See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/321148061

Double-loop PI controller design of the DC-DC boost converter with a

proposed approach for calculation of the controller parameters

Article  in  Proceedings of the Institution of Mechanical Engineers Part I Journal of Systems and Control Engineering · November 2017
DOI: 10.1177/0959651817740006

CITATIONS READS

25 11,156

2 authors:

Ayhan Özdemir Zekiye Erdem

Sakarya University Ontario Tech University
10 PUBLICATIONS   60 CITATIONS    13 PUBLICATIONS   104 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

107E165 View project

A New Approach for Calculation of PID Parameters with Model Based Compact Form Formulations View project

All content following this page was uploaded by Zekiye Erdem on 15 January 2018.

The user has requested enhancement of the downloaded file.

Original Article

Proc IMechE Part I:

J Systems and Control Engineering
1–12
Double-loop PI controller design of the Ó IMechE 2017
Reprints and permissions:
DC-DC boost converter with a sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0959651817740006

proposed approach for calculation of journals.sagepub.com/home/pii

the controller parameters

Ayhan Özdemir and Zekiye Erdem

Abstract
Parameters of digital proportional–integral/proportional–integral–derivative controllers are usually calculated using com-
monly known conventional methods or solution of discrete-time equations. In literature, a model-based compact form
formulation for calculation of discrete-time proportional–integral/proportional–integral–derivative controller parameters
has not been come across yet. The proposed model-based compact form formulations are introduced to calculate the
proportional–integral parameters in discrete time as a new approach. Generally, different types of control techniques
are chosen in similar studies for double-loop control for direct current–direct current boost converter control except
proportional–integral/proportional–integral. In this study, double-loop proportional–integral controller is used as a differ-
ent control method from literature. By this way, the most important advantages of the proposed study are to reduce dif-
ferent design methods to a unique proportional–integral design method and shorten all calculations. The accuracy of the
double-loop proportional–integral controller’s parameters calculated using the model-based compact form formulations
is validated both in simulation and experimental studies under various disturbance effects. Satisfactory performance of
the proposed controller under model uncertainty and other cases are comparatively shown with the predefined perfor-
mance criteria.

Keywords
Compact form formulation, direct current–direct current boost converter, double-loop control, electrical parameter
variation, proportional–integral control

Date received: 27 September 2016; accepted: 2 October 2017

Introduction Boost converter has a Right Half Plane Zero

(RHPZ) structure. Dynamic response is very important
Boost-type direct current–direct current (DC-DC) con- while designing the controller. However, heuristic
verters are used mostly in industry such as wind energy, methods do not use prior information on dynamic
solar energy and electrical vehicle systems where output structure of controlled systems thus also may lead to
voltage needs to be higher than the input. Controller stability problems. Besides, these methods may cause
design of DC-DC boost converters is more complex extreme overshoots and sharp rises and falls in the
and difficult than the buck converters because of their transient and steady-state response of the output.20
non-minimum phase behavior.1 Although fuzzy logic–based controllers have a good
proportional–integral/proportional–integral–deriva- adaptation to non-linear time variant systems, an expert
tive (PI/PID) control methods are widely used and pre- understanding is needed in order to design a reliable
ferred in most of control applications. PI/PID
controller parameters are calculated in literature using
heuristic methods such as Ziegler Nichols, particle Department of Electrical and Electronics Engineering, Sakarya University,
swarm optimization (PSO), genetic algorithm (GA) Sakarya, Turkey
and simulated annealing (SA)2–7 or analytic methods
Corresponding author:
such as frequency response, Bode plot and root-locus Zekiye Erdem, Department of Electrical and Electronics Engineering,
technique8–14 or intelligent methods15–19 such as neural Sakarya University, Sakarya 54040, Turkey.
network (NN) and fuzzy logic. Email: [email protected]
2 Proc IMechE Part I: J Systems and Control Engineering 00(0)

