Institute for Automotive Engineering (ika)

RWTH Aachen University

Mechatronics in Vehicle Engineering
Lecture - Analog Signal Processing


Aachen, 07. May 2025 Univ. Prof. Dr.-Ing. Lutz Eckstein 



© ika 2025 • All rights reserved --- Version 25.04.01 1

 Signal and Data Processing
Mechatronic Systems

Where does analog signal and processing fit into a mechatronic system?





© ika 2025 • All rights reserved --- Version 25.04.01 2

 Signal Processing
Tasks of Analog Signal Processing

Analogue signal processing concerns the signal pre-

processing of temporally continuous signals.
Signals are usually present in the form of time-
continuous voltage or current waveforms
Signal Pre-Processing

Conditioning of sensor signals

Compensation of zero-point fluctuations
Filtering of interference signals
Amplification and linearisation of measurement
signals
Measuring range adjustment and switching
Normalisation of the output signal





© ika 2025 • All rights reserved --- Version 25.04.01 3

 Analog Signal Processing
The Transfer System

A general transfer system with the input variable x(t) and the output variable y(t) is considered

Problem definition

How can the system be described mathematically?

Is there a general relationship between input x(t) and output y(t)?

How can the relationship between input x(t) and output y(t) be described?





© ika 2025 • All rights reserved --- Version 25.04.01 4

 Analog Signal Processing
Important Functions

In the following, some key functions for the analysis of transmission systems will be considered

Heaviside-Function (Step-Function) Dirac-Function (Impulse-Function)

is also called a step function is also referred to as a impulse function
takes the value 0 for all values less than 0 and the value 1 takes the value 0 for all non-zero values
for all values greater than or equal to 0 area under the function is 1

σ(x) = 0 x<0 δ(x) = 0​

x
=0 ​

=1 x≥0
​ ​

∞
∫ ​ δ (x) dx = 1
−∞ 



© ika 2025 • All rights reserved --- Version 25.04.01 5

 Analog Signal Processing
Linear Time Invariant System

In the following, a linear time-invariant system (LTI) will be considered as a special case of the general transfer system

Linear System

All occurring functions within the system are linear

f (a ⋅ x) = a ⋅ y f (x1 + x2 ) = y1 + y2
​ ​ ​ ​

Time-invariant system

The behaviour of the system is completely independent of time

An event at time t1 is results in the same behavior as an event at t2 if the system is in the same state.
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 6

 Analog Signal Processing
Impulse Response of a LTI-System

In general, the input function can be represented with the help of the Dirac function
∞
x(t) = ∫ ​ x(τ ) ⋅ δ(t − τ ) dτ
−∞

Let the system behavior be described by the operator H

∞
y(t) = H[x(t)] = H [∫ ​ x(τ ) ⋅ δ(t − τ ) dτ ]
−∞

For a linear system + time-invariant system, the following applies

∞ ∞
y(t) = ∫ ​
x(τ ) ⋅ H[δ(t − τ )] dτ y(t) = ∫ ​
x(τ ) ⋅ h(t − τ ) dτ
−∞ −∞

The term h(t) = H[δ(t)] is called the impulse response.

Any output function of an LTI system can be represented by the impulse response h(t)

y(t) = x(t) ∗ h(t) 




© ika 2025 • All rights reserved --- Version 25.04.01 7

 Analoge Signalverarbeitung
Laplace Transformation

The Laplace transformation is an integral transformation

Transfers functions from the time domain into the image domain
Enables the transformation of time-dependent differential equations into easily solvable algebraic functions
Convolution in the time domain can be represented as a simple multiplication in the image domain

∞
F (s) = L(f (t)) = ∫ f (t) ⋅ e −st
dt s = σ + jω
0

e ⋅ F (s) ds = {
γ+j∞
1 f (t) t ≥ 0
−1
f (t) = L (F (s)) = ⋅∫ st
2πj γ−j∞ 0 t<0
​ ​ ​ ​

Example - Displacement y(t) for damped spring-mass pendulum with excitation u(t)

1 2D
2
⋅ ÿ (t) + ⋅ ẏ (t) + y(t) = u(t)
ω0 ω0
​ ​ ​ ​

​ ​

1 2 2D
2 ⋅ s ⋅ Y (s) + ⋅ s ⋅ Y (s) + Y (s) = U (s) 

ω0 ω0
​ ​

​ ​


© ika 2025 • All rights reserved --- Version 25.04.01 8

 Laplace-Transformation
Operations and Correspondences
Description f (t) = L−1 (F (s)) F (s) = L(f (t)) f (t) = L−1 (F (s)) F (s) = L(f (t))

Linearity a ⋅ f1 (t) ± b ⋅ f2 (t)

​ a ⋅ F1 (s) ± b ⋅ F2 (s)
​ ​ δ(t) 1

1 s e−sT
Ähnlichkeitssatz f (at) ⋅F ( ) ​ ​
δ(t − T )
a a

−sT σ(t) 1/s

Shifting the Original Function f (t − T ) e ⋅ F (s)

1 −sT
Shifting the Image Function e −tT
⋅ f (t) F (s + T ) σ(t − T ) ​⋅e
s

1st derivation of the original function

df
= f˙(t) ​ ​
s ⋅ F (s) − f (0) t 1/s2
dt

2
tn−1 1
df
2nd derivation of the original function = ¨(t) s2 ⋅ F (s) − s ⋅ f (0) − f˙(0) (n − 1)!
​

sn
​

f
dt2
​

​ ​

n −at 1
d f e
nth Derivation of the original function = f (n)
(t) sn ⋅ F (s) − sn−1 ⋅ f (0) − sn−2 ⋅ f˙(0) − ⋯ − s ⋅ f (n−2) (0) − f (n−1) (0) s+a
​

​
​

dt n

a
d F (s) n sin(at)
nth Derivation of the image function (−1) ⋅ t ⋅ f (t) n n s2 + a2
​

dsn
​

a
∂f (t,a) ∂F (s,a) sinh(at)
s2 − a2
​

Partial derivation according to 2nd variable

∂a ∂a
​ ​

s
cos(at)
t
1 s2 + a2
​

Integration of the original function ∫ f (τ ) dτ ​

⋅ F (s) ​

0 s
s
cosh(at)
s2 − a2
​

∞
f (t)
Integration of the image function ​
∫ ​ F (ω) dω
t s
−at
a
1−e
s ⋅ (s + a)
​

a2 a2
∫ ∫
​ ​

Integration according to 2nd variable ​ f (t,a) da ​ F (s,a) da

a1 a1
​ ​

−at 1
t⋅e
(s + a)2
​

t
Convolution ∫ f1 (τ ) ⋅ f2 (t − τ ) dτ
​ ​ ​
F1 (s) ⋅ F2 (s) ​ ​

0 e−at − e−bt 1
=b
a
(s + a) ⋅ (s + b)
​

b−a
​


γ+j∞
1
⋅∫

Multiplication of Original Functions f1 (t) ⋅ f2 (t) f1 (w) ⋅ F2 (s − w)dw
2πj γ+j∞
​ ​ ​ ​ ​ ​

