Top Banner
Where we’re going Speed, Storage Issues Frequency Space
27

Where we’re going Speed, Storage Issues Frequency Space.

Dec 25, 2015

Download

Documents

Joleen Mills
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Where we’re going Speed, Storage Issues Frequency Space.

Where we’re going

Speed,StorageIssues

Frequency Space

Page 2: Where we’re going Speed, Storage Issues Frequency Space.

Sine waves can be mixed with DC signals, or with other sine

waves to produce new waveforms. Here is one example of a

complex waveform:

V(t) = Ao + A1sin1t + A2sin 2t + A3sin 3t + … + Ansin nt--- in this case---V(t) = Ao + A1sin1t

Ao

A1

Fourier Analysis

Just an AC component superimposed on aDC component

Page 3: Where we’re going Speed, Storage Issues Frequency Space.

More dramatic results are obtained by mixing a sine wave of a particular frequency

with exact multiples of the same frequency. We are adding harmonics

to the fundamental frequency. For example, take the fundamental frequency and add 3rd

harmonic (3 times the fundamental frequency) at reduced amplitude, and subsequently add

its 5th, 7th and 9th harmonics:

Fourier Analysis, cont’d

the waveform begins to look more and more like a square wave.

Page 4: Where we’re going Speed, Storage Issues Frequency Space.

This result illustrates a general principle first formulated by the

French mathematician Joseph Fourier, namely that any complex waveform

can be built up from a pure sine waves plus particular harmonics of the

fundamental frequency. Square waves, triangular waves and sawtooth waves

can all be produced in this way.

...)7sin(7

1)5sin(

5

1)3sin(

3

1)sin(

1

1)(

,

tttttf

thatshownbecanitwavesquarethefor

oooo

(try plotting this using Excel)

Fourier Analysis, cont’d

Page 5: Where we’re going Speed, Storage Issues Frequency Space.

Spectral Analysis• Spectral analysis means determining the

frequency content of the data signal• Important in experiment design for

determining sample rate, fs - sampling rate theorem states: fs max fsignal to avoid aliasing

• Important in post-experiment analysis- Frequency content is often a primary experiment result. Experiment examples:

- determining the vibrational frequencies of structures

- reducing noise of machines- Developing voice recognition software

Page 6: Where we’re going Speed, Storage Issues Frequency Space.

Spectral analysis key points

Any function of time can be made up by adding sine andcosine function of different amplitudes, frequencies, and phases.

These sines and cosines are called frequency components or harmonics.

Any waveform other than a simple sine or cosine has more than one frequency component.

Page 7: Where we’re going Speed, Storage Issues Frequency Space.

Example Waveform• 1000 Hz sawtooth, amplitude 2 Volts

1000 Hz Sawtooth

-3

-2

-1

0

1

2

3

0.00E+00

2.00E-04

4.00E-04

6.00E-04

8.00E-04

1.00E-03

time

am

pli

tud

e (

volt

s)

tnbtbtbb

tnatatatfgeneralIn

n

nn

002010

0201

cos...2coscos

sin...2sinsin)(,

Fundamental frequency term Harmonic terms

b0 is the average value of thefunction over period, T

Period, T = .001 sec

Page 8: Where we’re going Speed, Storage Issues Frequency Space.

Fourier Coefficients, an and bn

• These coefficients are simply the amplitude at each component frequency

• For odd functions [f(t)=-f(-t)], all bn= 0, and have a series of sine terms (sine is an odd function)

• For even functions [f(t)=f(-t)], all an= 0, and have a series of cosine terms (cosine is an even function)

• For arbitrary functions, have an and bn terms.

• Coefficients are calculated as follows:

functionsoddfordttntfT

b

functionsevenfordttntfT

a

T

n

T

n

0cos)(2

0sin)(2

0

0

0

0

Page 9: Where we’re going Speed, Storage Issues Frequency Space.

An odd function (sine wave)

Page 10: Where we’re going Speed, Storage Issues Frequency Space.

More odd functions

Fundamentalor First Harmonic

Third HarmonicSine series orPure imaginary amplitudes

Page 11: Where we’re going Speed, Storage Issues Frequency Space.

