EEE 101 Experiment 2: Feedback Implementation and Error Signals

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 6

EEE 101 Experiment 2: Feedback Implementation

and Error Signals


Edgardo Callagon Jr., Jazzrine Tagle
Electrical and Electronics Engineering Institute
University of the Philippines Diliman
Quezon City, Philippines
edgardo.callagon@eee.upd.edu.ph
jazzrine.tagle@eee.upd.edu.ph

Abstract—The main objective of this experiment is to errors that were measured, and the implementation of each
construct a closed-loop system with an additional system gain to controller is discussed in the succeeding sections.
the first-order, type-0 plant developed in the previous
experiment. The error signal to variable controller values was A. Closing the loop
tested and observed. The variable controllers used are the
The plant obtained from experiment 1, with modelled
proportional controller and the integral controller. The error
results and step responses are shown in this paper. transfer function, G(s), was implemented in a closed-loop
system shown in Fig. 1.
I. INTRODUCTION
A closed-loop control system, also known as a feedback
control system is a control system which uses the concept of
an open loop system as its forward path but has one or more
feedback loops or paths between its input and its output. The
reference to “feedback” simply means that some portion of the
output is returned to the input to form part of the system’s Fig. 1. Closed-loop system
excitation. [1] Vdesired is a DC voltage input with value equal 0.8*(V TEMP-
Closed-loop systems are designed to automatically achieve ), where VTEMP-MAX is the stop condition value taken from
MAX
and maintain the desired output condition by comparing it the data in experiment 1. In this case, V desired = 0.8*(4.54) = 3.6
with the actual condition. It does this by generating an error V. A UA741 single op-amp was used to implement the
signal which is the difference between the output and the summing block in the closed-loop system. The schematic
reference input. [1] diagram of the operational amplifier for the unity gain
When the closed-loop system is properly designed, the feedback (i.e. E(s)= Vdesired - VTEMP) is shown in Fig. 2.
output will be less sensitive to disturbances, less sensitive to
plant changes, and will have a lesser steady-state error with
respect to a desired value.
Another way to get a more accurate output (i.e. smaller
steady-state error) is to increase the system type of the plant.
This technique was examined using an integral controller,
wherein given an input step function, the steady state error
theoretically must approach zero.

II. EXPERIMENT, RESULTS, AND DISCUSSION


The experiment was divided into three parts. The first part
focused on converting the open-loop plant of Experiment 1
Fig. 2. Summing Block (Difference Amplifier)
into a closed-loop system. A proportional controller was then
cascaded in the forward path of the closed-loop system. The Gc(s) in Fig.1 is a controller configuration which was
behaviour of the proportional controller was analysed by varied in the next parts of the experiment.
taking the output voltage (VTEMP)with respect to time and the
steady-state errors of the system given a varying constant of
proportionality (kp) equal to 1, 2, 5, and 10. Afterwards, the B. Proportional Controller
proportional controller was replaced by an integral controller. The controller (in this case a proportional controller) with a
The output voltage with respect to time and the steady-state transfer function, Gc(s), is inserted in closed-loop system
modelled in Fig.1. To implement the proportional controller, a
UA741 single op-amp was used. The schematic diagram is
shown in Fig. 3.

The error signals for each gain that were obtained are
plotted in Fig. 4.

Error Results of the Proportional Controllers


1.2

Error (Vdesired - Vtemp)


Fig. 3. Proportional Controller (Non-Inverting Amplifier) 0.8
kp = 1
Given that the gain for the non-inverting amplifier in Fig.3 0.6 kp = 2
R1 R kp = 5
is given by the equation Av =1+ =1+ , 0.4
R2 1k kp = 10
the resistor R was decided to have values 0 (or shorted) for 0.2
kp = 1, 1kΩ for kp = 2, 3.9kΩ for kp = 5, and 9.1kΩ for kp = 10.
0
The behaviour of the proportional controller was observed 0 50 100 150 200 250 300 350
by changing the constant of proportionality (k p) – the gain of -0.2
the amplifier. The output voltages (VTEMP) of each gain was
Time in Seconds
measured every 30 seconds until steady-state and the results
are shown in Table 1. Then, the steady-state errors for each
gain is measured at the output of the summing block. The Fig. 4. Error Results of the Proportional Controller
steady-state error, ess, which from Fig.1. can be defined as
Vdesired - VTEMP,STEADY-STATE, that were obtained for kp = 1, 2, 5,
and 10 are 1.09 V, 0.91 V, 0.33 V, and 0.28 V respectively. Theoretically, the steady-state error can be computed as
follows:
TABLE I
OUTPUT VOLTAGES WITH RESPECT TO TIME
With the proportional controller providing simply a
constant multiplier kp to the forward gain, the plant will
TIME VTEMP (V) remain to be a first-order type 0 system.
(min:sec) kp = 1 kp = 2 kp = 5 kp = 10 Hence,
0:00 2.52 2.71 2.52 2.52 1
0:30 2.51 2.71 3.15 3.72 e ss = wherek =lim Gc ( s ) G(s) .
1:00 2.50 2.70 3.26 3.42 1+ k s →0
1:30 2.49 2.69 3.27 3.28 From the original open loop plant in the previous
2:00 2.50 2.69 3.27 3.30 experiment,
2:30 2.51 2.69 3.27 3.35
3:00 2.51 2.69 3.28 3.36 0.055
3:30 3.27 3.34 G ( s )= ∧Gc ( s )=k p
1
4:00 3.34 s+
4:30 3.33 144.155
5:00 3.33
5:30 3.32 For k p =1, ess =¿ 0.112
k p =2, ess =¿ 0.059
k p =5, ess =¿ 0.025
k p =10,e ss =¿ 0.012.