1 1 
⋅ e−Dω0 t ⋅ sin(ω0
​

1− D2 t)
1 − D2 s2 + 2Dω0 s + ω0
​ ​ ​
​

ω0
9
​ ​
​ ​

© ika 2025 • All rights reserved --- Version 25.04.01

 Analog Signal Processing
Fourier Transformation

The Fourier transformation is an integral transformation

Converts functions from the time domain into the frequency domain
Enables the transformation of time-dependent differential equations into easily solvable algebraic functions

∞
1
F (jω) = F(f (t)) = ∫ f (t) ⋅ e −jωt
dt
2π −∞
​ ​

Unlike the Laplace transformation, the transformed system does not describe a transient system behaviour
The system is viewed in steady state





© ika 2025 • All rights reserved --- Version 25.04.01 10

 Analog Signal Processing
Time Domain and Laplace Domain





© ika 2025 • All rights reserved --- Version 25.04.01 11

 Analog Signal Processing
Transfer Function

The transfer function specifies the relationship between the output value Ua and the input value Ue of a component.​

The transfer behavior of a component can be described by the complex transfer function

The general transfer function with transient behavior is defined as F (s) with s = σ + jω
Ua
F (s) =
​

Ue
​

The frequency dependent complex transfer function is defined as

Ua
F (jω) =
​

Ue
​

If the Fourier transform is used instead of the Laplace transform to determine the transfer function, the description of the transient 

behavior (e.g. transient response) is lost. 

© ika 2025 • All rights reserved --- Version 25.04.01 12

 Transfer Function
Poles and Zeros

Poles and zeros of the transfer function form an important basis for the analysis
and synthesis of transfer systems.
The representation is done in the pole-zero diagram in the complex number
plane
Example Calculation
2
s
F (s) = 2
s + 2s + 1
​

Zeros
2
s =0 → sn1,2 = 0 (Double zero)
​

Poles
2
s + 2s + 1 = 0 → sp1 = −1 ± j
​
(Conjugate complex pole pair)





© ika 2025 • All rights reserved --- Version 25.04.01 13

 Analog Signal Processing
Amplifier Circuits

Definition and Task

Amplifier circuits are electronic assemblies for processing incoming analogue signals.

The goal, in most cases, is for the output of the assembly to be greater than the input to the assembly. This process is called
amplification.
The input and output quantities of the module are usually voltage signals that are continuous in time.

In most cases, the module should have a linear amplification behaviour. A multiplication of the input variable by the factor n should
therefore also result in a multiplication of the output variable by the factor n.

Components of Amplifier Circuits

In order to achieve a signal amplification, active components are used in the amplifier circuits.
Commonly used components are
Bipolar transistors
Field-effect transistors
Operational amplifiers




© ika 2025 • All rights reserved --- Version 25.04.01 14

 Amplifier Circuits
Operational Amplifier
Operational amplifiers are electronic components for amplifying The ideal Operational Amplifier
electrical voltage
Infinitely high amplification
Very high amplification factor A
A→∞
Inputs:
Inverting input (-) Infinitely high input resistance. No current flows into the inputs
Non-inverting input (+)
Ri → ∞
High input resistance
​

Output: The output current has no influence on the output voltage, i.e.
you can drive infinitely high loads
Low output resistance
Ra = 0
Ua = A ⋅ (Up − Um ) = A ⋅ Ud
​

​ ​ ​ ​

In this course, all operational amplifiers are considered to be

ideal operational amplifiers





© ika 2025 • All rights reserved --- Version 25.04.01 15

 Operational Amplifier
Schematics

Differential amplifier with high input resistance

and Constant current control at the emitter.
Voltage amplification with pnp transistor in
emitter circuit
Impedance converter in collector circuit design,
leads to low output resistance
Supply voltage (e.g. +/- 15 volts)





© ika 2025 • All rights reserved --- Version 25.04.01 16

 Amplifier Circuits
Comparator

Input of the operational amplifier has a very high impedance

9 15
10 ≤ Re ≤ 10 Ω ​

1000 ≤ Ra ≤ 1000Ω
For an ideal operational amplifier the input current is I =0
Very high gain factor

For an ideal operational amplifier, the gain is a =∞

Determine the the mesh equation

−Ue + Ud = 0
​ ​

Ua = a ⋅ Ud
​ ​

This results in

Ua
=a
​

Ue
​





© ika 2025 • All rights reserved --- Version 25.04.01 17

 Amplifier Circuits
Feedback and Differential Voltage

For the output value y of a general operational amplifier with amplification factor A the following applies

y = A ⋅ (Up − Um )
​ ​

If we now define the positive input Up as input value x and a general feedback transfer function f we get
​

y = A ⋅ (x − f y)
y
Convert to transfer function F = x
​

y A
F = =
x 1 + Af
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 18

 Amplifier Circuits
Asymptotic Transfer Characteristic

The general transfer function of an operational amplifier is considered

y A
F = =
x 1 + Af
​ ​

If we now assume an infinitely high gain A → ∞ we get

1 1
limA→∞ ( ) = limA→∞ ( ) = limA→∞ ( 1 )=
y A
x 1 + Af + f
​ ​ ​ ​ ​ ​ ​

A f​

This corresponds to the transfer function for an operational amplifier with negative feedback loop.
For very high amplification factors, the transfer function is thus independent of the amplification factor itself.





© ika 2025 • All rights reserved --- Version 25.04.01 19

 Feedback and Differential Voltage
Asymptotic Differential Voltage
To obtain a finite output signal in case of infinite gain A → ∞ with negative feedback on a finite input signal, the factor (Up − Um )
​ ​

must converge to zero.

limA→∞ (A ⋅ (Up − Um )) = UOut

​ ​ ​ ​

This relationship can be easily illustrated by an iterative observation

Time step n
Differential voltage is not equal to zero
Output voltage is therefore also non-zero
Feedback on negative input leads to increase of Um ​

Difference becomes smaller

time step n+1
Differential voltage is not zero
Output voltage decreases due to decreasing differential voltage but is not zero
Feedback on negative input leads to further increase of Um ​

Differential voltage decreases further

As long as the differential voltage is not zero, the voltage level Um of the negative input is increased. Asymptotically, the input voltage
​

difference tends to zero. For a stable system in equilibrium with a negative feedback loop, Ud ​ = 0 can therefore be assumed. 



