Sei sulla pagina 1di 9

DSP First

Lab 13: Numerical Evaluation of Fourier Series


Pre-Lab and Warm-Up: You should read at least the Pre-Lab and Warm-up sections of this lab
assignment and go over all exercises in the Pre-Lab section before going to your assigned lab
session.
erification: The Warm-up section of each lab must be completed during your assigned Lab
time and the steps marked Instructor erification must also be signed off during the lab time.
!ne of the laboratory instructors must verify the appropriate steps by signing on the Instructor
erification line. When you have completed a step that re"uires verification# simply demonstrate
the step to the T$ or instructor. Turn in the completed verification sheet to your T$ %hen you
leave the lab.
Lab !eport: It is only necessary to turn in a report on &ection ' %ith graphs and explanations.
You are asked to label the axes of your plots and include a title for every plot. In order to keep
track of plots# include your plot inlined %ithin your report. If you are unsure about %hat is
expected# ask the T$ %ho %ill grade your report.
1 Introduction & Objective
The goal of the laboratory pro(ect is to sho% ho% to calculate the )ourier &eries coefficients#
*a
k
+# %ithout doing the integrals by hand. Instead# %e %ill use LabI,W-s numerical integration
capability to evaluate the integrals numerically. $nother approach %ould be to use a symbolic
algebra package such as Mathematica or .aple to derive formulas for the )ourier &eries
coefficients.
1.1 Background: Fourier Series Analysis and Synthesis
The )ourier &eries representation applies to periodic signals. The )ourier synthesis e"uation for a
periodic signal x/t0 1 x/t 2 T
3
0 is

k
t jk
k
o
e a t x

0 /
# /40
%here
3 3
5 6
is the fundamental fre"uency. To determine the )ourier series coefficients from a
timedomain formula for the signal over one period# %e must evaluate the analysis integral for every
integer value of k7

To
o
t jk
k
dt e t x
T
a
3
0 /
3
4

/60
%here
3 3
5 6 T
is the fundamental period. If necessary# %e can evaluate the analysis integral over
any
period8 in /60 the choice 93# T3: is sometimes a convenient one# but integrating over the interval 9-; T3
# ;T3: %ould also give exactly the same ans%er.
" Pre-Lab
In this lab# %e %ill use a feature of LabI,W that allo%s you to evaluate integrals by numerical
approximation. LabI,W has several numerical integration functions# such as <Lobatto
=uadrature.vi> or <4? numeric integration.vi># that are very po%erful methods for evaluating
integrals of functions specified by formulas. The <formula> for the integrand must be %ritten
similarly to a .ath&cript function. Then the numerical integration function %ill adaptively figure
out the best %ay to approximate the integral %ith a sum. We can use these functions to evaluate
)ourier &eries integrals and then %e %ill plot the resulting )ourier &eries coefficients *ak+ to
display the spectrum.
2.1 Numerical Integration
The basic idea of numerical integration is to approximate an integral %ith a sum# and the
simplest %ay to do this is %ith the @iemann sum7

,
_

L
a b
t f dt t f
b
a
L
n
n
4
0 / 0 /
%here the <sampling points>
n
t
are ( ) +
6
4
n a t
n
# and
L
a b
. The parameter is
actually the %idth of subintervals defined by breaking the interval 9a# b: into L e"ual siAed
subintervals. The sampling points are taken to be the mid-points of those subintervals.
4
)igure 47 $pproximating the area under f/t0 %ith the area of many narro% triangles.
The theory of @iemann integration /in calculus0 tells us that taking the limit as L %ill make
the sum on the right converge to the integral on the left in /B0. )urthermore# if %e think of the
integral as calculating area# then the @iemann sum is approximating the area %ith the sum of
areas of many rectangles %hose %idth is and %hose height is
( )
n
t f
. We could get a better
4
We could define tn to be at one of the endpoints of the interval# and the same results %ould be obtained /in the limit0 for
approximating the integral %ith a sum.
approximation to the area if %e used trapeAoids instead of rectangles to compute the area under
the curve
( ) t f
. Carrying this idea to the next step# %e can use polynomial approximations to
( ) t f
to get even more accurate area approximations.
2.2 Integrate a Simple Function with La!I"#
&uppose that %e %ant to calculate the value of the definite integral
6
.
( ) D cos
E
3

