Automatic Commutation of Stepper Motors

using low cost standard stepper drivers

Dr. Norbert Veignat, GVP Marketing & Sales, DMM & BLDC Motors
Portescap, La Chaux-de-Fonds, Switzerland
_________________________________________________________________________________________

this purpose because, contrary to position

CONTENTS

1 Introduction, principle autocommutation

of
2 Encoder resolution same as motor full-step
resolution
control circuits, the sensor resolution may be quite
3 High resolution encoder with a motor running
low and only needs to indicate certain discrete rotor
in microstep mode
positions, depending on the desired commutation
4 Implementation in speed and position control
angle.
5 Conclusion

Figure 1 shows a motor model having two

phases and one pair of poles. The encoder has two
1 INTRODUCTION, PRINCIPLE OF
AUTOCOMMUTATION Hall sensors and one pair of poles, its resolution is 4
counts per rev.
Stepper motors are mostly driven in open loop
Sensor a
mode. Timing of the step pulses is defined by the Phase B

controller, the motor is supposed to follow without Sensor b

loosing synchronism.
Stepper motors are in fact natural positioning
devices which do not require position feedback. The
Phase A
drawback of their design are important iron losses
when running at high speed. Therefore steppers are Fig. 1

mainly used for low speed positioning. Model of a two phase stepper motor

and encoder
However, the escap disc magnet stepper motor is an

exception to this rule : its design is totally different

With both phases energised, the target positions
from conventional stepper technologies and results in
at electrical angles of 45°, 135°, 225° and 315° (see
rather low iron losses. It is therefore capable of fast
fig. 2).
incremental motion and can compete with traditional
Phase B
DC servo and BLDC motors if an optical encoder and
a small circuit is added for automatic commutation.
This solution provides comparable performance at Phase A

lower overall cost.

Automatic commutation requires knowledge of
the rotor position. Hall sensors are very often used for
Fig.2.1.3

______________________________________________________________________________________________
____
Portescap Page 1 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 The four full-step target positions encoder channels (example: 01 means a = 0, b = 1). α
per electrical period
is the phase advance angle introduced between motor
and sensor signals, its influence on motor
The magnetic encoder indicates the rotor
performance will be seen later.
position within an electrical angle of 90° (fig. 3).

Figure 5 shows the back-EMF of each phase

Sensor a Sensor b

and the corresponding sensor output signals over one

electrical period, which for the model of figure 1
equals one motor revolution.
0 90 180 270 360 0 90 180 270 360
0
Electrical angle Electrical angle

Fig. 3
Encoder output signal (2 channels) versus position of BEMF Phase A
the encoder magnet BEMF Phase B
Sensor a
0 45 90 135 180 225 270 315 360
Electrical angle Sensor b

2. ENCODER RESOLUTION SAME AS MOTOR

FULL-STEP RESOLUTION Fig. 5
Hall sensor signals versus EMF
2.1 Influence of the commutation angle of each phase, over one period
The angular position of the sensors versus back-
EMF of the phase windings is set according to figure
4. Depending on the logic equation (relation
Logic status of sensors a and b between sensor signal status and phase
Example 01, means a=0, b=1
00
01 energization) the motor can behave in different ways:
10 as a positioner, a BLDC motor with phase advance α,
A+B+
11 α
A-B+ a BLDC motor with phase advance 90+α, or in an

A-B- A+B-
Sensors unstable oscillating mode. Table 1 gives examples of
the motor working mode for various phase currents
Motor Stable position with current
Example A+B+ means Ι A = Ι 0, Ι B = Ι0 (hence rotor positions) if in each case the sensor is
aligned to give the same signal state 01 (a = 0, b = 1,
Fig. 4 α < 45°).
Rotor target positions with phase currents,
and corresponding logic signals of
encoder channels
logic states vs initial dir. of initial dir. of
phase currents rotation = CCW rotation = CW
01 for A+ B+ position mode position mode
01 for A- B+ BLDC mode oscillating
The first circle (lower left) shows the four motor phase advance mode
target positions and corresponding states of phase =α
01 for A- B- at low speed: at low speed:
energization (example: A+B+ means: IA = IB = +I0). The
oscillating mode oscillating
upper right circle shows the logic states of both at high speed: at high speed:
BLDC mode, BLDC mode
______________________________________________________________________________________________
____
Portescap Page 2 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 phase advance phase signal CCW CW
= 90° +α advance = 90- state
α
01 A- B+ A+ B-
01 for A+ B- oscillating mode BLDC mode
00 A- B- A+ B+
phase
advance =α 10 A+ B- A- B+
11 A+ B+ A- B-
Table 1 logic equations logic equations
Motor behaviour resulting from various sensor A = a A =a
positions for logic state 01
B = b B =b