© ika 2025 • All rights reserved --- Version 25.04.01 20

 Amplifier Circuits
Impedance Converter

Determine the mesh and node equations

I1 = 0
​ I2 = 0
​

−Ue + Ud + Ua = 0
​ ​ ​

Ud = 0​

This results in

Ua
=1
​

Ue
​





© ika 2025 • All rights reserved --- Version 25.04.01 21

 Amplifier Circuits
Inverting Amplifier

Determine the mesh and node equations

I1 − I2 = 0
​ ​

−Ue + I1 ⋅ R1 + Ud = 0
​ ​ ​ ​

−Ud + I2 ⋅ R2 + Ua = 0
​ ​ ​ ​

Ud = 0 ​

This results in

Ua R2
=−
​ ​

Ue R1
​ ​

​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 22

 Amplifier Circuits
Non Inverting Amplifier

Determine the mesh and node equations

−Ue + Ud + R2 ⋅ I = 0
​ ​ ​

−(R1 + R2 ) ⋅ I + Ua = 0
​ ​ ​

Ud = 0 ​

This results in

Ua R1 + R2 R1
= =1+
​ ​ ​

Ue R2 R2
​ ​ ​

​ ​ ​

The achievable gain is ≥ 1





© ika 2025 • All rights reserved --- Version 25.04.01 23

 Amplifier Circuits
Non Inverting Schmitt-Trigger
Setting up the mesh equations and node equations

−Ue + I ⋅ R1 + Ud = 0
​ ​ ​ → I = (Ue − Ud )/R1
​ ​ ​

Hysteresis caused by positive feedback

−Ud + I ⋅ R2 + Ua = 0
​ ​ ​ → I = (Ud − Ua )/R2
​ ​ ​

Ud 
=0
The operational amplifier switches when Ud =0
​

The system knows only two stable states depending on the positive
Ue − Ud Ud − Ua Ue −Ua R1
supply voltage UCC and negative supply voltage −UCC of the = → = → Ue = −Ua ⋅
​ ​ ​ ​ ​ ​ ​

R1 R2 R1 R2 R2
​ ​ ​ ​ ​ ​
​ ​

operational amplifier
​ ​ ​ ​ ​

To determine the threshold values for the two stable states, a case
distinction is necessary

Case 1 : Ud ​ < 0 → Ua = −UCC ​ ​

Case 2 : Ud ​ > 0 → Ua = +UCC ​ ​

R1
Uth1 = −UCC ⋅
​

​ ​ ​

R2 ​

R1
Uth2 = +UCC ⋅ 
​

R2
​ ​ ​

​


© ika 2025 • All rights reserved --- Version 25.04.01 24 . 1

 Non Inverting Schmitt-Trigger
Hysteresis





© ika 2025 • All rights reserved --- Version 25.04.01 24 . 2

 Non Inverting Schmitt-Trigger
Simulation
File Edit Draw Scopes Options Circuits

Reset

RUN / Stop

Simulation Speed

Current Speed

Power Brightness

Current Circuit:





© ika 2025 • All rights reserved --- Version 25.04.01

Simulator by Paul Falstad/Iain Sharp https://github.com/sharpie7/circuitjs1 24 . 3
 Amplifier Circuits
Inverting Schmitt-Trigger
Setting up the mesh equations and node equations

−Ue − Ud + I ⋅ R2 = 0
​ ​ ​
→ I = (Ue + Ud )/R2 ​ ​ ​

Hysteresis caused by positive feedback −I ⋅ (R1 + R2 ) + Ua = 0

​ ​ ​ → I = Ua /(R1 + R2 ) ​ ​ ​

Ud 
​ =0 The operational amplifier switches when Ud ​ =0
The system knows only two stable states depending on the positive (Ue + Ud )/R2 = Ua /(R1 + R2 ) → Ue /R2 = Ua /(R1 + R2 )
​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​

supply voltage UCC and negative supply voltage −UCC of the

​ ​

operational amplifier Ue = Ua ⋅ R2 /(R1 + R2 )

​ ​ ​ ​ ​

To determine the threshold values for the two stable states, a case
distinction is necessary

Case 1 : Ud < 0 → Ua = −UCC

​ ​ ​

Case 2 : Ud > 0 → Ua = +UCC

​ ​ ​

R2
Uth1 = −UCC ⋅
​

R1 + R2
​ ​ ​

​ ​

R2 

Uth2 = +UCC ⋅
​

R1 + R2
​ ​ ​


​ ​

© ika 2025 • All rights reserved --- Version 25.04.01 25 . 1

 Inverting Schmitt-Trigger
Hysteresis





© ika 2025 • All rights reserved --- Version 25.04.01 25 . 2

 Inverting Schmitt-Trigger
Simulation
File Edit Draw Scopes Options Circuits

Reset

RUN / Stop

Simulation Speed

Current Speed

Power Brightness

Current Circuit:





© ika 2025 • All rights reserved --- Version 25.04.01

Simulator by Paul Falstad/Iain Sharp https://github.com/sharpie7/circuitjs1 25 . 3
 Amplifier Circuits
Differentiator

