ME 211 – Modeling and Analysis of Dynamic Systems

MOTOR MODELING COURSE PROJECT SOLUTIONS

Part 1
The static equation relating the command voltage from the computer processor to the voltage
applied to the motor is
e ( t ) = K a ec ( t ) (1)

where Ka = 2.4. Applying KVL around the electrical loop, the differential equation relating the
voltage applied to the motor to the current and back EMF voltage is
−e ( t ) + LIm ( t ) + RI m ( t ) + eb ( t ) =
0 (2)

where R = 8.33 Ω and L = 6.27 mH. The static motor equation relating the current to the torque
applied to the motor shaft is
Ti ( t ) = K t I m ( t ) (3)

where Kt = 3.95∙10–2 N∙m/A. The static motor equation relating the back EMF voltage to the
motor shaft angular velocity is
eb ( t ) = K vωm ( t ) (4)

where Kv = 3.95∙10–2 V/(rad/s). Applying D’Alembert’s Law to the motor shaft

− J mω m ( t ) − Bmωm ( t ) − T f sgn ωm ( t )  + Ti ( t ) =
0 (5)

where Tf = 2.5·10–3 N·m, Jm = 9.9∙10–7 kg∙m2, and Bm = 1.0∙10–6 N∙m∙s. Combining equations (1),
(2), and (4)
R K K
Im ( t ) =
− I m ( t ) − v ωm ( t ) + a ec ( t ) (6)
L L L
Combining equations (3) and (5)
B Tf Kt
ω m ( t ) =
− m ωm ( t ) − sgn ωm ( t )  + Im (t ) (7)
Jm Jm Jm
Angular position is related to angular velocity by
 Motor Modeling Course Project Solutions

θm ( t ) = ωm ( t ) (8)

Equations (6)–(8) are two first order differential equations describing the motor dynamics.

Part 2
Taking the Laplace transform of equation (6), with zero initial conditions, and solving for the
current
Kv Ka
Im ( s ) =
− Ωm ( s ) + Ec ( s ) (9)
sL + R sL + R
Taking the Laplace transform of equation (5), with zero initial conditions and ignoring the
Coulomb friction, and rearranging
( J m s + Bm ) Ωm ( s ) =
Kt I m ( s ) (10)

Taking the Laplace transform of equations (8), with zero initial conditions
sΘ m ( s ) =
Ωm ( s ) (11)

Combining equations (9)–(11)

Θm ( s ) K a Kt
= (12)
Ec ( s ) J m Ls + ( Bm L + J m R ) s 2 + ( Bm R + K t K v ) s
3

Inserting numerical parameters into equation (12)

Θm ( s ) 5.918 ⋅10−2
= (13)
Ec ( s ) 6.207 ⋅10−9 s 3 + 1.199 ⋅10−5 s 2 + 7.629 ⋅10−4 s

The transfer function does not have any finite zeros. The transfer function poles are located at s =
0, –65.90, and –1865. The pole located at 0 does not have a time constant, rise time, or 2%
settling time associated with it. The pole located at –65.90 has a time constant of 1.517∙10–2 s, a
rise time of 3.338∙10–2 s, and a 2% settling time of 6.070∙10–2 s. The pole located at –1865 has a
time constant of 5.362∙10–4 s, a rise time of 1.180∙10–3 s, and a 2% settling time of 2.145∙10–3 s.

Part 3
The Simulink diagram is shown in Figure 1 and the results for the step, ramp, and six sine inputs
are shown in Figures 2–9, respectively. For each case the nonlinear simulation is closer to the
experimental results than the linear simulation is to the experimental results. The linear

2
 Motor Modeling Course Project Solutions

simulations over predict the experimental results since the linear simulations do not include
Coulomb friction. For the sinusoidal cases, the nonlinear and linear simulations deviate from the
experimental data more as the command voltage frequency increases. This is possibly due to
unmodeled dynamics. Also, for the sinusoidal experiments, the simulated motor positions
oscillated about a mean while the experimental motor positions drifted slightly from this mean.
This indicates the motor friction characteristics are direction dependent.

Kv/L

Back EMF Gain

-K-

Viscous
Friction
2
omega
Coulomb
Friction
R/L

Current Gain 1 1
1
s s
theta
Angular Velocity Angular Position
1 Integrator Integrator
-K-
s
1 Ka/L
Current Current Current Gain
Vc
Command Integrator Saturation
Voltage Gain
3
Im

Figure 1: Motor Simulink Diagram.

3
 Motor Modeling Course Project Solutions

4
2

voltage (V)
0
-2
-4
0 0.5 1 1.5 2 2.5 3

400
angular position (rad)

200

0
nonlinear linear measured
-200
0 0.5 1 1.5 2 2.5 3
time (s)

Figure 2: Simulation and Experimental Results for Step Command Voltage Input.

