Professional Documents
Culture Documents
ANALOG SYSTEMS
(Version 2.1)
Network U1 ( p ) = Excitation
with linear, U 2 ( p ) U 2 ( p) = Answer
U1 ( p )
time-independent U 2 ( p)
H L ( p) =
elements U1 ( p )
i N ( jω ) = N (− jω ) → N ( jω ) is an even function
i ϕ (−ω ) = ϕ (ω ) → ϕ (ω ) is an odd function
i The system function is derived from quantities of the same dimension:
U2
H L ( p) =
U1
Additional notes:
H L ( jω )
H L ( jω ) = e − a (ω ) − jb (ω ) = H L ( jω ) e j with
U 2 ( jω ) U ( jω )
H L ( jω ) = ⇒ a (ω ) = 20 log 1 dB is the damping ratio
U1 ( jω ) U 2 ( jω )
and b(ω ) = −ϕ (ω ) =- H L ( jω ) is the damping angle
ANALOG SYSTEM
3.2 Basic Properties of a System
Transmission
Source Transmitter Receiver Drain
Channel
x (t ) y (t )
Input excitation Output answer
Cause effect
L R
ia ( t )
U1 (t ) C U2 (t ) Ra
Input-Output connections:
u1 ( t ) → u2 (t ) or u1 ( t ) → ia (t )
m s1 ( t ) → s2 ( t )
s2 ( t )
u1 ( t )
u2 A
B
u1 ( t ) A u2 ( t )
u1 t
−A
u2 ( t )
D B
u1 ( t ) R u2 ( t )
u2 (t )
A
u1 ( t )
If a bipolar impulse chain is used as an input for the rectifier, it can not be
figured out whether the system has a saturation or not !
u2 ( t ) u1 ( t ) u2 ( t )
u1 ( t ) C R
u2 ( t )
u1 ( t )
T t t
Different Circuits with the same Input-Output relation
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 13
Fachgebiet N T S
Nachrichtentechnische Systeme
3.2.1 Definition of a System and General Remarks
Benefits from the system theory:
2. Additivity:
A system is called additivity if:
s2 (t ) → g 2 (t )
s1 (t ) + s2 (t ) → g1 (t ) + g 2 (t )
s1 (t ) → g1 (t )
3. Homogeneity:
s (t ) → g (t ) ⇒ as (t ) → ag (t )
Example: let’s assume an LTI system with its answer to the unit-step as
followings:
⎧⎪0 t<0
⎧ 0 t < 0 answer
s (t ) = ε (t ) = ⎨ g (t ) = ⎨ −
t
⎩1 t ≥ 0 ⎪⎩1 − e τ t ≥ 0
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 16
Fachgebiet N T S
Nachrichtentechnische Systeme
3.2.2 Basic System Properties
⎧⎪0 t<0
s (t ) g (t ) g (t ) = ⎨
⎧0 t < 0 t
s (t ) = ε (t ) = ⎨ −
⎪⎩1 − e τ t≥0
⎩1 t ≥ 0
t t
Answer of a system to an Unit-step function
x (t )
x(t ) = ε (t ) − ε (t − T )
1 ?
T t
The solution goes by exploiting the linearity and time-invariance of the system:
• First step: separation of the input function into 2 steps using rect-function
• Second step: find out the 2 answers for each rect-function
• Third step: superposition of the answer
T t t
⎧0 t<0
x2 ( t ) ⎧0 t < 0 y2 (t ) ⎪
y2 (t ) = ⎨ ⎛ −
t −T
⎞
x2 (t ) = −ε (t − T ) = ⎨ −
⎪ ⎜ 1 − e τ
⎟ t≥0
T ⎩−1 t ≥ 0 ⎩ ⎝ ⎠
t t
−1
y (t ) = y1 (t ) + y2 (t )
x (t)
x(t ) = x1 (t ) + x2 (t ) y (t) ⎧
⎧0 t<0 ⎪ 0 t<0
⎪ ⎪
A = ⎨1 0≤t≤T ⎪ −
t
= ⎨1 − e τ
0≤t≤T
⎪0 t >T ⎪ −t
⎩
⎪ e τ ⋅ ⎛⎜ e τ − 1⎞⎟
T
t>T
T t t ⎪⎩ ⎝ ⎠
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 19
Fachgebiet N T S
Nachrichtentechnische Systeme
3.2.3 Realisable Systems
1. Real systems:
A system is called real when it follows:
Real input Real output
2. Causal system:
A system is causal if the output signal g(t) until an arbitrary time depends
only the input signal s(t) until this time.
