The root locus plots in the s-plane the locus of all the roots of 1+Go(s)=0 as k varies. The OLTF will contain the controller Gc(s) and the poles of the CLTF. A method can be used to predict the CLTF poles given the zeros and poles of OLTF.
The root locus plots in the s-plane the locus of all the roots of 1+Go(s)=0 as k varies. The OLTF will contain the controller Gc(s) and the poles of the CLTF. A method can be used to predict the CLTF poles given the zeros and poles of OLTF.
The root locus plots in the s-plane the locus of all the roots of 1+Go(s)=0 as k varies. The OLTF will contain the controller Gc(s) and the poles of the CLTF. A method can be used to predict the CLTF poles given the zeros and poles of OLTF.
3.1 THE PROBLEM From the performance specifications: know the dominant 2 nd order poles that the Closed Loop Transfer Function (CLTF) is to have We know poles and zeros of G(s) and H(s). Must design the controller Gc(s) to give the required CLTF poles (plus any other non-dominant ones). We need a method (quick and easy) to predict the CLTF poles given the zeros and the poles of the Open Loop Transfer Function (OLTF). The OLTF will contain the controller Gc(s). We guess the form of this which we think will do the job. We then can apply easy methods to calculate the coefficients of Gc(s) to give the required CLTF poles. 3.2 THE ROOT LOCUS Given the OLTF Go(s) with Go(s)=kGc(s)G(s)H(s) (where k is normally a gain included in the controller), the root locus plots in the s-plane the locus of all the roots of 1+Go(s)=0 as k varies. Since the denominator of the CLTF is going to be 1+Go(s), then the root locus plote the locus of all the poles of the CLTF for a variable loop gain constant k. i.e. it tells you the value of the CLTF poles as k varies. Note: The poles and hence the dynamic response of error abd disturbance transfer functions are the same as that of the CLTF. Only the zeros will be different. So we have: (k included in the controller) G(S) kGc(s) - H(s) + G(S) kGc(s) c(s) r(s) - H(s) + Go(s)=kGc(s)G(s)H(s) Denominator of CLTF 1+ kGc(s)G(s)H(s) = 0 There are 3 methods for obtaining the root locus: 1) By Maths 2) By Computer 3) By some sketching rules 3.2.1 By Maths Example: Let Gc(s)=1 and have only a gain as controller (i.e. we are not designing anything!) ) ( 1 ) ( ) 7 )( 1 ( 4 1 ) 7 )( 1 ( 4 ) ( ) ( s kG s kG s s k s s k s r s c + = + + + + + = ) 4 7 ( 8 4 2 k s s k CLTF + + + = Root locus from 1+kG(s)=0 0 ) 7 )( 1 ( 4 1 = + + + s s k 0 ) 4 7 ( 8 2 = + + + k s s For different values of k: k 0 1.25 2.25 2.5 8.5 k CLTF denominator 7 8 2 + + s s 12 8 2 + + s s 16 8 2 + + s s 17 8 2 + + s s 41 8 2 + + s s A s s + + 8 2 A Closed loop poles S 1 =-1 S 2 =-7 S 1 =-2 S 2 =-6 S 1/2 = -4 S 1 =-4+j1 S 2 =-4-j1 S 1 =-4+j5 S 2 =-4-j5 S 1 =-4+j|, S 2 =-4-j1| | Characteristics e 0 =2.645 ,=1.5 e 0 =3.464 ,=1.15 e 0 =4 ,=1 e 0 =4.123 ,=0.97 e 0 =6.4 ,=0.62
0 e 0 ) 7 )( 1 ( 4 + + s s k c(s) r(s) - + G(s) Plot the roots (poles) in the s-plane: System closed loop poles moves as the value of k changes. Obviously this method is very slow. What about drawing the root locus if: ) 8 )( 4 )( 3 ( 30 4 ) ( 2 + + + + + = s s s s s s s G Note If s=x+jy lies on the root locus, then s=x+jy is a solution to 1+kG(s)=0 o jy x s s kG 180 1 1 ) ( Z = = + = Example: Is s=10 on the previous root locus? o k k k 0 27 4 27 4 ) 7 10 )( 1 10 ( * 4 Z = = + + The answer is no because k>0. Is s4+j4 on the previous root locus? o o o k k j j k j j k 180 25 4 13 . 53 5 * 87 . 126 5 4 ) 4 3 )( 4 3 ( 4 ) 7 4 4 )( 1 4 4 ( * 4 Z = c c = + + = + + + + The answer is now yes! The value of k for which the closed loop poles are -4j4 is: 1 25 4 = k k=6.25 3.2.2 By Computer We can use software like Matlab or codas and tha root locus will be obtained simply by typing the relevant transfer functions. 