2.2 Examples Table 2

Motor working with zero phase advance Relation between logic states and
phase currents with logic equations

Direction
Back-EMF Phase A +5v GND
Back-EMF Phase B Sensor b
Sensor a
Sensor a
0 45 90 135 180 225 270 315 360
Sensor b Vm
Electrical angle out 1 out 2 out 3 out 4

Phase A Phase B

SGS
298

1 2 3 4

Fig. 6
In A In B

Sensor signals and back-EMF

versus rotor position + 5V

Vm + 5V In A
7805
Start/ In B

With the sensor adjusted according to fig. 6, we Stop

have α = 0.

Figure 7 illustrates the logic equations for sensor Fig. 8

signals and phase currents as per Table 2, where a Circuit for driving the motor
with zero phase advance
positive phase current is taken to be equivalent to
logic 1 and a negative current equivalent to logic 0.
Motor working with 45° phase advance
Figure 8 shows the motor drive circuit required for
If the sensor remains adjusted as before, the
zero phase advance.
logic equations need to be changed. If the sensor is
rotated by -45°, the same drive electronics can be
A+B +
A - B+
used.
00 01

Motor working with 90° phase advance

10 11

. With α = 0, the logic equations are:

A -B - A +B -

for CCW rotation: A =b , B = a

Fig. 7 for CW rotation:A =b , B = a

Angular relation between phase The motor can rotate in either direction, low speed
currents and logic states torque is about nil, as commutation takes place at an
unstable rotor position.
sensor ab motor rotates motor rotates

______________________________________________________________________________________________
____
Portescap Page 3 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 Note that this approach also
is valid with only
Torque
(mNm)

120

one phase energised. 80

60

Measurements 40

20

Measurements were made with an escap® type 0

Speed

0 2000 4000 6000 8000 10000 (step/s)

P310-158-005 disc magnet stepper motor. Phase 0 1200 2400 3600 4800 6000 (rev./min)

Dynamic torque for different phase

windings were connected in parallel, and an escap®

advance, U = 36V

type D15 magnetic encoder was added. Torque

(mNm)
Power
(W)

The driver was a double bridge SGS 298 with a

100

80

current limitation added to avoid overcurrent at low 60

speed. Three different voltages (7, 12 and 24V) and 40

Dynamic torque,
phase advance 45°
Power
phase advance angles were used. The results are 20

0 0 Speed

shown in figure 9 for angles of 0° and 45°. 0

0
2000
1200
4000
2400
6000
3600
8000
4800
10000 (step/s)
6000 (rev./mn).)

Dynamic torque for different voltage,

phase advance = 45°

Phase advance 0° Phase advance 45°

Fig. 10
16 16

Dynamic torque of the PP520 at 36V and different

14 14
12 12
Torque (mNm)

Torque (mNm)

10 U=7V

phase advance angles, and at 45°

10 U=7V
8 U=12V 8 U=12V

U=24V 6 U=24V

and different voltages

6
4
4
2
2
0
0 1 2 3 4 5 6 7 8 9 10
0
0 1 2 3 4 5 6 7 8 9 10 1112 131415 3 HIGH RESOLUTION ENCODER WITH A MOTOR
Speed in thousand rpm
Speed in thousand rpm

RUNNING IN MICROSTEP MODE

Fig. 9
Dynamic motor torque for different voltages 3.1 Motor working in half-step mode
and phase advance angles

Use of a sensor having the same resolution as We now consider a motor working in half-step

the motor is relatively inexpensive. The 100 steps per mode, with an encoder giving four counts per full step.

rev. escap ® motor type PP520 has built-in Hall

sensors which are excited directly from the disc
magnet of the motor rotor.
When using such a procedure of
autocommutation, a torque constant with some third
harmonics is preferable to a purely sinusoidal one,
as torque ripple will be somewhat lower.
The limitation of such systems is the fact that for
maximum torque the phase advance angle should be
constantly adapted to the speed. In view of the short Fig. 11
electrical time constant and low iron losses of small The driver generates the motor target positions
n° 1, 3, 5, 7, 9, 11, 13 and 15. The encoder detects
motors using the disc magnet technology, a phase
rotor positions at n°. 1, 2, 3, 4, 5, ...
advance of 45° seems to be an excellent compromise
as illustrated in figure 10. With a motor having a sinusoidal torque
constant, the low speed commutation angle should

______________________________________________________________________________________________
____
Portescap Page 4 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
be 90 + 22.5 = 112.5° ahead of the position of stable V

equilibrium (see fig. 12).

