Nichols Chart
Nichols Chart
Nichols Chart
Introduction
If you are trying to control a dynamic system, it can be represented by a transfer function in 's'. One way
to interpret a gain, G(s), is to let s → jω. This results in the system behaving as a filter with the gain of
G(jω).
If a dynamic system has feedback, as shown below, two questions arise:
"How large can you make the gain and have a stable system?"
"How large should the gain be for some desired damping ratio?"
If interpreted using root-locus techniques, stability results from keeping the poles in the left-half plane. If
interpreted using frequency domain techniques, stability results from the closed-loop gain being finite.
R Y
G(jw)
-
∞ a≥
Similarly, from a frequency-domain standpoint,
For a system to be stable, the gain must be less than 1.000 when the phase of the open-loop
system is 1800 .
In addition to being stable, a system should also behave well. 'Well' typically means the damping ratio
must be at least some value (meaning the resonance is less than XdB, the overshoot is less than Y%, etc.)
To find the gain which keeps the resonance less than XdB, a Nichols chart is used.
ECE 402 Controls Lab
JSG - November 30, 2001
Nichols Charts
The relationship which converts the gain of the open-loop system to the closed-loop gain is
For example, if the gain of G is 0.9∠ − 150 0 at some frequency, the closed-loop gain at that frequency will
be 1.7958∠ − 86.11 0 = 1+0.9∠−150
0.9∠−150 0
0 . Since this is a nonlinear mapping, a Nichols Chart is useful to show
G
how G maps into 1+G .
Gain (dB)
20
G(jw)
15 1dB
3dB
10
0dB
5 6dB
9dB
12dB
15dB
0 A
-5
-10
-15
-20
-220 -210 -200 -190 -180 -170 -160 -150 -140 -130 -120 -110 -100
degrees
The resonance can be read by finding the point closest to -1 as read by the M-circles. (0dB,
-1800 .) For this system, the largest closed-loop gain is close to point, 'A', with a resonance of
about 5dB closed-loop.
Gain (dB)
20
G(jw)
15 1dB
3dB
10
0dB
5 6dB
9dB
12dB
15dB
0 A
-5
GK
-10
K
-15
-20
-220 -210 -200 -190 -180 -170 -160 -150 -140 -130 -120 -110 -100
degrees
ECE 402 Controls Lab
JSG - November 30, 2001
Bode Plots:
A Bode Plot is a plot of gain and phase vs. frequency. Ideally, you would like the gain to be 1.000 at al
frequencies, meaning that the output tracks the input. Unfortunately, this often is not the case. To make
the gain 1.000, feedback can be used with high gains.
With feedback, the gain becomes
This is approximately:
G(jω) ≈ 1 G >>
1+G(jω) G <<
G
Hence, if you make the gain large, the closed-loop gain will be approximately 1.000 as desired.
From previous results, there is a limit on how much gain you can tolerated, however. As a result, the trick
to designing a feedback system using frequency domain techniques is to
Crank up the gain as much as possible. A Nichols chart is a useful tool to determine how large
the gain can be.
This results in the closed-loop system having a gain of about 1.000 when the open-loop gain is
greater than 1. This will likewise be the 'passband' of your closed-loop system.
30
Open-Loop
20
10
dB
0
Closed-Loop
-10
-20
-30
0.01 0.1 1 rad/sec 10 100 1000
ECE 402 Controls Lab
JSG - November 30, 2001
1M R
100k
10k
R 0.01
2
100k
100k
100k 741 Y
6
741
3
0.01 0.01 0.01
100k
R
k =
1M
10k
Open in Step 2
10k
Close in Step 4+
10k 2
6
741
3
Write-up:
If you design a compensator using frequency domain techniques, what information do you need?
Do you need to know the transfer function of the system to design a feedback compensator using
frequency domain techniques?
Was the gain vs. frequency 'flatter' with feedback or without? Why was this so?
How does designing a compensator using a Nichols chart compare to a similar design using Root Locus
techniques?
Gain (dB)
20
15 1dB
10 3dB
0dB
5 6dB
9dB
12dB
15dB
0
-5
-10
-15
-20
-220 -210 -200 -190 -180 -170 -160 -150 -140 -130 -120 -110 -100
degrees