No effects before the cause
s (t ) ≡ 0 for t < tν g (t ) ≡ 0 for t < tν
3. Stable system:
s (t ) g (t )
t
s (t ) → g (t ) = ∫ s(τ )dτ
s (t ) −∞
g (t )
t t
0
If for anytime t, the value of the output signal g(t) of a system depends
exclusively on the value of the input signal s(t) at the same time t, then the
system is called memory-less
ANALOG SYSTEM
s0 (t ) → g 0 (t ) = T [ s0 (t )]
t t
⎛t⎞
s0 (t ) = rect ⎜ ⎟
⎝T ⎠
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 24
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.1 Convolution Integral and Impulse
Response
The answer of the system to an arbitrary input signal can be calculated
approximately if the properties LTI are used:
s (t )
sa ( t )
iT0
−T0 0 T0 2T0
Step 3: because the system is linear, one can use the superposition:
∞
g (t ) ≈ g a (t ) = ∑ s(iT ) ⋅ g (t − iT )
i =1
0 0 0
s ( 0 )i g 0 (t )
s (T0 )i g0 ( t − T0 )
0a
0 T0 2T0 t
s ( iT0 )i g0 ( t − iT0 )
Some remarks:
t
= ∫ s(τ )h(t − τ )dτ
−∞
Impulse response
Dirac's delta function
1 h (t )
RC
δ (t )
t RC t
0
s (t ) ⎛ T0 ⎞
⎛ t 1⎞ ⎜t− 2 ⎟
s (t ) = a ⋅ rect ⎜ − ⎟ = a ⋅ rect ⎜ ⎟
⎝ T0 2 ⎠ ⎜⎜ T0 ⎟⎟
a
⎝ ⎠ ?
T0 t
0 RC
τ
h ( −τ )
1 τ
1 RC
RC
h ( −τ ) = (τ ) e
RC
− RC 0 τ
h (t − τ ) t −τ
1 − RC
h (t − τ ) = s (t − τ ) e
1 s (τ ) RC
RC a
h ( t0 − τ ) F1 F2
h ( t2 −τ )
t = t0 0 t = t1 t = t2
τ
d
Moreover: h(t ) = w(t )
dt
du (τ )
s (τ ) = u (τ ) ⇒ = s′(τ ) ⇒ du = s′(τ )dτ
dτ
dv
h(t − τ )dτ = dv ⇒ = h(t − τ ) where t − τ = x → − dτ = dx
dτ
dv
= − h( x ) ⇒ v( x) = − ∫ h( x)dx = − w( x) = − w(t − τ )
dx
Thus:
+∞ +∞
s (t ) → g (t ) = − s (τ ) w(t − τ ) + ∫ w(t − τ ) s′(τ )dτ
−∞ −∞
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 36
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.3 The Transfer Function H(ω)
Let the exponential function be the excitation of a LTI-system:
∫ h(τ ) ⋅ s0 e jω ( t −τ ) dτ
jωt
g (t ) = s0 e ∗ h(t ) =
−∞
+∞
∫
jωt