Direction Encoder Motor
Stepper driver for
CK half step mode

G ND Additional circuitry used during

Torque

initialisation and phase advance

adjustment
Torque A

Torque B

Torque -A Start pulse "killer" (to decrease lead angle)

Torque -B
pulse "adder" (to increase lead angle)
Torque A+B
0 45 90 135 180 225 270 315 360
Torque -A+B
Electrical angle
Torque -A-B
Torque A-B

112,5° Fig. 13
22.5° 90°

Autocommutation drive circuit with a

Fig. 12 stepper motor and encoder

Motor torque functions with half-step mode. The

optimum low speed commutation angle is 112.5°. At switch-on or after a reset, the circuit is in
Angles are always electrical
stepper mode with, say, phase A powered. In case of
half-step mode, a logic signal at the Start input will
generate 2 pulses at the maximum frequency the
With the rotor being in position 2 for example,
sequencer accepts, and consequently phase B is
the stator field should target position 7, and also as
powered and the rotor starts turning.
the rotor moves to "3". Once it is in position 4, a
The first encoder pulse is sent to the sequencer,
commutation is triggered which moves the stator field
the second one is suppressed, the third one is used
vector to position 9.
again, and so forth. The motor runs with the optimum
We can increase the phase advance by
static phase angle of 112.5°, and we are using 1 out
increments of 22.5° which correspond to the encoder
of every N pulses, with N being equal to 2 in our
resolution. With higher resolutions the phase advance
example.
may be adjusted more accurately. Obviously, with
If the phase advance should be increased by one
autocommutation, torque ripple depends on the motor
encoder increment, the system uses the number "N-
torque function vs rotor position and on the way the
1" pulse and then returns to the original rhythm. That
phases are energized. The size of torque angle
way it has gained one pulse, or 22.5°, and now runs
increments depends on encoder resolution.
with 112.5 + 22.5 = 135°.

Practical implementation
As a simple solution for autocommutation a If the phase advance should be decreased

stepper motor with encoder and a standard stepper instead, the process is similar but now uses the

driver are used, see figure 13. With this cost-effective "N+1" pulse before going back to the original rhythm.

system the encoder pulses, through an electronic That way it has lost one pulse and now runs with a

interface, are fed to the clock input and trigger the phase advance of 112.5 - 22.5 = 90°.

full- or half-step commutations. Some precautions

must be taken at system initialisation and also when The result is a motor running on a standard

changing the phase advance. stepper driver but with automatic commutation, a
convenient solution for spindle drives, pumps, mirror
drives and so forth.

______________________________________________________________________________________________
____
Portescap Page 5 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 equilibrium :
3.2 Motor working in microstep mode
In this case the commutation angle is not
We now consider a motor working in microstep optimum for low speed, and torque ripple will slightly
mode and making 4 microsteps per full step, with an increase. With the rotor in position 1, the stator field
encoder giving four counts per full step. may target either "5" or "6", resulting in a phase
advance of 90° or 112.5° instead of the ideal 101.25°.
The error can be corrected by calculating the
motor speed and, after detection of position 1, letting
the rotor move approximately 11.25° before
commutating the stator field to position 6.

- Encoder pulses differ from positions of stable

equilibrium :
In this case the encoder must be shifted by an
angle of one half the motor resolution. In the example,

Fig. 14 a shift of 11.25° allows the detection of rotor positions

The driver generates the motor of 11.25°, 33.75°, 56.25° and so forth, resulting in the
target positions n° 1, 2, 3, 4, ..., 15 and 16. optimum low speed commutation angle.
The encoder detects rotor positions at
n°. 1, 2, 3, 4, ..., 15 and 16 When using an encoder of twice the resolution,
giving 8 pulses per full step, the solution proposed in
paragraph 3.1 may be applied.
With a motor having a sinusoidal torque
constant, the low speed commutation angle should 3.3 Motor working in sine-cosine mode
be 90 + 11.25 = 101.25° ahead of the position of
stable equilibrium (see fig. 15). This particular case makes it easy to obtain a
torque motor.
Let TA = torque of phase A
1
TB = torque of phase B
Torque

