0% found this document useful (0 votes)
48 views15 pages

Dead-Time Compensation of A Current Controlled Inverter Using The

Uploaded by

hojes12163
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
48 views15 pages

Dead-Time Compensation of A Current Controlled Inverter Using The

Uploaded by

hojes12163
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

This article was downloaded by: [Korea University]

On: 01 January 2015, At: 22:51


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,
UK

International Journal of
Electronics
Publication details, including instructions for authors
and subscription information:
http://www.tandfonline.com/loi/tetn20

Dead-time compensation
of a current controlled
inverter using the space vector
modulation method
W. S. OH
Published online: 10 Nov 2010.

To cite this article: W. S. OH (1996) Dead-time compensation of a current controlled


inverter using the space vector modulation method, International Journal of
Electronics, 80:2, 277-289, DOI: 10.1080/002072196137462

To link to this article: http://dx.doi.org/10.1080/002072196137462

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness,
or suitability for any purpose of the Content. Any opinions and views
expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the
Content should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages,
and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the
Content.
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan,
sub-licensing, systematic supply, or distribution in any form to anyone is
expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions
Downloaded by [Korea University] at 22:51 01 January 2015
int. j. electronics, 1996, vol . 80, no . 2, 277± 289

Dead-time compensation of a current controlled inverter using the


space vector modulation method²
W. S. OH³ , K. P. HONG§, Y. T. KIM§ and H. J. KIM§

The switching dead-time, avoiding a bridge leg short-circuit in a PWM voltage


source inverter, produces distortions of the controlling inverter output performance
such as current waveform, voltage vector and torque. In this paper, the in¯ uence
of dead-time is investigated. The online space voltage vector modulation method
is used for the current controller. It is possible to compensate dead-time by space±
voltage vector modulation, which generates an additional pulse compensating
voltage distortion caused by the dead-time. In addition, a narrow pulse, which is
generally neglected, can be compensated. All the algorithms, including ® eld-oriented
control are performed by a one chip microprocessor 80C196MC and DSP
Downloaded by [Korea University] at 22:51 01 January 2015

TMS320C31. Experimental results prove that the proposed scheme provides a good
inverter output performance.

1. Introduction
Pulsewidth modulated (PWM) voltage source inverters ( VSIs) for three-phase
load are widespread, feeding variable-speed or high-dynamic-controlled asynchron-
ous motors. In this high dynamic control of an induction motor using an inverter,
torque ripple, voltage and current waveform distortion are problems. The main
reasons for these problems are dead-time in¯ uence, DC link voltage variation and
the voltage drop of the power devices. Among these reasons, many studies on dead-
time have been carried out (Murai et al. 1987, Sukegawa et al. 1991, Chung and Lee
1993, Van der Broeck et al. 1986, Heumann and Schroder 1988, Pedersen et al.
1993). Dead-time is the short time that elapses between switching one device in an
inverter leg o and switching the other one on: it ensures both switches do not
conduct simultaneously.
Dead-time, which is inevitable to prevent a bridge leg short-circuit, produces
distortion of inverter output voltage and low harmonics. Therefore, dead-time
contributes to the poor performance of induction motors.
Especially in low speed drives, its in¯ uence is increased. The current control
strategies of VSI can be divided into three categories: (1) tolerance band controllers
(hysteresis controllers) with variable pulse frequency for direct control of the load
currents; (2) PWM controllers with ® xed pulse frequency for controlling the load
currents through the average load voltage; (3) space voltage controllers (which are
used in the vector controlled inverter) for better dynamic response.
In this paper, in order to compensate for the dead-time e ect, the in¯ uence of
dead-time is investigated. As the use of vector controlled AC drives and the desire
for complete digital control increase, the fully digital controller, using the space

² A shorter version of this paper was presented at the International Conference on Power
Electronics and Drive Systems (PEDS’95), which was held at the Pan Paci® c Hotel, Singapore,
on 21± 24 February 1995.
³ Department of Electrical Engineering, Yuhan Junior College, Puchon, 185-34, South
Korea. Tel: + (82) 02 610 0735; Fax: + (82) 02 619 9771.
§ Power Electronics Control Lab. Dept. of Electrical Engineering, Hanyang University,
Seoul, 133-791, South Korea. Tel: + (82) 0345 400 5169; Fax: + (82) 0345 407 9930.
0020± 7217/96 $12.00 Ñ 1996 Taylor & Francis Ltd.
278 W. S. Oh et al.

voltage vector strategy, is considered. The narrow switching pulse of the space
voltage controller, which is smaller than dead-time, is considered. The proposed
algorithm is performed by software without additional hardware. An experimental
system composed of one chip microprocessor 80C196MC and TMS320C31 is
designed.

