Modelling of DC Drive PI Controllers

for the Speed and the Torque Loops

by Using Matlab-Simscape
Mohd Razani, Nik Syahiruddin, Norhaisam
Faculty of Electrical Engineering
Universiti Teknologi Malaysia
81310 UTM, Skudai
Johor Malaysia

Abstract - This report will present about the feature of a DC motor is its simple control
configuration of the DC drive system's requirements with excellence performance.
proportional-integral (PI) speed and torque
controllers and is measured using
MATLAB/SIMULINK. Where the inside loop is
the torque loop, the cascade control structure is
assumed. The model will begin with the torque
loop, followed by the speed loop. In this analysis,
the design of the Torque controller and speed
controller will use an estimated method for PI
controller design. Then, using a broad signal Figure 1: DC motor
simulation fed by the given speed reference
profile, the design will be tested. With the The torque is produced as a result of interaction
estimated method for the PI controller of field flux with current in armature
configuration, the torque profile should be the conductors and is given by
same. In addition, the design will use Unipolar
and Bipolar switching schemes and will
compare the torque ripples. Te = kt ia (1)

Keywords: DC Drive, Proportional-Integral (PI) where kt is a constant depending on motor

Speed and Torque Controllers, MATLAB- windings and geometry is the flux per pole due
SIMCAPE software. to the field winding

I. INTRODUCTION Due to the commutator action, the

torque is always quadrature to the field flux and
Like other rotating machines, a DC machine
hence the torque is directly controlled by the
consists of stator and rotor. Stator is stationary part
armature current. Ideally, no magnetic coupling
where the field flux is produced, while the rotor is
exists between the torque and field components.
rotating part where the armature winding is placed.
This means that any changes in the armature
Field flux is obtained either from permanent
current will result in an instantaneous change in
magnet or from field winding excitation. Field
torque.
fluxinteracts with current carrying conductors in
armature to produce torque.
When the armature rotates, the flux
Commutator in armature circuit will ensure
linking the armature winding will vary with time
that the torque production is always at its
and therefore according to Faraday’s law, an emf
maximum, regardless of rotor position. Due to the
will be induced across the winding. This
windings on the rotor, the brushes and the
generated emf, known as the back emf, depends
commutators, when compared to AC motors, DC
on speed of rotation as well as on the flux
motors are heavier, more expensive, less robust,
produced by the field and is given by:
require regular maintenance and cannot be
operated in explosive environment. The attractive
ea = kt Φω (2)
 Among the advantages of using a cascade
structure is that the control signal of the outer
loop becomes the reference input to the inner
loop. This means that the reference to the inner
loop can be clamped to its allowable range by
clamping this control signal. If the bandwidth of
the outer loop is much smaller than the inner
Figure 2: Armature circuit loop, the gain of the inner loop can be assumed
unity when designing the controller for the outer
loop thus simplifying the design process.
Using KVL around the armature circuit, we
can write The power electronic converters are used
to obtain an adjustable DC voltage applied to the
d ia armature of a DC motor. There are basically two
v t=i a Ra + La + ea (3)
dt types of converter normally employed in DC
drives which is controlled rectifier and switch–
In steady state condition, mode converter.

v t=I a R a + Ea (4) Switch–mode converters normally

operate at high frequency. As a result of this, the
In terms of torque and speed the steady state average output voltage response is significantly
equation can be written as: faster than the controlled rectifier, in other words
the bandwidth of a switch–mode rectifier is
T higher compared to the controlled rectifier and
v t= R + k Φω (5) the armature current ripple is relatively less than
kt Φ a t
the controlled rectifier circuit when the same
amount of inductance present in the armature
which gives:
circuit.
vt T
ω= − (6)
k t Φ ( k||tΦ )2 R a