= s0 e h(τ ) ⋅ e − jωτ dτ = s0 e jωt ⋅ H (ω )
−∞
where:
+∞
Transfer function of the system:
H (ω ) = ∫ h(t ) ⋅ e
− jωt
dt
−∞ The Fourier transform of the systems
impulse response
G (ω )
G (ω ) = S (ω ) H (ω ) ⇔ H (ω ) = ∀S (ω ) ≠ 0
S (ω )
Effectless systems:
Let‘s assume a combined system as follows:
s (t ) gz (t ) g (t)
h1 (t ) h2 ( t )
G (ω ) = S (ω ) H1 (ω ) H 2 (ω )
⎧ 1 − RCt
⎪ ⋅e for t ≥ 0
s (t ) g (t ) ⇒ h(t ) = ⎨ RC
C ⎪
⎩0 for t < 0
+∞ +∞
1 −
t
∫ ∫e
− jωt
H (ω ) = h(t )e dt = RC
e − jωt dt
−∞
RC 0
⎛ 1 ⎞ ⎛ 1 ⎞
1
+∞ −⎜ + jω ⎟t 1 −1 −⎜ + jω ⎟t +∞ 1
=
RC ∫e
0
⎝ RC ⎠
dt = ⋅
RC 1 + jω
e ⎝ RC ⎠
0
=
1 + jω RC
RC
U 2 (ω )
The same result can be obtained by using network analysis: H (ω ) =
R U1 (ω )
1
1
U 1 (t ) U2 (t )
jωC U 2 (ω ) jω C 1
= = = H (ω )
U1 (ω ) R + 1 1 + jω RC
jω C
RC-Circuit as complex voltage divider
⎛ Im { H (ω )} ⎞
ϕ (ω ) = arctan ⎜⎜ ⎟⎟ = − arctan(ω RC )
⎝ {
Re H (ω )}⎠
Because h(t) is real-valued function, one gets:
H (−ω ) = H ∗ (ω )
H (ω ) = H (−ω )
ϕ (−ω ) = −ϕ (ω )
1 1 ω
−
RC RC
ϕ (ω )
π
2
1 0 1 ω
−
RC RC
π
−
2
Magnitude and phase of an RC-Circuit (Low-pass filter)
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 44
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.4 Analog LTI-Systems, Describable by
Differential Equations
δ ( t ) ∗h( t ) = h( t )
+∞
δ (t ) ∫−∞ h( t ) δ (t −τ ) dτ = h(t )
h( t )
H ( ω)
1
1⋅ H (ω) = H (ω)
These kinds of system are LTI if the coefficient aν and cµ are constants.
The Fourier transform of both sides of the equation above yields:
m n
∑
ν
aν ( jω )
=0
ν
S (ω ) = ∑ cµ ( jω ) µ G (ω )
µ =0
m
⎧ m ν ⎫
G (ω ) ∑ aν ( jω ) ν
−1 ⎧ G (ω ) ⎫
⎪∑ ν
−1 ⎪ ν = 0
a ( jω ) ⎪
⎪
or H (ω ) = = ν=0 h(t ) = F ⎨ ⎬=F ⎨ n ⎬
S (ω ) ⎩ S (ω ) ⎭ ⎪ ∑ cµ ( jω ) µ ⎪
n
∑
µ
cµ ( jω ) µ
=0 ⎩⎪ µ =0 ⎭⎪
Most transmission channels are limited to a certain frequency range due to:
• natural property of the channel (cable, radio,...)
