NATIONAL UNIVERSITY OF SINGAPORE

EE3302 Laboratory
Report
Ziegler-Nichols & Relay Tuning Experiments

Name: Goh Kai Hee

Matric Number:
1 Objectives
The objectives of this experiment are:

a) To estimate a first-order dead-time process model from an open-loop step

response test for the Ziegler-Nichols open-loop method of tuning.
b) To obtain the ultimate gain and ultimate period from a relay experiment for
the Ziegler-Nichols ultimate cycling method of tuning
c) To estimate a transfer function model of the process from the ultimate gain
and ultimate period

2 Equipment
1. DIGIAC 1750 Transducer and Instrumentation Trainer
2. Connecting Wire
3. Digital Multi-meter
4. Personal Computer
5. Matlab

6 Questions
1. Obtain temperatures in degree Celsius using Equation (1) and subtract them from
the ambient temperature to give the temperatures above ambient in Table 1.
2. Divide the temperatures above ambient by the input power u12 (=8.492) and record
the step response, y, in Table 1.

Refer to following page for Table 1.

 Table 1: Open-Loop Step Test (Ambient Temperature = 24.9 °C)
Time(minute) 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Voltage v, (V) 2.979 2.982 2.993 3.009 3.027 3.044 3.061 3.076 3.089 3.1
Temperature above Ambient (°C) 0 0.3 1.4 3 4.8 6.5 8.2 9.7 11 12.1
Unit Step Response y (°C) 0 0.004162 0.019423 0.04162 0.066593 0.090177 0.113762 0.134573 0.152608 0.167869

Time(minute) 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5

Voltage v, (V) 3.11 3.119 3.127 3.133 3.139 3.144 3.149 3.153 3.157 3.16
Temperature above Ambient (°C) 13.1 14 14.8 15.4 16 16.5 17 17.4 17.8 18.1
Unit Step Response y (°C) 0.181742 0.194228 0.205327 0.213651 0.221975 0.228912 0.235849 0.241398 0.246947 0.25111

Time(minute) 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5

Voltage v, (V) 3.162 3.164 3.167 3.169 3.171 3.172 3.174 3.176 3.177 3.178
Temperature above Ambient (°C) 18.3 18.5 18.8 19 19.2 19.3 19.5 19.7 19.8 19.9
Unit Step Response y (°C) 0.253884 0.256659 0.260821 0.263596 0.26637 0.267758 0.270532 0.273307 0.274694 0.276082

Time(minute) 15 15.5 16 16.5 17 17.5 18 18.5 19 19.5

Voltage v, (V) 3.18 3.18 3.181 3.181 3.181 3.182 3.183 3.183 3.184 3.184
Temperature above Ambient (°C) 20.1 20.1 20.2 20.2 20.2 20.3 20.4 20.4 20.5 20.5
Unit Step Response y (°C) 0.278856 0.278856 0.280244 0.280244 0.280244 0.281631 0.283018 0.283018 0.284406 0.284406

Time(minute) 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5

Voltage v, (V) 3.185 3.185 3.185 3.185 3.185 3.186 3.186 3.187 3.188 3.188
Temperature above Ambient (°C) 20.6 20.6 20.6 20.6 20.6 20.7 20.7 20.8 20.9 20.9
Unit Step Response y (°C) 0.285793 0.285793 0.285793 0.285793 0.285793 0.287181 0.287181 0.288568 0.289955 0.289955

Time(minute) 25 25.5 26 26.5

Voltage v, (V) 3.189 3.189 3.189 3.189 Steady state voltage vss = 3.189 V
Temperature above Ambient (°C) 21 21 21 21
Unit Step Response y (°C) 0.291343 0.291343 0.291343 0.291343
 3. Use Table 1 to plot the experimental unit step response
4. From the experimental step response,
(a) fit a first-order plus dead-time model according to the Ziegler-Nichols step
response method

According to the Ziegler-Nichols step response method and using a first-order plus
dead-time model:
KP = y (∞) = 0.3
R = KP/T = 0.0486
T = KP/R = 0.3/0.0486 = 6.17
L = intercept of tangent line with x-axis = 0.63
Thus, the first-order plus dead-time model is KPe-sL/(sT+1) = 0.3e-0.63s/(6.17s +1)
Y(s) = 0.3e-0.63s/s(6.17s +1) for unit step input
y(t) = 0 for t < L
y(t) = 0.3(1 – e-(t-0.63)/6.17)

(b) superimpose the unit step response from the estimated first-order plus dead-time
model on the experiment data,
The graph obtained is shown below.