Thus three methods can be used to control

the speed: Vt, Φ and Ra. Figure 4: Four
Quadrant Converter
Three controlled outputs which are
common in electrical drive systems are the This paper will cover the designing of PI
position, speed, and torque. The typical control controllers for the speed and the torque loops for
structure of industrial drive systems is the cascade the DC motor. This DC Motor is to be fed by a 4-
structure as shown in Fig. 3. quadrant DC-DC converter with input DC
voltage Vdc of 100 V. It is desired that the motor
speed to follow the profile shown in Fig. 5. The
parameters of the DC motor are: Ra = 2Ω, La =
6mH, ke = 0.1V/rad/s, J = 150 x 10-6kgm2 and B

Figure 3: Cascade control structure

It consists of multiple loops: with the inner-

most loop being the fastest and the outer-most loop
= 10 x 10-6 kg-m2/s. The frequency and the peak
being the slowest. Typically, the innermost loop is
value of the triangular waveform for the
the torque loop, followed by speed and position
modulator are 15 kHz and 10 V, respectively.
loops. Depending on the controlled output, the
Figure 5: Motor Speed Profile
outer-loop(s) can be removed or added if needed.
 II. METHODOLOGY The subscript “I” refers to the current loop.
The open-loop transfer function, GI,OL (s) of the
Motion control systems often must respond simplified current loop in Fig. 6(c) is
k iL s 1/ Ra
to large changes in the desired (reference) values of
the torque, speed, and position. They must reject
large, unexpected load disturbances. For large
G I ,OL ( s )=
s [1+
]
k iI /k pI
k PWM
1+
s
1/τ e
changes, the overall system is often nonlinear. This (8)
nonlinearity comes about because the mechanical
load is often highly nonlinear. Additional To select the gain constants of the PI
nonlinearity is introduced by voltage and current controller in the current loop, a simple design
limits imposed by the power-processing unit and the procedure, which results in a phase margin of
motor. 90 degrees, Select the zero (kiI/kpI) of the PI
controller to cancel the motor pole at (1/τ e) due
The first step is to assume that, around the
to the electrical time-constant τ e of the motor.
steady-state operating point, the input reference
Under these conditions,
changes and the load disturbances are all small. In
such a small-signal analysis, the overall system can
k iI 1
be assumed to be linear around the steady-state = ∨k pL =τ e k iI (9)
operating point, thus allowing the basic concepts of k pI τ e
linear control theory to be applied.
Cancellation of the pole in the motor
Based on the linear control theory, once the transfer function renders the open-loop transfer
controller has been designed, the entire system can function to be
be simulated on a computer under large-signal
conditions to evaluate the adequacy of the k I ,OL
controller. The controller must be "adjusted" as G I ,OL ( s )= (10)
s
appropriate.
where
i. The Design of Torque Controller Loop
k iI k PWM
We will begin with the innermost loop in Fig. k I ,OL = (11)
6(a). In permanent-magnet dc motors in which Φf is Ra
constant, the current and the torque are proportional
to each other, related by the torque constant k T. In the open-loop transfer function of Eq.
Therefore, we will consider the current to be the 10, the crossover frequency ωcI = kI,OL. We will
control variable because it is more convenient to select the crossover frequency f cI =ωcI /2π of
use. Notice that there is a feedback in the current the current open-loop to be approximately one
loop from the output speed. This feedback dictates to two orders of magnitude smaller than the
the induced back-emf. Neglecting T L, and switching frequency of the power-processing
considering the current to be the output, E a(s) can be unit in order to avoid interference in the control
calculated in terms of Ia (s) in Fig. 6(a) as Ea(s) = loop from the switching-frequency noise.
kT k E
I (s). Therefore, at the selected crossover
s J eq a
frequency, from Eq. 11,
Therefore, Fig. 6(a) can be redrawn as shown
ω cI Ra
in Fig. 6(b). Notice that the feedback term depends k iI = (12)
inversely on the inertia Jeq. Assuming that the inertia k PWM
is sufficiently large to justify neglecting the
feedback effect, we can simplify the block diagram, This completes the design of the torque
as shown in Fig. 6(c). The current-controller in Fig. (current) loop, where the gain constants k pI and
6(c) is a proportional-integral (PI) error amplifier kiI can be calculated from Eqs. 9 and 12.
with the proportional gain kpI and the integral gain
kiI. Its transfer function is given by Eq. 7.

