ELECTROKINETICS 2

Summary:
Chapter 1: FILTERS (by myself, Marc Lethiecq) 6 x 1h20min
Chapiter 2: QUADRUPOLES (by Maxime Bavencoffe)

References
- Electronique - Théorie du signal et composants - Cours et exercices corrigés -
éditions DUNOD - Manneville / Esquieu
- Les fondements du génie électrique - éditions TEC&DOC - Laurent Henry
- Electronic Devices and Circuits, Schaum’s outline series – Jimmie J. Cathey

UE SCIENCES APPLIQUEES ET INDUSTRIELLES 2 1

EC Electrocinétique 2
 Course organisation

Lectures (Cours Magistral, CM) mixed with

exercises (Travaux Dirigés,TD), including
tests

Evaluation :
+8 h Lab. Work
(Travaux • Tests (40%)
• Lab Work (20%)
Pratiques, TP)
• Final exam(40%)

2
 Electrokinetics 2 (Electrocinétique 2)
Summary of chapter 1

1.1. Frequency analysis of linear circuits

1.2. Bode plots
1.3. First order filters
1.4. Second order filters

3
1.1. Frequency (harmonic) analysis of linear circuits

Here we will describe the methods used to analyse the behaviour in the frequency
domain of a linear circuit, i.e. how the output varies when the input frequency changes.

In order to study a linear circuit, we will apply a sinusoidal input signal (for instance
using a function generator in sinusoidal mode) and study the output (for instance using
an oscilloscope). The input, usually represented on the left side is also called the
excitation signal, while the output, generally represented on the right side is also called
the response.

Thanks to the linearity of the circuit, all signals will be sinusoidal and their frequency
will be equal to that of the input, so we will only use complex notations to represent
the signals (voltages and currents).

Input signal Linear circuit Output signal

(response)

4
1.1.1. Dipoles and quadrupoles
i
A dipole is an electrical element (or D
component) with two connections:
u

A quadrupole is an electrical circuit that has four connections: two input

connections and two output connections. There are four electrical quantities: input
voltage ue and input current ie, output voltage us and output current is.

ie is
Ae As
Source Quadrupole us Charge
ue
Be Bs

A quadrupole is made of dipoles (it can be represented as an assembly of

interconnected dipoles). It is said to be linear if and only if all of the dipoles it is made
of are linear.

5
 1.1.2. Harmonic transfer function of a linear quadrupole

The harmonic transfer function of a linear quadrupole, generally noted T(jw) (or H(jw)) is
the ratio of the complex representation of the output signal S to that of the input signal E:

T(jw) = S / E

This function depends on the characteristics of the quadrupole and on the frequency
(noted f or sometimes n) of the signals (generally voltages, but can also be currents)
which is imposed by the input. Often, instead of the frequency (in Hz), the angular
frequency w in rd/s is used, where w = 2 p f .
1.1.3. Magnitude, amplification/attenuation and phase shift
To completely characterise the response of a linear circuit to a sinusoidal input, one
needs to study its transfer function T as a function of frequency f or angular frequency w:
T being a complex function, we will need to study both
- its magnitude: A(w) = | T(jw) |
- its phase shift: ϕ(ω) =arg(T())
If A() > 1 the circuit amplifies the input (the output amplitude is higher than that of the input)

If A() < the circuit attenuates the input (the output amplitude is lower than that of the input)
6
 What can be the use of a linear quadrupole?

Amplify a
signal
Shift a
signal’s
phase
Quadrupole
Q

Change the shape of a

Filter (sort out) a mix of several complex signal by
signals: spectral analysis, acting differently on its
selection, elimination of noise… harmonics …

7
 1.1.4. The GAIN of a quadrupole
One can define the gain of a quadrupole (without dimension) using the decibel (dB)
as unit, by:
GdB = 20 log10(A(w)) = 20 log10 (|T(jw)|)

The gain is just another way to quantify the magnitude of the transfer function.

A few comments:

The decibel is a sub-multiple of the bel, a unit named after Alexander Graham Bell, the
inventor of the telephone. It was first - and still is - used to measure sound intensity (it takes
into account the fact that the human ear has a close to logarithmic response) and has been
extended to electrical quantities.

Originally the gain was was defined as a ratio of powers : GB = log (Pout/Pin) in bels, but a
unit ten times smaller is now preferred: the decibel defined as GdB = 10 log (Pout/Pin) .

Since the electrical power is proportional to the voltage (or current) square, the gain in
decibels can be calculated as GdB = 10 log (aUout2/aUin2) = 20 log (Uout/Uin).
8
Interpretation of the gain
1) What does it mean if the gain GdB positive ? Negative ?
The gain GdB is positive if & only if A > 1 Þ the output is amplified as
compared to the input .
The gain GdB is negative if & only if A < 1 Þ the output is attenuated as
compared to the input .

