# UNIVERSIDAD AUTÓNOMA DE AGUASCALIENTES CENTRO DE CIENCIAS BÁSICAS DEPARTAMENTO DE SISTEMAS ELECTRÓNICOS.

SECOND PARTIAL EXAM LINEAR SYSTEMS

NOVEMBER 22

ND

, 2010

STUDENT: JUVENTINO SAUCEDO NAVARRO Exercise 1 Here is the root locus for a simple control system. The block diagram is shown first, and the root locus follows below.

From the root locus, you should be able to determine the following.

The transfers function for the system being controlled. The transfer function can be formed just by knowing the poles and zeroes, which are shown on the root locus plot. Determine an adjusted system that has a DC gain of 1.0 Using your adjusted system as G(s), determine the gain, K, which will produce poles at -2.5 +/- j 2.5.

Solution: We can see that the root locus plot only has 2 poles (at -1 and -4) and the zeros are located at the infinity

1 k C(s) kG(s) (s + 1)(s+ 4) (s + 1)(s + 4) = = = 1 (s+ 1)(s + 4) + k R(s) 1 + kG(s) 1 + k (s + 1)(s+ 4) (s + 1)(s + 4) k

Simplifying it we have the following:

C(s) k(s+ 1)(s+ 4) k k = = = 2 R(s) (s+ 1)(s+ 4)[(s+ 1)(s+ 4) + k] (s+ 1)(s+ 4) + k s + 5s+ 4 + k

Using Matlab to obtain the transfer function

>> GS = zpk ([], [-1 -4], [1]) Zero/pole/gain: 1 ----------(s+1) (s+4) Taking into account the value of k = 1 and adjusting the system with a gain of 1 we have the following response >> k = 1; >> t = feedback (k*GS, 1); >> t Zero/pole/gain: 1 -------------------------(s+1.382) (s+3.618) >> rltool (t) Finally, the gain k obtained is 7.48 when the poles are in 2.5 +/- 2.5j

Exercise 3 Assume that you have a system with three poles at s = -1, -2 and -3. The G(s) for the system is:

G(s) =

5 (s+ 1)(s+ 2)(s+ 3)

The system is being controlled in a feedback system as shown below.

Predict step response performance for K = 1, 2, 5, 10, 20, 50 Solution: First we need to obtain the transfer function:

5k 5k C(s) kG(s) 5k (s + 1)(s+ 2)(s+ 3) (s+ 1)(s + 2)(s + 3) = = = = 5k (s + 1)(s+ 2)(s+ 3) + 5k (s + 1)(s+ 2)(s+ 3) + 5k R(s) 1 + kG(s) 1 + (s + 1)(s+ 2)(s+ 3) (s+ 1)(s + 2)(s + 3)

C(s) 5k 5k = 2 = 3 2 R(s) (s + 3s+ 2)(s+ 3) + 5k (s + 6s + 11s+ 6) + 5k

With the help of matlab we obtain the following

>> GS = zpk ([], [-1 -2 -3], [5]) Zero/pole/gain: 5 ----------------(s+1) (s+2) (s+3) For k = 1 >> k = 1; >> t = feedback( k*GS, 1 ); >> t Zero/pole/gain: 5 ------------------------------------------(s+3.904) (s^2 + 2.096s + 2.818) >> rltool() In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=1 in the SISO we have the following graph

By the previous plot we can see that the answer is Under-damped.

For k = 2

>> k = 2; >> t = feedback( k*GS, 1 ); >> t Zero/pole/gain: 10 -------------------------------(s+4.309) (s^2 + 1.691s + 3.713) >> rltool() In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=2 in the SISO we have the following graph

By the previous plot we can see that the answer is Under-damped. For k = 5

>> k = 5; >> t = feedback( k*GS, 1 ); >> t Zero/pole/gain: 25 ------------------------------------------(s+5.038) (s^2 + 0.962s + 6.153) >> rltool In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=5 in the SISO we have the following graph

By the previous plot we can see that the answer is Under-damped. For k = 10

>> k = 10; >> t = feedback( k*GS, 1 ); >> t Zero/pole/gain: 50 ---------------------------------------------

(s+5.774) (s^2 + 0.2255s + 9.698) >> rltool In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=10 in the SISO we have the following graph

In the plot we can see that the answer is Under-damped. For k = 20 >> k = 20; >> t = feedback( k*GS, 1 ); >> t Zero/pole/gain: 100 -------------------------------------------(s+6.713) (s^2 - 0.7134s + 15.79) >> rltool In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=20 in the SISO we have the following graph

In the plot we can see that the system’s answer is unstable. For k = 50 >> k = 50; >> t = feedback( k*GS, 1 ); >> t Zero/pole/gain: 250 -----------------------------------------(s+8.353) (s^2 - 2.353s + 30.65) >> rltool In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=50 in the SISO we have the following graph

In the plot we can see that the system’s answer is unstable.