One Phase on

0.5

Two Phase on with

k = torque constant
0
identical current in
0 45 90 135 180 225 270 315 360
-0.5
each Phase
θ = rotor position
Microstep

N = number of pole pairs

-1
90° Electrical angle
101.25°

iA = current in phase A
Fig. 15
iB = current in phase B
Motor torque functions withmicrosteps
4
per full step.
Then the torque equations are:
The optimum low speed commutation angle is
101.25°. Angles are always electrical TA = -k iA sin(Nθ) and TB = k iB cos(Nθ)

Let the phase currents depend directly on rotor

If motor and encoder resolution are the same,
position:
there are 2 options.
iA = -I0 sin(Nθ) and iB = -I0 cos(Nθ)

- Encoder pulses occur at positions of stable

______________________________________________________________________________________________
____
Portescap Page 6 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 The resulting motor torque is constant: Analog speed control
T = TA + TB = k I0 (sin²(Nθ) + cos²(Nθ)) = k I0 Similar to DC motors the speed of steppers
driven in BLDC mode may be controlled using a

However, with a stepper motor this theoretical encoder (see fig. 16).

model is difficult to achieve, mainly for 2 reasons:

ωc : Speed reference
- at high speed (>2000 steps/s), due to the electrical Power supply

time constant and the high commutation frequencies, F

good current regulation is difficult to achieve unless

V
Stepper motor
with encoder

the driver uses a high voltage and high chopper Sequencer Driver M

frequency. The resulting cost increase makes the

autocommutated stepper less attractive in
comparison to traditional DC servo type solutions.
- the second reason is linked to the encoder. With Fig. 16
the high number of pole pairs of the stepper motor it Speed control of the escap® stepper PP520 using
Hall sensors.
is difficult to generate two sinusoidal output signals in
Two Ericsson driver chips 3717 allow separate
quadrature with low enough distortion. adjustment of the current reference
This is illustrated by a test, where an escap®
This type of control requires the separate linear
P850 stepper motor was equipped with an auxiliary
adjustment of both phase currents, and a torque
magnet of the same number of pole pairs, rotating in
which is a linear function of current over the speed
front of analog Hall sensors. Their adjustment inside
range considered. For high speed operation the latter
the motor was extremely difficult and signal distortion
requires a high drive voltage. The solution is only
over one rev made it impossible to reduce torque
recommended if speed is controlled within a narrow
fluctuations to less than 5%. In fact, at low speed a
range.
much better result was obtained using a true
microstep drive mode.
Digital speed control
We can see that between the various
This type is more complex but offers the
possibilities of driving stepper motors in a BLDC
advantage of taking into account the motor torque
mode, the one using an encoder and a normal
non- linearities. The speed is measured digitally and
stepper driver is highly attractive in terms of cost and
the speed loop coefficients are set for the speed
performance.
range considered.
Such a speed control circuit was set up for the
escap ® PP520 motor. Each phase is powered by two
4 IMPLEMENTATION IN SPEED AND
POSITION CONTROL PWM drivers which convert a reference voltage into
phase current.
It was demonstrated that driving a stepper motor
The current references come from two DACs; a
in BLDC mode is quite simple. This is attractive for
microprocessor provides the digital values. The speed
about 70 % of applications which do not require the
is calculated by measuring the time between encoder
execution of a speed profile. Now let us look at speed
pulses. The phase advance and current amplitude for
control of stepper motors with
autocommutation.
each phase are then controlled depending on the
speed and the speed error (see fig. 17).

______________________________________________________________________________________________
____
Portescap Page 7 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 below 500 rpm sampling time is insufficient and
therefore results are better with the motor driven in
Ph B Ph B Ph B

normal open loop microstep mode rather than with

x x x

Ι1 Ph A k Ι1 Ph A k Ι1 Ph A

closed loop operation.

