Sunteți pe pagina 1din 6

National Power Electronics Conference 2010

HIGH PERFORMANCE BLDC DRIVE WITH DIGITAL POSITION ESTIMATOR

Sugantha Krishnamoorthy, Prof. V Ramanarayanan

Dept. of Electrical Engineering, Indian Institute of Science, Bangalore, India

ABSTRACT High resolution position information is essential


to drive the motor as a vector controlled drive. A
This paper presents a simple digital position
standard two wheeler drive motor has been taken
estimator. Standard BLDC motors come with
to demonstrate the vector controlled drive with
embedded hall sensors. The hall sensors provide
the position estimator. A simple torque controlled
sector information of shaft position. The
drive has been realized with a DSP platform.
proposed digital position estimator takes in the
The drive motor, modeling of the same, position
sector information; the estimator output is high
estimator, the torque controlled drive and the
accuracy position information. High resolution
performance results are all given in the following
position signal enables vector control of the
sections.
BLDC motor. Such a control reduces the torque
pulsations on the drive shaft. This work is readily
DRIVE MOTOR AND TEST SETUP
applicable to electric two wheelers. This paper
presents such a drive; the effectiveness of the A two wheeler motor (PMSM) has been taken to
estimator is demonstrated through a vector demonstrate the several features of this work.
control drive for such an application. Two identical motors (800W) are taken to put
together the test platform.
INTRODUCTION
BLDC motors have become standard for electric
vehicle applications. The main reasons are
economy and reliability. Recently Permanent
Magnet Synchronous machines (PMSM) are
becoming popular for this application.

The standard PMSM has a structure of outer


stator and inner permanent magnet rotor. Fig. 1: Laboratory Setup
Encoders can be readily used with such
machines. The machines used for two wheeler Figure 1 shows the machine setup. It consists of
drives employ a different construction. The rotor a 3 phase bridge inverter driving the test motor.
forms the hub. The stator is attached to the shaft. The test motor is coupled to another identical
Such a construction is well suited for low speed, machine through a belt drive. The output of the
gearless, direct drive applications. The pole second machine is rectified with a diode bridge
number will be typically 30-40. The speed will be and electrically loaded with rheostats. The dc
typically 300-600 rpm. In such machines high power is from a laboratory source (48V). The
resolution position encoders are not used. name plate details of the drive motor are as
Typically hall position sensors are standard. The follows.
position sensors provide low resolution sector
information only.
DC Voltage 48 [V]
This paper presents a digital position estimator No Load RPM 610 [rpm]
driven by the sector signals of the hall sensor.
The position estimator provides high resolution Power 0.8 [kW]
angle information of the rotor. This position Rated DC I 16 [A]
estimator has been evaluated to find out its
performance. The position error as measured at Table 1 : Name Plate Details
the end of every sector is found to be within 6 º.

1
National Power Electronics Conference 2010

The test setup was used to evaluate the 4. Continue integrating position with new speed
parameters of the machine ( La, Ra, J, B, Ke) information. The integration is done in discrete
and is given in the following Table II. steps of duration tstep.

Ra 0.06 [ohm]
La 150 [uH]
J/B 1 [sec]
Ke 0.0228 [V/rad/sec]

Table 2 : Machine Parameters

POSITION ESTIMATOR
Calculation of Speed and Theta
Figure 2 shows the hall sensor output of the
machine.

Fig 3: Calculation of theta

The method of evaluating the position consists


of the above simple steps of

• Speed calculation and position updating at


every 60 º

• Position integration every time step.

This position estimation is a feed forward


method. Ideally the position error is limited by the
correction added at every update points. This has
been measured to be 6 º.

The step time tstep determines the resolution of


theta and sets a limit on the machine speed up to
which the code can run efficiently. Here tstep of
Fig. 2 : Hall sensor Vs Back emf 200µs has been chosen. The base speed of the
machine is 600 rpm corresponding to 200 Hz.
Each edge of the hall sensor indicates the border With tstep = 200µs, the number of steps per
of a sector. Each electrical rotation consists of six electrical rotation is 25 and resolution of theta is
sectors. The purpose of the position estimator is 14.4º at rated speed. Thus resolution varies
to generate a position signal in the intermediate inversely as speed and the algorithm gives
regions of the sector. The strategy of position better resolution at low speed. Reducing the step
estimation is as in figure 3. time by half will increase the resolution twice.
With TMS2407 processor running at clock
1. Evaluate shaft speed at Tx. This is done by frequency of 40MHz, the loop time to execute
measuring the time interval between Tx and Tx-1 various operations like ADC conversion, running
2. The position signal in Tx to Tx+1 is obtained by PI controllers and updating PWM takes almost
integrating the speed found at Tx. 120µs. The processor speed hence sets a limit
3. At Tx+1, update the position signal to theta = on resolution of the rotor angle.
30º + x .60 º .