Exercise 5 Assume that you have a system with three poles at s = -1, -3, and -5, and a single zero at s = -2. The G(s) for the system is:

G(s) =

5(s+ 2) (s+ 1)(s+ 3)(s+ 5)

The system is being controlled in a feedback system as shown below.

You need to determine how the poles move in the s-plane as the gain, K, is varied. (i.e. plot the root locus)

Using analysis techniques, obtain an expression for the closed loop poles. Sketch the root locus on the plot below

Solution: Using the Matlab’s Workspace and the SISO tool we have the following: >> GS = zpk ([-2], [-1 -3 -5], [5]) Zero/pole/gain: 5 (s+2) ---------------------(s+1) (s+3) (s+5) >> feedback( GS, 1 ) Zero/pole/gain: 5 (s+2) ------------------------------------------(s+1.484) (s^2 + 7.516s + 16.85) >> rlocus(ans) >> rltool

In the Architecture Menu and the System Data submenu, importing G(s),we have the following graphs:

In the plot we can see that the system’s response is critically damped.

Exercise 7 Dr. Abner Mallity has been working for a small electronics design company. They need to produce a chip for a precise amplifier. Here the problems they have.

They have a design for an amplifier stage. o When they implement the design the gain changes due to component variability, temperature (and this is for a wide temperature range application).

They think that they can use a feedback configuration to reduce the variability. o In the feedback configuration they would use three stages. o Each stage has a nominal transfer function of 100/(10-5s + 1) o The propose using a feedback configuration like this one, but instead of unity gain in the feedback path, they propose using a gain of .01.

Using root locus analysis, determine if they can use this configuration. >> GS=tf([100],[10E-5 1]) Transfer function: 100 -----------0.0001 s + 1 >> k=0.01; >> tfr=feedback(k*GS,1) Transfer function: 1 -----------0.0001 s + 2 >>rltool In the Architecture Menu and the System Data submenu, importing G(s), and changing the compensator value in the Compensator Editor C=0.01 in the SISO we have the following graph

The System response is stable because it’s always the root ever stay in the left of the complex axis and incrementing (over damped), therefore, the Dr. Mallity can use this configuration. Exercise 9

Dr. Abner Mallity needs your help. He has been working on a system, and he got his two grad students Willy Nilly and Millie Farad involved. Here is the transfer function of the system, and a block diagram for the system.

G(s) =

5(s- sz ) (s+ 1)(s+ 3)(s+ 5)

The system is being controlled in a feedback system as shown below.

Mallity's problem is that due to various things changing within the system there is a zero that can move and be anywhere on the negative real axis. His grad students say that can present problems. He doesn't see that. He claims the following is true.

No matter where the zero is, the real axis segments are always in the LHP. Since there are three poles and one zero there are only two poles that go to infinity. o Since there are only two branches that go to infinity, they go at +90o and -90o. o Therefore - no problem.

Mallity is convinced that this system will never be unstable, and that it doesn't matter what gain value you have for K, or where the zero is. The trouble is, he can't prove it, and can't convince Willy and Millie that he is right. As a matter of fact, they claim that is not true and he is getting very angry about the whole deal. Help Mallity out. His personal pride is at stake. Show that the system is always stable for any gain or zero position. Solution: Using Matlab to solve this problem we have the following:

For use several values for Sz we need to define a function without zeros and after we define a zero for the 4 cases. >> GS = zpk( [], [-1 -3 -5], 5 ) Zero/pole/gain: 5 ----------------(s+1) (s+3) (s+5) >>rltool(GS) In the Architecture Menu and the System Data submenu, importing G(s), we have the root locus plot:

Adding a Zero between the pole that is in -1 and the imaginary axis, in the Root Locus Editor the system continues being stable:

Moving the zero between the pole that is in -1 and the pole that is in -3 the system continues being stable:

Moving the zero between the pole that is in -3 and the pole that is in -5 the system continues being stable:

Moving the zero between the pole that is in -5 and the negative infinite (-∞) the system continues being stable:

We have proved 4 possible cases of Sz for the System, and we conclude that the Dr. Mallity is at stake, the System is stable for all the values of Sz and the gain K because the response is damped and in the root locus graph the poles ever are left of the imaginary axis. Exercise 11:

In this problem, the system has three poles:

G(s) =

0.5 s(s+ 1)(s+ 2)

The system is in a feedback loop as shown below.

