INTRODUCTION TO RECONSTRUCTION OF

SAMPLED SIGNALS.
1. INTRODUCTION.
 A data-reconstruction device also known as a filter is often
used to interface between digital and analog components.
The hold circuit in the S/H device is the most common
filtering device in the discrete-data systems. Filtering
devices or data-reconstruction devices are used to sieve
out high-frequency harmonic components in a signal
resulted from the sampling operation. Although the S/H
device comes in a single unit, for mathematical
simplification, only the hold device is modeled.

 Firstly, assume that an ideal sampler is sampling at a

sampling frequency ωC, which is at least twice as large as
the maximum frequency component of the continuous
input signal being sampled. The reconstruction of
continuous data from digital data using an ideal low-pass
filter. The amplitude characteristics of an ideal filter are
shown in Figure 1. The amplitude gain is unity.

 Perfect reconstruction of the continuous signal is based on

the assumption that f (t) is band limited. Hence, it is
impossible to recover a totally perfect continuous signal
once it is sampled. The best way to reconstruct a signal is
to approximate the original time function as closely as
possible.

 The hold device is the simplest form of a general data-

reconstruction problem. The problem of data
reconstruction can be regarded as a given sequence of
numbers f (0), f (T), f (2T), . . . , f (kT), . . .. An analog
signal f (t), where t >0 is to be reconstructed from the
information contained in the sequence. This data-
reconstruction process may be regarded as an extrapolation
 process, since the analog signal is to be constructed based
on information available only at past sampling instants.

 Power series expansion of f (t) in the interval between the

sampling instants kT and (k +1)T is used to generate a
desired approximation. The approximation is:

 The higher the order of the derivation, the larger will be

the number of delayed pulses required. In general, the
number of delayed pulse data required to approximate the
value of f (n)(kT) is n+1. Thus, the extrapolating device
consists of a series of time delays, and the number of
delays depends on the accuracy of the estimate of the time
function f (t) during the time interval from kT to (k +1)T.

 Although utilizing a higher-order derivative produces a

more accurate extrapolation, it causes a reduction on the
stability of the closed-loop control systems and it also
makes the circuitry more complicated and expensive.

Figure 1 Amplitude characteristic of an ideal filter.

2. RECONSTRUCTION BY ZERO ORDER HOLD. (ZOH).

 The most widely used holding device is the zero-order

hold (ZOH). This is because it is less complicated and less
expensive. A simple ZOH is shown in Figure 3. It is called
zero-order extrapolator as its polynomial used is of the
zeroth order.
 It holds the value of the sampled value f (kT) for kT ≤t <(k
+ 1)T until the next sample f [(k + 1)T] arrives:

fk (t) = f (kT) ………………. (3)

Figure 3 A simple ZOH device

 This equation is used for approximation of f (t) during the

time interval kT ≤t < (k + 1)T. If a unit impulse input
signal is applied to a ZOH, the impulse response is shown
in Figure 4. The ZOH is a linear device as it satisfies the
principle of superposition.

 The impulse response of a ZOH is expressed as:

gh0(t) = us(t) − us(t − T) ……… (4)

Where us(t) is the unit-step function.
The response of the ZOH to the unit-impulse input is equivalent
to the difference between two unit-step function. By taking a
Laplace transform, the transfer function is obtained as:

Gh0(s) = ( 1 − e−Ts ) / s ……….. (5)

Figure 4 Responses of a ZOH: (a) unit-impulse input to and (b)

impulse response of ZOH.

 Figure 5 shows the output waveform of a ZOH with a

pulse sequence input. The output waveforms clearly
indicate that the accuracy of the ZOH greatly depends on
the magnitude of the sampling frequency ωS. As the
sampling frequency increases toinfinity or sampling period
T approaches zero, the output of the ZOH h(t) approaches
the analog signal f (t).

 The ZOH is a data-reconstruction or data-filtering device.

