1 Definitions and Symbols VI = V I e j I = V I I is the vector quantity corresponding to an input sinusoidal waveform ) 2 cos( ) ( I I I t f V t VI ; V I is the amplitude of the input sinusoidal signal; I is the phase of the input sinusoidal signal. VO = V O e j O = V O O is the vector quantity corresponding to the output sinusoidal waveform ) 2 cos( ) ( O O O t f V t VO ; V O is the amplitude of the output sinusoidal signal; O is the phase of the output sinusoidal signal. G(jw) is the transfer function of the filter, which depends on the input sinusoidal frequency f I =w I /2. w is the angular frequency of the input sinusoidal signal. G(jw) amplitude is |G(jw)| while its phase is G(jw) : G(jw)= |G(jw)| G(jw) =RC is the time constant of the first order RC filter f C is the cutoff frequency of the filter, which defines the pass band B of the filter. B is the pass band of the filter and it is the range of frequencies for which the input sinusoidal signal is not significantly attenuated. Bw is the bandwidth of the filter and it is the width of its pass band B. A Decade is the distance between two major ticks in a frequency logarithmic scale axis. Decibel (dB) is a measurement unit defined as 20*Log of the value of the quantity to be measured. Log is the logarithmic with base 10. The Bode plot is the transfer function amplitude diagram in full logarithmic scale: both |G| (transfer function amplitude) axis and f (frequency) axis are represented in logarithmic scale. OPAMP stands for OPerational AMPlifier, and it is an integrated circuit (IC) used to realize active filters.
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
1
Definitions and Symbols
VI = VIe
jI = VI
I is the vector quantity corresponding to an input sinusoidal waveform
)2cos()( III tfVtVI ; VI is the amplitude of the input sinusoidal signal; I is the phase
of the input sinusoidal signal.
VO = VOejO = VO
O is the vector quantity corresponding to the output sinusoidal waveform
)2cos()( OOO tfVtVO ; VO is the amplitude of the output sinusoidal signal; O is the
phase of the output sinusoidal signal.
G(jw) is the transfer function of the filter, which depends on the input sinusoidal frequency
fI=wI/2. w is the angular frequency of the input sinusoidal signal. G(jw) amplitude is |G(jw)|
while its phase is G(jw): G(jw)= |G(jw)| G(jw)
=RC is the time constant of the first order RC filter
fC is the cutoff frequency of the filter, which defines the pass band B of the filter.
B is the pass band of the filter and it is the range of frequencies for which the input sinusoidal
signal is not significantly attenuated.
Bw is the bandwidth of the filter and it is the width of its pass band B.
A Decade is the distance between two major ticks in a frequency logarithmic scale axis.
Decibel (dB) is a measurement unit defined as 20*Log of the value of the quantity to be
measured. Log is the logarithmic with base 10.
The Bode plot is the transfer function amplitude diagram in full logarithmic scale: both |G|
(transfer function amplitude) axis and f (frequency) axis are represented in logarithmic scale.
OPAMP stands for OPerational AMPlifier, and it is an integrated circuit (IC) used to realize
active filters.