It is desired that the closed loop system dominant poles (Every other pole is at least 5 or 10 times as far into the LHP so that any response from those poles dies out so quickly that the system looks second order - the dominant pole(s)) have a damping ratio of 0.7.

Determine the gain that produces poles with the required damping ratio. Determine if those poles are - in fact - dominant, i.e. the third pole is at least 5x as far into the LHP.

Solution: Using Matlab to solve this problem we have the following: >> GS = zpk( [ ], [0 -1 -2], 0.5 ); >> GS Zero/pole/gain: 0.5 ----------------s (s+1) (s+2)

>> rltool In the Architecture Menu and the System Data submenu, importing G(s),in the plot click right, on the popup menu selecting Design requirement>>New>> type search Damping ratio, and write on Design requirement parameters < 0.7, after we have the following: The plot below shows that the gain is 1.32 when the damping ratio is 0.7 and we move the poles :

The current location for the pinks points on the blacks lines is: -0.381 ± 0.389i, this value multiplied by 5 only in the real axis is; -1.905, and the current point for the pink point on the real axis is -2.24, this tell us that at least this 5 times away, therefore, these parallel poles are Dominant because the lowest poles determine the velocity of the control system. Exercise 13:

Here is the transfer function for a system to control.

G(s) =

4(s+ 0.5) s + 11s + 25.25s+ 65.25

3 2

This open loop system has three poles: -1 + j2.5, -1 - j2.5, -9. The open loop system has one zero: -0.5. The system is being controlled in a feedback system as shown above at the right. Here is the root locus for the system.

Determine the gain, K, which will produce the minimum overshoot in the system by finding the gain for the largest damping ratio in the complex roots. Estimate where the third pole is for the gain found in the first part.

Using Matlab to solve this problem we have the following: >> GS = tf( [4 2], [1 11 25.25 62.25] ) >>Transfer function: 4s+2 ----------------------------------------s^3 + 11 s^2 + 25.25 s + 62.25 >> rltool

In the Architecture Menu and the System Data submenu, importing G(s), in the plot click right, on the popup menu selecting Design requirement>>New>> type search Damping ratio>>OK, adjusting the black line to the minimum value to the exponential function, then moving the poles to these place, we have the following:

The gain observed is k=5.37, the third pole is localized in 3.7 in the Real- axis, with a damping ratio of 8.24, finally we find the minimum overshoot percentage on 194% as we can see in the below figures:

Information Sources: WEB:

E. J. Mastascusa. An Introduction To The Root Locus. Lewisburg, Pennsylvania, USA, URL[http://www.facstaff.bucknell.edu/mastascu/econtrolhtml/RootLocus/RLocus1A.html] consulted November 22nd, 2010. MC. Manuel Amarante Rodríguez. Práctica N° 7 Laboratorio de Ingeniería de Control Análisis de Sistemas de Control por Lugar Geométrico de las Raíces. Nuevo León México. URL[ amarante.fime.uanl.mx/ago-Dic-2010/Labs-ago-2010/ic.../pract7.PDF]. Consulted November 13, 2010 Source Code Matlab

**Practice performed on October 28, 2010, directed by MC Guillermo Ramirez Prado. Annex Source code 10/28/2010
**

%Control toolbox(Sistemas de control en Matlab) %tf transfer function tf([numerador],[denominador]) %zpk zero/pole/gain([ceros],[polos],[ganacia o k]) %impulse impulso %step escalón unitario funcion1=tf([1 2],[1 1 1]) funcion2=zpk([],[-2 -3],1) %grafica impulso impulse(funcion1) impulse(funcion2) %grafica escalon step(funcion1) step(funcion2) %poniendo en serie funcion3= funcion1*funcion2 %poniendo en paralelo funcion3= funcion1+funcion2 %Con retro alimentación funcion3=feedback(funcion1,funcion2) %lugar de las raíces rlocus(funcion1) rlocus(funcion2) %poner en lazo cerrado con k=1 funcion4=feedback(1*funcion1,1) rlocus(funcion4) %con k= 10 funcion4=feedback(10*funcion1,1) rlocus(funcion4) rltool funcion7=zpk([-1 -3],[2 4],1) funcion8=zpk([[],[2 4],1)