An even function (cosine wave)

Page 12: Where we’re going Speed, Storage Issues Frequency Space.

More even functions

Fundamentalor First Harmonic

Second HarmonicCosine series orPure real amplitudes

Page 13: Where we’re going Speed, Storage Issues Frequency Space.

Periodic, but neither even nor odd

Cosine and sine series orComplex amplitudes

Page 14: Where we’re going Speed, Storage Issues Frequency Space.

Sawtooth Fourier Coefficients• Odd function so:

• Using direct integration or numerical integration we find the first seven an’s to be:

• We can plot these coefficients in frequency space:

0

sin)(2

0

0

n

T

n

b

dttntfT

a

0000.0

0331.1801.

000.00000.0

0648.6211.1

4

73

62

51

a

aa

aa

aa

Fourier Coefficients

-0.5

0

0.5

1

1.5

2

0

1000

2000

3000

4000

5000

6000

7000

8000

frequency

ampl

itude

Our sawtooth wave is an ________ function. Therefore all ____ = 0

Page 15: Where we’re going Speed, Storage Issues Frequency Space.

Start with a sine wave...

Page 16: Where we’re going Speed, Storage Issues Frequency Space.

Add an odd harmonic (#3) ...

Page 17: Where we’re going Speed, Storage Issues Frequency Space.

Add another (#5)...

Page 18: Where we’re going Speed, Storage Issues Frequency Space.

And still another (#7)...

Page 19: Where we’re going Speed, Storage Issues Frequency Space.

Let’s transform a “Sharper” sawtooth

sec10.05.660)(

sec05..060)(

sec1./1

sec,/83622,10 .

tforttf

tforttf

fT

radfHzf

Sharp Sawtooth

-4

-2

0

2

4

0 0.02 0.04 0.06 0.08 0.1

time

f(t)

6366.)83.623sin()660()83.623sin(601.

23sin)(

2

9549.)83.622sin()660()83.622sin(601.

22sin)(

2

9098.1)83.62sin()660()83.62sin(601.

21sin)(

2

05.

0

1.

05.0

03

05.

0

1.

05.0

02

05.

0

1.

05.0

01

dtttdtttdtttfT

a

dtttdtttdtttfT

a

dtttdtttdtttfT

a

T

T

T

Even or odd?

Page 20: Where we’re going Speed, Storage Issues Frequency Space.

Frequency Domain Plot of Fourier Coefficients

Sharp Sawtooth Fourier Coefficients

-2-10

123

0 5 10 15 20 25 30 35

Frequency

Ampli

tude

Get “powerspectrum”by squaringFouriercoefficients

"Power Spectrum"

0

1

2

3

4

0 10 20 30 40

Frequency

Relat

ive Po

wer

Page 21: Where we’re going Speed, Storage Issues Frequency Space.

Construction of Sharp Sawtooth by Adding 1st, 2nd, 3rd

Harmonic

Third Harmonic

First HarmonicSecond Harmonic

Page 22: Where we’re going Speed, Storage Issues Frequency Space.

Spectral Analysis of Arbitrary Functions

• In general, there is no requirement that f(t) be a periodic function

• We can force a function to be periodic simply by duplicating the function in time (text fig 5.10)

• We can transform any waveform to determine it’s Fourier spectrum

• Computer software has been developed to do this as a matter of routine. - One such technique is called “Fast Fourier Transform” or FFT- Excel has an FFT routine built in

Page 23: Where we’re going Speed, Storage Issues Frequency Space.

Voice Recognition

The “ee” sound

Page 24: Where we’re going Speed, Storage Issues Frequency Space.

Voice Recognition (continued)

The “eh” sound

Page 25: Where we’re going Speed, Storage Issues Frequency Space.

Voice Recognition (continued)

The “ah” sound

Page 26: Where we’re going Speed, Storage Issues Frequency Space.

Voice Recognition (continued)

The “oh” sound

Page 27: Where we’re going Speed, Storage Issues Frequency Space.

Voice Recognition (continued)

The “oo” sound