1

DSM2 week 2 2011 SPFfirst Chapter 4 (p 83-p. 96 ) is finished by the end of the week.
Ex. 1 The continuous function x(t) may be written x(t)=Acos( t+ ) formula 1

x(t) is sampled at fs=1000Hz and x[n] (shown below, fig. 1) is created. Assume Shannon Sampling Theorem is satisfied and n=0 correspond to t=0. fig.1

1) Determine 2 values of t where x(t) has max. 2) Determine . 3) Determine A, and in formula 1 above. Ex. 2 Three continuous signals s1, s2, s3 is given by s1=cos( 4002t) s2=cos( 6002t) s3=cos(24002t) Sample frequency fs =1000 samples/second.

Determine the normalized radian frequency and the normalized cyclic frequency f for s1, s2 and s3. 2) Does Shannon Sampling Theorem apply? 3) Compute the frequencies of the output signals.
1)

Ex. 3 1) Suppose that the discrete-time signal x[n] is given by the formula: [ ] ( ) Sketch the spectrum (as we did last week Ex. 1.2) 2) If the input x(t) is given by the two-sided spectrum representation shown below, 7ej/3 5e
-j/2

7e-j/3

5ej/2

-3500

-1100 0

1100

3500

f(in Hz)

Determine the discrete-time spectrum for x[n] when fsi=4000 samples/second. Make a plot for your answer, but label the frequency, amplitude and phase of each spectral component. 3) Using the discrete-time spectrum from part b), determine the analog frequency components in the output y(t) when the sampling rate of the D-to-C converter is fso=8000 samples/sec. In other words, the sampling rates of the two converters are different. Ex. 4 MATLAB supports sound on PC platforms (on platforms with audio devices). soundsc(y,fs) sends the sampled signal in the vector y to the computers speaker at sample frequency fs. See MATLAB help soundsc for more information. a) With n=0:4000 try: soundsc(cos(n*0.2*pi),4400). Which frequency did you hear? The plot below is titled Spectogram, it is a plot of frequency versus time. In the interval [0, 0.2[ x(t) is a sinusoid of frequency 500 Hz, In the interval [0.2,0.4[ x(t) is a sinusoid of frequency 1000 Hz In the interval [0.4,0.6[ ....etc.

The signal x(t) is sampled with a sampling frequency fs=3200 samples/sec. b) Create the vector xn and check the sound using soundsc(xn,fs). Did you get the expected result? If not explain why. Help: xn=[x1n x2n ], where x1n=cos(500*2*pi*n*Ts), x2n=cos(1000*2*pi*n*Ts) etc. c) Try also the MATLAB function specgram(xn,[],fs). d) To avoid aliasing, change the sampling frequency. Ex. 5 In Ex. 4 the input signal x(t) was a staircase shaped signal. In this problem x(t) is a linear ( ) ( ) FM chirp signal: [ ] ( ( ) ) ( ) After sampling: MATLAB has a linear chirp function: Y = CHIRP(T,F0,T1,F1) generates samples of a linear swept-frequency signal at the time instances defined in array T. The instantaneous frequency at time 0 is F0 Hertz. The instantaneous frequency F1 is achieved at time T1. Now run the following MATLAB script:
% DEMO of aliasing by bej. Just run "aliastest" with sound turned on. fs=16000; f1=120; f2=24000; T=8; x = chirp(0:1/fs:T,f1,T,f2,'linear'); soundsc(x,fs) specgram(x,256,fs)

1) Determine the max. frequency 2) Sketch the spectrogram if fs = 8000 samples/second. 3) Calculate (by differentiation) the frequency as function of time of the signal xt cos(2f m t 2 )

Ex. 6 A block diagram representation of a sampling and reconstruction system is given below. x(t) Ideal C-to-D Converter x[n] Ideal D-to-C Converter y(t)

Ts = 1/fs

Ts = 1/fs

Suppose the continuous-time input x(t) to the above system is given as ( ) ( ) ( ) ( ) ( ) ( 1) What sampling rate is required such that no aliasing occurs for x(t)? 2) Given fs=10000 samples/second, plot the frequency spectrum for x[n]: )

-0.5

-0.4

-0.3

-0.2

-0.1

0 0.1

0.2

0.3

0.4

0.5

3) Given that fs=3000 samples/second, plot the frequency spectrum for x[n]:

-0.5 Given that

-0.4

-0.3

-0.2

-0.1

0 0.1

0.2

0.3

0.4

0.5

( )

4) and fs=10000 samples/second, write a simplified expression for the output y(t) in terms of cosine functions.