Hence it is useful to examine its frequency domain
characteristics. By replacing s by jω in the transfer
function, we get,

Gh0( jω) = (1 − e−Tjω) / jω …….. (6)

 Figure 5 Input and output waveforms of a ZOH in the time
domain: (a) input signal f (t) and sampled signal f ∗ (t), and (b)
output waveform of ZOH.

3. RECONSTRUCTION BY FIRST ORDER HOLD. (FOH).

 The first-order hold (FOH) uses the first two terms of the
power series (Equation 2) to extrapolate the time function f
(t) over the time interval kT ≤t <(k + 1)T. The equation for
the FOH is:

fk (t) = f (kT) + f (1)(kT)(t − kT) …….. (7)

where the first-order derivative of f (t) (f(1) (kT)) at t = kT is

approximated as:

f (1)(kT) = {f (kT) − f [(k − 1)T]} / T …….. (8)

 Substituting Equation (8) in Equation (7) gives:

  The output of the FOH between two consecutive sampling
instants is a ramp function. The slope of the ramp is equal
to the difference of f (kT) and f [(k - 1)T].

 By applying a unit impulse at t =0 as input, an impulse

response of the FOH is obtained. The corresponding
output is obtained by setting k =0, 1, 2, . . . for the various
time intervals.

 For k =0,

f0(t) = f (0) + {[f (0) − f (−T)]/T} t 0 ≤ t ≤ T .(10)

For a unit-impulse input, f (0)=1 and f (−T)=0, the impulse

response of the FOH for 0≤ t ≤T is:

gh1(t) = 1 + t / T

 For k =1,

 The impulse response of the FOH over time interval T ≤ t

≤2T is (see fig.6 f(0)=1 and f(T)=0):

gh1(t) = −(t – T) / T

 If a unit-impulse input signal applied to a FOH, the

impulse response is shown in Figure 6.
Figure 6. Responses of an FOH: (a) unit-impulse input signal
and (b) impulse response of the FOH.

 The transfer function of the FOH is obtained by taking the

Laplace transform of the FOH is

 The reconstruction of a continuous-time signal by means

of an FOH is shown in Figure 7.

Figure 7 Reconstruction of a continuous-time signal by means

of an FOH.
4. RECONSTRUCTION BY SECOND ORDER HOLD. (SOH).

 The Second-order hold (SOH) uses the first three terms of

the power series (Equation 2) to extrapolate the time
function f (t) over the time interval kT ≤ t <(k + 1)T. The
equation for the SOH is:

 The output of the second order hold between two

consecutive sampling instants may be a parabola function.
The simplest of the curve is a square-law to the difference
of f (kT) and f [(k + 1)T].

 By applying a unit impulse at t =0 as input, an impulse

response of the second-order hold is obtained. The
corresponding output is obtained by setting k =0, 1, 2….
for the various time intervals. For k =0, a unit impulse
input, f (0)=1 and f (−T)=0, the impulse response of the
second order hold for 0≤ t ≤T is

 For k =1, the impulse response of the second order hold

over time interval T ≤ t ≤ 2T is

.
 If a unit impulse input signal is applied to an SOH, the
impulse response is shown in Figure 8.
Figure 8 Responses of an SOH: (a) unit-impulse input to and
(b) impulse response of SOH.

 Although using a higher-order derivative produces a more

accurate extrapolation, it causes a reduction on the stability
of the closed-loop control systems and it also makes the
circuitry more complicated and expensive. Generally, say
that the impulse response of an SOH is a parabola function
as shown in Figure 8.

 The reconstruction of a continuous-time signal by means

of an SOH is shown in Figure 9.

Figure 9 Reconstruction of a continuous-time signal by means

of an SOH.

