rakhne ke liye. D term tumhe extra $0.1K_u$ "kharid ke" deta hai. > **Verify:** $0.5\times30=15$ aur $0.6\times30=18$ toh P-gain $15<18=$ PID-gain ✓, reasoning se match karta hai > ki D tumhe harder push karne deta hai. Limit exact hai: $K_i=K_d=0$ ke saath, $C(s)=K_p$ identically ✓. --- ## Ex 7 — Cell G: Real-world word problem (oven, units, detune) > [!example] Statement > Ek resistive oven ke step test (heater power → temperature) mein dead time $L=20$ s aur reaction > rate $R=0.5\,°\text{C·s}^{-1}$ per unit power (slope/step-height) dikhta hai. ZN reaction-curve PI > controller design karo (ovens derivative noise se nafrat karte hain), phir $K_p$ ko 30\% se **detune** karo > kyunki ZN ka 25\% overshoot product scorch kar deta. > > **Forecast:** lamba dead time — kya $K_p$ bada aayega ya tiny? **Step 1 — ZN reaction-curve PI table. Kyun PI, PID nahi?** Temperature sensors noisy hote hain aur D noise amplify karta hai; ovens slow hain toh D zyada add nahi karta. PI row hai $$K_p=\frac{0.9}{RL},\qquad T_i=\frac{L}{0.3}=3.333\,L.$$ **Step 2 — Plug in karo.** $$K_p=\frac{0.9}{0.5\times20}=\frac{0.9}{10}=0.09,\qquad T_i=3.333\times20=66.67\text{ s}.$$ $$K_i=K_p/T_i=0.09/66.67=1.35\times10^{-3}\ \text{s}^{-1}.$$ **Step 3 — Detune karo. Kyun?** ZN quarter-amplitude decay (~25\% overshoot) target karta hai. $K_p$ ko 30\% reduce karo: $$K_p^\text{tuned}=0.7\times0.09=0.063.$$ $T_i$ (integral time) fixed rakho taaki steady-state behaviour preserve ho; $K_i$ recompute karo: $$K_i^\text{tuned}=0.063/66.67=9.45\times10^{-4}\ \text{s}^{-1}.$$ > **Verify:** lamba $L=20$ s $RL=10$ banata hai, toh $K_p=0.9/10=0.09$ — **tiny**, forecast se match karta hai > (bada delay ⇒ timid controller) ✓. Detune: $0.7\times0.09=0.063$ ✓. *Detuned* > gains ka units check: $K_p$ ko detune karna dimensionless $0.7$ se multiply karna hai, toh $K_p^\text{tuned}$ abhi bhi > $K_p$ ke same units carry karta hai (power/°C); kyunki $T_i$ seconds mein unchanged hai, $K_i^\text{tuned}=K_p^\text{tuned}/T_i$ > abhi bhi [power/°C · s$^{-1}$] hai — detune dimensional consistency preserve karta hai ✓. --- ## Ex 8 — Cell H: Exam twist — $T_i,T_d$ vs $K_i,K_d$ trap > [!example] Statement > Ek exam mein $K_u=8,\ T_u=3.63$ s diya hai (parent ka $1/(s+1)^3$ plant) aur poochha hai: > "*PID gains $K_p,K_i,K_d$ report karo.*" Ek student likhta hai $K_p=4.8,\ K_i=T_u/2=1.815,\ K_d=T_u/8=0.454$. > Error dhundo aur fix karo. > > **Forecast:** student ne kaun si quantities confuse ki — times ya gains? **Step 1 — Confusion spot karo. Kyun?** ZN tables **$T_i,T_d$** list karte hain (integral/derivative *times*, seconds mein units), **nahi** $K_i,K_d$ (gains). Student ne $T_u/2$ aur $T_u/8$ seedha $K_i,K_d$ mein copy kar diya. Wrong units: $K_i$ [1/s]·gain hona chahiye, $K_d$ [s]·gain. **Step 2 — Correct conversion. Kyun?** Standard form se $C(s)=K_p(1+1/(T_i s)+T_d s)$ hume milta hai $K_i=K_p/T_i$ aur $K_d=K_p T_d$. $$T_i=T_u/2=1.815\text{ s},\quad T_d=T_u/8=0.4538\text{ s}.$$ $$K_i=\frac{K_p}{T_i}=\frac{4.8}{1.815}=2.645,\qquad K_d=K_p T_d=4.8\times0.4538=2.178.$$ **Step 3 — Sanity gap. Kyun yeh matter karta hai:** student ka $K_i=1.815$ vs correct $2.645$ — 30\% error jo visible steady-state offset chhod deta; $K_d=0.454$ vs correct $2.178$ — nearly $5\times$ too weak, toh overshoot intended se kahin zyada hota. > **Verify:** $K_i=4.8/1.815=2.645$ ✓ aur $K_d=4.8\times0.4538=2.178$ ✓ — parent ke > §2 example se exactly match karta hai. Student ke values unit analysis fail karte hain: tum $K_i$ (per-second > gain) ko *time* $T_u/2$ ke equal set nahi kar sakte ✓. --- ## Ex 9 — Cell I: Pathological plant — non-minimum-phase zero & unstable pole > [!example] Statement > Do short scenarios. > **(a)** Plant $G_a(s)=\dfrac{1-s}{(s+1)^2}$ mein **right-half-plane (RHP) zero** hai $s=+1$ pe > (ise *non-minimum-phase* kehte hain). Dikhao ki uska phase *extra-negative* hai aur explain karo ki yeh kyun > achievable crossover $\omega_c$ ko **cap** karta hai. > **(b)** Plant $G_b(s)=\dfrac{1}{s-1}$ mein **unstable pole** hai $s=+1$ pe. Dikhao ki pure > proportional control ke saath stabilize karne ke liye *minimum* gain ki zaroorat hoti hai. > > **Forecast:** (a) ke liye, kya RHP zero lead add karta hai ya extra lag? (b) ke liye, kya chota $K_p$ safe hai? **Step 1 (a) — RHP zero ka phase. Kyun?