• usage of filters to separate the information content and the noise
A system view is useful to:
• describe the basic properties of filters, transmission system or
their effect on given signals
• simplify the design of transmission and other systems
Low-pass and band-pass filters are most important system for theory and practice
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 47
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.5 Causal, Analog Low-Pass and Band-
Pass Systems
There are other ways of writing the transfer function of filters:
H (ω ) = H (ω ) e jϕ (ω ) = A(ω )e − jb (ω )
Hereby, it follows:
3. Phase delay: 1
τ Ph (ω ) = − ϕ (ω )
ω
is the time delay of a cosine signal
corresponding to a specific phase φ
The signal output is amplified by the factor A(ω 0 ) and is delayed about the
phase delay:
1
τ Ph (ω 0 ) = − ϕ (ω 0 )
ω0
h (t )
s (t ) g (t ) = s0 H ( ω 0 ) ⋅ e jω0 t
H (ω )
Linear filter: s (t ) H (ω ) = A (ω ) ⋅ e − jb (ω ) g (t )
b (ω )
S (ω )
A (ω )
−Ω 0 Ω ω0 − Ω ω0 ω0 + Ω
RC
U1 U2 R2
s (t ) g (t )
s0 h0
s0
t t0 t
thus: τ Ph = τ Gr = t0 frequency-independent
All deviations from these properties are called linear distortions:
when i A(ω ) ≠ const. : amplitude distortions
i ϕ (ω ) ≠ −ωt0 : phase distortions
−ωg 0 ωg 0 t0
ω t
1 1
t0 − t0 +
2 fg 2 fg
⎡ t0 t ⎤
= h0 ⎢ ∫ si (ω g (τ − t0 ))dτ + ∫ si (ω g (τ − t0 ))dτ ⎥ ( Si func. is symmetric)
⎢⎣ −∞ t0 ⎥⎦
A0
2
t0 t
1
2 fg
d A0ω g
A0 = te w(t ) = te h(t ) = te
dt t = t 0 t = t 0 π
⎡ ∆ω ⎛ ∆ω ⎞ jω0t ∆ω ⎛ ∆ω ⎞ − jω0t ⎤
h(t ) = A0 ⎢ si ⎜ t ⎟e + si ⎜ t ⎟e ⎥ ∗ δ (t − t0 )
⎣ 2π ⎝ 2 ⎠ 2π ⎝ 2 ⎠ ⎦
A ∆ω ⎛ ∆ω ⎞
= 0 si ⎜ (t − t0 ) ⎟ ⎡⎣ e jω0 (t −t0 ) + e − jω0 (t −t0 ) ⎤⎦
2π ⎝ 2 ⎠
A0 ∆ω ⎛ ∆ω ⎞
= si ⎜ (t − t0 ) ⎟ cos(ω0 (t − t0 ))
π ⎝ 2 ⎠
2π 2π
t0 − t0 t0 +
ω ω
⎛ ω − ω 0 ⎞ − jω t0 ⎛ ω ⎞ − jω t0
H 0 (ω ) = 2 A0 rect ⎜ ⎟e and H T (ω ) = 2 A0 rect ⎜ ⎟e
⎝ ∆ω ⎠ ⎝ ∆ω ⎠
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 62
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.5 Causal, Analog Low-Pass and Band-
Pass Systems
2A0
Aequiv.lp (ω )
b (ω ) = ω t0
A0
−ω0 ∆ω ∆ω +ω0
−
2 2 ω
∆ω
∆ω
∆ω
critical frequency ωg = of the
2
equivalent low-pass
∆ω ∆ω
Re {H ( ω )}
−ω0 ω0
Im { H (ω )}
H B − (ω ) H B + (ω )
H (ω ) = H + (ω ) + H − (ω ) with
H + (ω ) = H (ω ) ⋅ ε (ω ) and
H − (ω ) = H (ω ) ⋅ ε (−ω ) Re { H T (ω )}
H T (ω ) = Re { H T (ω )} + j Im { H T (ω )} Im {H T (ω )}
∆ω
hT (t ) = u (t ) + j ⋅ v(t )
Equivalent low-pass signal of the non-symmetrical band-pass
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 66
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.5 Causal, Analog Low-Pass and Band-
Pass Systems
Based on the formulas above, one can get the following relations:
H T (ω ) = 2 H + (ω + ω0 ) = 2 ⎡⎣ H − (−ω − ω0 ) ⎤⎦ ∗
1 1
or H + (ω ) = H T (ω − ω 0 ) = H 0 (ω )
2 2
1
H − (ω ) = H T (−ω − ω 0 )
2
with: H (ω ) = H +
(ω ) + H −
(ω )
{ }
h(t ) = Re hT (t )e jω0t = u (t ) cos ω 0t − v(t ) sin ω 0t
= u 2 (t ) + v 2 (t ) cos(ω 0t + ϕT (t )) = hT (t ) cos(ω 0t + ϕT (t ))
y (t ) → g (t ) = y (t ) ∗ h(t ) G (ω ) = Y (ω ) ⋅ H (ω )
1 1
GT (ω ) = YT (ω ) H T (ω ) gT (t ) = yT (t ) ∗ hT (t )
2 2
and thus:
⎧1 ⎫
{ }
g (t ) = Re gT (t )e jω0t = Re ⎨ [ yT (t ) ∗ hT (t ) ] e jω0t ⎬
⎩2 ⎭
H (ω ) = A0 e − jω t0 h(t ) = A0δ (t − t0 )
are:
i ideal delay elements for t0 > 0
i ideal predictors for t0 < 0 , because of:
s (t ) → g (t ) = s (t ) ∗ A0δ (t − t0 ) = A0 s (t − t0 )
Usage: often used to correct the phase of transmission channels and filters.
