Contents
Given a transfer function, such as
the question naturally arises: "How can we display this function?" In the previous document the argument was made that the most useful way to display this function is with two plots, the first showing the magnitude of the transfer function and the second showing its phase. One way to do this is by simply entering many values for the frequency, calculating the magnitude and phase at each frequency and displaying them. This is what a computer would naturally do. For example if you use MATLAB® and enter the commands
>> MySys=tf(100*[1 1],[1 110 1000]) Transfer function: 100 s + 100 ------------------------------ s^2 + 110 s + 1000 >> bode(MySys)
you get a plot like the one shown below. The asymptotic solution is given elsewhere.
However, there are reasons to develop a method for drawing Bode diagrams manually. By drawing the plots by hand you develop an understanding about how the locations of poles and zeros effect the shape of the plots. With this knowledge you can predict how a system behaves in the frequency domain by simply examining its transfer function. On the other hand, if you know the shape of transfer function that you want, you can use your knowledge of Bode diagrams to generate the transfer function.
The first task when drawing a Bode diagram by hand is to rewrite the transfer function so that all the poles and zeros are written in the form (1+s/ω_{0}). The reasons for this will become apparent when deriving the rules for a real pole. A derivation will be done using the transfer function from above, but it is also possible to do a more generic derivation. Let's rewrite the transfer function from above.
Now lets examine how we can easily draw the magnitude and phase of this function when s=jω.
First note that this expression is made up of four terms, a constant (0.1), a zero (at s=-1), and two poles (at s=-10 and s=-100). We can rewrite the function (with s=jω) as four individual phasors.
We will show (below) that drawing the magnitude and phase of each individual phasor is fairly straightforward. The difficulty lies in trying to draw the magnitude and phase of H(jω). We can write H(jω) as a single phasor:
Drawing the phase is fairly simple. We can draw each phase term separately, and then simply add them. The magnitude term is not so straightforward because of the fact that the magnitude terms are multiplied, it would be much easier if they were added - then we could draw each term on a graph and just add them. A method for doing this is outlined below.
One way to transform multiplication into addition is by using the logarithm. Instead of using a simple logarithm, we will use a deciBel (named for Alexander Graham Bell). (Note: Why the deciBel?) The relationship between a quantity, Q, and its deciBel representation, X, is given by:
So if Q=100 then X=40; Q=0.01 gives X=-40; X=3 gives Q=1.41; and so on.
If we represent the magnitude of H(s) in deciBels we get
The advantage of using deciBels (and of writing poles and zeros in the form (1+s/ω_{0})) are now revealed. The fact that the deciBel is a logarithmic term transforms the multiplication of the individual terms to additions. Another benefit is apparent in the last line that reveals just two types of terms, a constant term and terms of the form 20log10(|1+jω/ω_{0}|). Plotting the constant term is trivial, however the other terms are not so straightforward. These plots will be discussed below. However, once these plots are drawn for the individual terms, they can simply be added together to get a plot for H(s).
If we look at the phase of the transfer function, we see much the same thing: The phase plot is easy to draw if we take our lead from the magnitude plot. First note that the transfer function is made up of four terms. If we want
Again there are just two types of terms, a constant term and terms of the form (1+jω/ω_{0}). Plotting the constant term is trivial; the other terms are discussed below.
The discussion above dealt with only a single transfer function. Another derivation that is more general, but a little more complicated mathematically is here.
Following the discussion above, the way to make a Bode Diagram is to split the function up into its constituent parts, plot the magnitude and phase of each part, and then add them up. The following gives a derivation of the plots for each type of constituent part. Examples, including rules for making the plots follow in the next document, which is more of a "How to" description of Bode diagrams.
Consider a constant term,
Clearly the magnitude is constant
The phase is also constant. If K is positive, the phase is 0° (or any even multiple of 180°). If K is negative the phase is -180°, or any odd multiple of 180°. We will use -180° because that is what MATLAB® uses. Expressed in radians we can say that if K is positive the phase is 0 radians, if K is negative the phase is -π radians.
Consider a simple real pole
The frequency ω_{0} is called the break frequency, the corner frequency or the 3 dB frequency (more on this last name later).
The magnitude is given by
Let's consider three cases for the value of the frequency:
Case 1) ω<<ω_{0}. This is the low frequency case. We can write an approximation for the magnitude of the transfer function
The low frequency approximation is shown in blue on the diagram below.
