As the cost for permanent magnets are decreasing and the quality of the magnets are increasing the
Permanent Magnet Synchronous Machine (PMSM) are becoming more popular. The PMSM is
superior to induction machines both in torque per kilograms and efficiency. Typical applications for
these machines are applications where machine volume or efficiency is important. In order to
control a PMSM the rotor position needs to be known. This can be obtained by usage of sensor or
by usage of a sensorless control scheme. Sensorless control schemes typically divide into two
categories; schemes that give good controllability from 2-5 Hz and schemes that gives good
controllability at all speeds including zero speed.
The PMSM is a version of the synchronous machine, where the rotor coils and the slip rings are
replaced with permanent magnets. Instead of feeding current into the rotor by means of slip rings to
create magnetic poles, the permanent magnets have fixed poles that will try to align with the
rotating field of the stator.
There are different approaches of incorporating the magnets into the rotor design:
Surface mounted magnets (SPMSM)
Interior magnets (IPMSM).
A major difference of these two rotor layouts is found in the concept of saliency. Saliency means
non-uniformity of the air gap between the rotor and the stator. A SPMSM is non-salient because the
magnet has the permeability similar to air, and an IPMSM is a salient pole machine because the iron
and magnet parts of the rotor have different permeability. Having saliency in the machine adds
some advantages. One advantage is, that a salient pole machine utilizes the reluctance torque
instead of only the electro magnetic torque. Unfortunately, it also introduces cogging torque due to
the saliency and magnets of the rotor. A sketch of the two different rotor setups are shown in next
figure.
SPMSM
If the rotor is rotating at a certain speed with a constant AC voltage applied to the stator, the speed
will determine the produced back EMF, and the current going into the windings will determine the
torque. At a certain value for the back EMF (Back EMF = Counter EMF = CEMF, is a voltage
that occurs in electric motors where there is relative motion between the armature of the
motor and the external magnetic field), it will be equal in magnitude to the input voltage. This
means, that the machine is no longer capable of producing torque. The voltage equation of the
machine can be written as follows:
0 ias
vas Ras 0
as
d
v 0 R
0 ibs bs (volt)
bs
bs
dt
vcs 0
cs
0 Rcs ics
The matrix consists of the flux linkage. According to Faradays law:
e = d/dt (volt)
The flux linkage can be written as:
= L.i [Wb.turns]
This means, that e is a function of the flux linkage. But it is the inductance of the stator, and the
permanent magnets of the rotor, that give rise to this flux linkage. By having three phases, and
thereby three inductances and a permanent magnet rotating in the middle, mutual inductances has to
be considered.
sin( r )
as Laas Labs Lacs ias
stator
rotor, and this affects the flux linkage and thereby the inductance. The principle is shown on next
figure.
The motor inductances are function of the angular displacement rm (r = prm /2). The magnetizing
inductance varies sinusoidally as a function of 2 r about a positive average value. Let LA be the
average value of the magnetizing inductance and LB the amplitude of the sinusoidal variation about
this average value.
LA = (Ns/2)2..0.r.l.1 [Henry]
LB = 0.5.(Ns/2)2..0.r.l.2 [Henry]
where:
Ns is the number of turns. It is assumed equal for each phase.
l is the axial length of the rotor.
0 = 4..10-7 is the permeability of air.
r is the radius from center of rotor to the stator inner surface.
1 is the magnetic length between the stator and the iron part when the poles are aligned with the
iron part of the rotor, and 2 is the same length when the poles and the magnet are aligned.
1
1
1
(m)
g min g max
1
1
(m)
g min g max
Being gmin and gmax, respectively, the minimum and the maximum distances between stator and rotor
iron parts.
From the above equations it can be seen that if the rotor is non salient (gmin=gmax), it results 2 = 0
and LB = 0.
Using LA and LB, the matrix [L], which contains the self and mutual inductances for the stator, can
be written as follows.
Laas = Lls + LA - LB.cos(2r) [H]
Lbbs = Lls + LA - LB.cos(2r + 2/3) [H]
Lccs = Lls + LA - LB.cos(2r - 2/3) [H]
Where Lls is the stator leakage inductance. It is difficult to determine exactly and usually is
approximated from experimental data. The mutual inductances in the stator is as follows.
Labs = Lbas = -LA/2 - LB.cos(2r - 2/3) [H]
Lacs = Lcas = -LA/2 - LB.cos(2r + 2/3) [H]
Lbcs = Lcbs = -LA/2 - LB.cos(2r) [H]
So, the total stator flux linkage for the machine can be written as follows:
as
bs
cs
ias
Ls ibs
i
cs
flux _ due _ to _ stator
sin( r )
pm sin( r 2 / 3)
[Wb.turns]
sin( r 2 / 3)
Where:
LA / 2 LB .cos(2r 2 / 3 )
LA / 2 LB .cos(2r 2 / 3 )
Lls L A LB .cos(2r )
LA / 2 LB .cos(2r 2 / 3 )
LA / 2 LB .cos(2r )
Lls L A LB .cos(2r 2 / 3 )
Then,
vas Ras 0 0
v 0 R
0
bs
bs
vcs 0 0 Rcs
ias
sin( r )
ias
i d L i sin( 2 / 3)
pm
r
bs dt bs
ics
i
sin(
/
3
)
r
cs
[volt]
Now all inductances, and thereby flux linkage, is a function of the rotor position. In order to
simplify these equations, the rotating dq0 reference frame is applied to the equations.