 The typical second-order transfer function in the s-domain

is shown as follows:
Where M is the voltage-transfer gain; τ, the time constant; τd, the
damping time constant (in which τd = ξ τ) and s is the Laplace
operator in s-domain.
In order for the typical second-order transfer function to be
applicable to the digital modeling for power DC/DC converters,
a z-transformation is necessary.
MATHEMATICAL MODELING OF DIGITAL
POWER ELECTRONICS CONVERTERS

 All switching circuits including all AC/DC rectifiers,

DC/AC inverters, DC/DC converters and AC/AC
(AC/DC/AC) converters are working in the discrete-time
state.
 Therefore, they have to be described by digital control
theory rather than analog control.

1. CRITERIA FOR ANALOG AND DIGITAL

SYSTEMS.

 Analog control theory describe the systems working in the

continuous time process. All parameters in these systems
vary from time to time continuously.

 The dynamic process can be described by differential

equations and Laplace transform operations. If there is any
element in a control system that is discrete-time circuit,
then the whole system has to be treated as a discrete-time
system since the signal transferring is no longer
continuously varying.

 All switching circuits are working in the discrete-time

mode. They have not responded to the signal change in
the continuous state. Therefore, any system involved in a
switching circuit has to be treated as a discrete system.

 Most digital control systems have the analog output

functions. For example, digitally controlled variable-speed
motor drive systems have analog output speed in
continuous time mode although their controllers are digital
signal processor (DSP).
 The differences between analog and digital control
systems are in the following aspects:

o transfer function’s form,

o stability characteristics,
o unit-step responses and impulse responses.

Figure 10 Stability boundary in the s-plane.

 The transfer function of all analog control systems is

described by the Laplace transform in the s-domain. The
stability characteristics and criteria states that if all poles
of the transfer function are located in the left-hand half-
plane (LHHP) of the s-plane, the system is stable. If any
pole is located in the right-hand half-plane (RHHP) of the
s-plane, the system is unstable. The stability boundary is
shown in Figure 10. If any pole is located on the stability
boundary, it means that the system performs at the critical
stability state. Usually the system is considered unstable.
A stable analog control system has stable unit-step
response with or without oscillation. Correspondingly, the
impulse responses are also stable with or without
oscillation.
 The system stability can be modified and improved by a
closed-loop control and an optimization operation. A good
analog control system has the following step response with
the typical characteristics:

o fast response, i.e. the settling time is less than 4.7

times of the time constants;
o oscillation cycle number is not more than 2;
o the overshoot is not more than 5%.

 These technical features are available for the impulse

responses.

 The transfer function of all digital control systems is

described by either the Laplace transform in the s-domain
and/or the z-transform in the z-domain.

 The stability characteristics and criteria sates that if all

poles of the transfer function are located inside the unity-
cycle of the z-plane, the system is stable. If any pole of the
transfer function is located outside the unity-cycle of the z-
plane, the system is unstable. The stability boundary (the
unity-cycle in the z-plane) is shown in Figure 11. If any
pole is located on the stability boundary, it means that the
system performs at the critical stability state. Usually the
system is considered unstable.

 A stable digital control system has stable unit-step

response with or without oscillation. Correspondingly, the
impulse responses are also stable with or without
oscillation.
 Figure 11 Stability boundary in the z-plane.

 The system stability can be modified and improved by a

closed-loop control and an optimization operation. A good
analog control system has the following step response with
the typical characteristics:

o fast response, i.e. the settling time can be only one-

step delay;
o oscillation cycle number is not more than 2;
o the overshoot is not more than 5%.

 These technical features are available for the impulse

responses.
2. A ZERO-ORDER HOLD (ZOH) FOR AC/DC
CONTROLLED RECTIFIERS

 AC/DC rectifiers have many forms which are previously

presented.

 The devices of all types of the AC/DC controlled rectifiers

can be thyristors, transistors, bipolar transistors (BTs), gate
turn-off thyristors (GTOs) and Triacs. They are controlled
by the corresponding firing pulse with certain firing angle
α.