Hence, it can be observed that the steady-state error


decreases as the gain for the proportional controller increases.
This observation matches with the experimental result shown
in Fig.4. wherein the error V desired - VTEMP decreases as kp
increases.
To get the theoretical error reponse, we note that, 0.055
R (s) k p(
)
E( s)= 1
s+
1+Gc ( s ) G(s) Vtemp Gc ( s ) G( s) 144.155
= =
0.055 Vdesired 1+Gc ( s ) G(s) 0.055
G ( s )= 1+k p ( )
1 1
s+ s+
144.155 144.155
Since Gc(s) is a proportional controller,
Gc ( s ) =k p Vtemp 0.055 k p
V =
R(s)= desired , where Vdesired = 3.6 Vdesired s +0.0069+0.055 k p
s
So,
1 3.6
E( s)= ∙
0.055 s
1+k p ( )
1
s+
144.155
1 The theoretical step response obtained using the transfer
s+ function solved is shown in Fig.6.
144.155 3.6
E( s)= ∙
1 s
s+ + k (0.055)
144.155 p
The theoretical error response of the system using the
obtained E(s) is shown in Fig.5 where, again, error decreases
as kp increases.

Fig. 6. Theoretical Step Response of the System at different kp

Fig.7 shows the actual step response of the closed-loop


system with kp equal to 1,2,5 and 10. With the bulb not having
enough power to turn on, the output was measured to
Fig. 5. Error Step Response of Proportional Controller correspond to the ambient temperature which is at around 25-
27C for kp=1 and kp=2. With the bulb turning on at k p=5 and
kp=10, the output follows the expected first-order step
response curve.
Given that the system is still a first-order one, the expected
step response will be similar to the one obtained in the
From the data in Fig.7, it can be observed that the steady-
previous experiment wherein the curve resembles an
state output voltage of the system increases, which
exponential function.
consequently decreases the steady-state error, as k p increases.
This observation matches with the theoretical computations
Solving for the transfer function, we get, and analysis provided.

Fig. 7. Actual Step Response of the System at different kp


7:30 3.25 16:00 3.16
8:00 3.25 16:30 3.15

C. Integral Controller
The proportional controller was replaced with an integral
controller since the constant of proportionality (k p) needed is
only 1. A UA741 op-amp, a 1MΩ resistor and a 1µF capacitor
was used. An inverter was cascaded to the integral controller
to compensate for the negating effect of the integrating. The
inverter was implemented by also using a UA741 op-amp.
The schematic diagram of integral controller is shown in Fig.
8.
Fig. 9. Step Response of the Integral Controller

The error signal was also obtained in the experiment. The


plot is shown in Fig. 10.