2) What is the gain and the transfer function’s magnitude (also called the
amplification) of a quadrupole if we obtain a 1V output when applying a 5V input ?

A = Us / Ue = 1 / 5 = 0,2 and GdB = 20 log A = 20 log(0,2) = -14dB

3) What do the following sentences mean: « at 3 kHz, this quadrupole has a gain of
+20 dB / -20 dB / -40 dB / -3 / +6 dB » ?

GdB = + 20dB Þ 20 log (A) = + 20 Þ log (A) = 1 Þ A = 101 = 10

GdB = - 20dB Þ 20 log (A) = - 20 Þ log (A) = -1 Þ A = 10-1 = 0,1
GdB = - 40dB Þ 20 log (A) = - 40 Þ log (A) = -2 Þ A = 10-2 = 0,01
GdB = -3dB Þ 20 log (A) = -3dB Þ log (A) = -3/20 Þ A = 1/√2
9
GdB = +6dB Þ 20 log (A) = + 6dB Þ log (A) = 6/20 Þ A=2
 1.2. Bode plots
1.2.1. Characteristics of a Bode plot
In order to analyse the response of a linear quadrupole as a function of frequency, the
Bode plot is the most practical representation.
A Bode plot is a set of two curves as a function of frequency or angular frequency:
- The first one represents the gain G, in dB
- The second one represents the phase shift ϕ, in rad or degrees.
Very important specificity of the Bode plot:
a logarithmic scale is used for the frequency or angular frequency
Why was a logarithmic scale chosen?
A logarithmic scale allows a wide frequency range to be represented on a single graph. Each
frequency decade has the same length on the scale (e.g. 1Hz to 10Hz has the same length as
100Hz to 1kHz). The horizontal axis is thus in fact log (f), but only the values in Hz or Rad/sec
are written.
G(dB) ϕ(rad or deg)

1 10 100 1000 f(Hz) 1 10 100 1000 f(Hz)

0 1 2 3 log(f) 0 1 2 3 log(f) 10
A few definitions: decade, octave and choice of represented frequency range
Two frequencies f1 and f2 > f1 are separated by a decade if f2 = 10 * f1
Decades are already represented on sheets prepared for Bode plots
Two frequencies f1 and f2 > f1 are separated by an octave if f2 = 2 * f1
It is up to you to determine which values are represented on a Bode plot: you must
choose them so that the most interesting parts of the plot is around the middle of the
frequency axis.
In a Bode plot, the slopes of line segments are expressed either in dB/decade, or in
dB/octave. For linear quadrupoles, the slopes are always multiples of 20 dB/decade
(which practically corresponds to multiples of 6 dB/octave).

11
Construction of a logarithmic scale
A value of frequency f is chosen and placed according to its decimal logarithm log(f), f being
expressed in Hz.
- Construction of a decade:
The following table is filled-in:
f 1 2 3 4 5 6 7 8 9 10

log10(f) 0 0,3 0,477 0,6 0,699 0,778 0,845 0,903 0,954 1

- Then the positions of log10(x) are placed along the axis:

1 2 10 f
0 0.3 1 log(f)

To cover several decades, all you need to do is to reproduce this scale (on the left and/or on the
right) as many times as required. The frequency f is multiplied by 10 each time when moving
one decade towards the right and divided by 10 each time when moving one decade towards the
left.

Comment: for those who often need to trace Bode plots, specific sheets with a logartthmic
horizontal scale have been designed (semi-logarithmic paper), avoiding the need to build the
scale such as described above. 12
1.2.2. Asymptotic Bode plots of a few transfer functions.
The asymptotic Bode plot is an approximate plot that roughly describes the behaviour of
a quadrupole using only straight line segments.

It is obtained by considering the behaviour at the frequency limits: f or w ® 0 and f or w

® ¥, considering that in a sum, only the dominant terms are taken into account.

Some particular values then appear in the transfer function.

We will determine asymptotic Bode plots of a few transfer functions.

A helpful comment for some transfer functions: to obtain the Bode (or asymptotic Bode)
plot of the inverse of a transfer function, you just need to change the sign of its Bode (or
asymptotic Bode) plot.

A helpful comment for complicated transfer functions: the Bode (or asymptotic Bode) plot
of the product of two transfer functions is the sum of each of their Bode (or asymptotic
Bode) plots.

