palowseyeoe FREQUENCY RESPONSE ANALYSIS , 3.1 SINUSOIDAL TRANSFER FUNCTION AND FREQUENCY RESPONSE The response of a system for the sinusoidal input is called sinusoidal response. The ratio of sinusoidal response and sinusoidal input is called sinusoidal transfer function of the system and in general, it is denoted by Tj). The sinusoidal transfer function is the frequency domain representation of the system, and so it is also called frequency domain transfer function. The sinusoidal transfer, function T(j@) can be obtained as shown below, 1. Contruct a physical model of a system using basic elements/parameters. 2, Determine the differential equations goveming the system from the physical model of the system. 3. Take Laplace transform of differential equations in order to convert them to s-domain equation. 4, Determine s-domain transfer function, T(s), which is ratio of s-domain output and input. 3. Determine the frequency domain transfer function, T(jc) by replacing s by joo in the s-domain transfer function, T(s). | Note : If the s-domain transfer function, T(s) is known, then frequency domain transfer | function, Tj) can be obtained directly from T(s) by replacing s by jo. ie, Tis) "> To) Consider a linear time invariant system with frequency domain transfer function, T(jw) shown in fig 3.1. Let the system be excited by a sinusoidal signal frequency @, amplitude A, and phase 9. Now the response or output will also be a sinusoidal signal of same frequency ©, but the amplitude and phase of response will be modified by amplitude and phase of the transfer function respectively. Now, the amplitude of the response is given by the product of the amplitude of the input and transfer function. The phase of the response is given by the sum of the phase of the input and transfer function Let, Tjw)=|TGo)| Z TG) where, |T(jo)|= Magnitude of TG@), and, 2 T(@)= Phase of TG). Let, Input, 1(t)=A sin (ot + 0)=A 20 where, A= Amplitude of input, @ ~ Frequency of input, and = Phase of input. (t)x T(jo) =A 28 x |T(ja)] Z TG@) =Ax{TGo)| 2(@+ZTGe)) = B46 Ax{T(jo)| = Magnitude of response, and, = 6+ ZT(jo) = Phase of response. Now, Response, e( where, ett) eft) =B Zo jo 4TU@) |» where, B= Ax |T(jo)| 6=8+ ZIGw) r(t)=A sinfot +8) tem with sinusoidal transfer function The).3.2 |EQUENCY RESPONSE The frequency domain transfer function T(ja) is a complex function of o, Hence it can be seperated into magnitude function and phase function. Now, the magnitude and phase functions will be teal functions of a, and they are called frequency response. ‘The frequency response can be evaluated for open loop system and closed loop system. The frequency domain transfer function of open loop and closed loop systems can be obtained from the s-domain transfer function by replacing s by jw shown below. Open loop transfer function : G(s) —“# > G(jo) =|G(jo)| ZGGo) wen) Loop transfer function 2 G(s)H(s) (3.2) Ge) HGo) = |G(jo)HGo)| 26 Go) HG Closed loop transfer function: M(s) =" M(jo) =|M(jo)| ZMGo) 3.3) where, |G(jo)|, |M(jo)|, |Gj@) HG@)| are Magnitude functions 4G(jo), — | = 4 4e’ ral rz ech D) IMG lac 5 aml [o-wy 4g (39) iS a or 3.10) a= ZM(jo) = (jm) ie The resonant peak is the maximum value of M. The condition for maximum value of M can be obtained by differentiating the equation of M with respect to u and letting dM/du ~ 0 when u=u, where, u, = 2= = Normalized resonant frequency. e On differentiating equation (3.9) with respect to u we get, Mod fer eachep? od fcouty ae? ah cayoae 4u(1-u)?— 2fa-uty + age 2 [uty agar 4u (Iu) +807u] swe(3.11) Replace u by u, in equation (3.11) and equate to zero. 4u, (I-w)= La (3.12) 2[a- Ww)? sacs3.5 ‘The equation (3,12) will be zero if numerator is zero. Hence, on equating numerator to zero we get, 4u(1-u7)~80u,=0 => 4u,—4ui—8¢?u, <0 © dup=4u,-807u, => enee(3.13) = i = 1 [awe sactuz? [a-a—264y? +a? 2e) tt ag fca-aF Ke ~|Resonant peak, My 1 (3.14) 2Eyl-¢ | RESONANT FREQUENCY (a) Normalized resonant frequency,u,=t=Ji=2-? (@.15) The resonant frequency, , =o, y1-2¢ ssn(3.16) BANDWIDTH (o,) Let, Normalized bandwidth, u, =" When u = u,, the magnitude M, of the closed loop system is 1//2 (or -3db), Hence in the equation for M (equation 3.9), put u = u, and equate to 1/2. 1 1 Mew ee T [a-uby+40us © On squaring and cross multiplying we get, (l-u)P Haus =20 => T+ uf =2u? +40? u? =2 1.x? =2(1-207)x-1=0 20-207) + Jai-207 +4 2 Let us take only the positive sign, ek = 1-20? +240? +44 => ww d-c)-1 Uy _ 2-20?) 