2. Analysis of inverter switching


In order to investigate the in¯ uence of dead-time and switching times on voltage
distortion, the ® rst step is to analyse the behaviour of one phase leg of a PWM
inverter. Figure 1 shows one phase leg of a PWM inverter.
During the dead-time, inverter output voltages depend only on the current
direction. If ia is positive, then diode D4 is on and the output voltage depends on
switching signal G1, which is the switching pulse of the switch S1. In the alternative
case, the diode D1 is on and the output voltage depends on switching signal G4,
Downloaded by [Korea University] at 22:51 01 January 2015

which is the switching pulse of switch S4.


Figure 2 shows the switching waveforms of the inverter. When phase current i a

Figure 1. One phase leg of the inverter.

Figure 2. Switching waveforms of the inverter.


Dead-time compensation of current controlled inverter 279

is positive, the switching pulse is narrower than the reference pulse, therefore the
output voltage has a smaller value than the reference value (Pedersen et al. 1993).
Ta o n = Ta r e f Õ ((T d + T o n ) Õ To ff ) (1)
In the alternative case (i a < 0 ), that is, the phase current ia is negative, the switching
pulse is wider than the reference pulse, therefore the output voltage has a larger
value
T a o n = T a r e f + ((T d + T o n ) Õ To ff ) (2)
where T a r e f is the reference switching time for the output voltage, T a o n is the actual
switching time of the output voltage, T o n , T o f f are the turn on and turn o times of
switching device, respectively, and T d is the dead-time.
Equation (3) de® nes the switching time for three-phase
Tp o n = Tp refÕ sgn (i p )T e (3)
Downloaded by [Korea University] at 22:51 01 January 2015

where p = a, b, c and T e = (T d + T o n ) Õ To f f

1 for ip > 0
sgn (i p ) =
GÕ 1 for i p < 0
Thus, the average phase-to-centre voltage follows as (4)

A B
Tp o n 1
Up n = E d Õ p = a, b, c (4)
Ts 2

where T s is sampling time and E d is inverter DC link voltage.


There is a di erence between the reference and the actual inverter output voltage,
and it deteriorates the high performance drive of an induction motor.

3. Space vector current controller


The three-phase voltages are represented by a voltage space vector. There are
eight space vectors available, according to eight switching positions of the inverter.
It can be represented by (5) ( Van Broeck et al. 1988)
2
E d exp ( j(k p /3 )) for k = 1, 2, . . . , 6
G
3
V k (G a , G b , G c ) = (5)
0 for k = 0, 7
We de® ne a mean space vector V S , which is almost constant during a switching
period. This vector generates the fundamental behaviour of the machine, e.g. currents
and torque. V S is obtained from the speed controller, e.g. the PI speed controller.
V S is also modulated to the optimum space voltage vectors in the SVCC (Space
Vector Current Controller). Optimum space voltage vectors are met in general if
only the three switching states adjacent to the reference vector are used and the
cycle (wherein the average voltage vector becomes equal to the reference vector)
consists of three successive switching states only. Figure 3 represents the space
vector modulation technique. Figure 3(a ) shows the reference voltage vector V S ,
which is obtained from the adjacent two voltage vectors V A , V B and zero vectors.
G a , G b and G c are switching signals of the inverter.
For su ciently high switching frequency, the reference space vector V S is assumed
280 W. S. Oh et al.

(a) (b)
Figure 3. Space vector modulation: (a ) voltage vector; (b ) pulse pattern of the space vector.
Downloaded by [Korea University] at 22:51 01 January 2015

constant during one switching cycle. We can obtain


V S T Z = V 0 T0 + VA TA + V B TB + V7 T7 (6)
Here, T Z = T 0 + T A + T B + T 7 , T 0 , T 7 , T A and T B are the switching times of the output
voltage vectors. If the space vectors in this equation are described in rectangular
coordinates, it follows that
cos c cos 60ß
Csin D CD Csin 60ß D
2 2 1 2
TZ Ed a = TA Ed + TB Ed (7)
3 c 3 0 3

|VS |
a= 2 (8)
3 Ed

hence
sin ( 60 ß Õ c )
TA = TZ a (9)
sin 60ß
sin c
TB = T Z a (10)
sin 60ß
T0 + T7 = TZ Õ (T A + T B ) (11)
This results in a de® nite switching order according to Fig. 3(b ), which shows
exactly the same symmetry as derived for established sinusoidal modulation.

