IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 16, NO. 3, MAY 2001 351
Digital Scalar Pulse-Width Modulation: A Simple
Approach to Introduce Non-Sinusoidal Modulating
Waveforms
Cursino Brandão Jacobina, Senior Member, IEEE, Antonio Marcus Nogueira Lima, Member, IEEE,
Edison Roberto Cabral da Silva, Senior Member, IEEE, Raimundo Nazareno Cunha Alves, and Paulo Fernando Seixas
Abstract—The digital scalar pulse-width modulation (DSPWM)
gathers the characteristics of simplicity of implementation found
in the regular sampling with the flexibility of manipulation of
the switching patterns in the space vector modulation (SVPWM).
This paper establishes a correlation between the SVPWM and
DSPWM techniques. It also shows how to make the DSPWM
strategy equivalent to the SVPWM technique without loosing its
simplicity of implementation. By using such equivalence concept a
microprocessor-based scheme, which uses standard timer circuits
and a simple software algorithm, is proposed to implement
the DSPWM technique. The introduction of the “distribution
ratio” in this technique, allows the development of a systematic
approach for implementing of either conventional or any modified
vector strategies without changing the modulator scheme. This
corresponds to generate any attractive nonsinusoidal modulating
signals (NSMS) in the carrier-based modulation techniques.
Furthermore, the simple digital blocks can be easily implemented
as an specialized integrated circuit. Simulated and experimental
results demonstrate the validity of the proposed methods.
Index Terms—Pulse-width modulation, three-phase inverter.
I. INTRODUCTION
T
HE classical sine-triangle modulation, or natural sam-
pling modulation (NSPWM), compares a high frequency
triangular carrier with three reference signals, known as mod-
ulating signals, to create gating pulses for the switches of the
power converter [1]. This technique is basically an analog
domain technique and its digital version led to the tech-
nique named as regular-sampled PWM (RSPWM) [2]. In the
RSPWM technique, the modulating signal is sampled at each
period (symmetric regular sampling) or at every peak (asym-
metric regular sampling) of the triangular signal to produce
a sampled-hold modulating wave. Its digital comparison to a
triangular signal, generated by up-down counters, define the
switching instants. In other words, the corresponding time in-
tervals are computed in real time from the respective sampled
value [3]. Differently from the previous methods, the space
vector pulse-width modulation (SVPWM) technique [4], [5]
does not consider each of the three phases as a separate entity.
The three-phase voltages are simultaneously performed within
Manuscript received September 1, 1998; revised January 4, 2001. Recom-
mended by Associate Editor F. D. Tan.
The authors are with the Departamento de Engenharia Elétrica, Universi-
dade Federal da Paraíba, Campina Grande, Paraiba 58109-970, Brazil (e-mail:
jacobina@dee.ufpb.br).
Publisher Item Identifier S 0885-8993(01)04026-1.
a two-dimensional reference frame ( plane), the complex ref-
erence voltage vector being processed as a whole. Because its
flexibility of manipulation, the SVPWM technique is widely
employed, nowadays [3].
It is well-known that the addition of proper zero-sequence
components to the modulating signals generates nonsinusoidal
modulating signals (NSMS). Several different waveform pro-
files can be used as modulating signals [3], [6]. These NSMS
improve the performance of both NSPWM and RSPWM [7],
[8]. On the other hand, the same effect obtained by the use
of NSMS in carrier-based techniques, is achieved with both
conventional SVPWM and modified SVPWM techniques. It
should be noted that the modified SVPWM strategy is known
in the literature under names such as “two-phase modulation”
[9], “bus-clamping modulation” [10], or “discontinuous modu-
lation” [11].
Several authors have discussed the correlation among the
NSPWM, RSPWM, and SVPWM techniques, under different
focuses. With that purpose, the concept of “ordering of the ref-
erence voltages” has been used to establish the analogy among
the sectors defined by the active vectors and the segments of
60 , existing in a period of the references. Also, the concept
of “distribution ratio” [12], named as “apportioning factor” in
[8] and defined as the relation between the time of application
of one of the two null vectors and the total null-vector time,
has been employed. In fact, properly choosing the distribution
factor determines both the distribution of the zero voltage
vectors inside the sampling period and its correspondence to
the modified SVPWM [8], [12]–[14].
The alternative “digital scalar pulse-width modulation”
(DSPWM) technique imposes, to the pole voltage of an inverter
leg, an average value that corresponds to each reference phase
within the sampling interval [15]. Such strategy is of simpler
implementation than the SVPWM technique, reducing the
effort of calculation [16]. The technique introduced in [17] has
a similar treatment by using the concept of reallocation of the
“effective time.” This “effective time” is in fact the sum of the
times of application of the active vectors. The pulse-widths are
ordered and the sum of times is appropriately moved within the
sampling period.
Different methods have been employed to implement the re-
sultant modulators of these recent studies. In [17], an algorithm
is provided and implemented with a DSP TMS320C31. How-
ever, because of the correspondence between of the carrier-
0885–8993/01$10.00 © 2001 IEEE