42V d+ 4t? 407) 41 2 1 But, u,=vx : sea Ve [1-207 aac? act ; Also, u,=222 | andwiath, @y=04 uy = PHASE MARGIN (y) The open loop transfer function of second order system, aaa | (18) G@)=—*2— 3.19) Oe 2tm,) me The sinusoidal transfer function GGo) is obtained by letting s = ja o oo, 1 G(jo) = — = +. = —___ oe(3.20) a Foe aT per @, o,) 0, ®, Let normalized frequency, u= ofa, On substituting u = ov, in equation (3.20) we get, : 1 G(jo)=————— s(3.21) Ow ’ 1 1 Magnitude of G(j@) =|G(§0)|=$——ae =m (3.22) {S60} uae? +u? ener Phase of Gijo)=-90" ~tan'— +823) At the gain cross-over frequency «,,, the magnitude of G(ja) is unity. Let normalized gain cross over frequency, u,, = 0,/o, On substituting u by u,, in the equation (3.22) and equating to unity, we get, 2 AtU= Ug, |GGo)| = Let, x=uZ. 50 2x7 +40?x joins a2 fae Let us take only the positive sign, wes 24 faced But, u,.= vx ; cup vk -[-2t ep (3.24) ‘The phase margin, 7 =180+ ZG(jo)| Substituting for 7G(ja) from equation (3.23) in equation (3.25) we get, 20? + fact + if 26 Note : The gain margin of second order sysiem is infinite | (3.26) = 180¢{-s0 tS) =90= tan ze5-2 3.4 CORRELATION BETWEENTIME AND FREQUENCY RESPONSE The correlation between time and frequency response has an explicit-form only for first and second order systems. The correlation for second-order system is discussed here. Consider the magnitude and phase of.a closed loop second order system as a function of normalized frequency, as given by equations (3.9) and (3.10). Phase of closed loop system, a= ZM(a) = The magnitude and phase angle characteristics for normalized frequency u, for certain values of C are shown in fig 3.2 and 3.3. The frequency at which M has a peak value is known as the resonant frequency. The peak value of the magnitude is the resonant peak M, . At this frequency the slope of the magnitude curve is zero. The frequency corresponding to M, is u,, which is the normalized resonant frequency. From equations (3.14) and (3.15) we get, 1 Resonant peak, M, = = yl-o? Resonant frequency, «, =,1-2¢? Wheng=0, 0, =0,,1-22 G27) Wheng=0,. M=—aee (3.28) M aa ad for =£,< 0.707 e 12 foro =o, fort > 0.707 | 08. os : ii | ; ~180° 027° | oO 1 o uy u Fig 3.2 : Magnitude, M as a function of u Fig 3.3 : Phase, o: as @ function of u. From equations (3.27) and (3.28), it is clear that as ¢ tends to zero, @, approaches «,, and M, approaches infinity. When 1-2¢?= 0, @, = 0, which means there is no resonant peak at this condition. 1 1 Let, 1-2¢7=0; .@=— > = et, 1-26 ay ed For 0<¢<1//2 , the resonant frequency always has a value less than @,, and the resonant peak has a value greater than one.3.8 For €>1W2., the condition (dM/du) = 0, will not be satisfied for any real value of @. Hence when ¢> I//2 the magnitude M decreases monotonically from M = 1 at u = 0 with increasing u. It follows that for £ > V/V? there is no resonant peak and the greatest value of M equals one. The frequency at which M has a value of 1/ V2 is of special significance and is called the cut-off frequency w,. The signal frequencies above cut-off are greatly attenuated on passing through a system. For feedback control system, the range of frequencies over which M > 1/'V/2 is defined as bandwidth ,, Control system being low-pass filters (at zero frequency M = 1), the bandwidth @, is equal to cut-off frequency o,. In general the bandwidth of a control system indicates the noise-filtering characteristics of the system. Also, bandwidth gives a measure of the transient response. 1 The normalized bandwidth, maha [tate feat P From the equation of u, itis clear that u, is a function of ¢ alone. The graph between u, and C is shown in fig 3.4. The expression for the damped frequency of oscillation ©, and peak overshoot M, of the step response, for 0<¢< | are, xs Damped frequency, @,=0,j1-@ and Peak overshoot, Mae Comparison of the equation of M, and M, reveals that both are functions of only The sketch of M, and M, for various value of, are shown in fig 3.5. The sketches reveals that asystem with a given value of M, must exhibit a corresponding 1.5-1 value of M, if subjected to a step input. For ¢> 14/2. the resonant peak M, does not exist and the correlation 1.0 breaks down. This is not a serious problem as for this range of, the step response oscillations are well damped 95. and M, is negligible. “The comparison of the equation of ©, and 0, x reveals that there exists a definite correlation between 5 te car t h them. The sketch of fa, with respect to Gis shown in 24. armas bandwith asa fiction af M, M, iy a 3.0. 1.0 EVs 22 Mg 05 7 Pao + > ° 05 0.707 1 6 0 0.5 0.707 6 Fig 3.5: Mand M, asa fimetion of - Fig 3.6 : 0,10, as a function of