fuzzy-based controller. Otherwise, a design error may continuous time. Previous studies28–30 present a tuning
occur. Moreover, a few tools are available for designing formula derived especially for PID parameter calcula-
the fuzzy-based controllers. Similarly, NN needs well- tions using phase and gain margins for only continuous
chosen data and a good network design to get desirable time. These formulations are derived for a specific
performance from the controller. In addition, intelligent plant and are not generalized for all types of systems.
methods are more complex than conventional control MBCF formulations have been derived and imple-
methods in terms of implementation in embedded mented for a double-loop PI DC-DC boost converter
systems. controller design.31 In the following sections, first,
PI/PID controllers are implemented around the com- transfer functions of the inner and outer loops are writ-
plex conjugate dominant poles ensuring the predefined ten from small signal analysis. Second, MBCF formula-
performance of system response. Stability is also guar- tions are obtained. Implementation of the MBCF
anteed with these controllers. formulations for the calculation of the PI-PI para-
In literature, because of its easy implementation, the meters is also explained. Accuracy of proposed MBCF
PID control is highly desirable in industrial applications formulations is validated in three cases with simulation
likewise DC-DC Converters.9,21,22 However, single PID and experimental studies under parameter variations
control is not enough to ensure the dynamic response and disturbance effects. The satisfactory responses of
of the voltage and current outputs simultaneously. This the double-loop PI controllers are comparatively shown
burden can be eliminated using a double-loop control, with the predefined performance criteria.
which provides the control of both voltage and current
outputs.
Generally, different types of cascaded discrete-time Model and control of DC-DC boost
control techniques are given in literature such as PI– converter
sliding mode controller (SMC)1,23 and PI-Fuzzy.24
Closed-loop analysis and cascaded control of a non- Proposed double-loop control diagram given in
minimum-phase boost converter are applied to regulate Figure 1 provides instantaneous output voltage and
the output voltage of a double-loop DC-DC boost con- current control simultaneously. Because of its simple
verter. Among all these techniques, a double-loop dis- and failure tolerant structure and adequate perfor-
crete-time PI-PI controller technique has not been mance in both industrial applications and literature
studied yet. PI-PI-type controller method has been cho- studies,2,32–36 PI-type controllers are decided to be
sen for this study due to its easy implementation and applied for both loops.
design structure. Furthermore, the PI parameter calcu- The simplified block diagram of the experimental
lations of the controllers are getting one step easier setup consisting of digital signal processing (DSP)-
with the proposed model-based compact form (MBCF) based double-loop controller together with DC-DC
formulations. Each cascaded controller could be boost converter whose output are connected to feed-
designed with both proposed compact form formula- resistive loads in parallel is given in Figure 1.
tions and conventional design methods. In PI-SMC or The operating parameters, switching frequency and
PI-Fuzzy–type double-loop controllers, which are used sampling frequency of the DC-DC boost converter, are
in literature have to be different theorems and design adjusted to 100 and 50 kHz, respectively. Insulated
steps for each loop. Therefore, these bring extra design Gate Bipolar Transistor (IGBT) current signal io is the
and calculation burden for application engineers and envelope of the inductor current signal iL .
researchers. Since the proposed MBCF formulations are model
Output voltage control of DC-DC Boost converter based, the model of the system to be controlled must be
has two different topologies in literature, one of this obtained.
topology is entitled by voltage-mode control, and the
other one is current-mode control. Although voltage-
mode control has a single-loop topology, current-mode Open-loop transfer functions of the DC-DC boost
control has the double loop, which is the industry stan- converter
dard method of controlling switching power.25
Considering Figure 1, during the double-loop DC-DC
Therefore, in this study, current-mode control is
boost converter design, the output voltage and inductor
selected, and the advantages of this control topology
current have to be measured. Therefore, the open-loop
could be investigated from industrial applications in
transfer functions ‘‘G1 (s)’’ and ‘‘G2 (s)’’ are derived,
literature.25,26
respectively, in the following equations.
There are only a few studies about formulations for
The transfer function G1 (s) between inductor
PID parameter calculation. Pai et al.27 present a direct ~
current and duty ratio ‘‘~iL (s)=d(s)’’ is obtained as given
synthesis design (DS-d) formulation for the systems
in equation (1)
with dead time and inverse response in continuous
time. However, these formulations are restricted by a ~iL (s) Vo Cs + 2(1  D)IL
specific type of a process with delay, and the DS-d for- G1 (s) = = ð1Þ
~
d(s) LCs2 + RL s + (1  D)2
mulations of PID parameters are only available in
Özdemir and Erdem 3

Figure 3. Control block diagram of PI controller system.