d
abcs
dt
2
(vas a.vbs a 2 .v cs )
3
iabcs
2
(ias a.ibs a 2 .i cs )
3
abcs
2
(as a.bs a 2 . cs )
3
The flux linkage is more complex to handle compared to the voltage and the flux linkage, is to
divide the matrix [L] into two because only LB is position dependent.
Lls LA
Ls LA / 2
LA / 2
LA / 2.
Lls LA
LA / 2
LA / 2
cos(2r )
cos(2r 2 / 3 ) cos(2r 2 / 3 )
LA / 2. LB cos(2r 2 / 3 ) cos(2r 2 / 3 )
cos(2r )
cos(2r 2 / 3 )
Lls LA
cos(2r )
cos(2r 2 / 3 )
Now, only the second part of the matrix depends on the position of the rotor. This part will be
further reduced by using the Euler identity.
4
cos
1 j 1 j
e e
2
2
Because of
ex+y = ex.ey
Then a and a can be used to describe the phase shift and,
2
1 j 2 r 2 1 j 2 r
e
.a e
.a
2
2
1
1
cos(2 r 2 / 3) e j 2 r .a e j 2 r .a 2
2
2
cos( 2 r 2 / 3)
Lls LA LA / 2. LA / 2
LB j 2r
Ls LA / 2 Lls L A L A / 2. e
2
LA / 2 L A / 2 Lls LA
1 a2 a
a 2 a 1 e j 2r
a 1 a2
1 a a2
a a2 1
a 2 1 a
Also the flux linkage from the permanent magnet can be simplified using the Euler identity. Using
the Euler identity on the permanent magnet flux linkage leads to:
sin( r )
1
1
pm j r 2
j r
pm sin( r 2 / 3)
e a e a
2
2
sin( r 2 / 3)
a
a
Now the space vector representation of the flux linkage abc is as follows:
1 a2 a
1 a a 2 ias
as Lls L A L A / 2. L A / 2
pm jr
L /2 L L
j 2 r
j 2 r
2
2
B
L A / 2. e a
a 1 e
a a
1 ibs
e
A
ls
A
bs
2
2
2
2
a 1 a
a
cs L A / 2
L A / 2 Lls L A
1 a ics
According to [Chandana. Power Electronics: Converters, Applications and Design. ISBN 0-4714-2908-2,John Wiley & Sons 2003, p.19] ,
this expression can be further simplified by using the space vector notation.
2
abcs (as a.bs a 2 . cs )
3
2
3
( Lls .LA ) (ias a.ibs a 2 .i cs ) LB (ias a 2 .ibs a.i cs ) e j 2 r pm e j r
3
2
2
3
3
( Ll , s .LA ) iabcs LB i *abcs e j 2 r pm e j r
3
2
2
Now, the voltage equation is as follows:
vabcs R iabcs
d
3
3
1
a 2 e j
a
Illustration of the dq reference frame together with the permanent magnets. The permanent magnets are aligned with the d axis.
This leads to
vdq 0 s R iabcs e j r
d
abcs e j r [volt]
dt
From the chain rule, and from the fact, that = t the derivative can be written as
d
vdq 0 s R iabcs e j r abcs e j r j dq 0 s e j r e j r [volt]
dt
Now finally the voltage equations are represented as,
3
3
dq 0 s ( Lls .LA ) iabcs e j r LB i *abcs e j 2 r e j r pm e j r e j r
2
2
It results,
3
3
dq 0 s ( Lls .LA ) iabcs e j r LB i *abcs e j r pm
2
2
[Wb.turns]
The currents are transformed the same way as the voltages, so:
iabcs e j r idq 0 s ,
i *abcs e j r i *dq 0 s
3
3
.LA ) idq 0 s LB i *dq 0 s pm [Wb.turns]
2
2
3
( L A LB )
2
Lm , q
Lm , d Lm , q
LB
3
( L A LB )
2
Lm , d Lm , q
3
Lm , d Lm, q Lm , d Lm, q
ds ( Lls
Now the flux linkages d, q and 0, the currents id, iq and i0 and voltages are combined together in
next equation.