Error Results of the Integral Controller


1.5
Error (Vdesired - Vtemp)

1
Fig. 8. Integral Controller (Integrator and Inverting Amplifier)
0.5
The output voltage (VTEMP) was measured every 30 seconds.
The results are shown in Table 2. 0
0 90 180 270 360 450 540 630 720 810 900 990
TABLE III -0.5
OUTPUT VOLTAGES WITH RESPECT TO TIME
Time in Seconds
TIME
TIME VTEMP
VTEMP (V) (min:sec
(min:sec) (V)
)
0:00 2.43 8:30 3.25 Fig. 10. Error Results of the Integral Controller
0:30 3.93 9:00 3.22
1:00 3.46 9:30 3.39
1:30 3.41 10:00 3.34 With the integral controller in cascade to the forward loop
2:00 3.39 10:30 3.37 gain of the closed-loop system, the overall transfer function
2:30 3.34 11:00 3.26 can be computed as follows:
3:00 3.20 11:30 3.29
3:30 3.24 12:00 3.29
4:00 3.22 12:30 3.17
4:30 3.22 13:00 3.15
5:00 3.2 13:30 3.12
5:30 3.23 14:00 3.11
6:00 3.21 14:30 3.11
6:30 3.20 15:00 3.12
7:00 3.25 15:30 3.13
R (s )
0.055 1 E ( s )=
)( 1 0.055
1 s 1+( )( )
s+ s 1
Vtemp Gc ( s ) G( s) 144.155 s+
= = 144.155
Vdesired 1+Gc ( s ) G(s) 1 0.055 s( s+ 0.0069) 3.6
1+ ( ) E ( s )= 2 ∙
s 1 s +0.0069 s +0.055 s
s+
144.155
0.055 0.055 At steady-state:
¿ =
1 1
(
s s+
144.155 ) 2
+ 0.055 s +
144.155
s+ 0.055
e ss =lim sE ( s ) =lim
s ( s ( s+0.0069 ))
3.6
=
0
∙ =0
s→0 s →0
2
s +0.0069 s +0.055 s 0.055
The system is now second-order and type 1 with a transfer
function shown in Fig.11. Due to the increase in system type,
overshoot can be observed in the output step response.
Oscillations are present in the output that decrease and reach
steady-state at a value equal to Vdesired. This implies that the
steady-state error can be computed to be 0.

Fig. 12. Theoretical and Actual Step Response

The plot in Fig.12 shows the actual and theoretical step


response of the system. From the computation of the error
response of the plant presented and as the plot labelled
‘Theoretical’ in Fig.12 shows, the steady-output should reach
the value Vdesired=3.6. The plot labelled as ‘Actual-initial’ is the
Fig. 11. Theoretical Step Response of the System with Integral Controller
response of the plant with integral controller at first trial. The
response matches with the theoretical one in terms of the
To get the theoretical error reponse, we note that, expected overshoot, oscillations, and estimated steady-state
value.
R (s)
E( s)=
1+Gc ( s ) G(s) However, on the succeeding trial, labelled as ‘Actual-final’
in Fig.12, the system’s response deviated from the expected
From experiment 1, theoretical response. Though overshoot was still present,
0.055 oscillations were not evident and the steady-state output
G ( s )= deviated from Vdesired. The observed discrepancy may be
1
s+ caused by a change in the plant itself. A component may have
144.155 been defective, the ambient conditions may have changed, and
some parameters in the system may have been modified. For
example, the VTEMP-MAX (steady-state value taken from the data
Since Gc(s) is an integral controller, in experiment 1) may have changed which in effect changes
1 the value of Vdesired.
Gc ( s ) =
s
V desired
R(s)= , where Vdesired = 3.6
s
Fig. 13. Theoretical and Actual Error Response

The plot in Fig.13 shows the theoretical and actual error


response of the system. As can be computed using the
equation for E(s) solved in the preceding section, theoretical
steady-state error should be 0 which matches the plot in
Fig.13 labelled as ‘Theoretical’.The plot labelled as ‘Actual-
initial’ is the error response of the system on the first trial
which matches the theoretical response. The plot labelled as
‘Actual-final’, is the error response of the system on the
succeeding trials, as explained before. The steady-state error
deviates from the expected value, which is theoretically 0, and
experimentally should be less than that of the proportional
controller. The cause of this deviation is the same as the
explanation given with the step response deviation.

III. CONCLUSION
The goal of the experiment was to modify the previously
constructed fully functional plant of a temperature-controlled
system into a closed-loop system. This was implemented
using a unity feedback gain, designed using a difference
amplifier. An additional forward gain was also introduced
using a proportional controller (a non-inverting amplifier) and
an integral controller (cascade of integrator and inverting
amplifiers). The latter one increases the type of the system
from 0 to 1, and hence decreasing the steady state error,
theoretically to 0, given a step input. Some significant
deviations from the theoretical computations were obtained in
the integral controller most likely because of the change in the
behavior of the original open loop system. This may be caused
by non-idealities, defective components, and/or the change in
ambient conditions.

REFERENCES
[1] (2018) Electronics Tutorials Closed-loop Control Systems [Online].
Available: https://www.electronics-tutorials.ws/systems/closed-loop-
system.html
[2] (2018) Electronics Tutorials Op-Amp [Online]. Available:
https://www.electronics-tutorials.ws/opamp/opamp_6.html
[3] (2018) Electronics Tutorials Op-Amp [Online]. Available:
https://www.electronics-tutorials.ws/opamp/opamp_8.html

You might also like