The system response Y(s) fulfills the desired transient

parameters, overshoot and settling time, when the digi-
tal PI controller assigns the two dominant poles of the
closed-loop characteristic equation to the predefined
Figure 1. Double-loop control of DC-DC boost converter. poles given in equation (4).
Characteristic equation of closed-loop control sys-
tem in Figure 3 and redefined poles are given in equa-
tions (3) and (4), respectively
F(z) = 1 + GPI (z)Gp (z) = 0 ð3Þ
pffiffiffiffiffiffiffi2ffi
z1, 2 = es1, 2 T = eT(jwn 6jwn 1j ) = sz1, 2 + jwz1, 2 ð4Þ
Equation (4) is the definition of s domain and z
domain.38 Where j is the damping ratio and wn is the
natural angular frequency.
The control pole ‘‘z1’’ in equation (4) is replaced by
Figure 2. Closed-loop control block diagram of the double- ‘z’ in characteristic equation (3), and the characteristic
loop control. equation (3) is arranged as follows
F(z1 ) = GPI (z1 )Gp (z1 ) + 1 = 0 ð5Þ
Similarly, transfer function between output voltage z1 is written in polar coordinates as given in the
and the inductor current ‘‘~ vo (s)=~iL (s)’’ is derived as following
follows
z1 = sz1 + jwz1 = jz1 je jb ð6Þ
v~o (s) (1  D)Vo  LIL s
G2 (s) = = ð2Þ From equation (6), the magnitude and angle of com-
~
iL (s) Vo Cs + 2(1  D)IL
plex value are written as follows
The derivation of the inner-loop G1 (s) and outer-loop qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G2 (s) transfer functions is given in Appendix B. jz1 j = s2z1 + w2z1 ð7Þ
 
wz1
b = tan1 ð8Þ
s z1
Double-loop control
Double-loop control shown in Figure 2 using proposed Similarly, Gp (z1 ) complex value can be written in
method is explained in two parts; first, proposed com- polar coordinates as follows
pact form formulations are introduced; second,  
Gp (z1 ) = Gp (z1 )e jc ð9Þ
discrete-time PI controllers’ parameters have been
calculated with the proposed compact form c = \Gp (z1 ) ð10Þ
formulations.1,23,24,37 If equations (7)–(10) are substituted in equation (5) and
When inner and outer-loop controller parameters rearranged, following equation is obtained
are calculated, the interaction between two loops must
be taken into account. Therefore, in this part, the cal- jz1 je jb 1
Kp + Ki =    ð11Þ
culation method of the proposed formulations for the jb
jz1 je  1 Gp (z1 )e jc
double-loop control system is also explained.
Proposed MBCF formulations36 for calculation of
the PI parameters are obtained from equation (11) and
Proposed MBCF formulations. Proposed MBCF formula- given in equations (12) and (13), respectively. The deri-
tions using open-loop transfer function are introduced vation of equations (12) and (13) is given in Appendix
to calculate the PI controller parameters in Figure 3. 1 in detail
4 Proc IMechE Part I: J Systems and Control Engineering 00(0)

Figure 4. Closed-loop control block diagram of inner loop

controller. Figure 5. Closed-loop control block diagram of outer-loop
controller.
1
sin c jz1 j  2 cos b + jz1 j
Ki =    ð12Þ
Gp (z1 ) sin b the outer-loop response. Settling time, which is a
performance parameter, should be selected by consider-
cos c sin c jz1 j  cos b
Kp =   +
 