V c (s) k k s
E (s ) s s [
=k p + i = i 1+
k i /k p ] (7)
 (a) ki
(b) k p=
100
(c) 1
Figure 6: Design of Torque Control Loop k i=1; k p=
ki 100
k p+ ; k =1 , k =0
s p

sk p+ k i
s
20 log k i =75.1
kp
[( )
ki s
ki
+1
] log k i =
75.1
20
75.1
s k i=10 20
=5.6885 x 103

To maintain zero at 100 rad/s;

k i ( sτ +1 )
5.6885 x 10 3
s k p=
100
ii. The Design of Speed Loop
kp 1
20 log k i +20 log s
ki ( )
+ 20 log
s
We will select the bandwidth of the speed
loop to be one order of magnitude smaller than
G that of the current (torque) loop. Therefore, the
closed-current loop can be assumed to be ideal
for design purposes and represented by unity, as
shown in Fig. 8. The speed controller is of the
proportional-integral (PI) type. The resulting
G open-loop transfer function, GΩ,OL (s) of the speed
loop in the block diagram of Fig. 8 is as follows,
where the subscript “Ω” refers to the speed loop.

ki Ω kT
G Ω , OL ( s )= [ 1+ s / ( k i Ω /k p Ω ) ] 1 (13)
s sJ eq
k i /k p

−20 d i /d x Figure 8: Block Diagram of Speed Loop

Eq. 13 can be rearranged as

Figure 7: Frequency Response of ki Ω kT 1+ s / ( k i Ω /k p Ω )

Torque Control Loop GΩ , OL ( s )=
( J eq )( s2 ) (14)

k i=1; This shows that the open-loop transfer

function consists of a double pole at the origin.
ki At low frequencies in the Bode plot, this double
=ω z
kp pole at the origin causes the magnitude to decline
ki 1 at the rate of – 40 db per decade while the phase
k p= = angle is at −1800. We can select the crossover
ωz ωz
frequency ωcΩ to be one order of magnitude
ki smaller than that of the current loop. Similarly,
=100
kp we can choose a reasonable value of the phase
 margin Φpm,Ω. Therefore, Eq. 14 yields two 1 1
equations at the crossover frequency:
= =0.1
v t 10

s 3. vd

|( |
1+
k iΩ v d=100
ki Ω kT ( )
k pΩ (15)

J eq ) s2 s= jωc Ω
=1 4. Subsystem

Ra = 2Ω, La = 6mH, ke = 0.1V/rad/s, J =

150 x 10-6kgm2 and B = 10 x 10-6 kg-m2/s
and

s Analyze using linear analysis for bode plot and

|
1+ pole zero map
ki Ω
k k ( ) k pΩ (16)

( )
∠ iΩ T
J eq s2 s = jω cΩ
o
=−180 + ϕ pm ,Ω

The two gain constants of the PI controller can be

calculated by solving these two equations.

Set PID = 1/0.0667 and 1

20 log k i =−40.5

−40.5
log k i =
20
−40.5
20
k i=10 =9.4406 x 10−3

III. MATLAB/SIMULINK SIMULATION

& DISCUSSION FOR PI CONTROLLER
DESIGN
Figure 10: Bode Plot
Simulation of the MATLAB/SIMULINK will be
detailed out in this subtopic for PI controller Figure 11: Pole Zero Map
design.
i. Torque Controller From pole zero map
ω p 1=−296
ω p 2=37.6
ω z =−0.0 6 67
−B −10
z= = =−0.0667
J 150

Figure 9: Small Signal Model At k i=1 ;