 In a sampling interval, usually in the commutation period

the firing angle can be changed only once, i.e. the output
voltage of an AC/DC controlled rectifier is changed
period-by-period. Therefore, all types of the AC/DC
controlled rectifiers are working in the discrete state.

 In general situation, the load of the rectifiers is an L–R

circuit with the time constant τ =L/R. If the output current
is continuous, then the average value of the output DC
voltage is:

Vd = Vd-max cos α ………………..…. (16)

where α is the firing angle of the applied firing pulse. Vd is

the output DC average voltage of the rectifier. Vd-max is
the maximum output DC average voltage of the rectifier
corresponding to the firing angle α=0.

 Refer to the single-phase fully controlled AC/DC rectifier

with an R–L load shown in Figure 12. The corresponding
input and output voltage and current waveforms are shown
in Figure 4. If the AC input power supply with the
frequency f =50 Hz (T =1/f =20 ms), then each device is
 conducted in half a cycle, i.e. the conduction angle is 180◦
(or π rad) or in the interval of 10 ms. The current
commutation happens twice a cycle.
 The output voltage is out of control in a half-cycle once
the firing pulse is applied.

 Therefore, it is the element to keep the output voltage in a

period of T/2=1/2f . By per unit system, the voltage
transfer gain is unity (1) in a sampling interval T/2=10 ms.
That is:

G(t) = VO / Vin|per-unit = 1 …………. (17)

 Analogously, a single-phase half-wave AC/DC rectifier

has the sampling interval to be 20 ms. A three-phase half-
wave AC/DC rectifier has the sampling interval to be 6.67
ms (i.e. T/3) and a three-phase full-wave AC/DC rectifier
has the sampling interval to be 3.33 ms (i.e. T/6). No
matter how the multi-phase rectifier is, the output voltage
is expressed by Equation (16), its voltage transfer gain in
per-unit system is unity (1) as described in Equation (17).

Figure 12 A single-phase fully controlled AC/DC rectifier.

Figure 13 The input and output voltage and current waveforms.

2.1 Traditional Modeling for AC/DC Controlled Rectifiers

 Traditional modeling for AC/DC controlled rectifiers is a

time-delay element in the s-domain. The delayed time is
statistically as the sampling interval T or commutation
period σ.

 For example, if the power supply frequency f =50 Hz, a

single-phase half wave controlled rectifier has the time
delay T = σ =20 ms. The corresponding transfer function
of this rectifier in per-unit system is:
 G(s) = e−Ts = e−σs …………………… (18)

 If the rectifier is used in a current control system or speed

control system, and the current/speed responses are in the
stage 0.1–1 s, we can consider the variable in a sampling
interval of 20 ms is comparably small, i.e. Ts=σs→0.
Hence, Equation (3) can be rewritten as:

G(s) = e−σs = 1 – σs = 1 / (1 + σs) …….. (19)

This mathematical modeling is widely used in industrial

applications.

2.2 A Zero-Order Hold for AC/DC Controlled Rectifiers in

Digital Control

 The mathematical modeling given in Equation (19) was

very popular since 1960s to 1980s. Many large machinery
systems have slow-time responses and use analog
proportional-plus-integral (PI) controllers. This modeling
is good enough to describe the system characteristics.
Since 1980s the digital processors were applied in research
and industrial application, where the systems performed
faster. The rectifiers cannot be considered only a time-
delayed element. Since its output voltage is out of control
once the firing pulse was applied, it should be looked as a
sample-and-hold element.

 Therefore, its mathematical modeling in per-unit digital

control system should be a zero-order hold (ZOH) in both
the s- and z-domains:
  It means that the AC/DC controlled rectifier performs a
sampling time delay in the s-domain and one-step delay
(T) in a digital control system.

3. A FIRST-ORDER TRANSFER FUNCTION FOR

DC/AC PULSE-WIDTH-MODULATION INVERTERS

 Pulse-width-modulation (PWM) DC/AC inverters have

many forms which for example : DC/AC single phase,
DC/AC three phase, multi-level PWM, space vector
modulation, … etc.