2
1
Lesson 1: RC low pass filter schematic and transfer function (2hrs) 2
3
Objectives 4
To know the schematic of an RC low-pass filter. 5
To calculate the transfer function of an RC low-pass filter. 6
To calculate the time constant of an RC low-pass filter. 7
To draw the schematic of a circuit with CAD tools in English. 8
9
10
Today we're going to study a new circuit: the RC low-pass filter in the frequency domain. We‟ll start 11
with the schematic of the filter, shown in figure 1.1. 12
13 Fig. 1.1: schematic of a first-order passive RC low pass filter. 14
15
We are going to study the frequency response of this simple circuit. Since we are interested in the 16
frequency response the input voltage VI is a sinusoidal waveform with amplitude VI, frequency fI and 17
phase I: )2cos( III tfVVI . The output voltage VO is sinusoidal too with frequency fI because 18
there are only linear components in the circuit (a resistor R and a capacitor C) that cannot cause 19
distortion. So VO is a sinusoidal waveform with amplitude VO, phase O and the same frequency fI as 20
VI: )2cos( OIO tfVVO . We can represent both VI(t) and VO(t) with the corresponding vector 21
quantities: VI=VIejI and VI=VOe
jO. Assuming the input signal VI is known (the circuit stimulus) the 22
only unknown variables of the output signal VO are its amplitude VO and its phase O. How can we 23
find the mathematical relation between VO and VI and O and I? 24
25
The RC filter can be seen as an impedance divider, as shown in figure 1.2: the input signal VI is 26
divided between the resistor impedance (Z1=R) and the capacitor impedance (Z2=1/jwC). 27
VO(jw) VI(jw)
R
C
stimulus or input sinusoidal voltage
output sinusoidal voltage
3
1 Fig. 1.2: the schematic of a first-order passive RC low pass filter can be seen as an impedance divider. 2
3
Thus, applying the simple impedance divider formula we get: 4
Eq. 1.1) VIZR
ZVO
2
2 , whereCfjCwj
Z
2
112 is the capacitor impedance. As it can 5
be seen, Z2 depends on w=2f, the frequency of the input signal. 6
7
The transfer function G(jw) of the RC filter is defined as: 8
Eq. 1.2) VI
VOjwG )( . 9
Note that by definition )()()( jwVIjwGjwVO . We can therefore determine the amplitude and phase 10
of VO(jw) using the properties of complex numbers: 11
Eq. 1.3.a) Amplitude of VO(jw): IO VjwGV )( (the amplitude of the output is the product of the 12
amplitude of the input and the amplitude of the transfer function at the stimulus given frequency). 13
Eq. 1.3.b) Phase of VO(jw): IO jwG )](arg[ , where arg[G(jw)]=G(jw), is the phase of the 14
transfer function (the phase of the output is the sum of the phase of the input and the phase of the 15
transfer function at the stimulus given frequency). 16
Using equations 1.3 a) and b) we can calculate VO and O once VI, I and G(jw) are known. Note that 17
VI, I are the given inputs, so the only unknown variable of our problem is G(jw). However, using 18
equations 1.1 and 1.2 G(jw) is given by: 19
Eq. 1.4) 2
2)(ZR
Z
VI
VOjwG
20
By substituting Z2=1/jwC into equation 1.4 we get: 21
1
1
1
11
1
1
)(
jwRC
jwCR
jwC
jwCR
jwCjwG ; =RC is defined as the time constant of the RC 22
filter; hence we can write the transfer function of the low-pass RC filter as: 23
Eq. 1.5)
fjjw
jwG21
1
1
1)( 24
VO(jw)
VI(jw)
Z1=R
Z2=1/jwC
4
Equation 1.5 shows that G(jw) is a complex number whose value depends only on the filter time 1
constant and the frequency f of the input signal. From equation 1.5, using the properties of complex 2
numbers, we can easily calculate the amplitude )( jwG and the phase G(jw)= arg[G(jw)] of the filter 3
transfer function: 4
Eq. 1.6.a) 2)(1
1)(
wjwG , 5
Eq. 1.6.b) )()( wartgjwG 6
By substituting equation 1.6.a into 1.3.a and equation 1.6.b into 1.3.b we reach our final resolution 7
