Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Linear Systems

Input Signal Linear Output Signal


x(t) System y(t)
h(t)  H( f )

Useful Signal Characterizations

X(f ) Voltage Spectrum (via FT) Y( f )


Rx ( ) AutoCorrelation Function R y ( )
Px ( f ) Power Spectral Density Py ( f )

ECE 4710: Lecture #5 1


Linear Systems
 Linear Time Invariant (LTI) Conditions:
 Linear  Superposition holds
 Time Invariant  Shape of system response, H( f ), is
same no matter when input is applied to system
» Does not apply for most mobile (wireless)
communication channels
 Impulse Response = h(t)
 h(t) = 0 for t < 0  causal
 y(t) = h(t) when x(t) = d(t)  delta impulse function at
input can measure system response

ECE 4710: Lecture #5 2


System Output
 Output is result of convolution integral between input
and impulse response function

y(t )   x( )h(t   )d  x(t )  h(t )

 Convolution integral is difficult to evaluate

 Preferred approach is to find system transfer


function H( f )
 Convolution in time is multiplication in frequency
y(t )  x(t )  h(t )  Y ( f )  X ( f ) H ( f )
Y( f )
so that H( f ) 
X( f )

ECE 4710: Lecture #5 3


System Transfer Function
 H( f ) is the FT of h(t) and is called the frequency
response or the transfer function of the system
h(t )  H ( f )
 In general H( f ) is a complex function with magnitude
and phase response:
H ( f )  | H ( f ) | exp j H ( f )
 Magnitude response is even function in frequency
 Positive and negative frequencies have same amplitude
 Phase response is an odd function in frequency

ECE 4710: Lecture #5 4


System Transfer Function
 H( f ) can be measured by using sinusoidal test input
signal and sweeping the frequency over the desired
range
 Spectrum analyzer uses this approach
 How is power content of input signal affected by the
system?
Given Px ( f ) and H ( f ) what is Py ( f ) ??
 Py ( f )  | H ( f ) |2 Px ( f )

ECE 4710: Lecture #5 5


Power Transfer
 Power Transfer Function
Py ( f )
Gh ( f )   | H ( f ) |2
Px ( f )
 Example: RC Low Pass Filter (LPF)  Find Gh( f )

+ +
i(t) R
x(t) C y(t)
 

ECE 4710: Lecture #5 6


RC LPF
 KVL around loop: x(t )  R i(t )  y(t )
 Capacitor current related to voltage drop:
dv(t ) dy(t ) dy(t )
i (t )  iC (t )  C C  x(t )  RC  y (t )
dt dt dt
d n w(t )
 Table 2-1, pg. 50 : n
 ( j 2 f ) n
W( f )
dt
 Take FT of both sides: X ( f )  RC( j 2 f ) Y ( f )  Y ( f )
 Y ( f )[1  ( j 2 f ) RC]
 Solving for transfer function: H ( f )  Y ( f )  1
X ( f ) 1  j (2 RC ) f

ECE 4710: Lecture #5 7


RC LPF

et / T t  0 T
 Table 2-2, pg. 62:  
0 t 0 1  j 2 f T
 1
 e t / 
1 t 0
 So H( f )   h(t )  
1  j (2 RC) f  t 0
 0
where  = RC is the time constant of the LPF
1
h(t) = Impulse Response
RC
Function of ILPF

e-1
RC
t
 = RC

ECE 4710: Lecture #5 8


RC LPF
 Power Transfer Function:
Gh ( f )  | H ( f ) |2  H ( f )  H * ( f ) and
from Table 2 - 1 on page 50 H * ( f )  H ( f ) so
1 1 1
| H( f ) | 
2
   Gh ( f )
1  j (2 RC) f 1  j (2 RC) f 1  (2 RCf ) 2

 Define fo = cutoff frequency = 1 / 2RC so


1
Gh ( f ) 
1  ( f / fo )2
 At f = fo  Gh( fo ) = 0.5
 Power @ fo attenuated by half  half power or 3 dB BW

ECE 4710: Lecture #5 9


Distortionless Transmission
 Distortionless channel is very desirable in a
communication system
 Output is simply delayed replica of input : y(t) = A x (t - Td)
where A : channel gain (or loss for A < 1)
Td : time delay in channel
 In frequency domain a distortion free response is
Y ( f )  A X ( f ) e  j 2 f Td
 Thus, a distortion free channel has Note that there is
no frequency
Y( f )
H( f )   A e j 2 f Td  A e j ( f ) dependence for
amplitude but
X( f ) there is for phase

ECE 4710: Lecture #5 10


Distortionless Transmission
 LTI system will have no distortion if
1) Amplitude response is flat  | H ( f ) |  A
2) Phase response is linear function of
frequency   ( f )  2 f Td
 Distortion classified as either 1) amplitude or 2)
phase distortion
 Looking at phase distortion from time delay
standpoint: Td  constant value  Td ( f )
 Time delay of channel/system must be independent of
frequency, otherwise phase distortion will occur

ECE 4710: Lecture #5 11


RC Filter Distortion
1
 Filter Transfer Function: H ( f ) 
1  j (2 RC ) f
 Amplitude response:
1 1
Gh ( f )  | H ( f ) |2  so | H ( f ) | 
1  ( f / fo )2 1  ( f / fo )2
 ImH ( f )
 Phase Response:  ( f )  tan1     tan 1
( f / fo )
 ReH ( f )
1 1
 Time Delay: Td ( f )  ( f )  tan1 ( f / f o )
2 f 2 f
 Not a constant value

ECE 4710: Lecture #5 12


RC Amplitude Distortion

For f < 0.5 fo the amplitude distortion is < 0.5 dB (~12%)

ECE 4710: Lecture #5 13


RC Phase Distortion

For f < 0.5 fo the phase distortion is < 2.1 (~8%)

ECE 4710: Lecture #5 14


RC Time Delay

High frequency signal components have less delay


For fo = 1 kHz  delay is ~0.2 msec

ECE 4710: Lecture #5 15


Distortion
 Most communication systems have both amplitude
and phase distortion of H( f )

 Distortion can be acceptable depending on


 Distortion type  amplitude, phase, or both
 Distortion magnitude  e.g. how bad?
 Type of information signal  audio, video, or data

ECE 4710: Lecture #5 16


Audio Distortion
 Human ear is sensitive to amplitude distortion in f
 Audio spectrum is 300 Hz to 15 kHz
 15 phase distortion causes time delay of ~ 3 msec
 Duration of spoken syllable is 10-100 msec
 3 msec delay is imperceptible (< 0.05 % of 10 msec syllable)
 3 dB amplitude error is very noticeable to ear
 High-fidelity audio amplifiers
 Focus on maintaining flat spectral response
 Phase distortion is not concern

ECE 4710: Lecture #5 17


Video Distortion
 For analog video transmission the phase distortion
is the primary concern
 Amplitude variations will cause variations in image
intensity
 Phase variations will cause time delays which will
cause objects in the image to blur at edges
 Human eye is more sensitive to phase variations

 Analog video filters require excellent phase


linearity

ECE 4710: Lecture #5 18


Data Distortion
 For digital data both amplitude and phase distortion can
have serious affects
 Rectangular data pulse train:
Bandlimited

0 1 0 1 0 System or 0 1 0 1 0

Channel
Ts

 Pulse smearing into adjacent symbol time slots


 Inter-Symbol Interference (ISI)
 Increase probability that bit errors will occur (BER )
 Special filters designed for digital data to minimize ISI

ECE 4710: Lecture #5 19

You might also like