Output of transfer function given impulse response and input the

2 replies · 2K views · last reply
jaus tail
Messages
613
Reaction score
48

Homework Statement


Multiple-choice question: continuous-time LTI system with h(t) = e^(-5t)u(t) and x(t) = e^(5t)u(-t), asking for y(t) at t = 10 msec. Options A through D with values 0.035, 0.053, 0.026 and 0.095; option D is selected.


Homework Equations


Find laplace of x(t) and h(t)
Multiply the laplaced values to get Y(s)
Find inverse laplace to get y(t)

The Attempt at a Solution


Handwritten derivation of the output y(t) of an LTI system via Laplace transforms, using h(t) = e^(-5t)u(t), H(s) = 1/(s+5), and X(s) = -1/(s-5) from the time-reversed transform, ending with y(t) = (1/10)(e^(-5t) - e^(5t)).

In book the circled term isn't there. Why does that go away?
Book gets y(t) as
Handwritten-style worked solution showing convolution of x(t) = e^(5t)u(-t) and h(t) = e^(-5t)u(t), evaluating integrals to give y(t) = (1/10)e^(-5t), then y(10 ms) = 0.095 boxed, with the e^(-50x10^-3) term underlined in red.

In laplace why does the e5t term go away?
 
Attachments
  • Multiple-choice question: continuous-time LTI system with h(t) = e^(-5t)u(t) and x(t) = e^(5t)u(-t), asking for y(t) at t = 10 msec. Options A through D with values 0.035, 0.053, 0.026 and 0.095; option D is selected.
    upload_2018-2-9_10-5-47.png
    10.6 KB · Views: 1,225
  • Handwritten derivation of a system output using Laplace transforms: h(t) = e^(-5t)u(t) with H(s) = 1/(s+5), x(t) = e^(5t)u(t), time-reversal step f(-t) and F(s), X(s) = -1/(s-5), then Y(s) = H(s)X(s) and y(t) = (1/10)(e^(-5t) - e^(5t)).
    upload_2018-2-9_10-8-11.png
    50.9 KB · Views: 637
  • Handwritten-style worked solution showing convolution of x(t) = e^(5t)u(-t) and h(t) = e^(-5t)u(t), evaluating integrals to give y(t) = (1/10)e^(-5t), then y(10 ms) = 0.095 boxed, with the e^(-50x10^-3) term underlined in red.
    upload_2018-2-9_10-8-58.png
    12.2 KB · Views: 727
  • Handwritten derivation of the output y(t) of a system via Laplace transforms, using h(t) = e^(-5t)u(t), x(t) = e^(5t)u(t), H(s) = 1/(s+5), X(s) = -1/(s-5), and Y(s) = H(s)X(s).
    upload_2018-2-9_10-9-33.png
    49.6 KB · Views: 582
  • Handwritten notes on lined paper deriving an output y(t) by Laplace transforms: h(t) = e^(-5t)u(t), H(s) = 1/(s+5), x(t) = e^(5t)u(t), and a time-reversed signal with transform 1/(-s+5), ending in y(t) = (1/10)(e^(-5t) - e^(5t)), with the e^(5t) term circled in red.
    upload_2018-2-9_10-10-5.png
    47.9 KB · Views: 612
  • Handwritten derivation of the output y(t) of an LTI system via Laplace transforms, using h(t) = e^(-5t)u(t), H(s) = 1/(s+5), and X(s) = -1/(s-5) from the time-reversed transform, ending with y(t) = (1/10)(e^(-5t) - e^(5t)).
    upload_2018-2-9_10-17-34.png
    50.9 KB · Views: 1,106
You will fin that the answer derives from the convolution integral. That method is valid for all time -∞ < t < ∞. The Laplace transform you're familiar with is most likely the one-sided Laplace transform which assumes x(t) = 0 for t < 0, so not applicable here.

There are two other alternatives: the Fourier transform and the two-sided Laplace transform. The latter is seldom encountered (see footnote) while the Fourier is appropriate and gives the same answer.

[Footnote: Then Brooklyn Poly's Professor John G. Truxal's venerable "Automatic Feedback Control System Synthesis" invokes it in preference to the Fourier].
 
  • Like
Likes   Reactions: jaus tail
Thanks. Convolution is better than the Fourier route.