2
National Power Electronics Conference 2010

Sources of Error and Error Analysis

Figure 4 shows the waveforms of the hall position


sensors and the estimated position under two
different operating speeds. It is seen that the
estimates of the position deteriorate with the
speed.

Fig 5 : Percentage Error in theta on account of


machine dynamics

As seen, the error in theta estimation is affected


by mechanical time constant and for the machine
under consideration this term is negligible
especially above 50 Hz.

Rounding Off and Truncation Errors

The evaluation of rotor angle is also affected by


Rounding off and Truncation Errors due to usage
of fixed point processor. Figure 6, shows the plot
Fig 4 : Estimated Theta at different frequencies of error in theta for different speeds.
(a) At 300 rpm (b) At 600 rpm

Two of the sources of the error are machine


dynamics and error arising out of truncation &
round off.
Machine Dynamics

Machine dynamics is one of the sources of error.


In every sector we assume that the speed is the
same as what was measured in the previous
sector. This is not true while the machine is
accelerating or decelerating. This error is related
to the inertia (J) of the machine.
Coast down test was performed to get
mechanical time constant of the machine
(J/B=1 second).
Fig 6: Percentage Error in Theta on account of
During coasting,
rounding off and truncation
−t
ω (t ) = ω0e τ (1) As seen from the figures, the position estimator
has considerable error at low speeds while at
dω − ω0
= (2) high speed resolution is poor. But any error that
dt max
τ occur due to the above factors are reset every
60º by the hall sensor interrupts.
The theta error (½ dω/dt t2) arising out of
neglecting acceleration is as shown in figure 5. Thus, with the above method of calculating theta,
vector control can be satisfactorily employed
between frequency range of 50 to 200 Hz.
Above 200 Hz vector control will still work, but
torque pulsations are present. But below 50 Hz,

3
National Power Electronics Conference 2010

the machine can lock out of synchronism and


should be avoided. It is for this reason, machine id = iα cos θ + iβ sin θ (3)
is started in BLDC mode of operation and when
the PLL is fully synchronized with the hall i q = i β cos θ − iα sin θ (4)
sensors, mode is switched over to PMSM.
Alternate estimation algorithms with better
performance are needed at low and high speeds. The motor under current control is a simple first
These could be feedback estimators with order system. It can be effectively controlled with
asymptotic error attenuation. a simple PI controller [4].

The block diagram of the proposed drive system


VECTOR CONTROLLED PMSM is as shown in figure 8. The control flow is as in
figure 9. As the performance of the estimator is
It has been shown that we have a satisfactory poor at low speeds, machine is initially started in
position estimator with better resolution in the BLDC mode of operation.
range of 50 Hz to 200 Hz. In electric vehicle
application, this frequency range and the
corresponding speeds are the predominant
operating conditions. It is therefore proposed that
this position estimator may be used with electric
two wheelers to obtain smooth torque
characteristics. Vector control of the machine will
achieve this.

The principle of operation is as follows. Figure 7


shows the motor current in the field co-ordinates.

Fig 8. Proposed Drive System

Fig 7 : Transformation to d-q frame

The id , iq currents are the motor currents


transformed to the synchronously rotating Fig 9: Control Flow
reference frame. In this frame of reference, id
represents the field in line with the permanent In BLDC mode, the hall sensors signals are fed
magnets. Normally this current is set to zero for to the sector identification table. The table
control. iq is the quadrature axis current and is determines which of the phases to be turned on.
used for torque control [3]. At any instant of time, only two phases are
conducting and each phase conducts 120º in
The transformation from stationary frame to positive and negative direction. Torque is
synchronous reference frame is given by the controlled by controlling current on the DC side of
following equations [3]. the inverter [1]. A DC current sensor on the

4
National Power Electronics Conference 2010

battery current path is used for this purpose. The


error (Ibatref - Ibatfb) is fed to a PI controller. The
output of the PI controller sets the duty cycle of
the MOSFET switches. The switches are
operated at a frequency of 5kHz.