 ð13Þ ing this restriction.
Gp (z1 ) Gp (z1 ) sin b The complex conjugate dominant poles are calcu-
lated from equation (4) using defined j and wn values
Calculation of discrete-time PI controller parameters. This z1, 2 = 0:973160:0262j ð15Þ
section guides the designer about how to calculate the
PI parameters of inner and outer loops using MBCF The values of expressions in equations (7)–(10) are
formulations.36 calculated as follows
The PI controller parameters in the loop design are
calculated in two steps: jz1 j = 0:9735
  b = 0:0269
ð16Þ
1. The expressions jz1 j, b, Gp (z1 ) and c given in jG1h (z1 )j = 15:2049
equations (7)–(10) are calculated using defined c =  2:5065
transient parameter overshoot and settling time.
2. PI parameters are calculated from equations (12) The Kp and Ki parameters of D1 (z) controller in the
and (13) using the expressions in the first step. inner loop are obtained using the MBCF formulations
(12) and (13). The results are as follows
Inner loop controller design. The inner loop controller
‘‘D1 (z)’’ is designed due to calculation method men- Kp1 = 0:0909
ð17Þ
tioned above using G1 (z) and z1 . Ki1 = 0:0021
First, discrete-time transfer function of G1h (z) is
obtained using the DC-DC boost converter parameters
and given in equation (14). These parameters are used Outer-loop controller design. Design of the outer-loop
in both simulation and experimental studies. Inductor controller D2 (z) according to calculation method men-
value L = 620 mF, capacitor value C = 1640 mF, duty tioned above is slightly different than the inner loop
cycle D = 0.33, desired output voltage Vo = 18 V, input design. First, designer has to be calculated Gd (z) trans-
voltage Vi = 12:2 V, steady-state inductor current fer function which is given in Figure 5.
IL = 1:289 A, load at operating point R = 20:6 O and G12 (z) transfer function is calculated as
sampling time T = 20 ms  
1  esT
( ) G12 (z) = Z ðG1 (s)G2 (s)Þ =
1  esT V0 Cs + 2(1  D)IL s
ð18Þ
G1h (z) = Z 0:01286z2 + 0:3142z  0:01855
s LCs2 + RL s + (1  D)2 ð14Þ
z3  2:998z2 + 2:996z  0:9982
0:5808z  0:5801
= It can be seen that the order of the Gd (z) is equal to
z2  1:99z + 0:994
seven when Table 1 and Gd (z) transfer function are
Second, dominant pole z1 is calculated using perfor-
taken into account
mance criteria; settling time ts1 = 4 ms and the percent-
age overshoot OS = 4:33% and sampling time vo (z) D1 (z)G12 (z)
T = 20 ms. Inductor current is controlled in the inner Gd (z) = =
iref (z) 1 + D1 (z)G1h (z)
loop. Since the reference current Iref is produced by the
outer loop, the inner-loop response must be faster than the discrete-time closed-loop transfer function of the
outer loop is given as follows

vo (z) 0:0012z6 + 0:0078z5  0:020z4 + 0:028z3  0:022z2 + 0:009z  0:001

= ð19Þ
iref (z) z7  6:94z6 + 20:6z5  34:1z4 + 33:8z3  20:1z2 + 6:6z  0:94
Özdemir and Erdem 5

Table 1. Order of the transfer functions.

Transfer function Order

G12 (z) 3
G1h (z) 2
D1 (z) 1
Gd (z) 7

The performance criteria of the outer loop are deter-

mined as settling time ts2 = 16:2 ms, percentage over-
shoot OS = 4:33% and sampling time T = 20 ms. In Figure 6. Experimental setup.
this study, ts2 = 4:05ts1 has been chosen.
The values of expressions in equations (7)–(10) are
calculated using the defined performance criteria.
Table 2 shows that the simulation and experimental
The Kp and Ki parameters of the outer-loop D2 (z)
result errors are lower than 1%. This shows that the
controller are obtained using the MBCF formulations
model used for controller design is appropriate.
(12) and (13). The results are given as follows
Moreover, the robustness of the DC-DC boost conver-
Kp2 = 1:2863 ter against electrical parameter variations is given in
ð20Þ case 3.
Ki2 = 0:0134

Experimental and simulation results of the open-

loop transfer function verification and the transient Performance analysis of the proposed method
and steady-state analysis of the output voltage/current DC-DC boost output voltage has been regulated via PI
are given in the following section. In the experimental controllers whose coefficients are calculated using the
studies, since IGBT current signal io is the envelope of proposed method for inner and outer loops. Because of
the inductor current signal iL , instantaneous value io is its analytical calculation methodology, which is based
measured instead of instantaneous value iL . on pole placement, MBCF formulations are also
ensured of the stability of the control system. Further
Simulation and experimental results details could be found in Ozdemir and Erdem.31 The
performance analysis is given in detail with both simu-
To demonstrate the performance of the proposed lation and experimental results for the three cases as
MBCF formulations, switching frequency and sam- given below.
pling frequency of 50kHz and 100kHz, respectively, are
chosen. The DC-DC boost converter is implemented Case 1: dynamic response performance. In this section,
using the nominal values: 18 V, 2 A, L = 624 mH, dynamic responses of the DC-DC boost converter’s
C = 1640 mF, Diode 30CTQ060STRLPBFCT and output voltage are analyzed as follows:
IGBT IRFR3607TRPBFCT DPAK. Measurement
and control algorithms are executed by 80 MHz DSP
1. Step changes in the reference input.
F28M35H52C1–based development kit. Experimental
2. The input voltage sag during step changes in the
setup is given in Figure 6.
reference input and step load changes for both
Experimental and simulation studies have three
cases.
stages:
Step changes in the reference input. While DC-DC
1. Model verification.
boost converter is operated at the 14 V steady-state
2. Performance analysis of the proposed method.
value, the input voltage is step changed to 18 V with
3. Performance in electrical parameter variations.
load change in t = 2 ms. Related waveform for the
responses is depicted in Figure 8.
Model verification The values related to transient and steady-state
Model verification of the open-loop transfer function responses are obtained by considering the simulation and
~ shown in Figure 2 is performed under a
of v~o (s)=d(s) the experimental waveforms shown in Figure 8. These
defined duty ratio operating point using the given elec- results are tabulated and comparatively given in Table 3.
trical parameters. Simulation and experimental results, Two different assessments can be made for the val-
which validate the model accuracy of the DC-DC boost ues given in Table 3:
converter, are given in Figure 7.
The proposed method is model based. Hence, the 1. Simulation results are validated by experimental
error should be lower than the acceptable value. results.
6 Proc IMechE Part I: J Systems and Control Engineering 00(0)

