Professional Documents
Culture Documents
Page | 1
1) Show how the open-loop system responds to step changes in the motor voltage v(t) and disturbance torque
1] Response to step changes in motor voltage v(t)
The following time responses were obtained by setting the load torque to zero, and stepping the input voltage from 0V to 5V at 0 seconds.
Open Loop Current Response (Step Voltage Input)
16 14 12 10 8 6 4 2 0 0
Current(A)
0.5
1.5
2.5
3.5
4.5
Time(s)
20
Voltage(V)
15
10
0 0
0.5
1.5
2.5
3.5
4.5
Time(s)
Open Loop Tachometer Voltage Response (Voltage Step Input)
2.5
1.5
Voltage(V)
0.5
0 0
0.5
1.5
2.5
3.5
4.5
Time(s)
Page | 2
Current(A)
0.3 0.25 0.2 0.15 0.1 0.05 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time(s)
-0.1
-0.2
Voltage(V)
-0.3
-0.4
-0.5
-0.6
-0.7 0
0.5
1.5
2.5
3.5
4.5
Time(s)
Voltage(V)
0.5
1.5
2.5
3.5
4.5
Time(s)
Page | 3
3] Response of the system to step changes in disturbance torque with constant input voltage
The following graph shows the system reaction to a load torque of 100V for a constant input voltage of 5V. The disturbance torque is set to input into the system at 5 seconds. It is clear that each of the plots follows the same trend as shown above:
Open Loop Responses for Step Input Voltage and subsequent Step Input Disturbance Torque
35 Current Potentiometer Voltage Tachometer Voltage
30
25
20
15
10
0 0
10
Time(s)
Figure 4- Open Loop Time Responses for constant input voltage and stepped disturbance torque - There is an initial spike in current in the motor due to the step input voltage, which gradually tends to zero. The disturbance torque acts to lift the current level in the system following a waveform which could be approximated as ( ).
-The potentiometer voltage follows a waveform which can be approximated as a ramp function. There is an initial transient period from 0-1.5 seconds however. The load torque step input acts to reduce the slope of said ramp. -The tachometer voltage follows a second order sigmoidal trend due to the input voltage step and levels off at a steady state value. However when the disturbance hits the system, the tachometer voltage drops and stays at this lower value for an infinite time period. The Disturbance Rejection Capability of the open loop system is poor.
N.B. The tachometer voltage and potentiometer voltage are directly linked to the blade speed and blade pitch. By dividing each response by K and K respectively, the responses for blade speed and pitch can be easily obtained.
Page | 4
2) Determine by experiment the frequency response a Bode Plot, a Nyquist Plot and a Nichols Chart
Data Gathering and Experimental Bode Plot of
. Plot this as
However, clearly as these transfer functions are merely first order approximations, it is necessary for to achieve a truer approximation to the actual transfer functions and
. To do this, the input voltage was set to a sinusoidal wave of 1V amplitude. The amplitude and phase change of the output wave was then determined. This process was carried out for numerous frequencies of input voltage. Hence in this way, data points were gathered for the experimental bode plot of below.
Experimental Data Time Input Time Output Wave (s) Wave (s) 785.4 157.0795 78.5405 47.1225 44.5058 49.087 45.553 47.9093 48.6947 49.0874 49.323 49.4801 49.5923 48.891 49.0437 49.166 49.7163 49.68952 49.7681 49.8924 49.5272 49.48795 49.6843 49.7039 19.9334 19.9727 786.289 157.97 79.4281 47.996 45.328 49.867 46.288 48.4518 49.1138 49.4274 49.6091 49.7274 49.8104 49.0865 49.2211 49.3295 49.809 49.756333 49.8072 49.9236 49.5532 49.5022 49.6941 49.7114 19.9395 19.9778
Frequency (rad/s)
Gain (V) 0.01 0.4948 0.05 0.4943 0.1 0.4929 0.3 0.4786 0.6 0.4382 0.8 0.4055 1 0.3725 2 0.2456 3 0.1761 4 0.1358 5 0.11 6 0.09223 7 0.07927 8 0.069343 9 0.0617 10 0.05546 20 0.02675 30 0.01679 60 0.006608 80 0.004195 100 0.002872 200 0.0007994 300 0.0003634 400 0.0002061 500 0.0001324 600 0.00009215