dt t
In LabI,W# this can be done as a numerical approximation %ith functions such as the Lobatto
=uadrature.
2.2.1 Numerical Integration with Loatto $uadrature
In order to use the Lobatto =uadrature# %e must pass three arguments7 the function definition /of
the integrand0# and also the lo%er and upper limits of integration. The limits of integration are
scalar constants. You can pass the integrand in text in the same format as the .ath&cript node.
Fere is an example7
3 Warmup
In the %arm-up you %ill use the "uadrature to do more complicated integrals# such as those
needed in )ourier &eries analysis /60 to extract the )ourier coefficients#
k
a
.
%.1 &iecewise 'e(inition o( a Signal
.any signals are defined by giving several cases that define pieces of the overall signal. )or
example# the s"uare %ave is one for part of its period and Aero for the rest. There are a couple of
%ays of dealing %ith cases %hen %riting the function definition for "uadG. The tricky part is
making sure that the auxiliary function can handle a vector input. Pay careful attention to the
%arning in the help for "uadG that states7
= 1 =H$?G/-)-#$#I0 approximates the integral of )/J0 from $ to I... -)-
is a string containing the name of the function. The function must return
a vector of output values if given a vector of input values.
In other %ords# you must have a vector output if called by "uadG.E It helps to vectoriAe the cases8
perhaps by using the find function or %ith vectoriAed logical tests.
2
This easy example %as chosen so that you can check the result by hand.
/a0 Consider the problem of plotting the triangle %ave of )ig. 6 In order to define the
%aveform so that you can plot it over any interval# it is necessary to figure out t%o things7
ho% to plot one period of the signal# and then ho% to plot other periods. In this part# %rite
an auxiliary function that %ill define the triangle %ave over one period from t 1 3 to t 1
3.3'. In other %ords# convert the follo%ing mathematical statement into vectoriAed
.$TL$I code7
There are t%o %ays to approach this /use either one to %rite your triangle %ave function07
)igure 67 Triangle Wave re"uires a piece%ise definition.
/i0 Hse the find or search 4? array function to identify the subset of inputs associated
%ith each case. )or example# jkl = find(tt>=0 & tt<=1); yy =
exp(tt(jkl))8 %ill define a signal that is
t
e over the interval 3 K t K 4.
/ii0 ?efine one period of the triangle %ave by using unit-step signals to isolate
different regions of the signal. )or example# you can define
( ) ( ) ( ) [ ] 4 t u t u e t f
t
by %riting
function out = myfunc(tt)
out = exp(tt).*((tt>=0)-(tt>=1));
The logical expression ((tt>=0)-(tt>=1)) evaluates to 4 if the condition /3
K t K 40 is true and it evaluates to 3 if the condition is false. Thus# multiplying by
((tt>=0)-(tt>=1)) has the effect of <s%itching on> the exponential
function over the desired set of values of tt. You can use this techni"ue to s%itch
on and off various pieces of the function definition as re"uired in /'0. Lotice that
the t%o cases of the triangle %ave define in /'0 can be %ritten as7
/b0 !ne simple %ay to make the function definition periodic is to use the mod/0 function
%hich is defined in mathematics as a remainder function that gives a positive ans%er#
e.g.# mod/4.4E# 3.E0 is 3.4E because %hen 4.4E is divided by 3.E the integer "uotient is 6
and the remainder is 3.4E. )or negative numbers# the example mod/M3.N# 3.E0 1 3.B is
true because M3.N 1 3.E/M60 2 3.B. Consider the follo%ing vector example7
tt = -1:0.01:1; plot( tt, co(pi*mod(tt,0.!)) )
Hse this idea to define the periodic triangle %ave# but put the mod/0 operation inside the
.-file that defines the triangle %ave.
/c0 Hse your triangle %ave function and fplot to make a plot of the triangle %ave over the
time interval 3.OE K t K 4.3E.
#nstructor erification /separate page0
/d0 &ho% that you can calculate the ?C value of the triangle %ave by integrating over any
one period of the signal. )or example# you can do the integral from t 1 M3.36 to t 1 23.368
or from t 1 3 to t 1 3.3'. ?o the integral over both ranges and compare.P @emember to
divide by the period. Compare your ans%er to the correct ans%er %hich you can obtain
by hand.
#nstructor erification /separate page0
%.2 )sing &arameters in the Integrand
$ Lab E%ercises: Fourier Series &oefficients
In this lab# the ob(ective is to create a set of functions that %ill enable us to do the follo%ing7
4. ,valuate the )ourier &eries coefficients for the follo%ing periodic signal %hich is defined over
one period to be7
0 433 sin/ 6'3 0 / t t x
for
433
4
3 t /P0
The period is 45433 seconds. This signal is called a full-%ave rectified sinusoid# because it
contains only the positive lobe of the sinusoidal function. &uch signals and their )ourier &eries
are often used in ?C po%er supply design.
6. &ynthesiAe approximations to x/t0 using a finite number of )ourier &eries coefficients *a
k
+.