formulas: 8
Eq. 1.7.a) 22 )(1)(1
1
w
VV
wV I
IO 9
Eq. 1.7.b) )()( wartgwartg IIO 10
VI, I and w=2f are the amplitude, phase and angular frequency of the input signal and are all known 11
quantities. =RC depends on the RC filter components and can be calculated from the filter schematic. 12
Hence, eq. 1.7.a and 1.7.b are the resolution formulas that allow us to calculate amplitude and phase of 13
the output for any given input and any given RC filter. 14
For instance, let‟s consider an RC low pass filter with R=1K and C=159.24nF. We can easily 15
determine its constant time sss 3693 1015924.01024.159101024.1591 , i.e. about 16
0,16ms. Now let‟s consider two different cases: 17
Case A) input signal VI with amplitude 1V, phase 0 degrees (00) and frequency 1KHz; 18
Case B) input signal VI with amplitude 1V, phase 0 degrees (00) and frequency 15KHz. 19
In case A the input frequency is 1KHz, hence the value of the amplitude of the transfer function is (see 20
eq. 1.6.a): 21
707,02
1
)005,1(1
1
)16,02(1
1
)16,012(1
1
)(1
1)(
2222
msKHzwjwG22
Therefore, the amplitude of the output voltage is VVVV IO 06.15.1707.0707.0 . 23
The phase of the transfer function at 1KHz is instead (see eq. 1.6.b): 24 0
)( 45)005.1()16.012()( artgmsKHzartgwartgjwG . 25
Therefore, the phase of the output voltage is 000 45450 O . 26
Considering now case B the input frequency is 15KHz, so the values of the amplitude and of the phase 27
of the transfer function are: 28
066.0)08.15(1
1
)16.0152(1
1)(
22
msKHzjwG
29
0
)( 2.86)08,15()( artgwartgjwG . 30
Consequently the amplitude of the output voltage is VVVV IO 15.1066.0066.0 , while the 31
phase of the output voltage is 000 2.862.860 O . 32
Of course different results are obtained if we change input frequency (w value changes) and/or the 33
resistor or the capacitor of the RC filter ( value changes). 34
35
English is an important language for electronics since modern electronics was born in the United States 36
with the invention of the transistor in 1958. English is the common language of the electronic engineers 37
community and all electronics components are described by datasheets written in English. Commonly 38
5
used CAD (Computer Aided Design) tools, useful to design electronic circuits, are in English language, 1
too. Hence, the knowledge of technical English is a necessary prerequisite to design any innovative 2
electronic circuit. Just to supply an example we will now use a CAD tool called LTSpiceIV that can be 3
downloaded freely from the Internet (http://www.linear.com/designtools/software) to build the 4
schematic of the RC low-pass filter we have just analyzed. 5
6
7
6
1
Lesson 2: Frequency response graphical representation. (2 hrs) 2
3
Objectives 4
To be able to use the frequency response graphs of a filter to calculate the output voltage. 5
To know the cutoff frequency (fC) and the low-pass filter pass-band (B) and the relationship between 6
fC and . 7
To be able to recognize and draw the graphs of the frequency response of a low-pass filter in 8
logarithmic scale. 9
To be able to design a first-order RC low-pass filter with a given pass-band B. 10
To draw logarithmic and semilogarithmic graphs of the frequency response of an RC low-pass filter 11
with MS-excel. 12
13
14
In the previous lesson we found the algebraic expressions of the amplitude and phase of the transfer 15
function G(jw) of a first order RC low-pass filter: 16
Eq. 2.1) 2)(1
1)(
wjwG , )()( wartgjwG 17
Both functions depend on the frequency
2
wf of the input signal VI. We also showed how to 18
calculate VO using Eq. 2.1). 19
In this lesson we will show a more straightforward approach for the calculation of VO, i.e. the 20
graphical approach. By definition of transfer function it is: 21