 The devices of all types of the DC/AC inverters can be a

transistor, BJT, MOSFET (all types), IGBT, IGCT, …. etc.
They are controlled by the PWM scheme with the certain
carrier (chopping) frequency fc. In a sampling interval T
=1/fc, the pulse-width angle can be changed only once, i.e.
the output voltage of an DC/AC PWM inverter is changed
period by period. Therefore, all types of the DC/AC PWM
inverters are working in discrete state.

 In general situation, the input reference signal vin(t) is a

sinusoidal waveform, that is

vin(t) = Vm sin ωt ……………….. (22)

 Normally, the load of the inverters is an R–L circuit with

time constant τ =L/R, or the delayed angle is:
 φ = tan−1 (ωL/R) ……………….. (23)

 If the output current is continuous, the output AC voltage

instantaneous value after the filter should be:

vO(t) = Vm sin(ωt − φ) ………… (24)

where υ is the delay angle of the output voltage with

reference to the input signal. vO(t) is the inverter output AC
voltage instantaneous value after the filter. Vm is the amplitude
of the output AC voltage of the inverter corresponding to the
angle (ωt −υ) = π/2.

 Refer to the single-phase PWM DC/AC inverter with an

R–L load shown in Figure 5, the corresponding output
voltage and current waveforms are shown in Figure 6 with
the carrier frequency fc =400 Hz for indication although
particular fc may be very higher. If the input reference AC
voltage signal with the frequency f =50 Hz (T =1/f =20 ms)
and the triangle waveform with carrier frequency fc = 400
Hz (TC =2.5 ms), each device is conducted in a conduction
period=mTC, where m is the modulation ratio. The
conduction period is less than a cycle Tc (sampling
interval) since the modulation ratio m is usually smaller
than unity (1) for linear modulation.

 The conduction angle is smaller than 360◦ (or 2π rad), or

<2.5 ms. The current commutation happens once a
chopping cycle.

 The output voltage is out of control in a half-chopping

cycle once the PWM pulse is applied. Therefore, it is the
element to keep the output voltage in a sampling period of
TC =1/fc. By per-unit system, the voltage transfer gain is a
linear element in a sampling interval Tc =2.5 ms. That is,
 G(t) = VO/ Vin|per-unit = 1 − e−t/Tc ….(25)

 Analogously, different chopping frequency only changes

the sampling interval. The mathematical modeling is not
changed. No matter how high or low the carrier frequency
is, the output voltage is expressed by Equation (24), its
voltage transfer gain per-unit system is a linear element as
described in equation (25).

Figure 14 The single-phase PWM DC/AC inverter with an L–R

circuit.

Figure 15 The output (a) voltage and (b) current waveforms.

3.1 Traditional Modeling for DC/AC PWM Inverters

 Traditional modeling for DC/AC PWM inverters is a first-

order element in the s-domain. The output voltage of a
DC/AC PWM inverter is naturally a periodic pulse train
with the repeating frequency fO as requested. This output
periodic pulse train has plenty of harmonics corresponding
to both the requested frequency fO and the carrier
frequency fc. Usually, the carrier frequency fc must be
much higher than the requested output frequency fo to
avoid yielding high total harmonic distortion (THD).

 Actually, no matter how higher the applied carrier

frequency fc is, THD cannot be zero.

 A low-pass filter must be set to filter the higher-order

harmonics in order to obtain pure output sinusoidal
waveform. The low-pass filter is usually an R–C circuit
with the time constant τ =RC or an R–L circuit with the
time constant τ =L/R.

 Normally, the time constant τ is much larger than the pulse

width (τ >> T =1/fc) to avoid the parasitic power losses
and additional distortion. The relations between them
should be:

(1 / fO) >> τ >> (1/fc) …… (26)

 For example, if the requested output frequency fO =50 Hz,

a modulation (chopping) frequency fC =10 kHz, usually the
filter time constant τ will set in the range of 500μs to 4 ms
(typically 2 ms).