N
N k
kt T ej
k N
a t x
0 5 6 /
3
0 /


%here 6L24 is the number of terms used to form the signal and T3 is the period.
B. &tudy and explain the convergence of the )ourier &eries as N .
Write do%n the )ourier &eries integral that must be evaluated for the x/t0 given in ,". /P0. It is
relatively easy to evaluate this integral using the techni"ues of calculus /re"uired in &ection '.'0#
but first %e are going to <do> this integral numerically.
*.1 Function (or Fourier Synthesis
In this pro(ect %e are going to determine )ourier series representations for periodic %aveforms#
synthesiAe the signals# and then plot them. In general# the limits on the sum in /40 are infinite# but
for our computational purposes# %e must restrict the limits to be a finite number L# %hich then
gives the 6L 2 4 term approximation7

N
N k
kt T ej
k N
a t x
0 5 6 /
3
0 /

/N0
*.1.1 Fourier Synthesis: Sum o( +omple, ",ponentials
This synthesis can be done %ith a <sum of complex exponentials> function similar to that %ritten
for Lab Q6. That %as called syn sin. The re"uired calling se"uence of the ne% function# called
syn fourier# is given belo%. This function implements the computation given in /N0. When %e use
syn fourier for )ourier synthesis# the vector of fre"uencies %ill consist of fre"uencies that are all
integer multiplies of the fundamental fre"uency. In addition# %e must include both the positive
and negative fre"uency components. Therefore# the input vector of complex amplitudes k
a
%ill
be a vector of length 4 6 + N L containing the
+ *
k
a
)ourier coefficients in the order
+ #...# # # 4 0#...# 4 / # *
4 3 N
a a a a N a N a
and the vector
k
f
should contain the harmonic
fre"uencies
+ #...# # 3 # #...# 0 4 / # *
3 3 3 3 3
Nf f f f N Nf
%here 3 3
4 T f
is the fundamental
fre"uency of the periodic signal.
You must %rite a I /called syn fourier0 that %ill synthesiAe a %aveform from complex
amplitude and fre"uency information.
&everal examples of )ourier synthesis of a s"uare %ave are given in the text. )or E3R duty cycle
s"uare-%ave case# the )ourier coefficients are given by the formula7
( )

'