Uni
t
Ste
p
Re
sp
on
se
y(t)
°C

Time (minutes)
(c) find a first-order plus dead-time transfer function model using the method of areas
Using the method of areas, we find A0 and A1 from the experimental unit step
response.
First, A0 is found with function trapz(t,y)
A0 = KP (T+L) ≈ 0.3*26.5 – trapz(t,y) = 0.3*26.5 – 6.191 = 1.759
From the graph. KP = 0.3 so T+L = 1.759/0.3 = 5.86 ≈ 6

Next, A1 is found by integrating y(t) from t = T+L to t = 0

A1 = trapz(t1,y1) where t1 amd y1 are data points from 0 to 5.5 minutes = 0.5348
A1 = KPTe-1 = 0.5348
Since KP = 0.3, T = A1 /(KPe-1) = 4.85
Since T+L = 5.86, L = 5.86 – T = 1.01

Thus, the first-order plus dead-time model we obtain is:

KPe-sL/(sT+1) = 0.3e-1.01s/(5.86s +1)
Y(s) = 0.3e-1.01s/s(5.86s +1) for unit step input
y(t) = 0 for t < L
y(t) = 0.3(1 – e-(t-1.01)/5.86)
(d) superimpose the unit step response from the estimated first-order plus dead-time
model on the experimental data.
The graph obtained is shown below.

(e) find a first-order plus dead-time transfer function using least-square estimation,
Using least square approximation,
Ɵ1 = 0.9233, Ɵ2 = 0.0232
KP = 2Ɵ2/1- Ɵ1 = 0.3026
T = h*Ɵ1/1- Ɵ1 = 6.0221
Thus the 1st order plus dead-time model obtained using least squares approximation is
KP (1 – e-t/T) = 0.3026 (1 – e-t/6.0221)
(f) superimpose the unit step response from the estimated first-order plus dead-time
model on the experimental data

The graph obtained is shown below.

5. Determine the ultimate period, Tu, the period of the relay oscillation from Table 2

Using steady state voltage vss = 3.189 V

Table 2: ON-OFF Interval of the Electronic Switch

No. tON (12V) tOFF (0V)

1 1:26:50 = 86.5 s 1:25:63 = 85.63 s

2 1:26:50 = 86.5 s 1:28:19 = 88.19 s
3 1:28:69 = 88.69 s 1:30:40 = 90.4 s
Average 87.23 s 88.07 s

Referring to Table 2, we can calculate ultimate period Tu

Tu = tON + tOFF = 87.23 + 88.07 = 175.93 = 2.93 mins

6. Determine a, half the peak-to-peak amplitude of the temperature oscillations in
Table 3.
Table 3: Maximum and Minimum Temperature of the Oscillation
No. Maximum Temperature (°C) Minimum Temperature (°C)
1 47.7 45.0
2 47.7 45.0
3 47.9 45.0
Average 47.8 45.0

From Table 3, we can calculate half the peak-to-peak amplitude, a:

a = (47.8-45)/2 = 1.4 °C
7. Calculate Ku, the ultimate gain
Ku = (4d)/(πa)
122
= 4*( )/ (π*1.4)
2
= 65.5

8. By inspecting the experimental data, give an estimate of the dead-time, L. Together

with Kp, Ku and Tu obtained previously, find a second-order plus dead-time transfer
function model:
Kp
G p ( s )= e−sL
(s T 1+1)(s T 2 +1)

From the second-order plus dead-time transfer function, substitute s = jω (system

oscillating at marginal stability)
Given that the conditions for oscillation are |G( j ωu)| = 1/ Ku and argG( jω u) = −π,

1 Kp 1
we have: =|G p ( s )|=|G ( jω u )|= 2 2 =
Ku √ ωu T 1+1 √ ωu T 2+1 K u
2 2

We have obtained previously Kp = 0.3, Ku = 65.5, Tu = 2.93 and by observing the

experimental data L = 0.63.
2π
Thus ω u= =2.14 and we obtain the equation:
Tu
 0.3 1
=
2
√2.14 T 2 2 2
+1 √2.14 T +1 65.5
1 2

Now, given that argG( jω u) = −π , we have

−ω u L−tan −1 ω u T 1−tan−1 ω u T 2=−π
−2.14 (0.63)−tan−1 2.14 T 1−tan−1 2.14 T 2 =−π

Solving the 2 equations simultaneously, we obtain

T 1=0.277
T 2=15.5
Thus the second order plus dead-time model is
0.3
G p ( s )= e−s 0.63
(0.277 s+1)(15.5 s +1)

9. Superimpose the unit step response from the estimated second-order plus dead-time
model on the experimental data.

The graph obtained is shown below.