 The purpose to set the filter is to retain the fundamental

harmonics in 50 Hz and filter the higher-order harmonics
out, and then obtain very low THD. Sometimes, the time
 constant τ of the low-pass filter is not small to cause the
fundamental harmonic in 50 Hz attenuated. The
corresponding transfer function (per unit) of the system is:

G(s) = 1/ (1 + τs ) ……………. (27)

 The transfer function of the DC/AC PWM inverters is

ignored. It is treated as a proportional element. Therefore,
the traditional mathematical modeling is not really used to
describe the DC/AC PWM inverters although it is
generally agreed by most experts.

3.2 A First-Order Hold for DC/AC PWM Inverters in

Digital Control.

 The mathematical modeling seen in Equation (26) was

very popular since 1960s to 1990s. Many large machinery
systems have slow time responses and use analog PI
controllers. This modeling is good enough to describe the
system characteristics.

 On the other hand, if the inverters used to drive an

induction motor, the stator circuit is a natural first-order
filter with time constant τ =L/R, which is usually lower
than the power supply cycle T =1/fO. Carefully selecting
the inverter carrier frequency fC, the relation (26) is easily
satisfied.

 Since 1980s the digital processors have been applied in

research and industrial application, and the systems
perform faster. The DC/AC PWM inverters cannot be
considered as only a proportional element. Since its output
voltage is out of control once the pulse width is applied, it
should be looked as a sample and linear-varying element.
Therefore, its mathematical modeling in per-unit digital
control system should be a first-order hold (FOH) in both
s- and z-domains:
where T is the sampling interval T =1/fC.
 It means that the DC/AC PWM inverter performs a first-
order inertial element with time constant T in the s-domain
and a linear element in one step (T) in a digital control
system in the z-domain. From Equation (29) the transfer
function has one zero and one pole z =1/e inside the unit-
cycle. Therefore, a DC/AC inverter is always a stable
element.

4. A SECOND-ORDER TRANSFER FUNCTION FOR

DC/DC CONVERTERS

 DC/DC converters have many forms which are listed

below:

1. fundamental converters such as buck, boost and buck–

boost converters;
2. voltage-lift converters;
3. super-lift converters;
4. other converters.

 The devices of all types of the DC/DC converters can be

BJT, MOSFET (all types), IGBT, IGCT, …. etc. They are
controlled by the PWM pulse with certain conduction duty
cycle k.

 In a sampling interval/period T, the conduction duty cycle

k can be changed only once, i.e. the output voltage of a
DC/DC converter is changed period by period. Therefore,
 all types of the DC/DC converters are working in a
discrete state.

 In general situation, the load of the power DC/DC

converters is a resistive load R. If the output current is
continuous, the output DC voltage average value in a
steady state is:

Vo = M VI ………………….. (30)
or
M = Vo/VI

where VI and Vo are the input and output DC voltages, and

M is the voltage transfer gain.

 If the switching frequency is f (the switching period T

=1/f) and the conduction duty cycle is k, the switching-on
period is kT and switching-off period is (1−k)T. The
voltage transfer gain M is usually dependent to the
conduction duty cycle k, and independent from the
switching frequency f . For example, buck converter has
the voltage transfer gain M =k, boost converter has the
voltage transfer gain M =1/(1−k) and buck–boost converter
has the voltage transfer gain M =k/(1−k).

 The output voltage is out of control in a period T once the

duty cycle k is applied. Therefore, it is the element to keep
the output voltage in a period T =1/f .

 By per-unit system, the voltage transfer gain is unity (1) in

a sampling interval. Then a power DC/DC converter is a
second-order element, and its transfer function is:
where τ is the time constant and τd is the damping time constant.