Determine the mesh and node equations

I1 − I2 = 0
​ ​

−Ue + Uc + Ud = 0
​ ​ ​

−Ud + I2 ⋅ R + Ua = 0
​ ​ ​

dUc dUe
I1 = C ⋅ =C⋅
​ ​

dt dt
​ ​ ​

This results in

dUe
Ua = −RC ⋅
​

dt
​ ​

The phase shift makes the circuit susceptible to oscillation

The use of differentiators is therefore often avoided





© ika 2025 • All rights reserved --- Version 25.04.01 26

 Amplifier Circuits
Integrator

Determine the mesh and node equations

I1 − I2 = 0
​ ​

−Ue + I1 ⋅ R + Ud = 0
​ ​ ​

−Ud + Uc + Ua = 0
​ ​ ​

dUc dUa
I2 = C ⋅ = −C ⋅
​ ​

dt dt
​ ​ ​

From this follows

dUa
Ue = −RC ⋅
​

dt
​ ​

1
Ua = − ⋅ ∫ Ue dt + Ua (t = 0)
RC
​ ​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 27

 Amplifier Circuits
Adder Amplifier

Determine the mesh and node equations

I1 + I2 − I3 = 0
​ ​ ​

−Ue1 + I1 ⋅ R1 + Ud = 0
​ ​ ​ ​

−Ue2 + I2 ⋅ R2 + Ud = 0
​ ​ ​ ​

−Ud + I3 ⋅ R3 + Ua = 0
​ ​ ​ ​

From this follows

Ua = − ( ) ⋅ R3
Ue1 Ue2
+
​ ​

R1 R2
​ ​ ​ ​

​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 28

 Amplifier Circuits
Subtractor/Difference Amplifier

Determine the mesh and node equations

I1 − I3 = 0
​ ​

−Ue1 + I1 ⋅ R1 + Ud + I2 ⋅ R4 = 0
​ ​ ​ ​ ​ ​

−Ue2 + I2 ⋅ (R3 + R4 ) = 0
​ ​ ​ ​

−I2 ⋅ R4 − Ud + I3 ⋅ R2 + Ua = 0
​ ​ ​ ​ ​ ​

From this follows

R2
Ua = ⋅ (Ue2 − Ue1 ) mit R1 = R3 und R2 = R4
​

R1
​ ​ ​ ​ ​ ​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 29

 Amplifier Circuits
Instrumentation Amplifier (In-Amp)

Operational amplifier circuit consisting of several

operational amplifiers

9 12
Very high input impedance (10 − 10 Ω)
Robust against noise/interference
High common mode rejection ratio
Low common mode gain
High amplification achievable
Gain factor is predefined by integrated resistors
Mostly used as integrated circuit

2R1
Ua = ( 1 + )⋅
R3
⋅ (Ue2 − Ue1 )
​ ​

RG R2
​ ​ ​ ​

​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 30

 Amplifier Circuits
Common-Mode Amplification





© ika 2025 • All rights reserved --- Version 25.04.01 31

 Analog Signal Processing
Filter

Task
Elimination of unwanted frequency components in a signal
Frequency-dependent weighting of a signal
Application
Low-pass filtering before analog/digital conversion (aliasing)
Elimination of interfering signals and noise
Filters are circuits that can be used to "isolate" frequency ranges
Implementation options

Passive filters (only passive components: L, R, C )

Active filters (using operational amplifiers, "analog computers")

Circuit networks model fractional rational functions

desired filter behavior must be approximated by fractional rational function
transfer function must satisfy damping scheme




© ika 2025 • All rights reserved --- Version 25.04.01 32

 Transfer Function
Amplitude Response and Phase Response

Amplitude Response Phase Response

The magnitude ∣F (jω)∣ of the transfer function F (jω) is The phase response φjω of the transfer function F (jω) is often
often also called amplitude gain or amplitude response also called phase shift or phase difference.
For a general complex transfer function of 1st order the For a general complex transfer function of 1st order the following
following applies applies

a + jb r1 ⋅ e r1 j(φ1 −φ2 ) jφ1 ​

a + jb r1 ⋅ e jφ1
r1 j(φ1 −φ2 ) ​

F (jω) = = = ⋅ e = r ⋅ e jφ
F (ω) = = = ⋅ e = r ⋅ e jφ
​ ​ ​

​ ​ ​ ​

c + jd r2 ⋅ e 2
jφ r2 c + jd r2 ⋅ e 2
jφ r2
​ ​ ​ ​ ​ ​

​ ​

​ ​ ​

φ1 = arctan ( ) φ2 = arctan ( )
r1 = ​ a2 + b2 ​ r2 =
​ c 2 + d 2 ​
b d
​ ​ ​ ​

a c
a + jb ∣a + jb∣ a2 + b2
∣F (jω)∣ = = =
​

c + jd ∣c + jd∣ 2 2 φ(ω) = φ1 − φ2
+
​ ​ ​ ​ ​

c d
​ ​

= arctan ( ) − arctan ( )
​
b d ​

a c
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 33

 Filter
Ideal Transfer Function





© ika 2025 • All rights reserved --- Version 25.04.01 34

 Filter
Bode Diagram
Usually simple logarithmic representation of the magnitude ∣F (jω)∣ of the transfer function F (jω) and the phase response φ(ω) as a
function of the angular frequency ω or the frequency f .
The amplitude response can be displayed as a direct gain factor or in the unit decibel





© ika 2025 • All rights reserved --- Version 25.04.01 35

 Filter
Cutoff-Frequency

The corner angular frequency ωg or the corner ​

frequency fg is the frequency at which the signal

​

amplitude falls below a certain value.

Often also referred to as cutoff frequency

In filter circuits, the cutoff frequency is usually the
frequency at which the gain has dropped to a
1
factor of the maximum gain ∣F (jωmax )∣.
2
​ ​

The following applies:

ωg = 2π ⋅ fg
​ ​

and

1
∣F (jωg )∣ = ⋅ ∣F (jωmax )∣
2
​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 36

 Bode Diagram
Simplified/Real Amplitude Response

Approximation of the amplitude response by applying the tangent to the filter flank and the maximum value of the amplitude
response.
Intersection of the two straight lines at the position of the corner angular frequency





© ika 2025 • All rights reserved --- Version 25.04.01 37

 Bode Diagram
Simplified/Real Phase Response
Variant 1

Approximation of the phase response by determining the maximum and minimum values one decade above ω = 10ωg and below
​

ω = 0.1ωg of the angular frequency ωg

​ ​

Variant 2

Approximation of the phase response by approximating the tangent at the location of the corner angular frequency ωg ​





© ika 2025 • All rights reserved --- Version 25.04.01 38

 Analog Signal Processing
Additional Information - Decibel

The decibel (dB ) is an auxiliary unit of measurement

Decadic logarithm of the ratio of two amplitudes

= 20 ⋅ log10 ( )
Ua
∣A(ω)∣dB
​

Ue
​ ​ ​ ​ ​

Ua
Ue ​
​

A(ω)dB ​

10 20dB
1 0dB
1/ 2 −3dB ​

1/2 −6dB
1/10 −20dB





© ika 2025 • All rights reserved --- Version 25.04.01 39

 Bode Diagram
Amplitude Response Approximation at the Corner Angular Frequency

Using the definition of the corner angular frequency ωg ​

1
∣F (jωg )∣ = ⋅ ∣F (jωmax )∣
2
​ ​ ​

and the equation for converting the amplitude response into decibels

∣F (jωg )∣dB = 20 ⋅ log(∣F (jω∣)

​ ​

results

1 1
∣F (jωg )∣dB = 20 ⋅ log ( ⋅ ∣F (jωmax )∣) = 20 ⋅ log ( ) + 20 ⋅ log(∣F (jωmax ∣)
2 2
​ ​ ​ ​ ​ ​