Eq. 2.2) IO VjwGV )( , )( jwGIO 22
The graphical approach uses Eq. 2.2) but does not calculate |G(jw)|and G(jw) analytically. On the other 23
hand, the amplitude and phase of the transfer function at the given input frequency are directly read 24
from the amplitude and phase diagrams of the filter. The amplitude diagram shows |G(jw)| vs. f while 25
the phase diagram shows G(jw) vs. f. An example of amplitude and phase plots for an RC low pass 26
filter with =1ms is given in figure 2.1. Hence, once we know the amplitude and phase plots of a given 27
filter it is very easy to determine the value of |G(jw)| and G(jw) graphically from the plot at the given 28
input frequency without any analytical calculation (without calculating Eq. 2.1). 29
For instance, let us assume the input is a sinusoidal waveform with the following characteristics: VI 30
=1.5V, I=00 and f=100Hz. 31
From the amplitude plot at f=100Hz we read |G|=0.8 while from the phase plot at 100Hz we read G=-32
350. 33
Hence, using Eq. 2.2 we derive that VO=0.8*1.5V=1.2V while O=00-35
0=-35
0. 34
Let‟s assume now that the same input sinusoidal waveform has a higher frequency, 1KHz instead of 35
100Hz. At 1KHz we read |G|=0.18 and G=-810 from the amplitude and phase plots, respectively. 36
Hence at 1KHz the output of the filter is given by VO=0.18*1.5V=0.27V while O=00-81
0=-81
0. 37
The examples above show two clear advantages of the graphical method: 38
1) it is valid for any filter topology: we need only to know the amplitude and phase plots 39
associated to the filter to determine its output voltage. 40
2) output voltage is very easily calculated for any input frequency since we need only to solve Eq. 41
2.2 where |G| and G are directly read from the filter plots. 42
43
44
7
1
2
3
4
5
6
7
8
9
10
11
12
Fig. 2.1: examples of amplitude and phase plots for a first order RC low-pass filter with =1ms. 13
14
Let‟s now analyze more in detail the filter amplitude diagram (|G| vs. f), shown in figure 2.2: 15
16
17
18
19
20
21
22
23
24
25
26
27
28
29 Fig. 2.2: example of amplitude diagram of a passive first order RC filter. Note the cutoff frequency fC 30
can be easily read and that the frequency scale of the diagram is logarithmic. Note also the pass band 31
(B) of the filter. 32
33
The cutoff frequency fC of the filter is defined as the frequency value at which |G|=0.7. So when input 34
signal frequency f is less than the cutoff frequency (f<fC) the output signal amplitude is equal or only 35
slightly lower than input amplitude (no attenuation or only small attenuation of the input). On the 36
contrary, when f>fC the output signal amplitude is significantly lower than input amplitude (significant 37
attenuation). To sum up we can roughly approximate an ideal low-pass filter behavior as follows: 38
- input with f<fC no attenuation (the input signal passes) VO=VI 39
- input with f>fC great attenuation (the input signal is cut) VO0 40
In fact, a low pass filter is typically used to eliminate (attenuate) all high frequency components of the 41
input signals. Its fC represents the frequency threshold at which the filter starts to be active. 42
The pass band B of the filter is the range of frequencies for which |G| 0.7, i.e. the input signal is not 43
significantly attenuated (the input signal “passes”). Of course, it is B= [0,fC] for the low pass RC filter. 44
The bandwidth Bw of the filter is the width of the pass band B, hence: Bw=fC-0=fC. 45
The following relationship between fC and can be easily demonstrated (do it yourself to exercise!): 46
47
Eq. 2.3) 48
Amplitude diagram
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
0,1
1 10
10
0
10
00
10
00
0
10
00
00
10
00
00
0frequency
|G(j
w)|