13
 T1(jw) = jw/w0 (w and w0 are always positive) T2(jw) = 1 / jw/w0 = 1 / T1(jw)
GdB = 20 logw - 20 logw0 = 20 logw + cst GdB = -20 logw + 20 logw0 = -20 logw + cst
GdB is a straight line with a slope of +20 dB/decade GdB is a straight line with a slope of -20 dB/decade
ϕ = + p / 2 rad or +90 deg ϕ = - p / 2 rad or -90 deg
Here the asymptotic Bode plot and the Bode plot The Bode plot of T2 is minus the one of T1
itself are identical. Here again the asymptotic Bode plot and the Bode
plot itself are identical.
db db
Magnitude Magnitude

40 40

30 30

20 20

10 10

0 0

-10 -10

-20 -20

-30 -30

-40 -40
-1 0 1 2 3 -1 0 1 2 3
10
w
10 10 10 10 10 10 10 10 10

degrees Phase
w degrees Phase

0 0

0 0

-1 0 1 2 3 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10

14
 T3(jw) = 1+ jw/w0 T4(jw) = 1 / (1 + jw/w0) = 1 / T3(jw)
When w® 0 GdB® 20 log 1 = 0 and ϕ® 0 The asymptotic Bode plot of T4 is minus the one
When w® ¥ 1 becomes negligible compared to of T3 (so is the Bode plot itself).
jw/w0, GdB® 20 logw - 20 logw0 = 20 logw + cst
and ϕ® + p / 2 rad or +90 deg.

db db
Magnitude Magnitude

40 40

30 30

20 20

10 10

0 0

-10 -10

-20 -20

-30 -30

-40 -40
-1 0 1 2 3 -1 0 1 2 3

w
10
w
10 10 10 10 10 10 10 10 10

degrees Phase degrees Phase

0 0

0 0

-1 0 1 2 3 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
w w 15
 T5(jw) = 1 / (1 + jw/w1)(1 + jw/w2) = T6(jw) = 1 / (1+2mjw + (jw/w0)2),
1 / (1 + jw/w1) x 1 / (1 + jw/w2)
If m > 1, T6 can be expressed as T5.
Let’s use the results of T4: dashed lines, first with
If m<=1, T6 cannot be expressed as T5, then
w1 second with w2. Then we just need to add
the asymptotic Bode plot is the one below
the two asymptotic curves (solid line). Slopes
below. Slopes are then 0 and -40 dB/decade.
are (from left to right): 0, -20 and -40 dB/decade.
db db
Magnitude Magnitude
0 0

-5 -5

-10 -10

-15 -15

-20 -20

-25 -25

-30 -30

-35 -35

-40 -40

-45 0
-45
10
-1
10
0
w 10
1
w 10
2
10
3
10
-1
10
0
w 10
1
10
2
10
3

degrees Phase degrees Phase

0 1 2 0 0

-90

-180

-1 0 1 2 3 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
16
 1.3. First order filters
1.3.1. 1st order electrical circuits
In general, a transfer function can be written as a rational fraction of polynomial
functions of jω:
Us N ( jω)
T ( jw ) = = with Deg (N(jω)) ≤ Deg (D(jω))
Ue D ( jw )
The order of the circuit is given by the degree of the polynomial function D(jω):
- if D(ω) is a 1st degree polynomial of jω, then the circuit is a 1st order one
- if D(ω) is a 2nd degree polynomial of jω, then the circuit is a 2nd order one
- etc.
2 K
Examples : T 1 (w ) = and T ( f ) = are 1st order circuits
w 2
1+ j × 1 + j × 2pfRC
6
1+ 3 j × f K'
T 3( f ) = and T 4 (w ) = are 2nd order circuits
1 + 2 j × f + (300 j × f )
2 2
æ R ö
1 + 3 j × wRC + ç j w ÷
è L ø
17
1.3.2 How to study simple quadrupoles / filters
ue Z1
Simple quadrupoles/filters can generally by us
represented as in the figure: Z2

Quadrupole Q
If the output current is zero, the transfer
function can be obtained by applying the H(jw) = Us / Ue = Z2 / (Z1 + Z2)
voltage divider formula:

Variations of the magnitude and phase shift

as functions of angular frequency can then Magnitude: Phase shift:
be studied separately.
A(w) and GdB(w) φ(w)
Often, an asymptotic study is sufficient:
- Low frequency behaviour (ω ® 0)
- High frequency behaviour (ω ® +¥)
- Specific values: maximum,…

18
Remark:
Most often, the order of a circuit can be determined by the number of capacitors and
inductors that it contains. Only 1 inductor or capacitor: 1st order circuit.
C

