Some of the work can be done using MATLAB:

» n=1:15;
» cn=-4*j./n/pi.*sin(pi*n/6).*sin(n*pi/2).*exp(-j*n*pi/3);
» n=-15:-1;
» c_n=-4*j./n/pi.*sin(pi*n/6).*sin(n*pi/2).*exp(-j*n*pi/3);
» cn=[c_n 0 cn];
» n=-15:15;
» subplot(221),stem(n,abs(cn))
» title('|c_n|')
» subplot(222),stem(n,angle(cn))
» title('angle(c_n) in rad')

To plot the Fourier series to check your answers:

T=6;
w0 = 2*pi/T;
t = -1.5*T:T/1000:1.5*T;
N = input(’Number of harmonics ’);
c0 = 0;
x = c0*ones(1,length(t)); % dc component
for n=1:N,
cn = -4*j/n/pi*sin(pi*n/6)*sin(n*pi/2)*exp(-j*n*pi/3);
c_n = conj(cn);
x = x + cn*exp(j*n*w0*t) + c_n*exp(-j*n*w0*t);
end
plot(t,x)
title([’ N = ’,num2str(N)])

|c n| angle(c n) in rad
0.8 4

0.6 2

0.4 0

0.2 -2

0 -4
-20 -10 0 10 20 -20 -10 0 10 20
N = 50
4

-2

-4
-10 -5 0 5 10
Some of the work can be done using MATLAB:

» n=-10:10;
» cn=cos(pi/2*n*w0)/5./(1-(n*w0).^2);
» subplot(221),stem(n,abs(cn))
» title('|c_n|')
» subplot(222),stem(n,angle(cn))
» title('angle(c_n) in rad')

will give first plots shown below. To check your answer, you can plot the truncated series and see if it
converges correctly.

T=10;
w0 = 2*pi/T;
t = -1.5:T/1000:1.5*T;
N = input(’Number of harmonics ’);
c0 = 1/5;
x = c0*ones(1,length(t)); % dc component
for n=1:N,
cn = cos(pi/2*n*w0)/5/(1-(n*w0)^2);
c_n = cn;
x = x + cn*exp(j*n*w0*t) + c_n*exp(-j*n*w0*t);
end
plot(t,x)
title([’ N = ’,num2str(N)])

|c n| angle(c n) in rad
0.2 4

0.15 3

0.1 2

0.05 1

0 0
-10 -5 0 5 10 -10 -5 0 5 10

N = 20
1

0.5

-0.5
-20 -10 0 10 20
Time (sec)
% trucated Fourier series for a staircase signal
T = 4;
w0 = 2*pi/T;
t = 0:T/1000:3*T;
N = input(’input N’)
nneg = -N:-1;
npos = 1:N;
c0 = 3/4;
cneg = 1/j/pi./nneg.*(1-.5*(-1).^nneg-.5*exp(-j*nneg*pi/2));
cpos = 1/j/pi./npos.*(1-.5*(-1).^npos-.5*exp(-j*npos*pi/2));
n = [nneg 0 npos];
c_n = [cneg c0 cpos];
x = c_n*exp(j*w0*n’*t);
x = real(x); % strips off negligible imaginary parts
plot(t,x)
xlabel(’Time (sec)’)
title([’x’,num2str(N),’(t)’])

The truncated series is plotted from top to bottom for N = 3, N = 10, and N = 40.

x 3(t)
4

-2
0 2 4 x 10(t)
6 8 10 12
4 Time (sec)

-2
0 2 4 x 40(t)
6 8 10 12
4 Time (sec)

-2
0 2 4 6 8 10 12
Time (sec)
This program uses a loop to compute the Fourier series rather than the large matrix multiplication.
One student had difficulty using the student version of Matlab with the other program (it took too
much memory). This one is less efficient because of the loop, but it uses less memory.

% computes exponential Fourier series for trapezoidal wave

Running this program for N = 3, N= 10 and N= 40 gives:

N=3
2

-1
-5 0 N = 10 5 10
2

-1
-5 0 N = 40 5 10
2

0
-5 0 5 10
Time (sec)