UNIVERSITY OF CALIFORNIA AT BERKELEY College of Engineering Department of Electrical Engineering and Computer Sciences

EE105 Lab Experiments

Bode Plot Tutorial

Introduction

Although you should have learned about Bode plots in previous courses (such as EE40), this tutorial will give you a brief review of the material in case your memory is rusty.

Bode Plots Basics

Making the Bode plots for a transfer function involve drawing both the magnitude and phase plots. The magnitude is plotted in decibels (dB) and the phase is plotted in degrees. For both plots, the horizontal axis is either frequency (f ) or angular frequency ( ), measured in Hz and rad/s, respectively. The horizontal axis should be logarithmic (i.e. increasing by powers of 10). Most of the transfer functions well deal with in this class can be separated into a general that resembles the following: H (j ) = A j/z1 (1 + j/z2 ) (1 + j/z3 ) . . . j/p1 (1 + j/p2 ) (1 + j/p3 ) . . . (1)

A is an arbitrary constant and j is 1. As you can see, the basic component of this transfer function appears to be 1 + j/c, where c is some constant (with the slight variation j/c). Lets analyze this basic component rst before we analyze the entire transfer function.

2.1

Magnitude

Recall that the denition of magnitude (measured in dB) is as follows: 20 log |H (j )| = 20 log [H (j )] + [H (j )]
2 2

Lets apply this denition to our basic transfer function component (this is called a zero when it appears in the numerator of a transfer function): 20 log |1 + j/c | = 20 log 1 + (/c)2

For small , we have 20 log |1 + j/c| 0 dB. For large , 20 log |1 + j/c | . When = c , the magnitude of the transfer function is approximately 3 dB. Since theres so little change from = 0 to = c , we approximate the magnitude in this region as a constant 0 dB. 1

BODE PLOTS BASICS

For > c , the (/c ) dominates the magnitude expression, allowing us to approximate the magnitude as 20 log /c . From this expression its clear that if we increase by a factor of 10, we increase the magnitude by 20 dB. Thus, our Bode plot approximation for the zero is a constant 0 dB for < c and a line constantly increase by 20 dB/decade for > c , illustrated in Figure 1. Figure 1 also illustrates the Bode plot for a DC zero of the form j/c. This diers only slightly from the normal zero in that is lacks the additional 1. Thus, instead of having the constant magnitude region for < c , is simply always increases at 20 dB/decade. We draw its intersection with the frequency axis where = c , since thats where the magnitude is 0 dB.

40

20 Magnitude (dB)

20 log |1 + j/c | 20 log |j/c|

20

40 103

104

105 (rad/s)

106

107

Figure 1: Bode plots (magnitude) for a normal zero and a DC zero for c = 105 rad/s (the plots overlap for > c ) The basic transfer function component 1 + j/c can also appear in the denominator (in which case it is called a pole). Although this may seem like an entirely dierent problem, recall that we take the logarithm of our transfer function because our result is expressed in decibels. Taking the logarithm of the inverse of a function simply gives the negative logarithm of the function, meaning we simply have to negate the results of our zero analysis to get the appropriate expressions for poles. The same argument applies with DC poles of the form j/c, so we can negate our DC zero analysis to get the DC pole results. A normal pole will have a constant 0 dB value for < c and will drop by 20 dB/decade for > c . A DC pole will drop by 20 dB/decade for any and will intersect the frequency axis (0 dB) at = c . The results are shown in Figure 2.

2.2

Phase
[H (j )] [H (j )]

Lets take a look at the phase of a zero, DC zero, pole, and DC pole. Recall the denition of phase: Arg (H (j )) = tan1

COMBINING POLES AND ZEROES

40

20 Magnitude (dB)

20 log 0 20 log

1 1+j/c 1 j/c

20

40 103

104

105 (rad/s)

106

107

Figure 2: Bode plots (magnitude) for a normal pole and a DC pole for c = 105 rad/s (the plots overlap for > c ) Lets apply this to the normal zero rst. Arg(1 + j/c ) = tan1 c

For = 0, Arg(1 + j/c ) = 0. For , Arg(1 + j/c) 90 . For = c , Arg(1 + j/c ) 45 . Thus, our approximation for the phase of a zero is 0 for < 0.1c , 45 for = c , and 90 for > 10c with a straight line connecting these points. We can also look at the phase of a DC zero, which is always 90 . These results are shown in Figure 3. Similar to our analysis of the magnitude, we can also consider poles and DC poles in our phase plots. It can be shown that tan1 = tan1 , meaning our phase plots for poles and DC poles will simply be negated versions of the zero plots. These are shown in Figure 4.

Combining Poles and Zeroes

Generally, a transfer function may involve many poles and zeroes (and their DC counterparts). In order to make it easier to draw Bode plots, your rst step should be to factor the transfer function into the canonical form shown in Equation 1. This makes it easy to identify all of the poles and zeroes. First, youll have to handle the constant A in front (if present). The magnitude of A will aect your magnitude plot, and the sign of A will aect your phase plot. Your magnitude plot must be shifted up by 20 log |A|. For example, if A = 10, then your magnitude plot must be shifted up by 20 dB. Similarly, if A = 1/10, then your magnitude plot must be shifted down by 20 dB. If A < 0, then your phase plot must be shifted up (or downits the same in this case) by 180. Second, you need to draw each pole and zero individually on the same set of axes (whether youre making a magnitude or phase plot).

COMBINING POLES AND ZEROES

100

80 Phase (degrees)

60 Arg (1 + j/c ) Arg (j/c ) 40

20

0 3 10

104

105 (rad/s)

106

107

Figure 3: Bode plots (phase) for a normal zero and a DC zero for c = 105 rad/s (the plots overlap for > 10c ) Third, you simply add the curves that youve drawn at each point to obtain the nal Bode plot. Remember to shift your plots accordingly based on the constant A as mentioned previously. This superposition principle is possible because of the decomposition of the transfer function into zeroes and poles. When adding the poles and zeroes in the nal plot, remember that in areas where two curves are constant, the result will just be the sum of the constant values. When one is a contant and one is linear, then the result will start at the constant value and have the slope of the linear curve. Finally, when both are linear, the sum will have a slope equal to the sum of the slopes of the individal curves.

COMBINING POLES AND ZEROES

20 Phase (degrees) 40 60 80 100 103

Arg Arg

1 1+j/c 1 j/c

104

105 (rad/s)

106

107

Figure 4: Bode plots (phase) for a normal pole and a DC pole for c = 105 rad/s (the plots overlap for > 10c )