​ ​

1
= 20 ⋅ log ( ) + ∣F (jωmax )∣dB = −3dB + ∣F (jωmax )∣dB
​ ​

2
​ ​ ​ ​ ​

At the point of the corner angular frequency ωg , the magnitude of the amplitude of the transfer function ∣F (jωg )∣ is approximately
​ ​

3dB less than the maximum magnitude value of the transfer function ∣F (jωmax )∣ ​





© ika 2025 • All rights reserved --- Version 25.04.01 40

 Bode Diagram
Filter Slope

Considering the magnitude of a general transfer function ∣F (jω)∣ of 1st order at the filter slope, the following applies for the
proportionality of the magnitude depending on the angular frequency

1
∣F (jω)∣ ∝ ω or ∣F (jω)∣ ∝
ω
​

The following applies for the conversion of the absolute value into decibels

∣F (jωg )∣dB = 20 ⋅ log(∣F (jω∣)

​ ​

From this follows for the magnitude change per decade (ω2 /ω1 ​ ​ = 10)

Δ∣F (jωg )∣dB = 20 ⋅ log(10) = 20dB

​ ​ oder Δ∣F (jωg )∣dB = 20 ⋅ log(1/10) = −20dB
​ ​

The filter slope for a 1st order filter is +20dB/decade for the rising filter slope and −20dB for the falling filter slope.

According to the derivation above, slopes of ±40dB/decade result for 2nd order filters or +60dB/decade for 3rd order filters and
so on.





© ika 2025 • All rights reserved --- Version 25.04.01 41

 Filter circuits
Passive Low Pass First Order

IR = IC
​ ​

1
−Ue + I ⋅ (R + )=0
jωC
​ ​

1
−I ⋅ + Ua = 0
jωC
​ ​

Ua 1
F (jω) = =
​

Ue 1 + jωRC
​ ​

1
∣F (jω)∣ =
1+ (ωRC)2
​

1
∣F (jωg )∣ =
2
​ ​

1
ωg =
RC
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 42

 Passive Low Pass First Order
Amplitude Response/Phase Response

Amplitude response at fg decreased by 3dB .

​

Falling straight line with slope of

−20dB/decade
Phase rotation of −90° in total
Phasing rotation over a total of two decades

Phase rotation starts one decade before fg ​





© ika 2025 • All rights reserved --- Version 25.04.01 43

 Filter circuits
Passive High Pass First Order

IR = IC
​ ​

1
−Ue + I ⋅ (R + )=0
jωC
​ ​

−I ⋅ R + Ua = 0 ​

Ua 1
F (jω) = =
​

1
Ue 1 + jωRC
​

1
∣F (jω)∣ = ​

1
1+ (ωRC)2 ​ ​

1
∣F (jωg )∣ =
2
​ ​

1
ωg =
RC
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 44

 Passive High Pass First Order
Amplitude Response/Phase Response

Amplitude response at fg decreased by 3dB .

​

Rising straight line with slope of

+20dB/decade
Phase rotation of +90° in total
Phasing rotation over a total of two decades

Phase rotation starts one decade before fg ​





© ika 2025 • All rights reserved --- Version 25.04.01 45

 Filter circuits
Passive Allpass First Order

1
−Ue + I2 ⋅ (R + )=0 (1)
jωC
​ ​ ​

1
−Ue + I3 ⋅ (R + )=0 (2)
jωC
​ ​ ​

1
−Ua + I2 ⋅ − I3 ⋅ R = 0 (3)
jωC
​ ​ ​ ​

From the mesh equations (1) and (2) follows I2 ​ = I3 . ​

1
Ua jωC −R 1 − jωRC
F (jω) = = =
​

1 1 + jωRC
Ue +R
​ ​ ​

jωC
​

(1)2 + (ωCR)2
∣F (jω)∣ = =1
​

(1)2 + (ωRC)2
​

−ωRC
φ(jω) = arctan ( ) − arctan ( ) = −2 ⋅ arctan ( )
ωRC ωRC 


1 1 1
​ ​ ​

© ika 2025 • All rights reserved --- Version 25.04.01 46

 Passive Allpass First Order
Amplitude Response/Phase Response





© ika 2025 • All rights reserved --- Version 25.04.01 47

 Filter Circuit
Passive Bandpass

Determine the mesh and node equations

1
−Ue + I ⋅ (R + ) + Ua = 0
jωC
​ ​ ​

−R ⋅ IR + Ua = 0
​ ​

1
− ⋅ I C + Ua = 0
jωC
​ ​ ​

I − IR − IC = 0 ​

This results in

Ua jωRC
=
​

Ue 1 + 3jωRC − (ωRC)2
​ ​

1
ω0 = ωl ⋅ ωh =
RC
​ ​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 48 . 1

 Passive Bandpass
Derivation transfer function (1)

From the node equation we get

I = IR + IC ​ ​

Convert the mesh equations to IR and IC ​ ​

Ua
IR =
​

R
​ ​

IC = Ua ⋅ jωC
​ ​

Substituting into the remaining mesh equation to eliminate I

1
−Ue + ( + Ua ⋅ jωC ) ⋅ (R + ) + Ua = 0
Ua ​

​ ​ ​ ​ ​

R jωC
Extracting Ua gives ​

1 1
−Ue + Ua ⋅ [( + jωC ) ⋅ (R + ) + 1] = 0
R jωC
​ ​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 48 . 2

 Passive Bandpass
Derivation transfer function (1)

Ua
Converting to U results in
​

e ​

Ua 1
=
​

Ue ( R + jωC ) ⋅ (R + )
​ ​

1 1
​

jωC ​ +1

1
= 1
1+ + jωRC + 1 + 1
​

jωRC ​

1
= 1
jωRC + 3 +
​

jωRC ​

This results in the transfer function

Ua jωRC
=
​

Ue 1 + 3jωRC − (ωRC)2
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 48 . 3

 Passive Bandpass
Amplitude Response/Phase Response





© ika 2025 • All rights reserved --- Version 25.04.01 49

 Filter circuits
Passive Low Pass Filter Second Order

Determine the mesh equations

−Ue + I ⋅ (jωL + R) + Ua = 0
​ ​

1
−I ⋅ + Ua = 0
jωC
​ ​

This results in

Ua 1
=
​

Ue 1 + jωRC − ω LC
2
​ ​

Falling straight line with slope of

−40dB/decade.
∘
Phase rotation of 180 in total





© ika 2025 • All rights reserved --- Version 25.04.01 50

 Passive Low Pass Second Order
Amplitude Response/Phase Response





© ika 2025 • All rights reserved --- Version 25.04.01 51

 Analog Signal Processing
Higher Order Filter

1
F1 (jω) = 2
1 + jωRC − ω LC
​ ​

jωRC
F2 (jω) =
1 + jωRC
​

F (jω) = F1 (jω) ⋅ F2 (jω)