For unity gain; 1

k p=
100
1. Torque Controller
Proportional ( P ) :1 Torque Controller;
Integral ( I ) :0 Proportional (P): 1/100

2. 1/vt
 Integral (I):1 Figure 15: Speed Unity Gain Pole Zero Map
Figure 12: Existing Bode Plot

At zero location;
To maintain zero location at 100rad/s, lift up
75.1dB to crossover at 1 decade lower at 1.5 kHz Speed Controller
3
5.6885 x 10 Proportional ( P ) :1/0.0667
k p=
100 Integral ( I ) :1
75.1
k i=10 20
=5.6885 x 103

Figure 13: Existing Bode Plot

Figure 15: Speed Bode Plot – Zero Location

ii. Speed Controller To maintain zero location, reduce 40.5dB to

crossover at 1 decade lower at 0.15 kHz
For unity gain;
Speed Controller −40.5

Proportional ( P ) :1 k i=10 20 =9.4406 x 10−3

Integral ( I ) :0 9.4406 x 10−3
k p=
0.0667
Analyze using linear analysis for bode plot and
pole zero map

Figure 15: Speed Bode Plot – Final

Figure 14: Speed Unity Gain Bode Plot

iii. Closed Loop Controller Va = Ia Ra + Eb = 7(2) +15 = 29V

Va = Vdc Vc / Vtri
Vc = 29 x 10 / 100 = 2.9V

Ia = ( Va - Eb ) / Ra
Eb = KE ω = 0.1 X 150 rad/s = 15 V
Ia = ( 29 - 15 ) / 2 = 7A
It is proven that, from the speed reference profile
Figure 16: Large Signal Model given in the study which is 150rad/s will produce
armature average current Ia equal to 7A.
First we setting up DC Machine parameters, v. Comparison of torque ripples by using
Bridge & PWM parameters, speed & torque unipolar & bipolar switching scheme
controller and speed reference profile

By using bipolar switching scheme, current ripple

calculation are as the following:

Ttri = 1/ftri = 1/15k = 66.67 µs

d = 0.5 + Va/2Vdc
= 0.5 + 29/2x100
= 0.645

Δt = d x T
Δt = 0.645 x 66.67 µ = 43 µs

Figure 16: Large Signal Simulation

VL = LdiL / dt = L Δi / Δt
Δi = VL Δt / L
Δi = 71x 43µ / 6m
Δi = 0.5088 A

Current ripple by calculation is just as same as

current ripple by using MATLAB/SIMULINK
simulation.

Figure 17: Close-up Torque & Current Simulation

By using unipolar switching scheme, current ripple
calculation are as the following:
From the simulation of close-loop controller,
Bipolar Unipolar d = Va/Vdc
Torque, Nm 0.7 0.7
= 29/100
Armature Current, A 7 7
Current Ripple, A 0.521 0.176 = 0.290

iv. Verification of torque result based on Δt = d x T

speed reference profile
Δt = 0.29 x 66.67 µ / 2 = 9.7 µs

Te = Kt Ia
VL = LdiL / dt = L Δi / Δt
Ia = Te / Kt = 0.7 / 0.1 = 7A
Δi = VL Δt / L
Δi = 71x 9.7µ / 6m
Δi = 0.0.114 A

Current ripple by calculation is just as same as current

ripple by using MATLAB/SIMULINK simulation.

V. CONCLUSIONS

The simulation of torque profile computed by

SIMULINK-Simscape had been compared and verified
and same as the manual calculation. It is shown here
that Te-Tl must be positive value in order to accelerate
in the forward direction. This result shows that the
electrical torque applied to the system designed must be
larger than the load torque. In order to decelerate, the
net torque of the motor must be negative value which
the electrical torque must be made more smaller than
the load torque and the motor need to be operates in
braking mode.

REFERENCES
1.0 Ned Mohan, Electric Drives an Integrative
Approach, 2003

APPENDIX

1. Simulation result from SIMULINK-Simscape