R V2
V1

1.3.2. Filtering circuits (or filters)

A filter is an electrical circuit that contains elements assembled in a way to

selectively transmit signals in a given frequency range. An ideal filter has one or
several pass-bands in which signals are transmitted without being attenuated and
one or several stop-bands in which signals are attenuated or stopped.
Filters can be characterised by:
- Their amplification or atténuation
- The frequency bands in which they operate: low-pass, high-pass, band-pass notch
- Their technology: active filters (can amplify signals because they contain at least
one active component) of passive filter (made only of capacitors, inductors and
resistors).
19
Ideal filters
A A

fc f fc f
High-pass filter Low-pass filter
A A

fc1 fc2 f fc1 fc2 f

Bandpass filter Notch filter

Definition of the cut-off frequency(ies) of a filter.

Cut-off frequencies are the frequency limits of a filter’s passband, they are defined
by the following condition on the magnitudes:

GdB (fc) = GdB max – 3 dB Û A(fc) = Amax / Ö2

20
 1.3.3. Bode plots of first order filters
A first order filter is a 1st order electrical circuit that allows input signals to be
attenuated (cut or stopped) in a certain range of frequencies. In this stopband, the
magnitude curve of the Bode plot of a first order filter always has a ±20 dB/decade
slope (or ±6 dB/octave).
Low-pass filter: its transfer function is given by
db Magnitude
0
-5
-10 Plot for K = 1 A low-pass filter:
-15
-20 • does not attenuate low frequency
-25
-30
signals, i.e. between 0 Hz et fc.
-35 • attenuates (filters out) high
-40
-45
-1 0 1 2 3
frequency
Hz signals, i.e. between fc
10 10 10 10 10 and+∞.
degrees Phase
0
-10
-20
-3dB bandpass: [0 ;fc]

]f ;+¥[
-30
-40
-50 stop-band: c
-60
-70
-80
-90
fc:Hzcut-off frequency
-1 0 1 2 3
10 10 10 10 10
21
T3 = 1/(1+j.f/fo)
 High-pass filter: its transfer function is given by

45
db Magnitude
A high-pass filter:
40 • does not attenuate high
35
30 frequency signals, i.e. between fc
25
Plot for K’ = 125 and +∞.
20
15
10 • attenuates (filters out) low
5
0
frequency signals i.e. between 0
-5
-1 0 1 2 3
Hz
Hz and fc.
10 10 10 10 10

]f ;+¥[
degrees Phase
90
80
-3dB bandpass: c
70
60
50 stop-band: [0 ;fc]
40
30
20
10
fc: cut-off frequency
0 Hz 22
-1 0 1 2 3
10 10 10 10 10
1.3.4 Qualitative analysis of filters

The qualitative analysis of a filter consists in the determination of its behaviour for
extreme frequencies, i.e. f ® 0 and f ® +∞. For this one needs only to replace the
dipoles by their equivalent at these frequency limits:
iR(t) ic(t) iL(t)

R uR(t) C uc(t) uL(t)

L

ZR = R ZC = 1 /jCw ZL = jLw

ZC ® ∞ ZL ® 0
ω®0
ZR = R Capacitor equivalent to an Inductor equivalent to a
open circuit short circuit

ZC ® 0 ZL ® ∞
ω ® +∞ ZR = R Capacitor equivalent to a Inductor equivalent to an
short circuit open circuit

23
 1.3.4 Qualitative analysis of first order filters

C R C1

C2
ue(t) us(t) ue(t) C us(t)
R ue(t) R us(t)

ω®0 C ω®0 ω®0

R C1

C2
ue(t) us(t) ue(t) C us(t) ue(t) us(t)
R R

us(t) = 0 us(t) = ue(t) us(t) = 0

ω ®∞ C ω ®∞ R ω® ∞ C1

ue(t) C2
ue(t) us(t) us(t) R us(t)
R ue(t) C

us(t) = 0
us(t) = ue(t) us(t) = 0 24
=> High-pass filter => Low-pass filter => Band-pass filter
 1.4. Second order filters

The 4 types of typical 2nd order filters are characterised by the following transfer
functions:

Low-pass (or high-cut) filters: High-pass (or low-cut) filters:

2
1 æ wö
T bas (w ) = K çç j ÷÷
2
T haut (w ) = K è w0 ø
w æ wö
1 + 2m. j +ç j ÷ w æ wö
2

w0 çè w0 ÷ø 1 + 2m. j +ç j ÷
w0 çè w0 ÷ø

Bandpass filters: Notch filters:

w æ wö
2
2m. j 1 + çç j ÷÷
w0 w
T bande (w ) = K 2 T réj (w ) = K è 0 ø
w æ wö w æ wö
2
1 + 2m. j + çç j ÷÷ 1 + 2m. j + çç j ÷÷
w0 è w0 ø w0 è w0 ø

