You are on page 1of 3

EEE 101 WFX/WFU October 31, 2018

Experiment 2: Closed-Loop,
Unity Feedback System
Herschel Faye Fabregas Victor Jann Boie Salanga Luis Chino Turdanes
2013-18657 2014-89412 2013-18447

Electrical and Electronics Engineering Institute


University of the Philippines, Diliman

​ In this experiment, a closed-loop, unity feedback


Abstract— of the difference between the input voltages. For this case, the
system was implemented with the circuit from the first resistance value we used is 100kΩ each, and the output of the
experiment used as the plant. Control blocks were used, plant is subtracted to the set point which is 3.96V.
particularly for proportional gain control and integral control.

I. INTRODUCTION
HIS experiment revolved around building a closed-loop,
unity gain feedback system and investigating the
performance of putting a proportional gain or an integrator as
the controller. Previously, an open-loop temperature control
system was modelled. This model was used in this experiment
as the plant which output will be the measured value that
served as the feedback of the system. This was then compared
to the desired value which was 80% of the highest recorded
voltage output of the open-loop control system from the first
experiment. The control variable of the system was the Fig. 2 Unity Gain Differential Amplifier [1]
temperature which was reflected by the output voltage
multiplied to 10.
B. The Controller Block
II. EXPERIMENT SETUP There were two controller blocks used. One is for a
The block diagram of the closed-loop, unity feedback proportional gain control and the other is for integral control.
system to be implemented for this experiment is shown in i. Proportional Gain Controller
Figure 1.

Fig. 1 Control System Setup


Fig. 3 Proportional Gain Controller
The system is composed of the step input voltage, the
controller, the plant and the output voltage which will also A non-inverting amplifier as shown in Figure 3 is used for
R
serve as the feedback to the system. The step input voltage this stage. With the equation for amplifier gain, k p = 1 + R2 ,
1
magnitude is 3.96V. The controller Gc(s) will either be a pure the resistor values are varied to get k​p = 1, 2, 5, and 10,
gain controller or an integrator controller. The plant will be the respectively.
open-loop control system derived from Experiment 1.
ii. Integral Controller
This controller, on the other hand, is a unity gain amplifier
A. The Summing Block
cascaded to an integrating amplifier, so that a gain of 1s can be
For the summing block, a unity-gain differential amplifier
achieved. The schematic diagram for this part is as shown in
as in Figure 2 is constructed. A unity-gain differential
Figure 4.
amplifier has equal values for R1, R2, R3, and R4 and a gain
EEE 101 WFX/WFU 2

Fig. 4 Integral Controller

C. The Plant
As have been mentioned in our pre-demo, we had to redo
the first experiment because we had to replace the LF353,
which means there was a change in the original plant thereby
possibly also changing the output. As such, our new data are
as recorded in the spreadsheet that we submitted. The transfer Fig. 6 Integral Controller Output
function for our new plant is now

22.4
G(s) = 370s + 1

III. RESULTS AND ANALYSIS


The figure below shows the step responses, both the
expected and actual, as a result of the application of a
proportional gain controller.

Fig. 7 Error vs. Time

The steady-state error for a proportional controller to a


1
step response, from our pre-demo, is numerator + 1 . The
numerator in this case would be our proportional control gain,
k​p​. To get the steady-state error to an input of 3.96V, we need
to subtract the input and the ambient temperature, giving us a
difference of 1.16V. The final formula for the E​ss would then
be E ss = k1.16
p +1
. This is used to compute for the theoretical
steady-state error. As for the actual error as a result of our
experiment, we use E ss = V ss − V desired .