​ ​

Higher order filters can be achieved by connecting several filters in series

Coupled by impedance converters the resulting transfer function is the product of the individual transfer functions




© ika 2025 • All rights reserved --- Version 25.04.01 52

 Analog Signal Processing
Group Delay

The group delay τgr is an important parameter of an LTI-System and describes the delay of the envelope of a narrowband signal.
​

The group delay τgr can be calculated from the phase response φ(ω) of the transfer function
​

dφ(ω)
τgr (ω) = −
dω
​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 53

 Higher Order Filter
Amplitude Response and Group Delay





© ika 2025 • All rights reserved --- Version 25.04.01 54

 Higher Order Filter
Important Filter Circuits

Butterworth-Filter:
Flat amplitude-frequency response in the passband
Amplitude response drops off sharply
Overshoots with step response (increases with increasing filter order)
Tschebyscheff-Filter:
Steeper falling frequency response
Ripple in the passband of the filter
Bessel (Thomson-Filter):
Optimal behavior of the frequency response (maximum smooth)
Low slope
Constant group delay in the passband
Cauer-Filter
Also known as elliptical filters
Very steep amplitude response between passband and stopband
Oscillating behaviour in the course of the transfer function
Strong phase distortion 



© ika 2025 • All rights reserved --- Version 25.04.01 55

 Analog Signal Processing
Example: Analog wide-range inputs on the IKA control unit

Input capacitor C504 filters high frequency components

Second order low pass filter R507 / R508 / C505 / C506 to further filter unwanted frequency components
voltage divider input low-pass filter / R510 to reduce the input voltage range
Via C502 and C503 certain frequency modes are filtered from supply voltage
Coupling with supply voltage via voltage divider R506 / R510 to set a positive offset
OP with supply connection to GND and 5V: sets maximum output value range to 0V-5V. Output impedance adjustment via R509 



© ika 2025 • All rights reserved --- Version 25.04.01 56

 Analog Signal Processing and Digital Signal Processing
Signal and Information

To understand the difference between analog signal processing and digital signal processing, a clear distinction between signal and
information is necessary
Signal

The signal describes the way in which a piece of information is transmitted

A distinction can be made between analog and digital signals
Analog signals are physical characteristic curves of a time-continuous quantity (e.g. voltage characteristic), which can assume
arbitrary values.
Theoretically, all data transmissions could be regarded as analog (the type of signal transmission is decisive).
Analog here means that the information can continuously take on any value
Digital signals are characteristic curves of a quantity that allow only certain values (value discretization) at a certain time (time
discretization). The allowed values correspond to the predefined number of possible states

In reality, deviations of the discrete values also occur with digital signals. If these deviations are too high, the transmitted information is
lost!

Information

Information is the amount of data that is transmitted by means of a signal 




© ika 2025 • All rights reserved --- Version 25.04.01 57

 Analog Signal Processing and Digital Signal Processing
Analog/Digital-Converter

Analog/digital converters are electrical components for converting an analog signal into a digital signal

Sampling (time discretization)

Sampling of the analog signal in a certain time interval

The frequency of the sampling is also called sampling rate or sampling frequency

Quantization (value discretization)

Since digital signals can only take previously defined discrete levels, values between these discrete levels must be rounded
The values determined from this are also referred to as code words
The number of valid discrete values can be defined for the representation in a binary number system via the number of necessary
bits
The number of bits required to represent all allowed states is also referred to as the resolution 



© ika 2025 • All rights reserved --- Version 25.04.01 58

 Analog/Digital-Converter
Quantization & Sampling





© ika 2025 • All rights reserved --- Version 25.04.01 59

 Analog/Digital-Converter
Sample-and-Hold Circuit

Sample-and-hold circuit enables a defined voltage value to be "held" for a short duration at a specific point in time.

The analog/digital converter needs a certain amount of time to quantize the signal
Sample-and-hold circuitry is therefore necessary as part of the analog/digital conversion process





© ika 2025 • All rights reserved --- Version 25.04.01 60

 Analog/Digital-Converter
Quantization
Quantization (value discretization) describes the conversion of the
continuous input signal into a value-discrete output signal.

For an AD converter with a resolution of n bits, a total of 2 output values n

can be represented

The limited number of possible output values leads to a

representation error, which is referred to as a quantitation error.

The voltage ULSB (Least Significant Bit) describes the input voltage
​

distance between two output code words and is related to the FS (Full
Scale) value.
The term ‘Full Scale’ is used to describe the highest possible digital
output code of an ADC with n-bit resolution.

The value LSB is often used instead of ULSB ​

UFSR = URef + ​ ​
(URef − = 0V)
​
UFSR = URef + − URef −
​ ​ ​

UFSR
ULSB = n
​

2 
​ ​




© ika 2025 • All rights reserved --- Version 25.04.01 61

 Analog/Digital-Converter
Dirac-Comb

The Dirac comb Ш(t) represents the sum of a periodic sequence of shifted Dirac-functions

Ш(t) = ∑ δ (t − nT )
​ T >0
n∈N

With the help of the Dirac comb, the sampling of an arbitrary function y(t) with a sampling rate T can be described. The sampling
process is represented as a multiplication with the Dirac-Comb

yS (t) = y(t) ⋅ Ш(t)

​





© ika 2025 • All rights reserved --- Version 25.04.01 62

 Analog/Digital-Converter
Classification

Analog/Digital-Converters can be divided into different categories

Functional Principle
Counting method
Delta Sigma Converter
Successive Approximation Register
Pipeline Method
Parallel Method
Characteristic Parameters

Resolution
Smallest quantization of the input voltage range that can be digitally represented by the converter.
Speed (sample rate)
Number of measured values per time unit
Input voltage range, cost
The selection of the required converter depends on the task at hand

See exercises 


© ika 2025 • All rights reserved --- Version 25.04.01 63

 Analog/Digital-Converter
Resolution and Sampling Rate





© ika 2025 • All rights reserved --- Version 25.04.01 64 . 1

 Analog/Digital-Converter
Parallel Method (Flash-Converter)





© ika 2025 • All rights reserved --- Version 25.04.01 64 . 2

 Analog/Digital-Converter
Quantization Error

Unlike analogue signals, digital signals can only assume discrete

values. This value discretisation means that not all analogue signals
can be represented exactly.