25
 1.4.1. Bode plots of second order filter
1
 Low-pass filters: T bas (w ) = K 2
w æ wö
1 + 2m. j + çç j ÷÷
w0 è w0 ø
K= 1 m>1 m<1 m < √2 / 2
db db
Magnitude Magnitude
0 0
-5 -5

-10 (−1) -10 !

m>
-15 -15
! (−2)
-20 (−1) -20

-25 -25

-30 -30
-35 -35

-40 (−2) -40

-45 -45
10
-1
10
0
[ω1] 10
1
[ω2] 10
2
10
3
10
-1
10
0
[ωR] [ω0] 10
1
10
2
10
3

degrees Phase degrees Phase

0 0

-90

-180

-1 0 1 2 3 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
26
 2
æ w ö
çç j ÷÷
è w0 ø
‚ High-pass (or low-cut) filters: T haut (w ) = K 2
w æ wö
1 + 2m. j + çç j ÷÷
w0 è w0 ø
K= 1
m>1 m<1 m < √2 / 2
db db
Magnitude Magnitude
0 0

-5 -5 !
-10 -10 m>
!
-15
(−1) (−1) -15

-20 -20 (−2)

-25
(+2)
-25
(+2)
-30 -30

-35 -35

-40 -40

-45
+180 -45
-1 0 1 2 3 -1 0 1 2 3

[ω 0] [ωR]
10 10 10
10 10
[ω1] [ω0] [ω2] 10 10 10 10 10

degrees Phase degrees Phase

0
+180

90 90

0 0
-1 0 1 2 3 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
27
 w
2m. j
w0
ƒ Bandpass filters: T bande (w ) = K 2
w æ wö
1 + 2m. j +ç j ÷
w0 çè w0 ÷ø
K= 1
m> 1 m< 1
20log(2m)
db Magnitude db
Magnitude
0 0

-5 -5

-10 (−1) -10 20log(2m)

-15
(+1)
m=1 -15
(+1)
-20
(−1) -20
(−1)
-25 m = 0,707 (−1)
-25

-30 -30

-35 -35

-40 -40

-45 -45
-1 0 1 2 3 -1 0 1 2 3
10 10 10
ω0 2mω0
10 10 10 10
2mω 10
ω0
10 10

degrees Phase degrees Phase

0
0

+90 +90

0 0

-90 -90
10
-1
10
0
ω
10
1
10
2
10
3
10
-1
10
0
10
1
10
2
10
3
28
 2
æ wö
1 + çç j ÷÷
„ Notch filters: T (w ) = K è w ø 0
=
N(w )
D(w )
réj 2
K= 1 w æ w ö
1 + 2m. j +ç j ÷÷
m>1: æ wö
2 w çè w 0 0 ø
1 + çç j ÷÷
è w ø
T (w ) = K
réj
w w
0

m<1 (+2) N(w )

(1 + j )(1 + j )
db
w 1
w 2
(+2)
db
Magnitude Magnitude
0 0

-5 -5

-10 (+1) -10

-15
-15
(−1)
-20 -20
(−2)
-25 -25

-30
(−1)
-30 w- > w 0
-35 -35 1
-40 -40 D(w )
-45 -45
-1 0 1 2 3 -1 0 1 2 3
10 10
ω1
10
ω0
10
ω2
10 10 10 10
ω0 10 10

degrees Phase degrees Phase

0
90 +90

0 0

-90 -90
-1 0 1 2 3 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
29
 1.4.2. Qualitative analysis of second order filters

us(t)
us(t)
L C L C

ue(t) ue(t)
R R

us(t)
L C L C

ue(t) us(t) ue(t)

R R

30
 1.4.2. Qualitative analysis of second order filters

us(t) us(t)

L L C
C Þ Low-pass Þ High-pass
ue(t) filter ue(t) R
R filter

us(t) us(t) ωà0 ωà∞ us(t)

ωà0 ωà∞ us(t)

L C L C L C L C
ue(t) R ue(t) R ue(t) R ue(t) R

us(t) = ue(t) us(t) = 0 us(t) = 0 us(t) = ue(t)

us(t) ωà0 us(t)

L
C Þ Band-pass L C
ue(t) us(t) filter L
R C ue(t) R
ue(t) R
us(t) = ue(t)
ωà0 ωà∞ us(t)
ωà∞
L C L C
us(t) us(t) Þ Notch filter L C
ue(t) R ue(t) R
ue(t) R
us(t) = 0 us(t) = 0 31
us(t) = ue(t)
31