Phase (degrees) -0.50935948 -2.551094583 -5.08557339 -15.01435902 -28.26515395 -35.75256642 -42.11239794 -62.16592077 -72.03798358 -77.92226014 -81.96161259 -85.01547764 -87.47346658 -89.61059916 -91.47844157 -93.6785995 -106.2263752 -114.8430875 -134.4158987 -143.0102657 -148.9690267 -163.2929716 -168.4495918 -171.8873385 -174.7521275 -175.3250853
Gain (dB) -6.11141 -6.12019 -6.14482 -6.40055 -7.16655 -7.84018 -8.57747 -12.1954 -15.0848 -17.342 -19.1721 -20.7026 -22.0178 -23.1799 -24.1943 -25.1204 -31.4535 -35.499 -43.5986 -47.5454 -50.8363 -61.9447 -68.7923 -73.7184 -77.5622 -80.7101
V /V
Page | 5
Pitch Adjustment Scheme Design Exercise 700 800 900 1000 0.00006779 0.00005194 0.00004106 0.00003328 19.9379 19.9118 19.9474 19.9884 19.9423 19.9156 19.9508 19.9915
Mark Roche 109479961 -176.4710009 -174.1791697 -175.3250853 -177.6169165 -83.3767 -85.69 -87.7316 -89.5563
Table 1- Experimentally Measured Gains and Phase Angles for The Time Input Wave and Time Output Wave columns refer to the time at which a certain peak occurs in the input/output sinusoid. This phase difference was then easily determined by: ( )
At first it was quite challenging to decide which peaks of the input/output waves should be measured for each frequency. However, a number of things soon became clear: At low frequencies, it was necessary to run the simulation for a lengthy time (in the case of 0.01rads-1, more than 700 seconds). This is due to the fact that the wave itself is travelling so slowly at these low frequencies. For these measurements minimizing the step size to optimize measurement accuracy was not a major goal as if the minimum step size was too small, it would result in a slowing down of the simulation (hence, maximum step size was set to 0.001). For these frequencies, measurements were taken from the second peak of the input/output waves, as the first peak of the output wave still incorporated transient characteristics. At medium to high frequencies, a time scale of 50 seconds was chosen for the duration of the simulation. This was thought to be a reasonable time scale as, for the final peaks of the output wave, transient behaviour was seen to have died away and also, a timescale of 50 seconds allowed for a small maximum step size to be set within the simulation configuration parameters (maximum step size 0.0001, relative tolerance 1e7). However, at very high frequencies, ( 500rads-1), the need to increase measurement accuracy heightens due to the fact that the wave is moving so quickly. Hence, it was necessary to reduce the maximum step size to 0.00001. However, it was observed that a step size of this size resulted in a major retardation of the simulation when used with a 50 second time scale. It is for this reason that all data above 400rads-1 was measured over a timescale of 20 seconds, which was chosen as it was seen to give optimum levels of accuracy coupled with a relatively quick simulation time. Subsequently, the data points in Table 1 above were plotted on a Bode Gain/Phase Plot as shown below:
Page | 6
Gain(dB)
10
10
10
10
10
Degrees
10
10
Frequency(rad/s)
10
10
10
Hence, in order to convert from gain and phase values for needed is to scale the gain (not in dB) by
reading of each datapoint. The resulting data points found are expressed on the table below:
Experimental Data Time Input Time Output Wave Wave 1413.75 911.052 1963.497 486.9474 243.474 198.3142 1414.6 911.92 1964.3812 487.8152 244.289 199.089
Frequency (rad/s) V/V 0.01 0.05 0.1 0.3 0.6 0.8 Page | 7
Pitch Adjustment Scheme Design Exercise 1 2 3 4 5 6 7 8 9 10 20 30 60 80 100 200 300 400 500 600 700 800 900 1000 0.745 0.2456 0.1174 0.0679 0.044 0.030743333 0.022648571 0.01733575 0.013711111 0.011092 0.002675 0.001119333 0.000220267 0.000104875 0.00005744 0.000007994 2.42267E-06 1.0305E-06 5.296E-07 3.07167E-07 1.93686E-07 1.2985E-07 9.12444E-08 6.656E-08 95.8185 3004.148 3003.8864 9004.1975 9004.119 9004.0665 9001.336 9001.645 9001.885 9009.6165 9004.51175 9001.7625 9001.841 9001.93925 9001.9353 900000.3611 9.3462 9.9 19.6444 19.6585 29.5601 49.7569 49.7576 49.5885 96.5535 3004.687 3004.3058 9004.536 9004.404 9004.3132 9001.551 9001.84 9002.0635 9009.779 9004.6035 9001.829 9001.8805 9001.9705 9001.96125 900000.3754 9.356 9.9074 19.6504 19.6636 29.5645 49.7608 49.761 49.5916 -132.1123979 -151.7648503 -162.0895498 -167.5784855 -171.6464858 -174.8092128 -176.2301482 -179.381416 -182.0456698 -183.1056417 -195.1377554 -204.3050801 -225.7909974 -233.2394488 -238.6825478 -253.9003076 -258.4495918 -259.5955074 -261.8873385 -265.3250853 -266.4710009 -268.7628321 -265.3250853 -267.6169165