⎛ t − t0 ⎞ A0ω g ⎛ t − t0 ⎞
hk (t ) = h(t ).rect ⎜ ⎟= si (ω g (t − t0 ))rect ⎜ ⎟
2
⎝ 0 ⎠ t π 2
⎝ 0 ⎠t
A0ω g ⎛ t ⎞
= si (ω g t )rect ⎜ ⎟ ∗ δ (t − t0 )
π ⎝ 2t0 ⎠
1 ⎛ ω ⎞ − jω t − jω t
H K (ω ) = A0 rect ⎜ ⎟⎟ ∗ 2t0 .si (ω t0 )e 0 = H K (ω ) e 0
2π ⎜ 2ω
⎝ g ⎠
0 t0 2t0
t
b ( ω ) = ω t0
−ω g 0 ωg ω
s (t ) = s0e pt = s0 eσ t e jω t where p = σ + jω
One gets:
+∞
g (t ) = s (t ) ∗ h(t ) = s0 e pt ∗ h(t ) = s0 ∫ h(τ )e p (t −τ ) dτ
−∞
+∞
= s0 e pt ∫
−∞
h(τ )e − pτ dτ = s0 e pt I The difference is I
∫ τ − pτ
g (t ) = s0 e pt
h ( ) e dτ = s0 e pt
H L ( p)
0
GL ( p )
GL ( p ) = S L ( p ) H L ( p ) ⇔ H L ( p) =
S L ( p)
h1 ( t ) h2 ( t )
s (t ) g (t )
H L1 ( p ) H L2 ( p)
5. H (ω ) does not automatically equal the system function H L (p) on the jω _ axis
When moving from H L ( p ) to H (ω ), the following cases can occur:
Case 1:
Re p = β 2 < 0, H L ( p ) jω -axis is covered by convergence area ensuring:
H (ω ) = H L ( jω )
∫ h(t ) dt < ∞
0
such case does not occur
Case 3:
Re p = β 0 = 0
H (ω ) = H L ( jω ) + function with suitable Dirac's delta functions
0 σ
β2 β0 β1
Convergence areas of the Laplace transform
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 79
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.6 The System function HL(p)
Example: Given is the following network
L
Ri = 0
u1 (t ) u2 ( t )
U1 ( p ) U2 ( p)
C
ω0π
H (ω ) = H L ( jω ) + [δ (ω − ω0 ) − δ (ω + ω0 )]
2j
m
H (ω ) = H L ( jω ) + π ∑ anδ (ω − ω n )
n =1
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 81
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.6 The System function HL(p)
When moving from H (ω ) to H L ( p ), one can conclude from the existance H (ω ):
∑ i
a p i
am p m + ... + a0
H L ( p) = N ( p) = i =0
=
n
bn p n + ... + b0
∑i
b p
i =0
i
µ poles
∑r 0i = Order of the numerator polynom = m
i =1 am
υ Real constant: K =
∑r
i =1
∞i = Order of the denominator polynom = n bn
With poles and zeroes, the transmission properties are fully described:
p02 p01
−1 1
p∞1 σ
ω0
p∞3 − j1
∏ p − p0i
r0 i
Thus:
H L ( p) = K i =1
ν
magnitude
∏ p− p
r∞i