d
vdq 0 s Rs idq 0 s dq 0 s j dq 0 s
[volt]
dt
Part _ 2
Part _ 1
Because the derivative of a constant is zero, then d ( pm )dt 0 . The voltage equations are reduced
to:
d ( Lls Lm, d )
vds Rs ids
ids r ( Lls Lm, q ) iqs [volt]
dt
d ( Lls Lm, q )
vqs Rs iqs
iqs r ( Lls Lm, d ) ids r pm [volt]
dt
The quantities (Lls + Lm,q) and (Lls + Lm,d) are also referred to as Lqs and Lds, so finally:
dL
di
vds Rs ids d ids Ld ds r Lq iqs [volt]
dt
dt
dLq
diqs
vqs Rs iqs
iqs Lq
r Ld ids r pm [volt]
dt
dt
dL
di
v0 s Rs i0 s ls i0 s Lls 0 s [volt]
dt
dt
Rs
0 iqs 0 Lqs
0 iqs r pm [volt]
qs r d
dt
v0 s 0
i0 s
0
0
Rs i0 s 0
0 L0 s Lls
If the rotor is rotating at a certain speed with a constant AC voltage applied to the stator, the speed
will determine the produced back EMF, and the current going into the windings will determine the
torque. At a certain value for the back EMF (fcem), it will be equal in magnitude to the input
voltage. This means, that the machine is no longer capable of producing torque. The voltage
equation of the machine can be written as follows:
vas Rs 0 0 ias
as
d
v 0 R
0 ibs bs (volt)
s
bs
dt
vcs 0 0 Rs ics
cs
Because the inductance is position dependent due to the saliency, the voltage equations are
represented in the rotating dq reference frame. The benefit of this is, that any position dependencies
in the electrical system are cancelled. The dq reference frame model is:
d
vdqs Rs idq 0 s dq 0 s j dq 0 s [volt]
dt
When separated in to the d and q axis equations, it yields:
di
vds Rs ids Ld ds r Lq iqs [volt]
dt
di
vqs Rs iqs Lq qs r Ld ids r pm [volt]
dt
di
v0 s Rs i0 s Lls 0 s [volt]
dt
Here the d and q axis are also shown, and the angle represents 90 in the electrical system. In the
mechanical system this is equal to (supposing the motor has 12 permanent magnets)
n
m
2
Te
n
m
2
n P
mec [N.m]
2 r
e r
Pmec
m
The relationship between electrical (e=r) and mechanical speed (m) is: e r m
n
[rad/s]
2
) [W]
2
dt 2
dt 2
3
3
3
Pmec r Real( j s i s* ) r ds i qs qs i ds r ( pm i qs ( Lds Lqs ) i ds i qs ) [W] (*)
2
2
2
Pin
This means that the amount of energy delivered in to the machine is converted to magnetic and
mechanical energy. The Ploss is the dissipated heat in the windings, and thereby the loss in the
boundary between the magnetic and electrical energy. Pstored is the amount of magnetic energy that is
present in the air gap to some instant in time, hence the derivative. The presence of flux in the iron
parts of the motor give rise to the iron losses Pfe, but these are initially neglected in the model.
Pmec is the amount of power transferred to the mechanical system of the motor. There are losses in
the mechanical system, so Pmec is not the power output of the motor, just the power delivered to the
shaft.
Pmec can be written as:
Pmec = Pfric + Pout [W]
Here Pfric is all the mechanical losses in the motor. The power flow is illustrated on next figure.
From equation (*), the electromagnetic torque Te can now be written as:
3 n
3 n
Real( j s i s* ) ds i qs qs i ds [N.m]
2 2
2 2
L
From equations ds
ds
ds
pm and
qs
qs i qs it results in,
3 n
3 n
Te pm i qs ( Lds Lqs ) i qs i ds pm i qs ( Lm , d Lm, q ) i qs i ds [N.m]
2 2
2 2
Te
Te
2 pm Lqs sin s ( Lds Lqs ) sin( 2 ) [N.m]
2 2 Lds .Lqs
From the above equation, for constant stator and rotor fluxes amplitudes, the electromagnetic torque
can be changed by control of the torque angle . This is the angle between the stator and the rotor
flux linkage when neglecting the rotor resistance. The torque angle, in turn, can be controlled by
controlling the position of the stator flux vector in respect to the permanent magnet flux vector
using the actual voltage vector supplied by the inverter.
In steady state, is constant and corresponds to load torque, whereas stator and rotor flux rotate at
synchronous speed. In transient operation, varies because stator and rotor flux rotate at different
speeds.
When operating as a motor the electrical torque is then delivered to the shaft:
d
2 d
i Ti J dtm i Ti J n dt r resulting in,
B 2
J .2 d r
Te
m r Tload Tdf
[N.m]
n
dt
n
Where:
J is the moment of inertia.
Bm is the viscose coefficient.
Tload is the torque applied to the shaft.
Tdf is the dry friction.
n is the number of poles.
r is the electrical angular speed.
In
failure
situation
io0.
What
happens
incorporating
for
example
Pin
3
(i ds u ds i qs u qs 2 i0 s u 0 s ) ans so on???
2
10
r
Te
d r
dt
3 n
pm iqs ( Lds Lqs ) iqs i ds
2 2
PMSM block
General PMSM:
Te
3 n
pm i qs ( Lds Lqs ) i qs i ds
2 2
induc tan ce _ torque
magnet _ torque
3 n
pm iqs
2 2
arctan
i ds
Torque angle
3 n
2 2
3 n
2 2
11
12