Fig 10 shows the phase current and iq during


BLDC mode of operation. In a PMSM, torque is
directly proportional to the quadrature axis
current [3]. Thus to measure torque, it is Fig 11.Torque and Phase current in PMSM mode
sufficient to measure iq .For this purpose, rotor
angle was calculated during BLDC mode also.
Phase currents were sensed and transformed to Hysteresis band of (ω2 - ω1) has been
synchronously rotating frame of reference. Fig 10 maintained in the controller to prevent the
shows the quadrature axis current. Sixth machine from switching back and forth between
harmonic torque ripple of 20-30% is evident from two modes.
the figure.
The phase current during switchover from one
mode to another is shown in figure 12. Overshoot
is observed during transition. This is because,
during BLDC mode, current control is done on
the battery current and in PMSM mode, current
control is suddenly shifted to iq and id .

Fig 10. Torque and Phase current in BLDC mode

The position estimator starts calculating speed ω Fig 12 : Phase current during transition
and theta as the hall sensor interrupts the
processor. When ω ≥ ω1, the mode of operation
is shifted to PMSM in vector control. Torque CONCLUSION
reference is set by fixing Iq =Iq* and Id is
maintained zero. As seen in figure 8, phase The error in theta goes as high as 6º, especially
currents are sensed and using the calculated at high speeds. If machine acceleration is taken
theta, converted to Id and Iq. The errors (Iq* - Iq) into account and loop time is made lesser, the
and (Id* - Id ) are passed through PI controllers. error will be less.
The output of the controllers Vq* and Vd* is
converted back to voltages in stationary On account of loss of synchronism at low speeds
reference frame and fed to the SVPWM block. and high torque pulsations at high speeds, the
This runs the motor in vector control. position estimator is found suitable in the mid
region (50Hz to 200Hz). Alternate estimation
Fig 11 shows the R phase current and iq during algorithms with better performance are needed at
PMSM mode of operation. iq is almost a low and high speeds. These could be feedback
constant, thus ensuring ripple free torque estimators.
performance. It was observed during testing, that
the machine noise comes down well in this If the rotor angle is calculated properly at high
mode. speeds, the machine can be run above rated
speed in Field weakening mode (id < 0).

5
National Power Electronics Conference 2010

This approach is not limited to electric vehicle in TVS Motor Company, Hosur, India as an
application. High resolution encoders can be Hardware Engineer for the development of
replaced with low cost hall sensors. The Hybrid two-wheelers. Her areas of interest
algorithm is very easy to implement unlike include power electronics and variable speed
complex sensor less controls. drives.

REFERENCES

[1] P. Pillay and R. Krishnan, "Modeling, analysis


and simulation of a high performance, vector
controlled, permanent magnet synchronous
motor drive," Proc. IEEE-IAS Ann. Mtg., 1987,
pp. 253-261.
[2] R. Wu, G.R. Slemon, "A permanent Magnet
Motor drive Without a Shaft Sensor," IEEE Trans.
on I.A., vol.27, no.5, pp.1005-1011,1991.
[3] W. Leonhard, "Control of Electric Drives",
Narosa Publishing House, New Delhi, 1974.
[4] Vas Peter, "Vector control of A.C. Machines",
Oxford University Press. New York.1990.

V. Ramanarayanan
received the B.E. degree
from University of Madras,
Madras, India, in 1970, M.E.
degree from the Indian
Institute of Science,
Bangalore, India, in 1975,
and Ph.D. degree from
California Institute of Technology, Pasadena, in
1986. He held positions in industry as Senior
Design Engineer with M/S Larsen and Toubro
Ltd. from 1970 to 1979 and as Chief of Research
and Development with NGEF Ltd. from 1979 to
1982. He is currently a Professor in the
Department of Electrical Engineering, Indian
Institute of Science. He is a Consultant to several
industries in related areas. He also currently
holds the position of Academic Coordinator at the
National Mission on Power Electronics
Technology (NaMPET), India. His areas of
interest are in power electronics, industrial drives,
switched-mode power conversion, and power
quality issues.

K. Sugantha received the B.E


degree from, College of
Engineering, Guindy,
Chennai, India in 2004. She is
currently pursuing the M.Sc in
Electrical Engineering, Indian
Institute of Science,
Bangalore, India. During 2004-2008 she worked