∞i
i =1
µ ν
ϕ ( p ) = arctan K + ∑ r0iϕ 0i ( p ) − ∑ r∞iϕ ∞i ( p ) phase
i =1 i =1
⎧0 K > 0
where: arctan K = ⎨
⎩±π K < 0
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 86
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.8 Pole Zero plots
j Im { p}
p − p0
i
ϕ0 i
p
p0
i
Re { p}
jω
p - plane
point p = jω
jω − p ∞i
wanted H L (ω ) = H ( ω )
ϕ∞i
p∞ i
jω − p0i
σ
ϕ 0i
p 0i
jω because
low-pass (3rd order)
p - plane
A(ω ) =0
ω → ±∞
p∞ p0 1
1
jω − p02
σ
jω − p∞ 2
p∞ 2
p02
jω
Minimum-phase-system All-pass
jω jω
1
σ σ
H LN ( p) = K i =1
ν
= Kω N( m − n ) i =1
ν
ω n
N ∏ ( P − P∞i )
i =1
r∞i
∏ ( P
i =1
− P∞i ) r∞i
µ ν
where ∑r
i =1
0i = m and ∑r
i =1
∞i =n
ln H LN ( jω ) = − aN p (Ω) − jb(Ω)
R
U1 C U2
One gets: 1
U pC 1 1
H L ( p) = 2 = = with p∞1 = −
U1 R + 1 pRC + 1 RC
pC
1
A suitable normalization frequency is: ω N =
RC
1 1
⇒ H L ( P) = = and P∞1 = −1
p
+1 P +1
ωN
ω
jω = j
ωN
p - plane
jω − p∞ 1
p∞1
−1 1
σ
σ=
ωN
0 .7 0 7
−2 −1 1 2
ω
ω′ =
ωn
⎛ ω ⎞
b (ω ′ ) = arctan (ω ′ ) = arctan ⎜ ⎟
π ⎝ωN ⎠
2
−1 0 1 ω
ω′ =
ωn
π
−
2
−2 −1 0 1 2
ω
ω′ =
ωN
Envelope delay of RC-Circuit
Prof. Dr.-Ing. I. Willms Signals and Systems 1 WS 03/04
S. 102
Fachgebiet N T S
Nachrichtentechnische Systeme
3.3.9 Minimum Phase systems
• An LTI system with zeros in the right half-plane is said to contain all-pass
• Such a system can be divided into:
The all-pass which only effect on the phase
The all-pass-less or minimum phase system which exhibits the same magnitude
frequency response but with a smaller phase than the original system.
jω
Minimum-phase-system All-pass
jω jω
1
σ σ
+∞ +∞ +∞
g (t ) = ∫ s(τ )h(t − τ )dτ ≤ ∫
−∞ −∞
s (τ ) h(t − τ ) dτ < M 1 ∫ h(t − τ ) dτ
−∞
∫
−∞
h(t − τ ) dτ = ∫
−∞
h(t ) dτ < ∞ for all t
∑a p i
i
∏(p − p 0i ) r0 i
H L ( p) = i =0
n
=K i =1
ν
∑b p
i =0
i
i
∏ ( p
i =1
− p∞i ) r∞i
or may have, in the simplest case, just single roots in the denominator polynomial:
n
Ai
H L ( p) = K ∑
i =1 p − p∞ ,i
Re { p∞i } = σ ∞i < 0
Theorem: The poles of the system function of a stable causal network are all
located in the left half-plane.