Anda di halaman 1dari 8

MOHAMMED HAFIZ BIN ABDUL AZIZ

U090018W
GROUP 3F1


FORMAL REPORT



ME2142-1 FREQUENCY RESPONSE






SEMESTER 5

2011/2012






Department of Mechanical Engineering
National University of Singapore



CONTENTS

Table of Contents 2

Introduction 3

Tables of Values 4

Bode Plot 5

Sample Calculations 6

Discussions 8

Conclusion 9


















INTRODUCTION

With this lab the usage use of frequency response method to identify the (linear)
dynamics of a plant was learnt, i.e. it was proven that it is possible to find the plant
dynamics just by inputting a sinusoidal signal and comparing it with the sinusoidal output.

Purposes
- To perform a frequency response test on an aircraft electro-hydraulic servo-
actuator and to determine the phase and gain margins of the servo.
- To verify that increasing the gain causes instability.

Equipment
1. Electro-hydraulic servo-actuator
2. Amplifier unit
3. Oscilloscope (CRO)
4. Function generator (FG)
5. Hydraulic power supply (outside laboratory)



























TABLES OF VALUES

Table 1
] (Hz) e
(rad/s)
=2H]
V
i
(V) V
o
(V) Gain (dB)
=20 log
i
o
V
V

Atime
(s)
Phase (deg)
=Atime x ] x
360
o

0.5 3.142 5.04 4.88 -0.280 0.120 -21.60
1. 6.283 5.12 4.88 -0.417 0.100 -36.00
2. 12.566 5.12 3.84 -2.499 0.088 -63.36
3. 18.850 5.12 2.88 -4.998 0.084 -90.72
4. 25.133 5.12 2.24 -7.180 0.072 -103.68
5. 31.416 5.12 1.76 -9.275 0.066 -118.80
6. 37.699 5.12 1.28 -12.034 0.064 -138.24
7. 43.982 5.12 0.97 -14.441 0.056 -141.12
8. 50.265 5.12 0.87 -15.435 0.050 -144.00
9. 56.549 5.12 0.70 -17.271 0.048 -155.52
10. 62.832 5.12 0.64 -18.062 0.045 -162.00
11. 69.115 5.12 0.59 -18.768 0.043 -170.28
12. 75.398 5.12 0.53 -19.700 0.042 -181.44
13. 81.681 5.04 0.48 -20.460 0.040 -187.20
14. 87.965 5.04 0.43 -21.399 0.038 -191.52
15. 94.248 5.04 0.38 -22.545 0.037 -199.80

Table 2
e (rad/s) Open Loop Gain (dB) Open Loop Phase
(Deg)
Log
10
e
3.142 9.0 -97.5 0.497
6.283 4.5 -103.0 0.798
12.566 -1.8 -107.0 1.099
18.850 -6.0 -115.0 1.275
25.133 -8.0 -125.0 1.400
31.416 -11.5 -132.5 1.497
37.699 -14.2 -144.0 1.576
43.982 -16.5 -146.0 1.643
50.265 -17.5 -149.0 1.701
56.549 -18.0 -156.0 1.752
62.832 -18.8 -162.0 1.798


SAMPLE CALCULATIONS

For Table 1

Using Frequency, f = 1.0 Hz,
Input voltage, V
i
= 5.12 V
Output voltage, V
o
= 4.88 V
Atime = 0.100 s


(1) e in (rad/s): e = t 2 f
= 1 2 x xt
= 6.283 rad/s

(2) Gain (in dB) for the closed loop system: Gain (in dB) = 20
|
|
.
|

\
|
i
o
V
V
10
log
= 20 |
.
|

\
|
04 . 5
88 . 4
log
10

= - 0.417 dB

(3) Phase (deg) for the closed loop system: Phase = time A x f x 360
= -0.100 x 1 x 360
= - 0 . 36
= 36.0
o
(lagging)

For Table 2

Corresponding values of Open Loop Gain and Phase are obtained from the Nichols Chart.
Using Frequency, f = 1.0 Hz,
Gain in closed loop (in dB) = - 0.417 dB
Phase in closed loop (in deg) = - 0 . 36
Gain in open loop (in dB) ~ 4.50 dB
Phase in open loop (in deg) ~ -103.0
o


(1) From Nichols Chart
Theoretical Gain Value = Intersection of graph with - 180 open loop phase line
= - 20.5 dB
Theoretical Gain Margin = [0 (-20.5)]
= 20.5 dB

