martin regehr [email protected] july 8,...
TRANSCRIPT
![Page 2: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/2.jpg)
Sample Application: Star Tracker
Star light
Steering mirror
To Beam Combiner
Star light
Star light is delivered to beam combiner by steering mirror.
Want to maintain alignment of direction (wave fronts) of star light beam for good fringe visibility.
![Page 3: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/3.jpg)
Star Tracker
Star light
Steering mirror
To Beam Combiner
Disturbance (d)
Star light
Error (e)
Imperfection changes direction of incoming beam of star light.
E.g., warping of mirror mount due to a change in temperature.
No longer have good alignment at beam combiner; fringe visibility is degraded.
![Page 4: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/4.jpg)
Star Tracker
Star light
Steering mirror
To Beam Combiner
Pick-off
Pick-off
Position detector
Disturbance (d)
Star light
Error (e)
Measure pointing error using partially reflecting beam splitter (“pick-off”), lens, and position detector
Position detector may be CCD, Quadrant photodiode, etc.
![Page 5: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/5.jpg)
Star Tracker
Star light
Steering mirror
To Beam Combiner
Pick-off
Pick-off
Amplifier
Position detector
Disturbance (d)
Star light
Correction (c)
Error (e)
Amplify error signal and feed back to steering mirror
![Page 6: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/6.jpg)
Block Diagram Representation
d
Ac
e
-
+
Misalignment of incoming starlight
Correction subtracted from misalignment at steering mirror
Correction applied to steering mirror
c = A e
Residual alignment error in light reaching beam combiner
A is loop gain or “open loop gain.” Response of mirror, lens, position detector, and amplifier (angle in, angle out).
![Page 7: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/7.jpg)
Gain Setting
Suppose amplifier gain is adjustable.
What is appropriate / optimum setting of amplifier gain?
Ade
Aede
+=−=
1
d
Ac
e
-
+
Note: this holds for any frequency, if we allow the quantities involved (A, e, d) to be complex
![Page 8: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/8.jpg)
Gain Setting
Ade
Aede
+=−=
1
The higher the gain, the smaller the error. But…
d
Ac
e
-
+
• Stability
• Other sources of error may not be suppressed
![Page 9: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/9.jpg)
Other Sources of Error
E.G. noise in position detector
Steering mirror
To Beam Combiner
Pick-off
Pick-off
Amplifier
Position detector
Star light
Error (e)
+
+
Noise (n)
![Page 10: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/10.jpg)
Other Sources of Error
n
H
e
+
+
Position detector noise
Residual alignment error in light reaching beam combiner
H is response of lens and position detector
-G
-G is response of amplifier and steering mirror
•Use –G to denote gain of amplifier because we know there is an inversion in the loop, and we like to keep the values of the gains positive
•With this convention, A = GH
GHGne
GHeGne
+−=
−−=
1
If GH ™1, increasing G doesn’t help
![Page 11: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/11.jpg)
Stability
Recall:
Don’t want A ≅ -1 at any frequency. (Necessary condition)
Ade += 1
d
Ac
e
-
+
![Page 12: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/12.jpg)
Stability
Typical frequency response for A
c
jAAωω
+=
10
“Single pole response”
left half-plane pole at
s = σ+jω = -ωc
Amplitude Response
-25
-20
-15
-10
-5
0
5
10
15
20
25
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
00 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Slope: -20 dB/decade
Phase: -90 degrees
![Page 13: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/13.jpg)
Stability
Bode stability criterion:
Phase at unity-gain frequency
> -180 degrees
Amplitude Response
-25
-20
-15
-10
-5
0
5
10
15
20
25
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
00 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Phase: -85 degrees; ok
Okay to turn up gain?
Yes. Phase asymptotes at –90 degrees
![Page 14: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/14.jpg)
Stability
Bode stability criterion:
Phase at unity-gain frequency
> -180 degrees
Amplitude Response
-25
-20
-15
-10
-5
0
5
10
15
20
25
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
00 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Phase: -85 degrees; ok
Rule of thumb: provide at least 30 degrees of phase margin and 6 dB of gain margin.
In this case, phase margin is 96 degrees and gain margin is infinite.
What if gain isn’t adjustable?
![Page 15: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/15.jpg)
Stability
Cascade three amplifiers: too many poles
Amplitude Response
-80
-60
-40
-20
0
20
40
60
80
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-250
-200
-150
-100
-50
0
50
100
150
200
250
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Phase: -255 degrees; unstable
3
30
1 ⎟⎟⎠
⎞⎜⎜⎝
⎛+
=c
j
AAωω
Insert attenuator to reduce gain for stability?
