Вы находитесь на странице: 1из 6

ECE 461: Solutions to Homework #5

Error Constants, Routh Criteria, Root Locus Water Level Control: 1) The dynamics of a 2-tank system with proportional (P=k) feedback are
0.1k 0.1k G(s) = (s+0.004875)(s+0.2051) = s 2 +0.21s+0.001

1a) Determine the system type, Kp, Kv, and the steady-state error for a unit step and unit ramp input. This is a type 0 system, so Kv = Ka = 0 The steady-state error for a unit ramp input is infinity

Kp = 100k

G(s=0) = Kp)
1 1 = 100k+1

The steady-state error for a unit step input is K p +1

1b) Use a Routh table to find the range of k which results in a stable system.

s 2 + 0.21s + 0.001 + 0.1k = 0


1 0.21 0.001+0.1k 0.001+0.1k 0 0 k > -0.01 0

To be stable, k > -0.01. 1c) Use the simulation WATER.EXE with proportional feedback. (Select PID control, set D=I=0. Adjust P with the cursor keys). Describe what happens as the proportional gain (k above) goes from zero to infinity. (What is the steady-state error, overshoot, and stability for k small, medium, and large.) small k medium k large k stable stable stable some steady-state error steady-state error gets smaller steady-state error gets even smaller 1st order response 1st order response 2nd order response (some overshoot) very slow a little faster faster

2) Water Level Control with Integral Feedback: (K(s) = s


k 0.1k 0.1k G(s) = s(s+0.004875)(s+0.2051) = s 3 +0.21s 2 +0.001s

I = s

2a) Determine the system type, Kp, Kv, and the steady-state error for a unit step and unit ramp input. This is a type-I system (one pole at s=0). Hence Kp = infinity No error for a unit step input

Kv = 100k

K v =lim (s G(s))
s0

1 1 Error for a unit ramp is K v = 100k

2b) Use a Routh table to find the range of k which results in a stable system.

s 3 + 0.21s 2 + 0.001s + 0.1k = 0


1 0.21 0.00021-0.1k 0.1k The system is stable for 0 < k < 0.00021 0.001 0.1k 0 0 k < 0.00021 k>0 0

2c) Use the simulation WATER.EXE with proportional feedback. (Select PID control, set P=D=0. Adjust I with the cursor keys). Describe what happens as the proportional gain (k above) goes from zero to infinity. (What is the steady-state error, overshoot, and stability for k small, medium, and large.) small k medium k large k stable stable unstable no steady-state error no steady-state error no steady-state error 1st order response 2nd order response (some overshoot) 2nd order response (oscillates) very slow a little faster unstable

Temperature Control: 3) The dynamics of a metal bar (similar to heating the tip of a soldering iron) are...
0.1975k 0.19753k GK = s 4 +4.667s 3 +6.667s 2 +2.963s+0.1975 = (s+0.6667)(s+1.565)(s+2.355)(s+0.08041)

3a) Determine the system type, Kp, Kv, and the steady-state error for a unit step and unit ramp input. This is a type 0 system, so Kv = Ka = 0 The steady-state error for a unit ramp input is infinity

Kp = k

G(s=0) = Kp)

1 1 The steady-state error for a unit step input is K p +1 = k+1

3b) Use a Routh table to find the range of k which results in a stable system.

s 4 + 4.667s 3 + 6.667s 2 + 2.963s + (0.1975 + 0.1975k)


1 4.667 6.0286 2.8101-0.15288k 0.1975+0.1975k 0 6.667 2.963 0.1975+0.1975k 0 0 0 0.1975+0.1975k 0 0 0 0 0 k < 18.381 k > -1

To be stable, -1 < k < 18.381

3c) Use the simulation BAR.EXE with proportional feedback. (Select PID control, set D=I=0. Adjust P with the cursor keys). Describe what happens as the proportional gain (k above) goes from zero to infinity. (What is the steady-state error, overshoot, and stability for k small, medium, and large.) small k medium k large k stable stable stable some steady-state error steady-state error gets smaller steady-state error gets even smaller 1st order response 2nd order response (some overshoot) 2nd order response (some overshoot) very slow a little faster faster

This wasn't asked for - but you can explan the Routh Criteria and the response in the above table using a root locus plot.

With k=1, you are at the breakaway point For k < 1, you have a single, real dominnat pole. For k < 1, the system has no overshoot For k =1, the dominan pole is -0.2, resulting in a 20 second settling time

With k=18.381, you have complex poles (you are at the jw crossing).

So, Very small k (k < 0.1) results in a slow, 1st order response Small gains (0.1<k<1) results in faster systems which have a single real dominant pole Larger gains (1<k<18.381) result in complex poles - but a stable system Large gains (k<18.381) results in an unstable system with complex poles.

4) The dynamics of a metal bar with integral feedback are...


0.1975k 0.19753k GK = s 5 +4.667s 4 +6.667s 3 +2.963s 2 +0.1975s = s(s+0.6667)(s+1.565)(s+2.355)(s+0.08041)

4a) Determine the system type, Kp, Kv, and the steady-state error for a unit step and unit ramp input. This is a type-I system (one pole at s=0). Hence Kp = infinity No error for a unit step input

Kv = k
1 1 Error for a unit ramp is Kv = k

4b) Use a Routh table to find the range of k which results in a stable system.

s 5 + 4.667s 4 + 6.667s 3 + 2.963s 2 + 0.1975s + 0.1975k


1 4.667 6.0317 2.8102 + 0.032734k 0.5550 - 1.3037k - 0.001385k 0.1975k
2

6.667 2.963 0.1975 - 0.04231k 0.1975k 0 0

0.1975 0.1975k 0 0 0 0 k > -85.82 -941 < k < 0.4255 k>0

This system is stable for 0 < k < 0.4255

4c) Use the simulation WATER.EXE with proportional feedback. (Select PID control, set P=D=0. Adjust I with the cursor keys). Describe what happens as the proportional gain (k above) goes from zero to infinity. (What is the steady-state error, overshoot, and stability for k small, medium, and large.) small k medium k large k stable stable unstable no steady-state error no steady-state error no steady-state error 1st order response 2nd order response (some overshoot) 2nd order response (oscillates) very slow a little faster unstable

Again - you can explain this using a root locus plot (which wasn't asked for....)

Small gains (k < 0.01897) results in a real dominant pole. This gives no overshoot (no complex part), no steady-state error (the system is type 1), and a slow system (the real part is small). (the part of the root locus near s=0 is shown here)

Larger gains result in complex dominant poles and some oscillation. The system is stable if k<0.4255.

If the gain is too large (k > 0.4255), you have an unstable system.

Вам также может понравиться