The resulting error is referred to as quantization error e(t).

The quantization error is signal-dependent.

The quantization error e(t) can be approximated by a

sawtooth function and fluctuates between q/2 and −q/2
depending on the quantization step q = LSB.

−q q
e(t) = st <t<
2s 2s
​ ​

−q/2s 2
s q
eMSE = (e(t)) = ⋅ ∫
2 2
(st) dt =
q q/2s 12
​ ​ ​ ​ ​

2
q
eRMSE = (e(t)) =
12
​ ​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 65 . 1

 Quantization Error
Example Function

The graph shows an example of the quantization error for a sinusoidal input signal





© ika 2025 • All rights reserved --- Version 25.04.01 65 . 2

 Analog/Digital-Converter
Characteristics

Resolution

The resolution r is defined as the number n of bits used to represent the output code words.
Dynamic Range

The dynamic range DR of an AD-converter is defined as the ratio between the largest and smallest signals that can be precisely
measured and is closely linked to the resolution r and is usually specified in decibels.

2 ⋅ LSB
n
DR = 20 ⋅ log10 ( ) = 6.02 ⋅ n dB
LSB
​ ​

In addition to the described characteristics such as resolution and dynamic range, there are various characteristics for describing the
dynamic behaviour of AD-converters.

The usual metrics such as SNR, THD, SINAD, ENOB, SFDR for comparing the dynamic behaviour are based on a Fast Fourier
Transformation (FFT) following the AD-converter for spectral analysis of the output.
The following equations show the theoretical relationships for determining the dynamic characteristics. These can be calculated for
the ideal AD-converter and then compared with the actual values of a real AD-converter to obtain a measure of the performance of
the analysed AD-converter.




© ika 2025 • All rights reserved --- Version 25.04.01 66 . 1

 Analog/Digital-Converter
Dynamic Characteristics 1/5

Test setup for the determination of dynamic characteristics

Assumption of a sine wave v(t) as input with a full scale amplitude

q⋅2 n
2
q ⋅ 2 n
v(t) = ⋅ sin(ωt) (v(t)) =
2 2 2
​ ​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 66 . 2

 Analog/Digital-Converter
Dynamic Characteristics 2/5

Signal/Noise-Ratio

The signal-to-noise ratio (SNR) is a measure of the noise behaviour of an AD-converter and is defined as the ratio of the signal power
PSignal to the noise power PN oise .
​ ​

2
RMSSignal
=( ) = 20 ⋅ log ( )
PSignal USignal,RMS
SNR =
​ ​ ​

PN oise UN oise,RMS RMSN oise

​ ​ ​

​ ​ ​

) = 20 ⋅ log ( ) = 20 ⋅ log(2 ) + 20 ⋅ log ( ) = 6.02 ⋅ n + 1.76dB

RMSSignal (q ⋅ 2 )/(2 2) n
3
SNR = 20 ⋅ log ( n
​ ​

RMSN oise q/ 12 2
​ ​ ​ ​

Total Harmonic Distortion

The total harmonic distortion (THD) is defined as the ratio of the power of all harmonics to the power of the fundamental.

2
∑ Uharmonics,RMS
∑ PHarmonics ​ ​

THD = =
​

PSignal USignal,RMS
​ ​

​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 66 . 3

 Analog/Digital-Converter
Dynamic Characteristics 3/5

Signal to Noise and Distortion Ratio

The Signal to Noise and Distortion Ratio (SINAD/SNDR) is a measure of the dynamic performance of AD-converters

PSignal
SINAD = SNDR = = 20 ⋅ log ( 10 −SNR/10 + 10 THD/10 )
​

PN oise + PDistortion
​ ​

​ ​

Spurious-Free Dynamic Range

The Spurious-Free Dynamic Range (SFDR) is used to measure the usable dynamic range of an AD converter and is defined as the ratio of
the root mean square value of the sinusoidal test signal and the RMS value of the peak noise signal of the output.

RMSSignal
SFDR = 20 ⋅ log ( )
​

RMSSpur,Largest
​





© ika 2025 • All rights reserved --- Version 25.04.01 66 . 4

 Analog/Digital-Converter
Dynamic Characteristics 4/5

Effective Number of Bits

The Effective Number of Bits (ENOB) is a measure of the quality of the data conversion.

For an optimal AD-converter, the effective number of bits should also correspond to the theoretical resolution of the AD-converter
Noise and distortion reduce the effective number of bits in reality
ENOB specifies the remaining effective number of bits

SINADMeasured − 1.76dB + 20 ⋅
​ log ( Input Amplitude )
Full Scale Amplitude
​

SINAD − 1.76dB
ENOB = ENOBFull Scale Input =
6.02 6.02
​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 66 . 5

 Analog/Digital-Converter
Dynamic Characteristics 5/5





© ika 2025 • All rights reserved --- Version 25.04.01 66 . 6

 Analog/Digital-Converter
Alias-Effect

The digital representation of a signal only allows sampling at discrete time values

The possible incorrect determination of a signal frequency in this case is called aliasing

For a sampling frequency that is too low fs lead to a too low assumed signal frequency
​

The formerly high frequency components appear as low frequencies (alias)

Reconstruction of the original signal from the now digital signal is no longer possible
If the loss of high-frequency signal components is accepted, the signal can be processed with a low-pass filter before sampling to
avoid the alias effect.





© ika 2025 • All rights reserved --- Version 25.04.01 67

 Alias-Effect
Time Domain and Frequency Domain

If the alias effect is observed from the perspective of the frequency

domain, sampling with the frequency f of a signal with the cutoff
frequency fg results in periodic repetitions of the frequency
​

spectrum

n ⋅ fs ​
n∈N

Aliasing results from the overlapping of frequency spectra,

Too low sampling frequencies fs (or 1/Ts ) lead to aliasing

​ ​

Overlapping of adjacent spectra

Sampling should be done with at least twice the signal
frequency fg contained in the signal spectrum
​





© ika 2025 • All rights reserved --- Version 25.04.01 68

 Alias-Effect
Nyquist–Shannon Sampling Theorem

In order to avoid undersampling and thus the development of the alias effect, the Nyquist-Shannon sampling theorem describes a
minimum sampling frequency that should not be undercut

fs ≥ 2 ⋅ b
​

The sampling frequency fs should correspond to at least twice the bandwidth b of the signal (Nyquist criterion)
​

The half of the sampling frequency fs /2 is called Nyquist frequency, which should not be exceeded by the signal
​

By adhering to the Nyquist-Shannon sampling theorem, an overlap of the frequency spectra is avoided





© ika 2025 • All rights reserved --- Version 25.04.01 69

 Digital Signal Processing
Digital/Analog-Converter

Function

Conversion of a digital signal into a proportional analog electrical signal (voltage/current)

Various methods possible
Binary Weighted method
Parallel method
Counting method
DA-converters often use reconstruction filters to smooth the output signal

Parameters

Resolution: 8-16Bit usual

Settling time: [ns − μs]

Output: current/voltage





© ika 2025 • All rights reserved --- Version 25.04.01 70

 Digital/Analog-Converter
Binary Weighted Method (1)

The switches Si must be closed whenever a logical

​
Example of a 4-bit DA-converter with a total of 16 possible states
one occurs at the relevant position of the binary
number.

