Lecture 7: Control of a DC Motor Drive

ELEC-E8405 Electric Drives (5 ECTS)

Marko Hinkkanen

Aalto University
School of Electrical Engineering

Autumn 2015

1 / 25
Learning Outcomes

After this lecture and exercises you will be able to:

I Draw the block diagram of a cascaded control system and explain it
I Calculate the controller gains based on the model parameters and the desired
bandwidth
I Implement the current and speed controllers in the Simulink software

2 / 25
Introduction

I Modern electric drives automatically identify motor parameters

(auto-commissioning, identification run)
I Model-based controllers are preferred, since they can be automatically tuned
based on the known (identified) motor parameters
I Cascaded control system is commonly used
I Current or torque controller (fast)
I Speed controller (slower), not always needed
I Various control methods exist: our approach is based on
two-degrees-of-freedom (2DOF) PI controllers
I Control systems are typically implemented digitally
(but can be often designed in the continuous-time domain)
I Controllers and tuning principles can be extended to AC drives

A cascaded control system may also include a position controller, but we will not cover it in these lectures.
3 / 25
Outline

Preliminaries

Current Control

Voltage Saturation and Anti-Windup

Cascaded Control System and Speed Control

4 / 25
Closed-Loop Control

I Closed-loop transfer function

Controller d Plant
y (s) L(s) r e u y
= H(s) = K (s) G(s)
r (s) 1 + L(s)

where L(s) = K (s)G(s) is the loop

transfer function r = reference
I Typical control objectives e = control error
I Zero control error in steady state u = control output
I Well-damped and fast transient d = load disturbance
response y = output

The stability of the closed-loop system H(s) is often evaluated indirectly via the loop transfer function L(s). For example, the gain and phase
margins can be read from a Bode plot or a Nyquist plot of L(jω). In these lectures, we mainly analyse the closed-loop transfer function H(s) directly.
5 / 25
PI Controller

I Most common controller type

I Time domain PI controller
Z
1
u = kp e + ki e dt ki
s
I Transfer function r e u
kp
e(s) k
= K (s) = kp + i
u(s) s y

6 / 25
Outline

Preliminaries

Current Control
PI Controller
IMC Tuning Principle
2DOF PI Controller

Voltage Saturation and Anti-Windup

Cascaded Control System and Speed Control

7 / 25
Current Control System

Udc

Closed-loop current control enables: qA , qB

ia,ref
1. Current limitation Current ua,ref
controller PWM
2. Precise and fast torque control
ia
ua ia

8 / 25
Block Diagram
Assumptions:
I Switching-cycle averaged quantities
I Ideal voltage production: ua = ua,ref

Motor model

ea ωM
kf
TL
ia,ref
Current ua 1 ia 1
controller kf
sLa + Ra sJ
TM

Switching-cycle averaged symbols will not be marked with overlining in these lectures.
9 / 25
Simplified Block Diagram

I Back-emf ea is a slowly varying ea

load disturbance ia,ref
I P controller cannot drive the Current ua 1 ia
controller sLa + Ra
steady-state error to zero
I PI controller suffices

10 / 25
PI Current Controller

I Time domain
Z Current controller ea
ua = kp e + ki e dt
ia,ref e ua ia
K (s) Ya (s)
I Transfer function
e(s) k
= K (s) = kp + i
ua (s) s
1
I How to tune the gains kp and ki ? Ya (s) =
sLa + Ra

11 / 25
Closed-Loop Transfer Function

I Closed-loop transfer function

ia (s) K (s)Ya (s)

= H(s) =
ia,ref (s) 1 + K (s)Ya (s)
ea
I Desired closed-loop system
ia,ref e ua ia
αc K (s) Ya (s)
H(s) =
s + αc
where αc is the bandwidth
I Time constant of the
closed-loop system is τc = 1/αc

12 / 25
Internal Model Control (IMC) Principle

I Let us equal the closed-loop transfer function with the desirable one

K (s)Ya (s) αc αc
H(s) = = ⇒ K (s)Ya (s) =
1 + K (s)Ya (s) s + αc s
I Controller K (s) can be solved

αc αc αc Ra
K (s) = = (sLa + Ra ) = αc La +
sYa (s) s s
I Result is a PI controller with the gains kp = αc La and ki = αc Ra
I Bandwidth αc should be (at least) one decade smaller than the angular
sampling frequency 2π/Ts