4. Inverter analysis on dead-time


Inverter analysis on dead-time is divided into two modes. In the case of mode 1,
a narrow pulse, which is a driving pulse less than dead-time, is not present. In the
case of mode 2, a narrow pulse is present.
The sum of the current vector in the three-phase inverter must be zero. Thus,
one current phase is di erent from the others. Figure 4 represents the structure of a
three-phase inverter and space voltage vectors.
The light line is the reference voltgae vector and the heavy line is the actual
voltage vector considering the dead-time e ect.
In the case of Fig. 4, we analyse mode characteristics of the dead-time e ect.
Dead-time compensation of current controlled inverter 281

(a)
Downloaded by [Korea University] at 22:51 01 January 2015

(b)
Figure 4. Structure of a three-phase inverter and space voltage vector: (a ) three-phase inverter;
(b ) phase pattern of the space vector.

4.1. Mode 1
We can de® ne the driving duration of the output voltage as (12) from (1)

Ta o n = Ta r e f Õ Te
T b o n = Tb ref Õ Te (12)
Tc o n = T c r e f + T e

The average phase voltage is obtained as (13)

A B
Ta refÕ Te 1
Ua n = E d Õ
Ts 2

A B
Tb r e f Õ Te 1
Ub n = E d Õ (13)
Ts 2

A B
T c r e f + Te 1
Uc n = Ed Õ
Ts 2
282 W. S. Oh et al.

Thus, line to line voltage is given by (14)

A B
Ta ref Õ Tb ref
Ua b = Ed
Ts

A B
Tb ref Õ Tc r e f Õ 2T e
Ub c = E d (14)
Ts

A B
Tc r e f Õ T a r e f + 2T e
Uc a = E d
Ts

If we rearrange (14), we obtain

Ua b = U a b r e f
Ub c = U b c r e f Õ Ue (15)
Downloaded by [Korea University] at 22:51 01 January 2015

U c a = Uc a r e f + U e

where

A B
2T e
Ue = E d (16)
Ts

and U a b r e f , U b c r e f , U c a r e f are average reference line-to-line voltages.


There is no di erence between a reference line-to-line voltage and the actual one
in phases a and b , whose current directions are the same. Line-to-line voltages
between di erent direction currents have di erence U e . Thus, the average voltage
vector is distorted.

4.2 Mode 2
Current directions are the same as Mode 1 and there is a narrow pulse in phase a .
The driving duration of the output voltage is given by

Ta o n = 0
T b o n = Tb ref Õ Te (17)
Tc o n = T b r e f + T e

and the average phase voltage is given by

Ed
Ua n = Õ
2

A B
Tb refÕ Te 1
Ub n = E d Õ (18)
Ts 2

A B
Tc r e f + Te 1
Uc n = E d Õ
Ts 2
Dead-time compensation of current controlled inverter 283

Thus, the average line-to-line voltage can be written as

A B
Õ Tb r e f + T e
Ua b = Ed
Ts

A B
Tb r e f Õ Tc r e f Õ 2T e
Ub c = E d (19)
Ts

A B
Tc r e f + Te
Uc a = E d
Ts

Also rearranging (19), we obtain

Ta r e f Õ Te
Ua b = Ua b r e f Õ Ed
Ts
Downloaded by [Korea University] at 22:51 01 January 2015

Ub c = Ub c r e f Õ Ue (20)
Ta r e f + Te
Uc a = U c a r e f + E d
Ts

where U a b r e f , U b c r e f , U c a r e f are the average reference line-to-line voltages.


The same approach can be carried out in another case.
(a) ia , ib < 0, ic > 0 and the narrow pulse is in phase a .

Ta r e f Õ Te
Ua b = Ua b refÕ Ed
Ts
Ub c = Ub c r e f + Ue (21)
Ta ref Õ 3T e
Uc a = U c a r e f + E d
Ts

(b) ia , i b > 0, ic < 0 and the narrow pulse is in phase c .