** Ek normal (left-half-plane) zero $1+s/z$ *positive* phase (lead) add karta hai. Ek RHP zero factor $1-s$ balki phase **subtract** karta hai: $$\angle(1-j\omega)=-\arctan\omega.$$ Toh $G_a$ ka phase hai $\angle G_a(j\omega)=-\arctan\omega-2\arctan\omega=-3\arctan\omega$ — zero phase ke liye ek *extra pole* ki tarah behave karta hai, ise $-270°$ ki taraf drag karta hai. **Step 2 (a) — Yeh bandwidth kyun cap karta hai. Kyun yeh matter karta hai:** loop phase ek low frequency pe $-180°$ tak race karta hai. Concretely $\angle G_a=-180°$ jab $3\arctan\omega=180°\Rightarrow\arctan\omega=60° \Rightarrow\omega=\sqrt3$. Tum apna crossover $\omega_c$ $\sqrt3$ ke paas ya upar **nahi rakh sakte** aur phase margin maintain karo — RHP zero speed ko hard-limit karta hai. (Rule of thumb: $\omega_c\lesssim$ half the RHP-zero location.) Koi bhi D-term lead RHP zero ko cure nahi kar sakta, kyunki deficit sirf worsens hota hai jab tum $\omega_c$ upar push karte ho. Dekho [[Root locus method]] ke liye ki RHP zero closed-loop poles ko rightward kaise attract karta hai. **Step 3 (b) — Unstable pole ko minimum gain chahiye. Kyun?** $C=K_p$ ke saath closed loop: characteristic equation $1+K_p\,G_b=0\Rightarrow s-1+K_p=0\Rightarrow s=1-K_p$. Stability ke liye pole ko left half-plane mein chahiye, $s<0$: $$1-K_p<0\Rightarrow K_p>1.$$ Toh *chota* gain yahan **unsafe** hai — usual intuition ka opposite. $K_p=1$ se neeche plant unstable rehta hai; pole ko left half-plane mein drag karne ke liye *enough* gain apply karna zaroori hai. > **Verify:** (a) $\angle G_a(j\omega)=-3\arctan\omega$ hits $-180°$ at $\omega=\sqrt3\approx1.732$ > ✓; RHP zero phase subtract karta hai (extra lag) not lead ✓. (b) closed-loop pole $s=1-K_p$; stable iff > $K_p>1$, toh $K_p=0.5$ se $s=+0.5>0$ (unstable) jabki $K_p=2$ se $s=-1<0$ (stable) ✓. > Lesson: minimum-phase intuition ("more gain = less stable", "koi bhi gain first-order plant stabilize karta hai") > **reverse** ho jaati hai RHP zeros aur RHP poles ke liye. --- > [!mnemonic] Koi case mat bhulo > **"CLOD-PILEX"** — **C**losed-loop ZN, **L**oop-shape lead, **O**pen-loop ZN, **D**egenerate > (flat-phase/integrator), **P**hase-surplus (no lead), **I**ntegrator limits, **L**iteral word > problem, **E**xam unit trap, **X** = pathological (RHP zero / unstable pole). Agar tumhara problem > in nau mein se kisi mein fit nahi hota, use dobara padho. > [!recall]- Matrix ke across self-test > "Phase deficit" ek line mein define karo. ::: Target loop phase $(-180°+\text{PM})$ aur $\omega_c$ pe plant phase ke beech ka gap; positive matlab hume lead add karni hogi. > Closed-loop ZN pure integrator ko tune kyun nahi kar sakta? ::: Uska phase sab $\omega$ ke liye $-90°$ hai, kabhi $-180°$ tak nahi pahunchta, toh yeh kabhi sustained oscillation maintain nahi karta ($K_u=\infty$). > Ek first-order plant $10/(s+1)$: kya ise kisi bhi $\omega_c$ pe PM $=60°$ ke liye lead chahiye? ::: Nahi — max lag $90°$ hai, toh PM hamesha $>90°$ hoti hai; gain (± lag) use karo, lead nahi. > ZN table PID column ke neeche $T_u/2$ list karta hai. Kya woh $K_i$ hai? ::: Nahi, woh $T_i$ (integral *time*) hai; convert karo $K_i=K_p/T_i$. > Reaction-curve ZN mein lamba dead time $L$ $K_p$ ko kis direction mein push karta hai? ::: Neeche — $K_p=1.2/(RL)$, toh bada $L$ matlab chota, timid gain. > RHP zero tuning kaise change karta hai? ::: Yeh phase subtract karta hai aur $\omega_c$ cap karta hai (≈ zero location ka aadha); koi bhi lead ise fix nahi kar sakta. > Kya unstable plant $1/(s-1)$ ke liye chota proportional gain safe hai? ::: Nahi — tumhe $K_p>1$ chahiye (ek *minimum* gain) sirf ise stabilize karne ke liye.