Figure 7. Open-loop transfer function response: (a) simulation and (b) experimental results (12.3 V input voltage, 20.6 O load and
0.2 duty cycle rate).

Figure 8. Step changes in the reference input (22%) and load (25%): (a) simulated output voltage and current responses and
(b) experimental output voltage and current responses (input voltage is stepped up to 14–18 V at t = 2 ms and the load is increased
by 25% at t = 20 ms).

2. Both simulation and experimental results ensured loads are switched on and off between time intervals of
the following predefined performance criteria. 40 and 74 ms. Steady-state values of output voltage are
same (16 V), and percentage overshoots are measured
Settling time tsper . tsreal , tssim ; as 4.3% and 3.75% for simulation and experimental
Percentage overshoot OSper . OSreal , OSsim ; peak study, respectively. It can be clearly seen in Figure 9
voltage value Vopper . Vopreal , Vopsim ; that the disturbance effect of voltage sag is suppressed
Desired output voltage Voper = Vosim = Voreal . by the controller, and thus, the system response does
not deteriorate.
Input voltage sag. Simulation and experimental resul-
tant waveforms for the voltage sag, output voltage and Case 2: soft start response. Soft start provides wide scale
the load current are shown together in Figure 9(a) in the DC-DC boost converter reference input voltage
and (b), respectively. change. In addition, soft start avoids over current and
The input voltage sag in Figure 9(a) is generated by protect power switching component. In order to
signal builder block in MATLAB/Simulink. The step demonstrate the performance of the proposed method,
Özdemir and Erdem 7

Table 2. Transient parameter comparison of the open-loop response.

Parameters Simulation Experimental Error (%)

Peak value Vopsim = 15:8 V Vopreal = 15:9 V 0.62

Settling time (2%) tssim = 7:5 ms tsreal = 7:44 ms 0.81
Percentage overshoot OSsim = 6:04% OSreal = 6:71% 0.67
Steady-state value Vsssim = 14:9 V Vssreal = 14:9 V 0

Table 3. Transient response variables under reference and load variation.

Parameters Simulation Experimental Performance criteria

Peak value Vopsim = 18:5 V Vopreal = 18:7 V Vopper = 18:77 V

Settling time tssim = 9 ms tsreal = 7 ms tsper = 16 ms
Steady-state value Vosim = 18 V Voreal = 18 V Voper = 18 V
Percentage Overshoot OSsim = 2:78% OSreal = 3:89% OSper = 4:33%

Figure 9. Input voltage sag and output voltage and load step change: (a) simulation and (b) experimental results—reference
voltage step changes from 12 to 16 V at 8 ms, load step change increased 25% (20–15 O) and reduced 25%, and voltage sag of 58%
(12–6.96 V).