Theoretical Phase Value = Intersection of graph with 0 open loop gain line
= - 5 . 102
Theoretical Phase Margin = 180 - ( ) 5 . 102
= -77.5
o
(2) Experimental Gain Margin:

Gain, K, at which system starts to attain instability = 62.8
Experimental Gain Margin (in dB) = ( ) K
10
log 20
= ( ) 8 . 62 log 20
10

= 34.96 dB


For Bode Plot Diagram:

(1) Gain (open loop) = 21.993 dB (from Y-intercept when, =0)
20log K = 21.993
K =
|
|
.
|

\
|
20
993 . 21
10
= 12.56

(2) Corner frequency, e = 50.0 rad/s (cut-off point at the Bode gain graph)
Time Constant, t =
e
1

=
50
1

= 0.02

(3) Transfer Function of Open Loop System:
Transfer Function G(s) =
( ) 1 + s s
K
t

Since K = 12.56 and t = 0.02,
Transfer Function G(s) =
( ) 1 02 . 0
56 . 12
+ s s
which is a 2
nd
order system.
















DISCUSSIONS

Frequency response of system as input frequency increases

As the input frequency is increased in steps from 0.5 Hz to 15 Hz while the V
i
is kept
constant, the output peak to peak voltage, V
o
decreases. This is because when a sinusoidal
input has reached its peak, the mechanical output requires some time before it can attain its
corresponding maximum possible value, which can be achieved if the input is held steady at its
peak value long enough. As all real world mechanical systems are unable to respond
instantaneously to such commands (or changes in input) and mechanical output responses tend
to lag the input (which is why the phase is negative). Since, the sinusoidal input is changing
continuously and will have decreased to a lower value before the output can reach its
maximum possible value which means that its corresponding peak value as well as its peak to
peak value V
o
will decrease.

System response as gain increases

When the magnitude of gain is slowly increased, the system becomes more unstable. This is
because the Atime is decreasing and the output signal does not have enough time to oscillate
from instability to equilibrium. This result in a series of overshoots in the output square wave
signal and the amplitudes of this oscillating overshoots keep increasing as the gain increases.
When the gain is further increased steady oscillations of peak values are observed and they
become quite constant. There is also a critical value for the gain constant, above which the
system output fluctuated continuously about the desired value without settling down. Hence,
any gain above this critical value makes the system unstable.

Comparison of experimental and theoretical values of gain margin

The system approaches instability when the open loop gain is increasing. The value of
theoretical gain margin obtained from the intersection of the curve and the reference open
loop phase angle -180
o
of the Nichols chart is 20.5dB. This is much lower than the experimental
gain margin of 34.96dB due to some factors.
Difference in the gain margin are caused by experimental errors like operating the actuator
system for a long time which increases the heat generation by friction, the viscosity of the fluid
in the hydraulic actuator system operating under pressure affecting the actual readings. Also
the point of instability is mainly determined by human judgment, even though the system may
not reach its point of instability yet. In addition, readings on the control knob do not have
proper minor divisions and thus only rough estimations can be done when obtaining the gain of
the point of instability. The values obtained from the Nichols chart are also approximated
values.



Transfer function and order of system

Order of system = 2
nd
since the gradient of the graph (22.656) is within the gradients of the -
40dB/dec
and -20dB/dec.

Hence, the open loop transfer function,

()

()
()

()


The gradient of the Bode plot diagram is approximately between the limits of the gradients
of -20dB/dec and -40dB/dec graphs. This shows that the system is estimated to be of second
order. The open loop gains and phases are approximately obtained from the plot of close loop
gains and phases. A smooth curve results on the Nichols charts thus showing that the close loop
gains and phases are stable.

Source of errors

o Both the readings off the Nichols chart and the points being plotted are not accurate at all
as the divisions are purely estimated.
o Experimental instable gain observed is also not accurate. There were no small divisions on
the knob that could help in estimation in between the major divisions.
o Different interpretation on instability leads to different readings of K value while turning the
knob.

CONCLUSION
After completing this experiment, we were able to perform a frequency response test
on the aircraft electro-hydraulic servo actuator and also able to determine the phase and gain
margins of the servo. An open loop transfer function of the system is obtained during the
process, by plotting the closed loop values on the Nichols Chart and then using the Bode Graph
to obtain the transfer function. The point of instability is also located and at the same time, also
verified that increasing the gain of the system causes instability.