2
voltage (V)

-2
-4
0 1 2 3 4 5

400
angular position (rad)

200

0
nonlinear linear measured
-200
0 1 2 3 4 5
time (s)

Figure 3: Simulation and Experimental Results for Ramp Command Voltage Input.

4
 Motor Modeling Course Project Solutions

2
1

voltage (V)
0
-1
-2
0 5 10 15 20 25 30

200
angular position (rad)

nonlinear linear measured

150

100

50

0
0 5 10 15 20 25 30
time (s)

Figure 4: Simulation and Experimental Results for Sine (2 rad/s) Command Voltage Input.

2
1
voltage (V)

0
-1
-2
0 5 10 15

100
angular position (rad)

nonlinear linear measured

50

0
0 5 10 15
time (s)

Figure 5: Simulation and Experimental Results for Sine (4 rad/s) Command Voltage Input.

5
 Motor Modeling Course Project Solutions

2
1

voltage (V)
0
-1
-2
0 1 2 3 4 5 6

40
angular position (rad)

nonlinear linear measured

30

20

10

0
0 1 2 3 4 5 6
time (s)

Figure 6: Simulation and Experimental Results for Sine (10 rad/s) Command Voltage
Input.

2
1
voltage (V)

0
-1
-2
0 0.5 1 1.5 2 2.5 3

20
angular position (rad)

nonlinear linear measured

15

10
5

0
0 0.5 1 1.5 2 2.5 3
time (s)

Figure 7: Simulation and Experimental Results for Sine (20 rad/s) Command Voltage
Input.

6
 Motor Modeling Course Project Solutions

2
1

voltage (V)
0
-1
-2
0 0.5 1 1.5 2
angular position (rad)
15
nonlinear linear measured
10

0
0 0.5 1 1.5 2
time (s)

Figure 8: Simulation and Experimental Results for Sine (30 rad/s) Command Voltage
Input.

2
1
voltage (V)

0
-1
-2
0 0.2 0.4 0.6 0.8 1 1.2

8
angular position (rad)

nonlinear linear measured

6

0
0 0.2 0.4 0.6 0.8 1 1.2
time (s)

Figure 9: Simulation and Experimental Results for Sine (50 rad/s) Command Voltage
Input.

Part 4
Using equation (12), the system analytical frequency response is

7
 Motor Modeling Course Project Solutions

Θ m ( jω ) K a Kt
= (14)
Ec ( jω ) J m L ( jω ) + [ Bm L + J m R ] ( jω )2 + [ Bm R + K t K v ] ( jω )
3

Equation (14) can be rewritten as

Θ m ( jω )
= Re (ω ) + Im (ω ) j (15)
Ec ( jω )

where
K a K t [ Bm L + J m R ] ω 2
Re (ω ) = − 2 2
(16)
( Bm L + J m R ) ω 2  + ( Bm R + K t K v ) ω − J m Lω 3 

K a K t ( Bm R + K t K v ) ω − J m Lω 3 
Im (ω ) = − 2 2
(17)
( Bm L + J m R ) ω 2  + ( Bm R + K t K v ) ω − J m Lω 3 

The analytical magnitude and phase frequency responses, respectively, are

( Re (ω ) ) + ( Im (ω ) )
2 2
M (ω )
= (18)

 Im (ω ) 
φ (ω ) = tan −1   (19)
 Re (ω ) 
The experimental and nonlinear simulation magnitude and phase frequency responses at 2, 4, 10,
20, 30, and 50 rad/s are determined graphically from Figures 4–9, respectively. The experimental
magnitude and phase frequency responses at 2 rad/s, respectively, are
 117.5 − ( 7.75 ) 
=M ( 2 ) 20
= log10   28.77 dB (20)
 2 − ( −2 ) 

φ ( 2) =
360
( 25.92 − 26.7 ) =
−89.38° (21)
2π / 2
The nonlinear simulation magnitude and phase frequency responses at 2 rad/s, respectively, are
 98 − ( 0 ) 
=M ( 2 ) 20
= log10   27.78dB (22)
 2 − ( −2 ) 

φ ( 2) =
360
( 25.92 − 26.7 ) =
−89.38° (23)
2π / 2
The experimental magnitude and phase frequency responses at 4 rad/s, respectively, are

8
 Motor Modeling Course Project Solutions

 57.1 − ( 2.2 ) 
=M ( 4 ) 20
= log10   22.75dB (24)
 2 − ( − 2 ) 

φ ( 4) =
360
(12.96 − 13.4 ) =
−100.8° (25)
2π / 4
The nonlinear magnitude and phase frequency responses at 4 rad/s, respectively, are
 48.8 − ( 0 ) 
=M ( 4 ) 20
= log10   21.73dB (26)
 2 − ( −2 ) 

