Quiz · Hard
Calculation-heavy problems on frequency response, inverse Laplace, DFT bins, Parseval and discrete-time stability.
For , what is the magnitude response at ?
0
1
sqrt(2)
2
; at , . At DC it is 0, so this is a high-pass (differencing) filter.
An LTI system has . The steady-state output for the input is
2
10
5
0.5
Final value: . The DC gain is 10/5 = 2.
is driven by . The steady-state output is
cos(100t - 45°)
0.707 cos(100t - 45°)
0.707 cos(100t + 45°)
0.5 cos(100t - 90°)
At omega = 100 the denominator is 1 + j, giving magnitude 0.707 and phase -45 degrees.
Find for .
Partial fractions: A at s=0: 4/2 = 2; B at s=-2: (-2+4)/(-2) = -1. So 2/s - 1/(s+2), giving 2 - e^(-2t).
Sequence . What is (the DC gain)?
0.8
1.25
infinity
5
Geometric series: 1/(1-0.8) = 5. The sum is finite, so the system is BIBO stable (pole at z = 0.8).
A discrete-time causal system has a pole at z = -1.2. It is
unstable
stable
marginally stable
stable but oscillating
Stability requires poles strictly inside the unit circle. |-1.2| > 1, so the response grows with alternating sign.
A 1024-point FFT is taken at kHz. The bin spacing is
1.56 Hz
15.6 Hz
16 Hz
62.5 Hz
Hz. 62.5 Hz would be the bin spacing for N = 256.
A cosine at 1.5 kHz is sampled at kHz with N = 8 points. Which DFT bin (0..7, positive-frequency side) holds the peak?
Bin 1
Bin 2
Bin 3
Bin 4
Bin k corresponds to k fs/N = k x 500 Hz. 1.5 kHz is bin 3 (and the mirror image is bin 5).
The signal is sampled at 8 kHz. What frequencies appear (in the 0 to 4 kHz range)?
3 kHz and 5 kHz
3 kHz and 1 kHz
5 kHz only
3 kHz only
5 kHz is above fs/2, so it aliases to 8 - 5 = 3 kHz, landing on top of the 3 kHz component. The two cannot be told apart afterwards.
A signal has Fourier series coefficients , , . Its average power is
11
6
9
12
Parseval: . Forgetting the negative-k terms gives 6.
An LTI system has h(t) = e^(-t)u(t). If the input is x(t) = u(t), then y(t) = ?
The step response is the integral of h: the integral of e^(-tau) from 0 to t = 1 - e^(-t).
Which two systems in cascade give the same overall response regardless of order?
Any two systems
Only two causal systems
Any two LTI systems
Only two linear systems
Convolution is commutative, so the order of LTI systems in cascade does not matter. For nonlinear or time-varying systems it can.
Why does a rectangular window cause more spectral leakage than a Hann window?
It has a wider main lobe
It adds noise to the signal
It changes the sampling rate
Its abrupt edges produce high side lobes
A rectangular window has the narrowest main lobe but side lobes that decay slowly. Hann tapers to zero, lowering leakage at the cost of a wider main lobe.