Case 2) ω>>ω_{0}. This is the high frequency case. We can write an approximation for the magnitude of the transfer function
The high frequency approximation is at shown in green on the diagram below. It is a straight line with a slope of -20 dB/decade going through the break frequency at 0 dB. That is, for every factor of 10 increase in frequency, the magnitude drops by 20 dB.
Case 3) ω=ω_{0}. The break frequency. At this frequency
This point is shown as a red circle on the diagram.
To draw a piecewise linear approximation, use the low frequency asymptote up to the break frequency, and the high frequency asymptote thereafter.
The resulting asymptotic approximation is shown highlighted in pink. The maximum error between the asymptotic approximation and the exact magnitude function occurs at the break frequency and is approximately 3 dB.
The rule for drawing the piecewise linear approximation for a real pole can be stated thus:
For a simple real pole the piecewise linear asymptotic Bode plot for magnitude is at 0 dB until the break frequency and then drops at 20 dB per decade (i.e., the slope is -20 dB/decade).
The phase of a single real pole is given by is given by
Let us again consider three cases for the value of the frequency:
Case 1) ω<<ω_{0}. This is the low frequency case. At these frequencies We can write an approximation for the phase of the transfer function
The low frequency approximation is shown in blue on the diagram below.
Case 2) ω>>ω_{0}. This is the high frequency case. We can write an approximation for the phase of the transfer function
The high frequency approximation is at shown in green on the diagram below. It is a straight line with a slope at -90°.
Case 3) ω=ω_{0}. The break frequency. At this frequency
This point is shown as a red circle on the diagram.
A piecewise linear approximation is not as easy in this case because the high and low frequency asymptotes don't intersect. Instead we use a rule that follows the exact function fairly closely, but is also arbitrary. Its main advantage is that it is easy to remember. The rule can be stated as
Follow the low frequency asymptote until one tenth the break frequency (0.1 ω_{0}) then decrease linearly to meet the high frequency asymptote at ten times the break frequency (10 ω_{0}). This line is shown above. Note that there is no error at the break frequency and about 5.7° of error at one tenth and ten times the break frequency.
The second example shows a double pole at 30 radians per second. Note that the slope of the asymptote is -40 dB/decade and the phase goes from 0 to -180°.
There is another approximation for phase that is commonly used. The approximation is developed by matching the slope of the actual phase term to that of the approximation at ω=ω_{0}. Using math similar to that given here (for the underdamped case) it can be shown that by drawing a line starting at 0° at ω=ω_{0}/e^{π/2}=ω_{0}/4.81 (or ω_{0}·e^{-π/2}) to -90° at ω=ω_{0}·4.81 we get a line with the same slope as the actual function at ω=ω_{0}. This approximation is slightly easier to remember as a line drawn from 0° at ω_{0}/5 to -90° at ω_{0}·5. The latter is shown on the diagram below.
Although this method is more accurate near ω=ω_{0} there is a larger maximum error (more than 10°) near ω_{0}/5 and ω_{0}·5.
The piecewise linear approximation for a zero is much like that for a pole Consider a simple zero:
The development of the magnitude plot for a zero follows that for a pole. Refer to the previous section for details. The magnitude of the zero is given by
Again there are three cases:
The rule for drawing the piecewise linear approximation for a real zero can be stated thus:
For a simple real zero the piecewise linear asymptotic Bode plot for magnitude is at 0 dB until the break frequency and then increases at 20 dB per decade (i.e., the slope is +20 dB/decade).
The phase of a simple zero is given by:
The phase of a single real zero also has three cases:
The rule for drawing the phase plot can be stated thus:
Follow the low frequency asymptote until one tenth the break frequency (0.1 ω_{0}) then increase linearly to meet the high frequency asymptote at ten times the break frequency (10 ω_{0}).
This example shows a simple zero at 30 radians per second. The low frequency asymptote is the dashed blue line, the exact function is the solid black line, the cyan line represents 0.
A pole at the origin is easily drawn exactly. Consider
The magnitude is given by
This function is represented by a straight line on a Bode plot with a slope of -20 dB per decade and going through 0 dB at 1 rad/ sec. It also goes through 20 dB at 0.1 rad/sec, -20 dB at 10 rad/sec...
The rule for drawing the magnitude for a pole at the origin can be thus:
For a pole at the origin draw a line with a slope of -20 dB/decade that goes through 0 dB at 1 rad/sec.
The phase of a simple zero is given by:
The rule for drawing the phase plot for a pole at the origin an be stated thus:
The phase for a pole at the origin is -90°.
This example shows a simple pole at the origin. The black line is the Bode plot, the cyan line indicates a zero reference (dB or °).