Phase diagram
-90
-75
-60
-45
-30
-15
0
0,1
1 10
10
0
10
00
10
00
0
10
00
00
10
00
00
0frequency
G(j
w)
0.8
0.18
-350
-810
Amplitude diagram
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
0,1
1 10
10
0
10
00
10
00
0
10
00
00
10
00
00
0frequency
|G(j
w)|
fC=160Hz
|G|=0,7
Note the
frequency
scale is
logarithmic
1Decade
B=[0,fC]
CRfC
2
1
2
1
8
1
Finally, note the frequency scale of the plot is logarithmic and not linear. In fact, the distance between 2
0.1 and 1Hz is the same between 100 and 1000Hz. The frequency logarithmic scale allows to read the 3
values of the whole amplitude waveform for a wide range of frequencies. The distance between f and 4
10*f in the logarithimic scale is called a Decade. So, when we move from 1Hz to 10Hz we increase the 5
frequency of a Decade. When we move from 1Hz to 100Hz we increase the frequency of 2 Decades. 6
Figure 2.3 shows the comparison between a linear and a logarithmic scale: the logarithmic scale is 7
much wider than the linear one. 8
9
10 Fig. 2.3: comparison between logarithmic and linear frequency axis scale. In the linear scale the 11
distance between two ticks is constant =1Hz. In the logarithmic scale the distance between 10*f and f is 12
constant = 1 Decade. Hence, the logarithmic scale features a much wider range of frequency values. 13
14
Figure 2.4 shows the comparison between the same two amplitude diagrams, one with logarithmic 15
frequency scale and the other with linear frequency scale. The frequency axis of the two plots are the 16
same (from 0.1Hz to 1000000Hz=1MHz). Only the plot with logarithmic scale can be used to read the 17
transfer function amplitude values at different frequencies. 18
19 Fig. 2.4: the same two amplitude diagrams with and without logarithmic scale are compared. The 20
values of fC and of the transfer function amplitude can be read only from the logarithmic plot. 21
22
Figure 2.5 shows the filter phase diagram (G vs. f). We can divide the plot into three different 23
frequency ranges: 24
- f<<fC G=00, the filter does not introduce any phase shift, i.e. the output waveform is not 25
delayed with respect to the input. 26
- f=fC G=-450, at the cutoff frequency the filter phase shift is -45
0, corresponding to a 27
delay=0.125/fC of the output waveform with respect to the input. 28
- f>>fC G=-900, the filter introduces the maximum phase shift. The output waveform exhibits 29
the maximum delay with respect to the input =0.25/f. 30
Finally, we can roughly approximate an ideal low-pass filter behavior as follows: 31
- input with f<fC no delay (the input signal passes without any delay) O=I 32
Amplitude diagram
0
0,2
0,4
0,6
0,8
1
0,1
1 10
100
1000
10000
100000
1000000frequency
|G(j
w)|
fC=160Hz
Amplitude diagram
0
0,2
0,4
0,6
0,8
1
0,1
100000
200000
300000
400000
500000
600000
700000
800000
900000
1000000
frequency
|G(j
w)|
linear scale
log. scale
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1Hz
1 2 3 4 5 6 7 8
9
10
1 decade
100
16 17
200
18 19 20 21 22 23 24 25 26
20
linear scale
logarithmic scale
9
- input with f>fC 900 delay (the input signal is delayed) O=I-90° 1
2
3 Fig. 2.5: example of phase diagram of a passive first order RC filter. The cutoff frequency fC is the 4
frequency value at which G=-450. The frequency scale is logarithmic. 5
6
Let‟s now rewrite Eq. 2.2 for the calculation of the output voltage of a filter: 7
Eq. 2.4) IO VjwGV )( . 8
The equation becomes dimensionless if we divide both its members by 1V: 9
Eq. 2.5) V
VjwG
V
V IO
1)(
1 . 10
We can rewrite equation 2.5 in decibel (dB) by applying the function 20*Log() to both its members: 11
Eq. 2.6) )1
)((20)1
(20V
VjwGLog
V
VLog IO . 12
Remembering the properties of the logarithm Log(a*b)=Log(a)+Log(b) we can rewrite equation 2.6 as 13