3 4
3 4
4
k
k e
k j a
jk
k

This s"uare-%ave makes a convenient test case %hen you debug your function.
2.1 *.2 'e(ining a -ecti(ied Sinusoid
In this part# use the signal definition given in ,". /P0.
/a0 Write the auxiliary function that %ill be needed to define x/t0. ?efine the parameters
carefully. Turn in the code for this function.
/b0 .ake a plot of x/t0 from ,". /P0 over t%o periods of the signal7 use the range M45433 K t K
45433. Fint7 you can use the <"uotient and remainder> I to force periodicity in a
definition. $ll you need to do is replace t by the remainder of t divided by T
3
so that you
al%ays evaluate the function over the base period of definition. Fere is an example that
sho%s the periodicity of a sa%tooth %ave7
fplot("mod(t,#).*(mod(t,#)>=0 & mod(t,#)<=1)",$-!,!%,&00);
%ith a period of T3 1 B secs. /The argument 633 is used to force L1633 or larger in fplot.0
/c0 ,valuate the ?C value of x/t0 using numerical integration. @emember that this is also the
a
3
)ourier &eries coefficient.
*.% Fourier +oe((icients (or a Full.#a/e -ecti(ied Sine #a/e
/a0 In this case# the integrand is
( )
( ) kt T j
e t x
3
5 6
so it is necessary to define another auxiliary function and introduce another parameter# the
parameter k. Write this auxiliary function /and turn in the code for it0.
/b0 Write a .$TL$I function that %ill evaluate the )ourier &eries coefficients for the rectified sine
%ave for
N N k #...# 6 # 4 # 3 # 4 #...#
. The function %ill contain a for loop to do all the
coefficients from k 1 ML to k 1 2L. It should return a vector containing 6L24 elements %hich are
the
{ }
k
a coefficients. Turn in the code for this function.
/c0 Hse the auxiliary function %ritten in the previous part to evaluate the )ourier &eries coefficients
for x/t0 from ,". /P0. )ind the { }
k
a coefficients for ME K k K E# and make a table of magnitude
and phase versus k.
/d0 In this part# you must synthesiAe a set of %aveforms using different numbers of )ourier
coefficients. Call the syn fourier synthesis function that generates xL/t0 from a finite
number of terms. ?o the cases for L 1 6# L 1 ' and L 1 G.N )or each of these synthesiAed
signals make a plot sho%ing the synthesiAed signal xL/t0 and the desired x/t0 on the same
plot. Hse a three-panel subplot to sho% the three cases on one page.
/e0 ,xplain ho% you are getting convergence as L increases. ?etermine the %orst-case
approximation error for each value of L in part /d0. In other %ords# determine %here the error
bet%een the true signal and the )ourier approximation is the largest# and also record the siAe of
that error.
/f0 Fo% large do you have to make L so that you cannot see any difference bet%een x/t0 and xL/t0 on
a plot sho%n on the computer screenD
/g0 ,xplain the empirical observations about convergence in the previous t%o parts by looking at the
magnitudes of the
k
a
coefficients. In other %ords# explain %hat happens %hen you truncate the
)ourier &eries sum in ,". /40 to get the finite sum in ,". /N0. ,xamine the siAe of the terms that
are omitted. Hse the siAe of the omitted terms to (ustify your choice of L in the previous part.
*.* 0athematical 'eri/ation
?erive the mathematical formula for the )ourier &eries coefficients for x/t0 in ,". /P0. erify that your
formula matches the numerical computation that you did in &ection '.B/c0.
Lab 13
#NS'!U&'(! E!#F#&)'#(N S*EE'
)or each verification# be prepared to explain your ans%er and respond to other related "uestions that the
lab T$-s or professors might ask. Turn this page in at the end of your lab period.
Lame7 SSSSSSSSSSSSSSSSSSSSSSSSSSS ?ate of Lab7SSSSSSSSSSSS
Part B.4/c0 Write a function that defines a periodic triangle %ave# and make a plot of the signal.
erified7 SSSSSSSSSSSSSSS ?ate5Time7SSSSSSSSSSSSS
Part B.4/d0 Hse numerical integration to get the ?C value of the periodic triangle %ave. &ho% that the
ans%er does not depend on %hich period is used in the integral. ?erive /via mathematics0 the numerical
value of a3 in the space belo%7
erified7 SSSSSSSSSSSSSSS ?ate5Time7SSSSSSSSSSSSS

Potrebbero piacerti anche