A zero at the origin is just like a pole at the origin but the magnitude increases, and the phase is positive.
The magnitude and phase plots of a complex conjugate (underdamped) pair of poles is more complicated than those for a simple pole. Consider the transfer function:
The magnitude is given by
Let's consider three cases for the value of the frequency:
Case 1) ω<<ω_{0}. This is the low frequency case. We can write an approximation for the magnitude of the transfer function
The low frequency approximation is shown in red on the diagram below.
Case 2) ω>>ω_{0}. This is the high frequency case. We can write an approximation for the magnitude of the transfer function
The high frequency approximation is at shown in green on the diagram below. It is a straight line with a slope of -40 dB/decade going through the break frequency at 0 dB. That is, for every factor of 10 increase in frequency, the magnitude drops by 40 dB.
Case 3) ω≈ω_{0}. It can be shown that a peak occurs in the magnitude plot near the break frequency. The derivation of the approximate amplitude and location of the peak are given here. We make the approximation that a peak exists only when
0<ζ<0.5
and that the peak occurs at ω_{0} with height 1/(2·ζ).
To draw a piecewise linear approximation, use the low frequency asymptote up to the break frequency, and the high frequency asymptote thereafter. If ζ<0.5, then draw a peak of amplitude 1/(2·ζ) Draw a smooth curve between the low and high frequency asymptote that goes through the peak value.
As an example For the curve shown below,
The peak will have an amplitude of 5.00 or 14 dB.
The resulting asymptotic approximation is shown as a black dotted line, the exact response is a black solid line.
The rule for drawing the piecewise linear approximation for a complex conjugate pair of poles can be stated thus:
For the magnitude plot of complex conjugate poles draw a 0 dB at low frequencies, go through a peak of height,
.
at the break frequency and then drop at 40 dB per decade (i.e., the slope is -40 dB/decade). The high frequency asymptote goes through the break frequency.
The phase of a complex conjugate pole is given by is given by
Let us again consider three cases for the value of the frequency:
Case 1) ω<<ω_{0}. This is the low frequency case. At these frequencies We can write an approximation for the phase of the transfer function
The low frequency approximation is shown in red on the diagram below.
Case 2) ω>>ω_{0}. This is the high frequency case. We can write an approximation for the phase of the transfer function
The high frequency approximation is at shown in green on the diagram below. It is a straight line at -180°.
Case 3) ω=ω_{0}. The break frequency. At this frequency
The asymptotic approximation is shown below, followed by an explanation
A piecewise linear approximation is not easy in this case, and there are no hard and fast rules for drawing it. The most common way is to look up a graph in a textbook with a chart that shows phase plots for many values of ζ. Three asymptotic approximations are given here. We will use the approximation that connects the the low frequency asymptote to the high frequency asymptote starting at
and ending at
If ζ<0.02, the approximation can be simply a vertical line at the break frequency. One advantage of this approximation is that it is very easy to plot on semilog paper. Since the number 10·ω_{0} moves up by a full decade from ω_{0}, the number 10^{ζ}·ω_{0} will be a fraction ζ of a decade above ω_{0}. For the example above the corner frequencies for ζ=0.1 fall near ω_{0} one tenth of the way between ω_{0} and ω_{0}/10 (at the lower break frequency) to one tenth of the way between ω_{0} and ω_{0}·10 (at the higher frequency).
The rule for drawing phase of an underdamped pair of poles can be stated as
Follow the low frequency asymptote at 0° until
then decrease linearly to meet the high frequency asymptote at -180° at
.
at the break frequency and then drop at 40 dB per decade (i.e., the slope is -40 dB/decade). The high frequency asymptote goes through the break frequency. Note that the peak only exists for0 < ζ < 0.5
Not surprisingly a complex pair of zeros yields results similar to that for a complex pair of poles. The differences are that the magnitude has a dip instead of a peak, the magnitude increases above the break frequency and the phase increases rather than decreasing.
The graph below corresponds to a complex conjugate zero with
The dip in the magnitude plot will have a magnitude of 0.2 or -14 dB.
.
at the break frequency and then rise at +40 dB per decade (i.e., the slope is +40 dB/decade). The high frequency asymptote goes through the break frequency. Note that the peak only exists for0 < ζ < 0.5
Brief review of page: This document derived piecewise linear approximations that can be used to draw different elements of a Bode diagram. A synopsis of these rules can be found in a separate document.
© Copyright 2005 to 2015 Erik Cheever This page may be freely used for educational purposes.
Erik Cheever Department of Engineering Swarthmore College