4.1 Traditional Modeling for DC/DC Converters

 Traditional modeling for DC/DC converters is a complex

element in the s-domain. The main method is the voltage
division formula. Therefore, the transfer function with the
orders is equal to the number of the passive energy-storage
parts: inductors and capacitors. The simplest fundamental
converters, such as buck, boost and buck–boost converters
which have one inductor and one capacitor, possess a
second-order transfer function. Other converters with
multiple inductors and capacitors must have high-order
transfer function.

 For example, a buck converter shown in figure (16) has

the transfer function as:

where M is the voltage transfer gain, τ is the time constant L/R

and τd is the damping time constant RC.
 Figure 16 Buck converter.

 For the DC/DC converters with two inductors plus two

capacitors, their transfer function is in the fourth order. For
example, the negative output Luo-converter elementary
circuit shown in Figure (8) has the transfer function:

This mathematical modeling is very complex for industrial

applications.

Figure 17 Elementary circuit of N/O Luo-converter.

4.2 A Second-Order Hold for DC/DC Converters
in Digital Control

 The traditional mathematical modeling was popular since

1940s till now. It is very difficult for the higher-order
converters such as Luo-converters, super-lift converters,
Cúk converter, and so on. In the period 2001–2004 Dr F.
L. Luo and Dr H.Ye proposed a new method to model the
power DC/DC converters.

 This new methodology was described before. A second-

order transfer function simulated all power DC/DC
converters, the mathematical modeling in per-unit digital
control system should be a second-order transfer function
in the s-domain:

where τ is the time constant and τd is the damping time constant.

 In general situation if τd is greater than the critical value

τ/4, then there is a couple of conjugated complex poles s1
and s2. Equation (20 ) is rewritten as:
  Correspondingly, a power DC/DC converter is a second-
order hold (SOH) in the z-domain:

 It means that the DC/DC converter performs a second-

order response with oscillation in the s-domain and one-
step delay (T) in a digital control system.

5 A First-Order Transfer Function for AC/AC (AC/DC/AC)

Converters.

 AC/AC (AC/DC/AC) converters have many forms as

studied before.

 The devices of all types of the AC/DC converters can be

thyristor (SRC), transistor, BJT, GTO and Triac.

 They are controlled by the corresponding firing pulse with

certain firing angle α or the PWM scheme with the certain
carrier (chopping) frequency fc. In a sampling interval T
=1/fc, the pulse-width angle can be changed only once, i.e.
the output voltage of a DC/AC PWM inverter is changed
period by period.

 Therefore, all types of the AC/AC PWM inverters are

working in discrete state. In general situation, the input
reference signal vin(t) is a sinusoidal waveform:

vin(t) = Vm sin ωt ………………. (38)

  The load of the converters is an L–R circuit with time
constant τ =L/R, or the delayed angle is υ = tan−1 (ωL/R)

 If the output current is continuous, the outputAC voltage

instantaneous value after the filter should be:

vO(t) = Vm sin(ωt − υ) …………….. (39)

where υ is the delay angle of the output voltage with reference

to the input signal. vO(t) is the converter output AC voltage
instantaneous value after the filter. Vm is the amplitude of the
output AC voltage of the inverter corresponding to the angle (ωt
−υ)=π/2.

 Refer to AC/AC cycloconverter shown in Figure 18. The

source AC voltage with frequency fS is a sinusoidal
waveform, and the output voltage should follow the input
reference voltage with frequency f . The corresponding
input and output voltage and current waveforms are shown
in Figure 19.

 If the input reference AC voltage signal with the frequency

f =10 Hz (T =1/f =100 ms), each bridge is conducted in a
conduction period=T/2=50 ms. The output voltage is out
of control in a half-chopping cycle once the PWM pulse is
applied. Therefore, it is the element to keep the output
voltage in a period of TS/2=1/2fS. By per-unit system, the
voltage transfer gain is a linear element in a sampling
interval TS/2=10 ms. That is,

G(t) = VO/Vin |per-unit = 1 − e−t/τ ……. (40)

 Analogously, different chopping frequency only changes

the sampling interval. The mathematical modeling is not
changed. No matter how high or low the carrier frequency
is, the output voltage is expressed by Equation (39), its
 voltage transfer gain in per-unit system is a linear element
as described in Equation (40).