follows: 14
Eq. 2.7) )1
(20))((20)1
(20V
VLogjwGLog
V
VLog IO . 15
Since by definition of dB it is: OO VV
VLog )
1(20 [dBV], |)(|))((20 jwGjwGLog [dB] and 16
II V
V
VLog )
1(20 [dBV], equation 2.7 gives the resolution formula to calculate the filter output in dB: 17
Eq. 2.8) ][][)(][ dBVVdBjwGdBVV IO . 18
The filter output voltage amplitude (VO) in dBV is equal to the filter input voltage amplitude (VI) in 19
dBV plus the value in dB of the filter transfer function amplitude (|G|) in dB calculated at the given 20
input frequency f=w/2. 21
Eq. 2.8 is almost the same of Eq. 2.2: the only difference is that the product in Eq. 2.2 becomes a sum 22
in Eq. 2.8. Since the addition is a mathematical operation much simpler than the multiplication, 23
engineers prefer to use Eq. 2.8 rather than Eq. 2.2. 24
This means that engineers prefer to plot |G(jw)| in dB vs. frequency. When |G| is expressed in dB the 25
transfer function amplitude plot of figure 2.2 becomes as shown in figure 2.6: 26
Phase diagram
-90
-75
-60
-45
-30
-15
0
0,1
1 10
10
0
10
00
10
00
0
1E
+0
5
1E
+0
6frequency
G(j
w)
frequency scale is logarithmic
fC=160Hz
10
1 Fig. 2.6: this is the same amplitude diagram of figure 2.2 with the only difference that the amplitude |G| 2
is expressed in dB. This is a full logarithmic plot and it is called Bode plot of the filter. 3
4
The transfer function amplitude (|G|) plot in dB is called the Bode plot of the filter (any system 5
described by a transfer function has its own Bode plot). The Bode plot is a full logarithmic scale plot, 6
since the scale in dB=20*Log() is logarithmic too. 7
The cutoff frequency fC in a Bode plot is the frequency at which |G|[dB]=-3dB. In fact, when f=fC 8
|G|=0.7 and hence |G|[dB]=20*Log(0.7)=-3 decibel. 9
When f>fC the Bode plot can be approximated with a straight line of slope -20dB/decade. In fact, 10
|G(jw)|[dB] decreases of 20dB for each increase of a factor of 10 of the frequency, i.e. for each decade 11
of increase of frequency. Thanks to this property the Bode plot graphical drawing is very simple since 12
it can be approximated with only two straight lines: 13
1) f<fC |G(jw)|=0 (horizontal line along x-axis) 14
2) f<fC |G(jw)|=-20*Log(f/fC) (straight line with slope -20dB/dec). 15
Once the filter is given, fC can be calculated using Eq. 2.3. When fC is known the drawing of the Bode 16
plot of the filter is straightforward, as shown in figure 2.7: 17
18 Fig. 2.7: example of simplified Bode plot drawing of a first order RC low pass filter. The only 19
significant approximation of this plot is that |G(fC)|=0dB while it is -3dB in reality. 20
21
Amplitude diagram in dB (Bode Plot)
-80
-70
-60
-50
-40
-30
-20
-10
0
0,1
1 10
10
0
10
00
10
00
0
10
00
00
10
00
00
0frequency
|G(j
w)|
[d
B]
fC=160Hz
1dec
20dB
-3dB
|G(jw)| [dB]
f (log)
fC 10fC 100fC 1000fC 0dB
-20dB
-40dB
-60dB
11
In a few words, the Bode plot of the filter presents the following advantages with respect to the 1
traditional plot with a linear scale along |G| axis: 2
1) very simple to draw using the -20dB/dec rule. 3
2) VO can be calculated as a sum of VI and the |G| value read from the plot. 4
For instance, let‟s consider an input signal of amplitude =0dBV (corresponding to 1V amplitude) at the 5
frequency f=10000Hz=10KHz. From the Bode plot of Fig. 2.5 we read that |G|=-35dB when f=10KHz. 6
Hence, we can immediately calculate VO=0dBV-35dB=-35dBV using equation 2.8). Engineers of some 7
years ago were familiar with voltage values in dB. For us it is a little more difficult to associate a 8
voltage value to -35dBV. In this case it is sufficient to remember the inverse formula: 9
Eq. 2.9) 20
][
10][
dBVV
O
O
VV , 10
from which VO=10-1.75
V=0.018V (remember these general rules: 0dBV corresponds to 1V. when dBV 11