Need to reduce gain by over 40 dB; using one amplifier is better.
Two amplifiers might be better still.
![Page 16: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/16.jpg)
StabilityWhy is this –255 degrees and not +105 degrees?
Amplitude Response
-80
-60
-40
-20
0
20
40
60
80
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-250
-200
-150
-100
-50
0
50
100
150
200
250
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Phase: -255 degrees; unstable
For Bode stability criterion, 0 degrees is defined as the low-frequency limit of the phase, where the slope of the amplitude response vanishesIf low-frequency limit of amplitude response slope is n * 20 dB/ decade, then low frequency phase is n * 90 degrees
See Thaler and Brown for details
![Page 17: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/17.jpg)
StabilityMore typical gain limitation: mechanicalresonancesOften have “forest” of resonances at frequencies above first resonance
Amplitude Response
-60
-50
-40
-30
-20
-10
00 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-250
-200
-150
-100
-50
0
50
100
150
200
250
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Phase response depends on mechanism. Often design loop so that all resonances are below unity gain.
Note phase roughly proportional to slope of amplitude. Bode integral theorem.
![Page 18: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/18.jpg)
Example of Benign Resonance: Delay line with voice coil
OPD actuator consisting of retro-reflector mounted on soft flexures, with magnet driven by coil (simple harmonic osc.).
Used to control optical path delay. First resonance typically at very low frequency; next resonance much higher in frequency.
Little phase margin because at high frequencies, - m ω2 x = F
Amplitude Response
-140
-120
-100
-80
-60
-40
-20
0
20
40
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
00 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
![Page 19: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/19.jpg)
“Lead” CircuitAmplitude Response
0
5
10
15
20
25
30
35
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
0
10
20
30
40
50
60
70
80
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Construct using op-amp with a capacitive network.
Produces phase lead over a range of frequencies.
In
Out
![Page 20: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/20.jpg)
Cascade of Voice Coil and LeadAmplitude Response
-100
-80
-60
-40
-20
0
20
40
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
20
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Range shown has good phase margin:
Amplitude Response
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Closed-loop response:
e/d
![Page 21: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/21.jpg)
DelayAmplitude Response
-70
-60
-50
-40
-30
-20
-10
00 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Mag
Res
pons
e (d
B)
Phase Response
-250
-200
-150
-100
-50
0
50
100
150
200
250
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
log (Frequency / 1 Hz)
Phas
e R
espo
nse
(deg
rees
)
Pole at 10 Hz and 1 ms delay
E.g., sampled-data (digital) system with 1 kHz sample rate
Rule of thumb: in a sampled-data control system,
Unity-gain frequency = sampling rate / 10
![Page 22: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/22.jpg)
Procedure for Simple Design
Make sure there is a broad range of frequencies around the desired unity-gain frequency over which the phase is > -150 degrees and the slope of the amplitude response is approximately –20 dB/decade. Provide a means to change the sign of the feedback.
Increase the gain gradually. If the error increases when the gain is increased, change the sign of the feedback.
Increase the gain until the system oscillates, then decrease it by a factor of 2.
![Page 23: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/23.jpg)
Multiple Loops
P is response of fast actuator
M is response of slow actuator
d
M
e
+
+ P
Back to star tracker example. Suppose range of steering mirror is too small. Add larger range, slower (e.g., motorized) actuator.Both actuators suppress disturbances.
Adjust gain and frequency response so that fast actuator suppresses high-frequency disturbances, and slow actuator suppresses low-frequency disturbances.
P+M must satisfy stability criterion. Suppose P already satisfies criterion; what are constraints on M?
![Page 24: Martin Regehr martin.regehr@jpl.nasa.gov July 8, 2003nexsci.caltech.edu/...MSS/...Intro_control_systems.pdfIntroduction to Control Systems Martin Regehr martin.regehr@jpl.nasa.gov](https://reader033.vdocuments.site/reader033/viewer/2022060922/60ad990d0a68425dca0d1d4a/html5/thumbnails/24.jpg)
Multiple Loops
Open-loop gain, for
d
M
e
+
+
P+M must satisfy stability criterion. Suppose P already satisfies criterion; what are constraints on M?
1>>P
Can be written as
PM
PMA
≅
+=
11
-
This means that the ratio of M to P must satisfy the stability criterion.
At “cross-over,” relative phase must be >-180 degrees
P