System properties: linearity ✗ → ☒ -51=17×3 Definition 6.1. A CT system T is linear if for any two input signals x, x̃ and any two constants –, – ˜ œ C it holds that T {–x + – ˜ x̃} = rT{ } a + IT { I } I in words: if the input is a linear combination of different signals, you can apply T to each piece and add the results (“superposition”) I for a linear system, zero input always produces zero output (why?) I superposition immediately extends to more general sums Y Z n n ]ÿ ^ ÿ T –j xj = –j T {xj }. [ \ j=1 Section 6: Fundamentals of Continuous-Time Systems j=1 6-210 Example: the squaring system is not linear I recall: the squaring system Tsq defined by y(t) = (x(t))2 I let x(t) and x̃(t) be two inputs with corresponding outputs y(t) = Tsq {x}(t) = (x(t))2 , ỹ(t) = Tsq {x̃}(t) = (x̃(t))2 . I for constants –, – ˜ , we have Tsq {–x + – ˜ x̃}(t) = ( 981+1+2 Elt) ) = a' ✗ Hi + ' 248×111×-11-1 + FEW ' I on the other hand –Tsq {x}(t) + – ˜ Tsq {x̃}(t) = 0×11-1 " t IE Iti Therefore, the system Tsq is not linear Section 6: Fundamentals of Continuous-Time Systems 6-211 Example: the RC circuit w/ zero I.C. is linear I recall: the RC circuit TRC {vs }(t) = st 1 ≠(t≠· )/RC e vs (· ) d· 0 RC I let vSc , ṽcS be two input signals and let –, – ˜ be constants I we calculate that c- - ⁄ t Kc A TRC {–vSc + – ˜ ṽcS }(t) = ce≠c(t≠· ) (–vc (· ) + – ˜ ṽSc (· )) d· 0 ⁄ t ⁄ t 1 ≠(t≠· )/RC 1 ≠(t≠· )/RC =– e v (· ) d· + – ˜ ṽSc (· ) d· c S RC RC e S 0 = 0 oTRc{ us }lH + IT,zc{ Is } / t ) Therefore, the system TRC is linear. Section 6: Fundamentals of Continuous-Time Systems 6-212 System properties: time-invariance I notation: let x· (t) = x(t ≠ · ) be short form notation for a time-delayed signal Definition 6.2. A CT system T is time-invariant if for any input x with output y = T {x}, it holds that y· = T {x· } for all time shifts · œ R. physical meaning: an experiment on the system tomorrow will produce the same results as an experiment on the system today. Section 6: Fundamentals of Continuous-Time Systems 6-213 Example: the squaring system is time-invariant I for the squaring system, let x be an input with corresponding output y(t) = Tsq {x}(t) = (x(t))2 I if we simply shift the obtained output, we obtain y· (t) = I# it 1×(1--4) ' I if we instead shift the input signal to be x· (t) = x(t ≠ · ), we compute I # 21 Tsq {x· }(t) = ( ✗ 11-15=1×11--1-15 * These two calculations agree, so Tsq is time-invariant. Section 6: Fundamentals of Continuous-Time Systems 6-214 Example: the system Tmod is not time-invariant I for the system Tmod let x be an input with output y(t) = Tmod {x}(t) = x(t) · A sin(Ê0 t) I if we simply shift the obtained output, we obtain 1¥11 y· (t) = ylt-i-1-xlt-c-1.Asinlw.lt -4 ) I if we instead shift the input signal to be x· (t) = x(t ≠ · ), we compute (# 2) Tmod {x· }(t) = XHt-A-simlwotl-xlt-i-l.A-sinlw.tl These two calculations disagree, so Tmod is not time-invariant. Section 6: Fundamentals of Continuous-Time Systems 6-215 w/ right-sided input Example: the RC circuit w/ zero I.C. is time-invariant s 1 ≠(t≠‡)/RC I for TRC we have y(t) = 0t RC e x(‡) d‡ I if we simply shift the obtained output by · œ R, we have ⁄ t≠· 1 ≠(t≠· ≠‡)/RC 11 # y· (t) = y(t ≠ · ) = x(‡) d‡. I RC e 0 I if we instead shift the input signal by · , the corresponding output is I # 2) TRC {x· }(t) = = ⁄ t≠· ⁄ 0 t 1 ≠(t≠›)/RC e x(› RC ≠ · ) d› 1 ≠(t≠· ≠‡)/RC e x(‡) d‡ RC ≠· change = ⁄ → because g- t≠· 0 ✗ (g) d- variables =3 - t 1 ≠(t≠· ≠‡)/RC e x(‡) d‡. RC = o for set 0 These two calculations agree, so TRC is time-invariant. Physically, the circuit is time-invariant because R and C are constant. Section 6: Fundamentals of Continuous-Time Systems 6-216 yltl is = 1×11-15 causal System properties: causality Definition 6.3. A CT system T is causal if for all t œ R, the output value y(t) depends only on the past input values {x(· )}· Æt . Tard present I physical meaning: the system does not “look into the future” You can often check causality by just inspecting the formula for y(t). s 1 ≠(t≠· )/RC I the RC circuit y(t) = 0t RC e x(· ) d· is causal, because the integral uses only the values {x(· )} for 0 Æ · Æ t I the differentiating system y(t) = y(t) = lim hæ0 d dt x(t) is not causal, because x(t + h/2) ≠ x(t ≠ h/2) , h and therefore y(t) depends on x(t + h/2) for (very) small h > 0. Section 6: Fundamentals of Continuous-Time Systems 6-217 System properties: causality (equivalent definition) I the following equivalent definition is easy to visualize Definition 6.4. A CT system T is causal if for any time t0 and any two inputs x, x̃ that satisfy x(t) = x̃(t) for all t Æ t0 , the corresponding outputs y = T {x} and ỹ = T {x̃} satisfy y(t) = ỹ(t) for all t Æ t0 . interpretation: if two inputs agree up to some time, then the corresponding outputs must also agree up to that time. Section 6: Fundamentals of Continuous-Time Systems 6-218