values are negative it means voltage amplitude is between 1 and 0. The positive dB values are for all 12
voltages >1V). 13
14
Eq. 2.3) is used to calculate fC when the filter is known. However, it is often required to design a filter 15
for a given fC or pass band B. First of all we can use the inverse formula of Eq. 2.3) to calculate the 16
time constant of the filter: 17
18 Eq. 2.10) . 19
20
Then we can design the filter following the four steps algorithm proposed below: 21
22
1) calculate using equation 2.10) 23
24
2) choose a capacitor among those available in the laboratory, for instance C=1F 25
26
3) calculate the corresponding value of R according to the formula: 27
28
4) if R is within the range [100,100K] OK, otherwise go back to point 2) choosing another 29
C value 30
31
It is suggested to choose the capacitor value among those available because almost all resistors values 32
are commonly available. It is recommended not to have R<100 because parasitic resistors may affect 33
filter parameters. R>100K is not recommended either because the filter impedance may become 34
comparable to that of multimeters, thus affecting filter behavior by multimeter insertion. 35
For instance, the desing of a filter with fC =1KHz can be done as follows: 36
37
1) ; 2) C=1F; 3) ; 4) R[100,100K]OK! 38
39
And now we‟ll conclude the lesson by showing how to use an electronic sheet as MS-excel to draw 40
amplitude and phase plots of first order low pass RC filters. The starting point is equation 2.1) with the 41
analytical expression of |G| and G. MS-excel is used to calculate those analytical expressions at the 42
various frequency values of interest and to plot the corresponding graphs. 43
44
BfC
2
1
2
1
CR
sfC
41059.12
1
15910
1059.16
4
F
sR
12
1 2
13
1
Lesson 3: Analysis of the CR high pass filter. (2 hrs) 2
3
Objectives 4
5
To be able to draw the schematic of a high-pass CR filter and to know its transfer function. 6
To be able to draw amplitude and phase diagrams of the high-pass filter. 7
To know the relationship among the time constant (), the cutoff frequency (fC) and the filter pass 8
band (B). 9
To be able to calculate the output voltage with the graphical method. 10
To be able to design the filter with a given cutoff frequency fC. 11
To be able to simulate the filters with CAD tools in English. 12
13
14
In the previous lesson we saw how to represent the transfer function of a low pass RC filter graphically. 15
Figure 3.1 shows the schematic of a first-order passive CR high pass filter: 16
17 Fig. 3.1: schematic of a first order passive CR high pass filter. 18
19
The schematic is very similar to that of a low pass filter (compare figure 3.1 with figure 1.1 of lesson 20
1). The only difference is that R and C are interchanged. So we can follow the same approach used for 21
the RC low pass filter to determine the high pass filter transfer function: 22
Eq. 3.1) 21
2
)(
)(|)(|
ZZ
Z
jwV
jwVjwG
I
O
, 23
where the only difference with respect to the low pass filter is that: Z1=1/jwC while Z2=R (Z1 and Z2 24
are interchanged). Substituting Z1 and Z2 expression into equation 3.1 after a few mathematical 25
passages we get the analytical expression of the high pass filter transfer function G(jw): 26
27
Eq. 3.2) 28
29
is the time constant of the filter and it is defined as in the low pass filter: =RC. Using the properties 30
of complex numbers, we can calculate the amplitude and phase of G(jw) from equation 3.2: 31
32
Eq. 3.3) , . 33
C
1µ
R
159
V1
SINE(0 1 1K)
OI
VO(jw) VI(jw)
VI=VIeji
VO=VOejo
wj
wjjwG
1)(
2)(1)(
w
wjwG )(900
)( wartgjwG
14
Equation 3.3 can be used to calculate the output of the filter once the input is known. However, 1