13 / 25
Example: Step Response

ia (A)
I Ra = 1 Ω, La = 10 mH 60 ia,ref Closed-loop control
I Constant ea = kf ωM = 100 V
40
I Open-loop controller
I Slow (τa = 10 ms) 20 Open-loop control
I Inaccurate even in steady
0
state (since Ra varies as 0 4 8 12 16 20 t (ms)
a function of temperature) ua (V)
I Closed-loop PI controller 400 Z
I Fast (τc = 2 ms) ua = kp (ia,ref −ia ) + ki (ia,ref −ia )dt
I Steady-state accuracy
depends on the current 200
measurement accuracy ua = Ra ia,ref + ea
0
0 4 8 12 16 20 t (ms)

14 / 25
Example: Step Response During Acceleration

ia (A) Some control error during accelaration

60
I IMC-tuned PI controller is 40
sensitive to the varying load
20
disturbance ea = kf ωM
I Causes control error during 0
0 4 8 12 16 20 t (ms)
accelerations ua (V)
I Load-disturbance rejection 400
can be improved with an
active resistance
200

ea increases 2 V/ms
0
0 4 8 12 16 20 t (ms)

15 / 25
2DOF PI Current Controller: Improved Disturbance Rejection

I Active resistance r is a
controller gain Current controller ea
I Affects like a physical
resistance (without ia,ref ua0 ua 1 ia
K (s)
causing losses) sLa + Ra

ia (s) r
= Ya0 (s)
ua0 (s)
1
=
sLa + Ra + r
Ra + r
I Bandwidth of Ya0 (s) can = αc ⇒ r = αc La − Ra
be selected to equal αc La

This controller structure could also be represented as a state feedback controller with integral action and reference feedforward.
16 / 25
2DOF PI Current Controller

I IMC principle gives

αc
K (s) = ea
sYa0 (s)
αc (Ra + r ) ia,ref e ua0 ia
= αc La + K (s) Ya0 (s)
s
I Controller gains

r = αc La − R a 1
kp = αc La Ya0 (s) =
sLa + Ra + r
ki = αc (Ra + r ) = αc2 La

17 / 25
Outline

Preliminaries

Current Control

Voltage Saturation and Anti-Windup

Cascaded Control System and Speed Control

18 / 25
Voltage Saturation: Control Loop Becomes Nonlinear

I Maximum converter output voltage is limited: −umax ≤ ua ≤ umax

I Reference ua,ref may exceed umax for large ia,ref steps
(especially at high rotor speeds due to large back-emf ea )

1
ki ea
s Saturation
ia,ref e ua,ref ua 1 ia
kp
sLa + Ra

19 / 25
Back-Calculation Anti-Windup Method

Current controller
This signal is nonzero
1/kp only if the voltage saturates

1
ki ea
s

ia,ref e ua,ref ua 1 ia
kp
sLa + Ra

20 / 25
Example: Step Responses With and Without Anti-Windup

ia (A)
60 No anti-windup

I Voltage saturates after the 40 Back-calculation anti-windup method

current reference step
20
(due to high ea )
I No overshoot if anti-windup 0
0 8 16 24 32 40 t (ms)
is implemented
ua (V)
I Rise time is longer than the umax = 400 V
400
specified one
(due to voltage saturation)
200 Higher speed ⇒ higher ea

0
0 8 16 24 32 40 t (ms)

21 / 25
Outline

Preliminaries

Current Control

Voltage Saturation and Anti-Windup

Cascaded Control System and Speed Control

22 / 25
Cascaded Control System

Udc
Control system
ωM,ref qA , qB
Speed TM,ref 1 ia,ref
controller Current ua,ref
kf PWM
controller

ia
ua ia
ωM

Current controller: Inner (faster) loop

Speed controller: Outer (slower) loop
Controllers can be designed separately due to different time scales

23 / 25
Block Diagram

I Ideal torque control can be assumed: TM = TM,ref

I What is the resulting system?

Motor model

ea ωM
kf
TM,ref ia,ref
ωM,ref ua,ref TL
Speed 1
Current 1 ia 1
controller kf kf
controller sLa + Ra sJ
TM

24 / 25
Simplified Block Diagram
I Assumption on ideal torque control TM = TM,ref
I Holds well if the current-control bandwidth is (at least) one decade faster than
the speed-control bandwidth
TL
ωM,ref
Speed TM 1 ωM
controller sJ

I Mechanical and electrical systems are analogous

1 1
↔
sJ + B sLa + Ra
i.e. J ↔ La , B ↔ Ra , TM ↔ ua , ωM ↔ ia , . . .
I Same control structures and tuning principles can be directly used
25 / 25