Mark Roche 109479961 -2.556874545 -12.19543275 -18.60663806 -23.36260451 -27.13094647 -30.24498092 -32.89918372 -35.22114729 -37.25854699 -39.09980278 -51.45352427 -59.02081126 -73.14102441 -79.58656052 -84.81571138 -101.9447171 -112.3141267 -119.7390401 -125.5210405 -130.2525183 -134.2580482 -137.7311609 -140.7958714 -143.5357337
Gain(dB)
10
10
10
10
10
Frequency(rad/s)
Page | 8
Phase(Degrees)
10
10
10
10
10
Frequency(rad/s)
90 120
100 60 80 60
150 40 20 180
30
X: -0.01551 Y: 1.172e-010
210
330
240 270
300
Page | 9
Nichols Chart
50
Phase Margin
0.25 dB 0.5 dB 1 dB 3 dB 6 dB 0 dB -1 dB -3 dB -6 dB -12 dB -20 dB -40 dB
Gain Margin
-50
-60 dB -80 dB -100 -100 dB -120 dB -140 dB -150 -160 dB -180 dB -200 -360 -200 dB -315 -270 -225 -180 -135 -90 -45 0
This is an approximation for the ultimate gain of the system.Hence the gain margin has been determined also form the Nyquist Plot to be 64.4745. The gain and phase margins can also be determined from the Nichols Chart as shown above. As the Open Loop dB Gain at -180o was found from the Nichols Chart to be -35.8dB, this implies that:
From the Nichols Chart the phase margin can also be determined as the phase difference between the critical point and the point on the plot which corresponds to 0dB Open Loop o Gain. Hence, from the Nichols Chart shown above:
Page | 10
3) Use the frequency response to determine an estimate of the openloop transfer function
The transfer function indirectly from the can be obtained both directly from the gain bode plot, or
discussed, however the indirect method will be explained here, as this was the method that the student used to approach the problem initially, and it was found to give an approximation of equivalent accuracy to that of the direct method. Effectively, how the student approached the problem was to obtain the transfer function from the corresponding bode plot, as a question very similar was covered in the course notes. The method of asymptotes was used as shown in the plot below:
-40 -50 -60 -70 -80 -90 -100 -2 10 Experimental Bode Plot 0dB/decade Asymptote 20dB/decade Asymptote 40dB/decade Asymptote 10
-1
10
10
10
10
which corresponds to an asymptote of 0dB/decade (low frequencies), that which corresponds to an asymptote of -20dB/decade and that which converges to an asymptote of -40dB/decade (high frequencies). Where these asymptotes cross corresponds to the corner frequencies of the transfer function. By analysing the above plot: and
Page | 11
diagram it can be seen that the circuit consists of 2 integrators up until the tachometer. Hence the transfer function will be of the form:
The system also has a constant gain for low frequencies of -6.1115dB This implies that: Therefore: =>
)) (
))
- merely scale
by . Hence:
By declaring this transfer function in Matlab, the actual gain and phase margins, and crossover frequencies of this system can be obtained (which for the gain and phase margins will obviously be slightly different to those obtained via the Nichols Chart). The code needed is relatively simple: [ ][ [ This Yields: Gainmargin =61.8161, Crossover2 =0.8042, Phasemargin = 53.6001, Crossover3 = 1 Crossover1 = 8.2083, ] ]
Hence, the gain and phase margins obtained from the Nichols Chart were clearly accurate, presuming that the frequency response data accurately describes the system.
Page | 12
4) A proportional controller with gain Kp is proposed to control pitch angle. Show through simulation in Simulink, how the closed-loop performance depends on the choice of gain Kp. Show how your Nichols Chart could have predicted these results.