equation 3.3 can also be used to represent the frequency response diagrams of the filter graphically by 2
means of MS-excel, as shown in the previous lesson. Figure 3.2 shows the amplitude and phase 3
diagrams of a CR high pass filter with =0.16ms obtained using equation 3.3 and MS-excel: 4
5 Fig. 3.2: amplitude and phase diagrams of a CR high pass filter with =0.16ms. 6
7
The filter cutoff frequency fC is defined as the frequency at which |G|=-3dB (corresponding to a 0.7 8
factor of amplitude attenuation), as for the low pass filter. However note that for the high pass filter 9
when f= fC it is G=450 and not -45
0 as for the low pass filter. This means that the high pass filter 10
output is phase led with respect to the input, and not delayed as occurs in the low pass filter case. 11
The high pass filter pass band B is defined as the frequency range for which |G|>-3dB (or 0.7), i.e. the 12
frequency range where filter amplitude attenuation can be considered negligible. This is exactly the 13
same definition used in the low pass filter case. However, B=[fC,] for the high pass filter, as it can be 14
seen from figure 3.2. Consequently, the bandwidth Bw (i.e. the width of B) of the filter =. 15
Another important property of the Bode plot of the high pass filter is that |G| increases of 20dB for each 16
decade of increase of frequency (i.e. for a factor 10 of frequency increase). 17
With regard to the phase plot, the maximum phase shift of +900 occurs when f<<fC and corresponds to 18
an output voltage with 0.25/f lead with respect to input voltage. On the contrary when f>>fC there is no 19
phase delay. 20
In conclusion, input „high‟ frequencies (where „high‟ means higher than filter fC) are not significantly 21
attenuated nor phase shifted – the output and the input of the filter are practically the same. On the 22
contrary, input „low‟ frequencies (lower than fC) are attenuated and phase led, so that the output 23
amplitude is negligible and it is phase led with respect to the input. In a high pass filter, input high 24
frequencies “pass”, input low frequencies are “cut”. 25
The relationship between and fC is the same as the low pass filter: 26
Eq. 3.4) , CR
fC
2
1
2
1 27
28 Hence, the design of the high pass filter can be done exactly the same way as that of a high pass filter 29
(the four step algorithm): 30
31
1) calculate corresponding to the required fC using equation 3.4) 32
33
2) choose a capacitor among those available in the laboratory, for instance C=1F 34
35
3) calculate the corresponding value of R according to the formula: 36
37
Phase diagram
0
15
30
45
60
75
90
0,1
1 10
100
1000
10000
100000
1000000frequency
G(j
w)
Amplitude diagram in dB (Bode Plot)
-80
-70
-60
-50
-40
-30
-20
-10
0
0,1
1 10
10
0
10
00
10
00
0
10
00
00
10
00
00
0frequency
|G(j
w)|
[d
B]
fC=1KHz 1dec
20dB B=[fC,]
fC=1KHz
G(fc)=+450
|G(fC)| = -3dB
Cf
2
1
CR
15
4) if R[100,100K] OK, otherwise go back to point 2) choosing another C value 1
2
The Bode plot of the filter is easy to draw, too. In fact, we can approximate it as two straight lines: 3
1) one horizontal line when f>fC: |G(f)|=0dB (=1) when f>fC 4
2) one straight line with slope 20dB/dec when f< fC: |G(f)|=20*Log(f/ fC) 5
Figure 3.3 shows the Bode plot of a generic high pass filter obtained following the two rules above: 6
7 Fig. 3.3: Bode plot drawing of a first order high pass filter. The only significant approximation is 8
|G|=0dB when f=fC (|G(fC)|=-3dB in reality). 9
10
Let‟s now show some examples of output calculation for the high pass filter of figure 3.1. We‟ll 11
calculate the ouput using the graphical method, based on the knowledge of the two graphs of figure 3.2 12
and of the output resolution formulas found in lesson 2 (valid independently of filter topology and 13