University of California, San Diego Spring 2014 ECE 45 Discussion 6 Solutions Topics • Fourier Transform • Properties of F.T. Fourier Transform The Fourier transform is a way of representing a function in terms of its frequency components. We can think of it as a Fourier series with an infinite period (i.e. it never repeats itself), so it can have frequency contributions from any frequency. Z ∞ 1 F (jω)ejωt dω f (t) = 2π −∞ Z ∞ f (t)e−jωt dt F (jω) = −∞ Fourier Transform as an Input to an LTI System: Since a Fourier transform is an integral (sum with infinitesimally small intervals) of sinusoidal components, we know how to analyze the output of a system with a periodic function as its input. For any input x(t). Z ∞ Z ∞ 1 1 jωt x(t) = X(jω)e dω −→ H(ω) −→ y(t) = |H(ω)|X(jω) ej(ωt+∠H(ω)) dω 2π −∞ 2π −∞ We can represent y(t) as a Fourier Transform: Y (jω) = X(jω) H(ω) The Fourier transform is a one-to-one mapping, so if we know either the time function or the frequency function, we know the other as well. Parseval’s Theorem: Power is the same whether we look at our signal in the time or frequency domain Z ∞ Z ∞ 1 2 |x(t)| dt = |X(jω)|2 dω 2π −∞ −∞ Time Shifting: f (t − t0 ) ↔ F (jω) e−jωt0 Frequency Shifting: f (t) ejω0 t ↔ F (j(ω − ω0 )) Linearity: if f1 (t) ↔ F1 (jω) and f2 (t) ↔ F2 (jω) then a f1 (t) + b f2 (t) ↔ a F1 (jω) + b F2 (jω) Conjugation: x∗ (t) ↔ X ∗ (−jω) Time Scaling: 1 f (c t) ↔ X |c| jω c Time Derivative: df (t) ↔ jωF (jω) dt Very useful exercise to derive the properties using the definitions: Z ∞ 1 f (t) = F (jω)ejωt dω 2π −∞ Z ∞ F (jω) = f (t)e−jωt dt −∞ Unit Step Function (Heaviside step function) Useful for representing a signal “turning on” or starting at a certain value 0 t<0 u(t) = 1 t≥0 u(t − t0 ) = 0 t < t0 1 t ≥ t0 for any function f (t) 0 t < t0 f (t) u(t) = f (t) t ≥ t0 Example 1: Determine the Fourier transform of f (t) = u(t) e−a t where a > 0 Z ∞ Z ∞ 1 −jωt e−a t e−jωt dt = f (t)e dt = F (jω) = a + jω 0 −∞ Example 2: Determine the Fourier transform of g(t) = u(−t) ea t where a > 0 g(t) = f (−t)so we can use the time scaling property with c = −1 G(jω) = 1 1 F (jω/(−1)) = F (−jω) = | − 1| a − jω Example 3: Determine the Fourier transform of h(t) = ea |t| We calculated this last week, but we can use the linearity property to confirm our result, since h(t) = f (t) + g(t) H(jω) = F (jω) + G(jω) = 1 1 2a + = 2 a + jω a − jω a + ω2 Example 4: Given the input to an LTI system is x(t) = u(t) e−t , determine the output y(t) when the e−jω transfer function of the is H(ω) = 1+jω X(ω) = 1 1 + jω Y (ω) = X(ω)H(ω) = 1 y(t) = 2π Z e−jω (1 + jω)2 ∞ Y (jω)ejωt dω −∞ Let G(jω) = 1 (1+jω)2 Note: Y (jω) = G(jω) e−jω From the table of Fourier transform pairs: G(jω) = 1 ↔ g(t) = t e−t u(t) (1 + jω)2 Using the time shifting property: y(t) = g(t − 1) = (t − 1) e−(t−1) u(t − 1) Example 5: Determine the inverse Fourier transform (time domain representation) of F (ω) = 19 − jω 10 + ω 2 + 3jω Use partial fractions to break up F (ω) F (ω) = A B 19 − jω = + (5 − jω)(2 + jω) 5 − jω 2 + jω A(2 + jω) + B(5 − jω) = 19 − jω 2A + 5B = 19 and A − B = −1 → A = 2 and B = 3 F (ω) = 2 3 + = 2u(−t) e5t + 3u(t) e−2t 5 − jω 2 + jω Example 6: What is the power of the signal f (t) = π5 sinc(5t) Z ∞ Z 25 ∞ 2 |f (t)| dt = 2 P = |sinc(5t)|2 dt =??? π −∞ −∞ Recall the rectangle function: 0 ) is a rectangular pulse with height A, width W , and centered at t0 . A rect( t−t W A (t0 − W/2) ≤ t ≤ (t0 + W/2) t−t0 A rect( W ) = 0 else Use Parseval’s to simplify calculation! From the transform pairs table (and from last week) 1 |ω| ≤ 5 F (jω) = rect(ω/10) = 0 else 1 P = 2π Z ∞ 1 |F (jω)| dω = 2π −∞ 2 Z 5 1 dω = −5 5 π Example 7: Determine the Fourier transform of t 0≤t≤2 g(t) = 0 else We could do the integration, but we’d have to use integration by parts, so let’s use the properties instead. 1 −1 ≤ t ≤ 1 Let f (t) = rect(t/2) → F (jω) = 2 sinc(ω) 0 else Note f (t) = d g(t dt + 1) so F (jω) = jω ejω G(jω) Thus G(jω) = F (jω) 2 e−jω sinc(ω) = jω ejω jω