3.2.3 Sketching root locus by hand Rules for drawing the root locus 1) The Root Locus (RL) starts on open loop poles (k=0) and finishes on open loop zeros (k=) If OLTF has np poles and nz zeros (normally np > nz), then RL will go to in np - nz directions (i.e. we have np nz zeros at ). E.g. if G(s)= ) 7 )( 6 ( ) 3 ( 14 + + + s s s Then 1+kG(s)=0 (s+6)(s+7)+14k(s+3)=0 When k=0 closed loop roots are -6, -7 When k is very large ( ) the roots are -3 and 1a) Closed loop poles (i.e. RL) will always try and get to the open loop zeros avoiding the open loop poles 2) The real axis to the left of an odd number of OL poles and zeros is part of the RL. (i.e. A segment of the real axis is part of the root locus if there are an odd number of poles and zeros counted together to the right of it) Ex. ) 5 4 )( 9 )( 1 ( ) 20 8 )( 5 ( ) ( 2 2 2 + + + + + + = s s s s s s s s G 3 zeros: -5, -4j2 5 poles: -1, 3, -2j1 np-nz=2 2 zeros at Ex. ) 3 ( 1 ) ( 2 + = s s s G np-nz=3 3 zeros at poles: 0, 0, -3 Ex. ) 2 )( 1 ( ) 3 ( ) ( + + + = s s s s G np-nz=1 1 zero at Observation: If two real poles and one real zero with zero to the left of both poles and the distances of the two poles from the zero are respectively a and b, then the RL includes a circle of radius ab centred on the zero. 3) If RL goes to , it will converge to straight line asymptotes 3a) Asymptotes lie in the s plane at angles: nz np i
+ t ) 1 2 ( rad (for i=0,1,2,3 etc.) 3b) Asymptotes intersect the real axis at nz np z p z p n j j n i i
= nz np sum
) zeros of sum - poles of (
Where p i are the poles (i=0,1,2.np) z j are the zeros (j=0,1,2.nz) Ex. G(s) = ) 6 ( ) 1 ( 1 2 + + s s np-nz=3 3 zeros at poles = -1, -1, -6 3) (i /3 7 2) (i /3 5 1) (i 0) (i 3 / 3 ) 1 2 ( = = = = = + t t t t t i Intersect at 3 8 3 ) 6 ( ) 1 ( ) 1 ( = + + Ex. G(s) = ) 2 )( 13 4 ( ) 3 ( 2 + + + + s s s s np-nz=2 2 zeros at poles = -2j3, -2; zero: -3 Same as the 1 st angle Asymptotes: 3) (i /2 7 2) (i /2 5 1) (i 2 / 3 0) (i 2 / 3 ) 1 2 ( = = = = = + t t t t t i Intersect at 2 3 3 ) 3 ( ) 3 2 ( ) 3 2 ( ) 2 ( = + + + j j 4) Graphical method to check that a given point is on the RL Consider G(s) = ) 2 )( 52 12 ( ) 4 ( 2 + + + + s s s s s poles = -6j4, -2, 0; zero: -4 Is point A on the RL? Let A be s=x+jy 4 4 3 3 2 2 1 1 1 1 * * * ) 2 )( 4 6 )( 4 6 )( ( ) 4 ( p p p p p p p p z z V V V V kV jy x j jy x j jy x jy x jy x k 0 0 0 0 0 Z Z Z Z Z = = + + + + + + + + + + A is on the root locus if it satisfies the following condition: 0 ) ( 1 = + s kGo o s kGo 180 1 1 ) ( Z = = Same as the 1 st two angles Which means: a) 1 * * * 4 3 2 1 1 = p p p p z V V V V kV b) o z p p p p 180 1 4 3 2 1 = Z Z + Z + Z + Z 0 0 0 0 0 a) is called the GAIN CRITERIA (the value of k can be obtained from here) b) is called the ANGLE CRITERIA These two criteria can be generalized for any number of poles and zeros in the transfer function. 5) Angle of departure (convergence) from a pole (to a zero) Consider the following situation: Assume that the RL has just left the pole P by an unknown angle u . We want to find u . Consider a point A very near to P (i.e. u P = u A ) From the angle criteria in rule 4 applied to point A we have: o z p p p p 180 1 3 2 1 = + + + 0 0 0 0 0 o p p p z p 180 3 2 1 1 + = 0 0 0 0 0 So if l represent the generic pole we have: Angle of departure from a pole For convergence onto a zero we will have: (l represent the generic zero) Note: for a multiple pole (zero) of order q we MUST count the angle from the pole (to a zero) q times having: 180 1 + =
= = l i pi n j zj pl z q u u u 180 1 + =
= = l i zj n j pi zl p q u u u 180 1 + =
= = l i pi n j zj pl z u u u 180 1 + =
= = l i zj n j pi zl p u u u 6) Breakaway points from real axis This rule is more complex than the others to apply. There are a graphical and a analytical techniques to determine the breakaway points. We`ll have only a look briefly to the analytical one. Given the characteristic equation 1+kG(s)=0 We can isolate the factor k obtaining k=p(s). Breakaway points will be the maximum of k=p(s) and they can be determined by imposing: 0 ) ( = = ds s dp ds dk Example: 4) 2)(s -(s p(s) k 0 ) 4 )( 2 ( k 1 kG(s) 1 ) 4 )( 2 ( 1 ) ( + + = = = + + + = + + + = s s s s s G -3 s 0 6) (2s - 0 ) ( = = + = = ds s dp ds dk There are other secondary rules, but those six are more than enough to sketch a good root locus. Re Im -2 -4 Breakaway point