Ua b = U a b r e f
Tc r e f Õ 3T e
Ub c = U b c r e f + E d
Ts (22)
Tc r e f Õ 3T e
Uc a = Uc a r e f Õ Ed
Ts

(c) ia , ib < 0, ic > 0 and the narrow pulse is in phase c .

U a b = Ua b r e f
T c r e f + Te
Ub c = Ub c r e f + E d
Ts (23)
T c r e f + Te
Uc a = Uc a refÕ Ed
Ts

The in¯ uence of the narrow pulse of each phase is not regular, like mode 1.
284 W. S. Oh et al.

5. Compensation of dead-time e ects


In space voltage vector modulation, the reference voltage is made by three
optimum switching pulses between sampling times. The output voltage depends on
the direction of the phase current. For the compensation of dead-time, optimum
switching pulses are compensated for with the di erence between the reference
voltage vector and the actual voltage vector according to mode.

5.1. Mode 1
When two phase currents ia , i b of the inverter are in same direction, and phase
current ib is in the di erent direction, the di ernce between the output voltage and
reference voltage is U e . Two phases having the same current sign have no dead-time
in¯ uence and another phase, having the opposite current sign, must be compensated.
U e is the voltage error accompanied by reduced time interval 2T e . Thus, compensating
time is obtained as (24)
Downloaded by [Korea University] at 22:51 01 January 2015

(a) (b)
Figure 5. Compensated waveforms of: (a ) mode 1; (b ) mode 2.
Dead-time compensation of current controlled inverter 285

Tc c o m = Tc r e f Õ 2T e (24)
where T c c o m is compensated by reference time T c r e f . If current directions are reversed,
compensating time is obtained as (25)
T c c o m = T c r e f + 2T e (25)
Figure 5(a ) represents the compensated waveforms of mode 1.

5.2. Mode 2
When current directions are equal to mode 1 and the narrow pulse is in phase a,
the reference driving pulse in phase a is cut. Thus, we must compensate for the cut
pulse. From (20), the other phases are compensated as
Tb c o m = Tb r e f Õ (T a r e f Õ
H
Te )
(26)
Tc c o m = Tc r e f Õ (T a r e f + T e )
Downloaded by [Korea University] at 22:51 01 January 2015

Figure 5(b ) represents compensated waveforms of mode 2. In Fig. 5(b ), T c o m is


the compensated time during half the sampling period. According to the current
direction and the phase generating narrow pulse, the compensations are as follows.
(a ) ia , ib < 0, ic > 0 and the narrow pulse is in phase a .
= Tb r e f Õ ref Õ

H
Tb co m (T a Te )
(27)
Tc c o m = Tc r e f Õ (T a r e f Õ 3T e )

(b ) ia , ib > 0, ic < 0 and the narrow pulse is in phase c .


T a c o m = Ta refÕ (T c r e f Õ
H
3T e )
(28)
Tb co m = Tb r e f Õ (T c r e f Õ 3T e )

Host
link Communication

PWM
Switching Protection
pattern

Encoder

Figure 6. Block diagram of the experiment apparatus.


Downloaded by [Korea University] at 22:51 01 January 2015 286 W. S. Oh et al.

(a)

(b)

(c)
Figure 7. Phase current waveforms: (a ) non-compensated; (b ) compensated; and (c ) compens-
ated considering narrow pulse.
Dead-time compensation of current controlled inverter 287

MAG
dBV

10 µB

± 50
0 PWR SP A hon LIN 500 Hz
X: 15.00 Hz Y: ± 10.78 dBV
Downloaded by [Korea University] at 22:51 01 January 2015

(a)
0

MAG
dBV

10 µB

± 50
0 PWR SP A hon LIN 500 Hz
X: 15.00 Hz Y: ± 9.90 dBV
(b)
0

MAG
dBV

10 µB

± 50
0 PWR SP A hon LIN 500 Hz
X: 15.00 Hz Y: ± 10.00 dBV
(c)
Figure 8. FFT analysis waveforms of phase current: (a ) non-compensated; (b ) compensated;
and (c ) compensated considering narrow pulse.
288 W. S. Oh et al.

(c ) ia , ib < 0, ic < 0 and the narrow pulse is in phase c .