φ ( 4) =
360
(12.96 − 13.36 ) =
−91.67° (27)
2π / 4
The experimental magnitude and phase frequency responses at 10 rad/s, respectively, are
 22.7 − (1.24 ) 
=M (10 ) 20
= log10   14.59 dB (28)
 2 − ( −2 ) 

φ (10 ) =
360
( 5.184 − 7.37 ) =
−106.6° (29)
2π /10
The nonlinear magnitude and phase frequency responses at 10 rad/s, respectively, are
 19.44 − ( 0 ) 
=M (10 ) 20
= log10   13.73dB (30)
 2 − ( −2 ) 

φ (10 ) =
360
( 5.184 − 5.345) =
−92.24° (31)
2π /10
The experimental magnitude and phase frequency responses at 20 rad/s, respectively, are
 11 − ( −1.45 ) 
=M ( 20 ) 20
= log10   9.862 dB (32)
 2 − ( −2 ) 

φ ( 20 ) =
360
( 2.592 − 2.7 ) =
−123.8° (33)
2π / 20
The nonlinear magnitude and phase frequency responses at 20 rad/s, respectively, are
 9.485 − ( 0 ) 
=M ( 20 ) 20
= log10   7.500 dB (34)
 2 − ( −2 ) 

φ ( 20 ) =
360
( 2.592 − 2.68) =
−100.8° (35)
2π / 20
The experimental magnitude and phase frequency responses at 30 rad/s, respectively, are

9
 Motor Modeling Course Project Solutions

 7.04 − ( −1.35 ) 
=M ( 30 ) 20
= log10   6.434 dB (36)
 2 − ( −2 ) 

φ ( 30 ) =
360
(1.728 − 1.816 ) =
−151.3° (37)
2π / 30
The nonlinear magnitude and phase frequency responses at 30 rad/s, respectively, are
 6.1 − ( 0 ) 
=M ( 30 ) 20
= log10   3.665dB (38)
 2 − ( −2 ) 

φ ( 30 ) =
360
(1.728 − 1.788) =
−103.1° (39)
2π / 30
The experimental magnitude and phase frequency responses at 50 rad/s, respectively, are
 2.61 − ( 0.805 ) 
M ( 50 ) = 20 log10   = −6.912 dB (40)
 2 − ( −2 ) 

φ ( 50 ) =
360
(1.037 − 1.097 ) =
−171.9° (41)
2π / 50
The nonlinear magnitude and phase frequency responses at 50 rad/s, respectively, are
 3.382 − ( 0.01) 
M ( 50 ) = 20 log10   = −1.483dB (42)
 2 − ( −2 ) 

φ ( 50 ) =
360
(1.037 − 1.0755) =
−110.3° (43)
2π / 50
The magnitude and phase frequency responses are given below in Figure 10. The nonlinear
magnitude frequency response is very close to the experimental magnitude frequency response,
with slight deviation for the data at 50 rad/s. The analytical magnitude frequency response
slightly over predicts the experimental magnitude frequency response for all frequencies, with
greater deviation seen in the data at 50 rad/s. The nonlinear and analytical phase frequency
responses were very close to the experimental phase frequency response at frequencies of 2 and
4 rad/s. For larger frequencies the nonlinear and analytical phase frequency responses under
predicted the experimental phase frequency responses. The deviations were greater for the
nonlinear phase frequency responses than the analytical phase frequency responses, and
increased with frequency. The deviations between the analytical and experimental data and

10
 Motor Modeling Course Project Solutions

between the nonlinear and experimental data for higher frequencies are possibly due to
unmodeled dynamics.

40

magnitude (dB) 20
analytical
0 experimental
nonlinear
-20 0 1 2
10 10 10

-50
phase (deg)

-100
analytical
-150 experimental
nonlinear
-200 0 1 2
10 10 10
frequency (rad/s)

Figure 10: Analytical, Nonlinear Simulation, and Experimental Magnitude and Phase
Frequency Responses.

Nomenclature
Bm : motor viscous friction coefficient (1.0·10–6 N·m·s)
eb : motor back emf voltage (V)
ec : command voltage (V)
Im : motor current (A)
Jm : motor inertia (9.9·10–7 kg·m2)
Ka : amplifier gain (2.4)
Kt : motor torque constant (2.74·10–2 N·m/A)
Kv : motor velocity constant (2.74·10–2 V/(rad/s))
R : motor electrical resistance (12.1 Ω)
Tf : motor Coulomb friction torque magnitude (2.5·10–3 N·m)
Ti : motor input torque (N·m)
θm : motor angular position (rad)

11
 Motor Modeling Course Project Solutions

θm : motor angular velocity (rad/s)

ωm : motor angular velocity (rad/s)

ω m : motor angular acceleration (rad/s2)

12