-Ι2 - k Ι2 - k Ι2

a b c
Position control
Fig. 17
Stepper motors being designed for open loop
Different ways of speed control by changing :
positioning, it makes little sense closing the loop and
a) the phase advance
b) the currentamplitude increase cost. Motors built for microstep operation
c) both parameters simultaneously
have low detent torque and a perfectly sinusoidal
torque function. When driven in microstep mode they
With the Hall sensors used in the PP520 motor easily provide the accuracy needed for many
the control algorithm is simple as compared to an applications. If necessary, a final position error is
optical encoder because the two Hall signals identify corrected by adding a few microsteps, in stepper
4 rotor positions per electrical period. mode.
The microprocessor controls only two 5 CONCLUSION
parameters which are the absolute values of each
phase current. Their polarities depend directly on the Stepper motors are often used for reasons of low
logic state of the Hall outputs (fig. 18). cost and easy implementation in relatively simple
applications.

00 New technologies like the disc magnet motor

x 45°
Ph B developed by Portescap allow use of steppers not
S2
S1
only for positioning but for high speed motion as well.
Ι1 Ph A
Due to low iron losses and low rotor inertia, they
- Ι2
01 10
successfully compete against traditional brushless
S3 S4
DC motors.

11 By adding an encoder with a simple electronic

interface between the encoder and a standard stepper
Fig. 18
driver, these motors can be autocommutated in order
The 4 rotor positions S1 to S4 and corresponding
logic states at the encoder outputs to take advantage of their high dynamic possibilities.
The results are competitive in every respect. With
With the Hall sensors detecting the positions
incremental motion the best solution is to drive the
S1, S2, S3 and S4, the phase advance may be
motor in microstep mode at low speed and in BLDC
changed from 45° to 45+x° by imposing absolute
mode at high speed.
current values of:
Design engineers should be aware of these new
I1 = I0 cosx and I2 = I0 sinx
technologies, which lead to new ways of optimising a
Each phase current is a function of the sensor
entire system including motor, driver, and load.
output logic states.
Because of phase jitter between both sensor
signals motor speed is determined by measuring the
frequency of just one signal. However, at speeds

______________________________________________________________________________________________
____
Portescap Page 8 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.
 Bibliography :
D. Regnier, C. Oudet, D. Prudham
"Starting Brushless DC Motors utilizing Velocity Sensors" 1985
IMCSD

M. Jufer and R. Osseni

"Back E.M.F. indirect
Detection for Self Commutation Synchronous
of Motors" 1987
EPF, Grenoble

C. Oudet, N. Veignat
"Considerations on Peak Speed of Head Actuators used in Winchester Disc Drives" 1987
IMCSD

M. Jufer
"Evolution des Moteurs Synchrones du Moteur Pas-à-pas au Moteur Synchrone
Autocommuté", 1988
JEMP, Nancy

R. Osseni
"Modélisation etAuto-asservissements de Moteurs SynchronesAutocommutés" 1988
Thèse, EPFL

M. Crivii, M.Jufer
"Auto-commutation de Moteurs Pas-à-pas sans Capteur" 1990
6e colloque sur les moteurs-pas-à-pas, EPFL, Juillet 4-5, 1990

M. Crivii, M.Jufer
"Autocommutation de Moteurs Pas-à-pas sans Capteur" 1990
"Positionnement Incrémental par Entraînement Electrique"
EPFL, Juillet 1990

A. Hamzaoui, C.Goeldel, L.Afilal

"Modèle dynamique d'un Moteur Pas-à-pas, Application à la Commande en Boucle Fermée" 1990
Colloque sur les moteurs pas-à-pas, EPFL, Juillet 1990

L. Antognini, C.Dayer
"Sensorless Drive of P.M.
Step Motors applied to Dynamic Torque Optimisation" 1991
IMCSD

N. Veignat
"Application &Possibilities for a BLDCMotor having a High Number of Poles 1991
Motion Control Technology Conference, Boston, Mars 1991

Thomas Erich Weber, travail de diplôme d'ingénieur.,ref. 04EL88

"Fachhochschule München 1992

M. Simon-Vermot and Dr. Karmous

"Optimisation ofthe Utilisation of aStepper Motor" 1994
Electric Drive Design and Applications, EPFL, Oct. 19-20, 1994, ICEM 1994

P. André, N.Veignat
"Optimised Use ofhigh Speed Stepping Motor by Implementation ofAdequate Speed Profile" 1995
IMCSD
_________________________________________________________________________________________________
_
Portescap Page 9 of 9 NV/mf-June 1996
(C) 1996, Portescap. All rights reserved.