Ta c o m = Ta r e f Õ (T c r e f Õ
H
Te )
(29)
T b c o m = Tb ref Õ (T c r e f + T e )

6. Experiment
In this paper, the dead-time compensating algorithm is accomplished with soft-
ware in the one chip microprocessor, 80C196MC and DSP TMS320C31 without
any other additional hardware circuits.
The 80C196MC is a 16-bit microcontroller designed primarily to control three-
phase induction and DC brushless motors and can perform pulse-width modulation
using a three-phase waveform generator with minimum CPU intervention. An IPM
(Intelligent Power Module)-based inverter is designed. An IPM is a three-phase 30 A,
600 V, current-sense IGBT type. Sampling time is 0´125 ms and dead-time is 8 m s.
The operating frequency of experiment is 15 Hz as, at low speed, the dead-time
Downloaded by [Korea University] at 22:51 01 January 2015

in¯ uence is increased. Figure 6 shows a block diagram of the experiment.


TMS320C31 accomplishes the vector control routine and outputs the reference space
voltage vectors. In the 80C196MC, the space vector modulation and dead-time
compensation algorithm is performed.
Figure 7 shows the resulting current waveforms. Figure 7(a ) is a non-compensated
current waveform. Figure 7(b ) is a compensated current waveform disregarding the
narrow driving pulse. Figure 7(c ) is the compensated current waveform considering
a narrow pulse. By comparison, dead-time in¯ uence is properly compensated.
Distorted waveforms are improved with a proposed algorithm. Figure 8(a ), (b ) and (c )
represent the resulting FFT analysis waveforms of the current. In Fig. 8(c) , the
compensated waveform considering a narrow pulse, harmonic components are more
reduced than in Figs 8(b) and 8(c) . Thus, considering the narrow pulse is reasonable
and it should be included in the compensation processes.

7. Conclusions
The demand for high dynamic performance AC motor drives leads to a fast
response of the current loop and low harmonic current components. The switching
dead-time increases and causes harmonic components of inverter outputs.
A space voltage vector modulation compensating dead-time is presented. This
modulator works online and the new voltage vector information is computed every
sampling period, allowing a fast response of the current loop and yielding a good
harmonic performance. In mode 1, among the space± voltage vector pulse patterns,
only one phase having a di erent direction current is compensated. In mode 2, two
phases having no narrow pulse are compensated.
The proposed algorithm is accomplished only with software; additional hardware
circuits are not required. All control tasks have been realized on a one chip micro-
processor 80C196MC and DSP TMS320C31, with the consequent savings in hard-
ware and with the possibility of introducing novel algorithms to improve the drive
e ciency.

R eference s
C hung, Y . T ., and L ee, S . Y ., 1993, The waveform compensation of dead time in PWM
inverters. Journal of the Korean Institute of Electrical Engineers, 42, 29± 35.
Dead-time compensation of current controlled inverter 289

H eumann, K ., and S chroder, H ., 1988, Design criteria for fast switching PWM inverters.
Power Electronic Specialists’ Conference (PESC’88), pp. 271± 276.
M urai, Y ., W atanabe, T ., and I wasaki, H ., 1987, Waveform distortion and correction circuit
for PWM inverter with switching lag times. IEEE T ransactions on Industry
Applications, 23, 881± 886.
P edersen, J . K ., B laabjerg, F ., J ensen, J . W ., and T hogersen, P ., 1993, An ideal PWM-
VSI inverter with feedforward and feedback compensation. European Conference on
Power Electronics and Application (EPE93), pp. 501± 507.
S ukegawa, T ., K amiyama, K ., M atsui, T ., and O kuyama, T ., 1991, Fully digital vector
controlled PWM VSI fed AC drives with an inverter dead time compensation strategy.
IEEE T ransactions on Industry Applications, 27, 552± 559.
V an D er B roeck, H . W ., S kudelny, H . C ., and S tanke, G . V ., 1986, Analysis and realization
of a pulse width modulator based on voltage space vectors. Conference Record of
IEEE Industry Applications Society (IAS86), pp. 244± 251.
V an D er B roeck, H . W ., S kudelny, H . C ., and S tanke, G . V ., 1988, Analysis and realization
of a pulse width modulator based on voltage space vectors. IEEE T ransactions on
Industry Applications, 24, 142± 150.
Downloaded by [Korea University] at 22:51 01 January 2015

You might also like