Figure 18 An SISO AC/AC cycloconverter.

Figure 19 Waveforms of an SISO AC/AC cycloconverter (50–

10 Hz) with R load. (a) Load voltage and load current; and (b)
input supply current.

5.1 Traditional Modeling for AC/AC Controlled Rectifiers

 Traditional modeling for AC/AC converters is a first-order

element in the s-domain.
 The output voltage of a AC/AC converter is naturally a
periodic pulse train with the repeating frequency fO as
requested. This output periodic pulse train has plenty of
harmonics corresponding to both the requested frequency
fO and the carrier frequency fC. Usually, the carrier
frequency fC must be much higher than the requested
output frequency fO to avoid yielding high THD. Actually,
no matter how higher the carrier frequency fC applied is,
THD cannot be zero.

 A low-pass filter must be set to filter the higher-order

harmonics in order to obtain pure output sinusoidal
waveform. The low-pass filter is usually an R–C circuit
with the time constant τ =RC or an R–L circuit with the
time constant τ =L/R. Normally, the time constant τ is
much larger than the pulse width (τ >>T =1/fC) to avoid the
parasitic power losses and additional distortion. The
relations between them should be:

(1/fO ) >> τ >> ( 1/fC ) ……….. (41)

 For example, if the output requested frequency fO =50 Hz,

a modulation (chopping) frequency fC =10 kHz, usually
then the filter time constant τ sets in the range of 500μs to
4 ms (typically 2 ms). The purpose to set the filter is to
retain the fundamental harmonic in 50 Hz and filter the
higher-order harmonics out, and then obtain very low
THD. Sometimes, the time constant τ of the low-pass filter
is not small to cause the fundamental harmonic in 50 Hz
attenuated. The corresponding transfer function of this
rectifier in per-unit system is:

G(s) = 1/ ( 1 + τs ) …………. (42)

 From this point of view, the first-order transfer function is

the feature of the first-order filter, but the characteristics
of the AC/AC PWM inverters. The transfer function of the
 AC/AC PWM inverters is ignored. It is treated as a
proportional element. Therefore, the traditional
mathematical modeling is not really used to describe the
AC/AC PWM inverters, although it is generally agreed by
most experts and it is widely used in industrial
applications.

5.2 A FOH for AC/AC Converters in Digital Control

 The mathematical modeling given in Equation (21) was

very popular since 1960s to 1980s. Many large machinery
systems have slow time responses and use analog PI
controllers. This modeling is good enough to describe the
system characteristics. On the other hands, if the
converters used to drive an induction motor, the stator
circuit is a natural first-order filter with time constant τ =
L/R, which is usually lower than the power supply cycle T
=1/fO. Carefully selecting the inverter carrier frequency fC,
the relation (41) is easily satisfied.

 Since 1980s the digital processors were applied in research

and industrial application, where the systems perform
faster. The AC/AC converters cannot be considered only
as a proportional element. Since its output voltage is out of
control once the pulse width is applied, it should be looked
as a sample and linear-varying element. Therefore, its
mathematical modeling in per-unit digital control system
should be a FOH in both the s- and z-domains:

G(s) = 1/ (1 + Ts ) …………….. (43)

G(z) = Z[G(s)] = Z [ 1/ (1 + Ts ) = (z/T)/(z − 1/e) .. (44)

where T is the sampling interval T =1/fC. It means that the

AC/AC PWM inverter performs a first-order inertial element
with time constant T in the s-domain and a linear rising/falling
element in one step (T) in a digital control system.