TABLE I
COMPARISON OF STEADY-STATE ERROR
k​​p Theoretical Ess Actual Ess
Fig. 5 Proportional Gain Control Results at Various Gains 1 0.58 1.13
2 0.387 0.91
Meanwhile, Figure 6 shows the resulting output from the 5 0.1933 0.49
application of an integral controller, and Figure 7 shows the 10 0.105 0.13
steady-state error vs. time of both the proportional (kp = 1, 2,
5, 10) and integral controllers. It can be observed from the data above that as k​p is
increased, the experimental results get closer to the computed
theoretical value.
EEE 101 WFX/WFU 3

Mathematically, applying a non-inverting amplifier into zero, we could theoretically produce an infinite loop gain
V R R1
the system, V out = R2 + 1 , let k = R +R and substituting to thus, zero steady-state error.
in 1 1 2
the equation we get
V. APPENDIX
V out
V in
= 1
k [ ] 1
1
A(s)k
+1
=
A(s)
1+A(s)k ​(1) TABLE II
OUTPUT VOLTAGES OF THE CLOSED-LOOP SYSTEM
Time Integrato
1 (sec) kp = 1 kp = 2 kp=5 kp=10 r
Dividing by A(s) then multiplying by k ,
0 2.8 2.8 2.8 2.8 2.8
V out
= 1
= 1 [ 1
1+ A(s)k 1
− A(s)k ] ​(2)
30 2.82 2.86 3.06 3.08 3.28
V in
[ 1
A(s)k +1 ] k
[ 1
A(s)k +1 ] 60 2.82 2.91 3.35 3.45 3.65

90 2.82 2.98 3.46 3.77 4.02


Then subtracting A(s)k to the numerator yields
120 2.82 2.99 3.47 3.83 4.33
V out
V in
= 1
k [1 − 1
1+A(s)k ] (3) 150 2.83 3 3.46 3.82 4.57

180 2.83 3.02 3.46 3.81 4.42


1 1
Where k is the ideal gain and 1 − 1+A(s)k is the gain 210 2.83 3.03 3.47 3.83 4.22
error. As A(s)k approaches infinity, the equation will be
V 240 2.83 3.05 3.47 3.81 4.11
reduced to V out = 1k , so higher gain, less error. In this case,
in
270 2.83 3.05 3.47 3.83 4.27
A(s)k is the numerator in our pre-demo.
300 2.83 3.05 3.47 3.83 4.3
As for the integrator, it has a theoretical Ess of 0, but 330 2.83 3.05 3.47 3.83 4.28
our experiment yielded a negative value, -0.3. We think that
this is just acceptable and understandable, because looking 360 2.83 3.05 3.47 3.83 4.24
at the graph, it can be seen that the theoretical response is 390 2.83 3.05 3.47 3.83 4.26
still in the process of oscillating. It might be that our system
420 2.83 3.05 3.47 3.83 4.22
is just too fast due to the materials we used, like the
capacitor or the operational amplifier. 450 2.83 3.05 3.47 3.83 4.26

480 2.83 3.05 3.47 3.83 4.23


One advantage for the proportional gain controller is
that increasing the gain decreases the error as presented in 510 2.83 3.05 3.47 3.83 4.26
the discussion above. However, higher k also gives a higher 540 2.83 3.05 3.47 3.83 4.26
chance of instability as we have learned in class. Moreover,
using a 2nd-order system such as the integrator drives the
system error to zero. But using this type also makes the REFERENCES
system prone to various non-idealities. [1] "Op-amp Varieties", Hyperphysics.phy-astr.gsu.edu, 2018. [Online].
Available:http://hyperphysics.phy-astr.gsu.edu/hbase/Electronic/opampvar
IV. CONCLUSION 6.html#c1. [Accessed: 19- Oct- 2018].
[2] Datasheets: LM35, LF353, TIP35C, uA741
Doing the first and second experiments, we further [3] EEE 101 Discussion Notes
observed the differences between a first-order and a
second-order system. The experiment results also confirmed
the lessons taught in EEE 101. We know from the
discussions in class that one cannot simply label a system
just based on the system response, but the experiment was
helpful in a way that we experienced firsthand and saw why
is that. Proportional controller is mostly used to decrease the
steady state error of the system wherein as the gain factor k
increases, the error decreases. An integral controller, on the
other hand, is expected to entirely eliminate the error in the
system since we are introducing a pole at the origin into the
system, and via the Final Value theorem, as “s” approaches

You might also like