In this part of the design, it was attempted to improve the performance characteristics of the system by closing the loop and introducing a proportional controller. Hence, the equivalent circuit
becomes:
Potentiometer
150
7 100 4 6
Voltage(V)
Current(A)
50
Voltage(V)
0 10 20 30
3 -50 -2 2
-100
-4 1
-150
10
20
30
-6
10
20
30
Time(s)
Time(s)
Time(s)
Tachometer
15
10 9
Potentiometer
400
10
300 8 200 7
Voltage(V)
Current(A)
100
Voltage(V)
-100
-5
-200 3 -300 2
-10
-400 1
-500
10
20
30
40
50
Time(s)
-15 0
20
40
60
10
20
30
40
50
Time(s)
Time(s)
Current 4
x 10
Tachometer
6
x 10
Potentiometer
3 4
2 2 2 1
Voltage(V)
Current(A)
Voltage(V)
0 5 10
-1 -2 -1 -2
-4 -2 -3
-3
10
-4
-6
10
Time(s)
Time(s)
Time(s)
Hence, it is clear that for low values of proportional gain (Kp=10), the time responses of the current, speed and pitch angle are stable waveforms. However, the peak overshoot and settle time are undesirable. These qualities generally disimprove with increasing Kp However, once Kp reaches the level of ultimate gain, the system becomes marginally stable and subsequent to this, when Kp is increased past the level of ultimate gain, the system will become unstable, which is demonstrated above for a proportional gain of 200. For stable choices of Kp there is zero steady-state error for step inputs, due to the free integrator in the system which makes the system itself Type 1, having a position error constant of infinity.
Current
1500 25 20
Tachometer
12
Potentiometer
10 1000 15
10 500
Voltage(V)
Current(A)
Voltage(V)
0 5 10
-5
-500 -10 2
-1500
10
-25
-2
10
Time(s)
Time(s)
Time(s)
Page | 15
5) Design a first-order phase-lead compensator to provide what you believe is the best trade-off between steady-state accuracy and dynamic performance. Demonstrate the performance of you pitch control scheme, through simulation in Simulink.
The aim of this section is to design a phase-lead compensator for the system, so that the system will track a ramp of the form
-with the best trade-off between steady-state error and dynamic performance. From the material studied in lectures, the design specifications for a well configured phase-lead compensator will follow:
Where PO% refers to the peak overshoot for a step change in the desired pitch angle, and ess is the steady- state error. For a ramp input waveform:
As we want PO%=10%- need to find the corresponding damping of the system, which can be found via the equation:
This equation is difficult to solve- therefore the damping required was obtained from the graph given in the course notes. The resulting value of is 0.6, which when subbed back into the above equation gives a peak overshoot of 10%. Hence, if the system is assumed to be second order dominant, Now the gain compensated bode plot is found (K=40.42078032):
Page | 16
Bode Diagram From: Subsystem1/1 To: Subsystem1/1 100 System: Model I/O: Subsystem1/1 to Subsystem1/1 Frequency (rad/sec): 6.69 Magnitude (dB): -7.68e-008
50
Magnitude (dB)
0 System: Model I/O: Subsystem1/1 to Subsystem1/1 Frequency (rad/sec): 12.3 Magnitude (dB): -10.6
-50
-100
-150 -90
-135
Phase (deg)
-180 System: Model I/O: Subsystem1/1 to Subsystem1/1 Frequency (rad/sec): 6.69 Phase (deg): -177
-225
-270 10
-2
10
-1
10
10
10
10
Frequency (rad/sec)
Figure 16- Gain Compensated Bode Plot for Phase -Lead Compensator Design
The value of can be found from this plot as the phase difference between the compensated gain crossover frequency and -180o. From the above plot:
Therefore:
has been determined on the above bode plot and is marked with the data cursor.
Page | 17
can be found to equal 0.0240824 All the necessary parameters for the transfer function of the phase lead compensator have been determined and hence:
Now the phase lead compensator is added to the system (the subsystem shown is the same as that shown in Figure 11) :
18
16
14
Figure 18- Time Response of System with Phase-Lead Compensation for Ramp Input
Voltage(V)
12
10
10
Time(s)
Page | 18
The potentiometer voltage does undergo an initial lock-in phase however during the first second of simulation (approximately), however this can be expected in any system, and should not be considered a reason to reconfigure the controller. This initial lock in is 2 Input Voltage shown in Figure19. But how does the Potentiometer Voltage 1.8 system react to a stepped input voltage? 1.6 Figure 20 shows the response of the potentiometer voltage to a input voltage of 5V stepped up at time= 1 second. The plot clearly shows that the output has zero tracking error relative to the input voltage, and has a steady-state value of 5V after the initial overshoot. The overshoot however- designed to be 10% of the setpoint- actually rises to 6.119V- an overshoot of 22.38%. This may or may not cause problems during use, however, the controller is easily re1.4
Voltage(V)
1.2
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1.2
Time(s)
calibrated.
Voltage(V)
Figure 20-Time Response of PhaseLead Compensated System for Step Voltage Input
0 0
Time(s)
Page | 19
Page | 20