−URef + I ⋅ Re = 0​ ​

Determine the total voltage depending on whether

switch Si is closed. Switch closed for zi = 1 and open
​ ​

for zi ​ = 0. Switch S0 : LSB , S3 = MSB

​ ​

I ⋅ R + Ua = 0 ​

1 1 1 1 1 16R
= z3 ⋅ + z2 ⋅ + z1 + z0 → Re =
Re 2R 4R 8R 16R z0 + 2z1 + 4z2 + 8z3
​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​

​ ​ ​ ​ ​

General for n-bit DA-converters

n−1
1 1 2 R
n
= ∑ zi ⋅ n−i → Re =
Re 2 R n−1
2 ⋅ zi
​ ​ ​ ​ ​ ​

i=0
​

∑i=0 ​
i ​





© ika 2025 • All rights reserved --- Version 25.04.01 71

 Digital/Analog-Converter
Binary Weighted Method (2)

Substituting and converting the equation to Ua ​

Example of a 4-bit DA-converter with a total of 16 possible states

URef
Ua = −R ⋅ I = −R ⋅
​

Re
​ ​

1 1 1 1
= −URef ⋅ (z3 ⋅ + z2 ⋅ + z1 + z0 )
​ ​

2 4 8 16
​ ​ ​ ​ ​ ​ ​ ​ ​

Overall, this results in

Z
Ua = −URef ⋅
ZM ax + 1
​ ​ ​

The following applies to the maximum number that

can be displayed

ZM ax = 2 − 1
​
n





© ika 2025 • All rights reserved --- Version 25.04.01 72

 Digital/Analog-Converter
Parallel Method

The parallel method is based on the principle of a voltage divider

A fixed voltage is generated when a switch is closed





© ika 2025 • All rights reserved --- Version 25.04.01 73

 Digital/Analog-Converter
Reconstruction Filter (1)

Reconstruction filters are often used to smooth out the abrupt behaviour of the output signal

Input UA changes the value abruptly

​

The resulting output voltage UA,TP can be regarded as a step response

​

−τ
t
UA,TP = UA ⋅ (1 − e
​ ​
​

) with time constant τ = RC





© ika 2025 • All rights reserved --- Version 25.04.01 74

 Digital/Analog-Converter
Reconstruction Filter (2)

Initial situation
Rampe abgetastet mit Sample-and-Hold-Schaltung
The "optimal" setting

Highest possible attenuation of the mirror spectra while

retaining the main spectrum (see system theory / control
engineering)
An ideal low-pass filter (rectangular frequency response) is
required for exact reconstruction. This cannot be realised in
reality.

UA = UDA ⋅ (1 − e −τ
)
t
​ ​
​

τ = RC

1 1
fg,opt = ⋅ fa (Shannon) fg =
2 2πRC
​ ​ ​ ​ ​

1 1 Ta
= → τopt =
​

2πτopt 2Ta π
​ ​ ​ ​


​ ​

© ika 2025 • All rights reserved --- Version 25.04.01 75

 Digital/Analog-Converter
Pulse Width Modulated Signals (PWM-Signals)

Pulse width modulation is a type of modulation in which the

signal alternates between two values

Typical output of microcontrollers

Duty cycle D between 0 and 100%

Characteristic Parameters

The PWM signal is clearly described by the duty cycle D ,

the base frequency f and the pulse voltage UP W M ​

tH tH
D= = [D ] = %
​ ​

tL + tH T
​ ​

​ ​

1
f=
T
​

Estimation of the achievable medium voltage

tH
UM = ⋅ UP W M
​

T
​ ​ ​





© ika 2025 • All rights reserved --- Version 25.04.01 76

 Digital/Analog-Converter
Pulse Width Modulated Signals (PWM-Signals)

Microcontrollers often have a stand-alone controller for

generating PWM signals that do not impact the CPU load
PWM - An element of a microcontroller
Initialisation of the pulse duration and pulse frequency
by the SFR (Special Function Registers: PTx, PPx and
PWx)
PWM module reads the data from these registers and
uses it to generate the corresponding signal
The base frequency is defined in the PPx register; once
the PTx counter has reached the value PPx, it is reset and
starts a new counting process
Pulse duration tH is defined by register PWx, if the
​

counter value PTx is greater than or equal to the value in

register PWx, the digital output bit of the microcontroller
is set
Smoothing through external low-pass filter
Low-pass must be set to the pulse period of the PWM
signal. 



© ika 2025 • All rights reserved --- Version 25.04.01 77

 Digital/Analog-Converter
Counting Method

A digital/analogue-converter using the counting method is a simple and cost-effective alternative to the binary weighted method

Generation of an analogue output signal from a PWM signal using a low-pass filter

Decoupling of input and output voltage using an operational amplifier circuit with the transfer function F (jω) =1

Control of the switch via a PWM signal

When the switch is closed, the capacitor C is charged via the resistor R

The time constant τ and therefore the charging curve depends on the dimensioning of the components

τ =R⋅C




© ika 2025 • All rights reserved --- Version 25.04.01 78

 Digital/Analog-Converter
Sources of Error

Ideal result

Optimal conversion of the digital signal into an analog signal

Zero Point Error

Determined by reverse currents flowing through open switches

Full Scale Error/Gain error

Determined by the on-resistance of the switches and the accuracy of the negative feedback resistor
Non-linearity
Indicates by how much a step is greater or less than 1 LSB in the most unfavourable case
Non-linearity is usually below 0.5 LSB





© ika 2025 • All rights reserved --- Version 25.04.01 79

 Contact

Christian Kehl, M.Sc.

Institute for Automotive Engineering (IKA)

RWTH Aachen University
Steinbachstraße 7
52074 Aachen
Germany

Phone
Fax

E-Mail [email protected]


Internet www.ika.rwth-aachen.de 


© ika 2025 • All rights reserved --- Version 25.04.01 80