two different disturbances are sequentially applied Table 4. Soft start results for both simulation and
within soft start period. In experimental application, experimental.
soft start is achieved with DSP-based software, and no
Parameters Simulation Experimental
additional hardwares are used. Transient and steady-
state responses are shown in Figure 10. Settling time tssim = 7:2 ms tsreal = 7 ms
The parameter values given in Table 4 and Figure 10 Steady-state value Vosim = 19 V Voreal = 19 V
shows that the controller performance under distur-
bance effects assures the desired performance in both
simulation and experimental studies. Case 3: Robustness against the electrical parameter
The transient parameters such as settling time, variations. In experimental applications, generally those
steady-state value of output voltage, overshoot and parameters (L and C) cannot be measured accurately,
undershoot illustrate that the system controlled by pro- or their nominal values may be changed by environ-
posed method has a satisfying dynamic response. mental effects. Proposed MBCF formulations are
8 Proc IMechE Part I: J Systems and Control Engineering 00(0)

Figure 10. Soft start response while input voltage is decreased (16%) and load step is changed (25%): (a) simulation results and (b)
experimental results—output voltage reference changes from 14 V to 19 V; soft start duration is 50 ms; source voltage changes from
12.2 to 10.2 V (decreased by 16%).

Figure 11. Experimental transient response of the DC-DC boost converter output voltage and current under 610% electrical
parameter change: (a) voltage output and (b) load current.

model based. In this case, the PI controller coefficients remains within 61% tolerance dependent on the para-
are calculated using MBCF formulations for nominal meter variation. This tolerance is within the 62% band
+ 10% and 210% electrical parameter L and C varia- which is predefined for calculating the settling time
tion combinations. Each calculated PI controller coeffi- value ts . As an example, nominal values of the L
cients are applied to the DC-DC boost converter in (inductor) and C (capacitor) are L = 620 mH and
real time and the results are compared. Comparison C = 1640 mF, respectively. Waveform number 1 shows
results show that the MBCF formulations are robust the DC-DC boost converter output voltage, and cur-
against the variation in the electrical parameter values. rent results with nominal value of the electrical para-
The L (inductor) and C (Capacitor) parameters on the meters and the PI parameters of the double-loop
DC-DC boost converter circuit shown in Figure 1 are controller are Kp1 = 0:0909 and Ki1 = 0:0021 for the
not changed during this case. Output voltage and load inner loop and Kp2 = 1:2868 and Ki2 = 0:0134 for the
current waveforms are obtained as shown in Figure 11 outer loop. However, to demonstrate the environmen-
for the step changes in load and reference input tal effect on the parameter variation, the electrical
voltage. parameters are assumed to be L = 558 mH and
Figure 11(a) waveform shows that the DC-DC boost C = 1476 mF while calculating the PI controller para-
converter’s steady-state value of the output voltage meters using MBCF formulations. The PI coefficients
Özdemir and Erdem 9

S+R

S+R

S+R
S+R
S+R
S2R
are recalculated for the inner loop as Kp1 = 0:1142 and

NA: not available; CCM: continuous conduction mode; DCM: discontinuous conduction mode; VSL: variable structure linear controller; ABC: adaptive back stepping current; S: simulation; R: real time; (o): overshoot;
S
S
Ki1 = 0:0033 and for the outer loop as Kp2 = 1:3075
and Ki2 = 0:0171. DC-DC boost converter output vol-
overshoot or

voltage step
tage and current waveform are shown with number 9
undershoot

6.67%(u)
via input
in Figure 11(a) and (b). Despite the change in the
Output

10%(o)
voltage

17%(u)
15%(u)
change

8%(o)

3%(o)
double-loop controller parameters, Figure 11 shows

NA
that system response is robust against variation in the
electrical parameter values or incorrect measurements
voltage

change

and environmental changes. Case 3 results also indicate

Input

step

20%

30%

10%
50%

16%
NA
NA
that the 610% parameter variation does not affect the
controller performance, nevertheless the MBCF formu-
lations are model based.
Overshoot–

current
output

2.78%
Benchmark with previous studies. A benchmark table is
12%

10%
35%

50%
NA
created with similar previous studies to compare pro-
0

posed study. Six different studies selected from litera-

time–output

ture and results are given in Table 5.

current
Settling

0.0025

0.023

0.018

0.009
0.15
NA

Conclusion
0.8

This study is presented to indicate that the double-loop

Overshoot–

PI controller’s parameters could be accurately calcu-

lated using the proposed MBCF formulations. Case 1–
voltage

15%(u)
output

10.7%

1.33%

2.78%

3 studies validate that the double-loop PI controllers

13%

12%

50%

fulfill the desired performance criteria despite step

changes in reference input and disturbances such as
time–output

voltage sag, input voltage step change, load step change

and soft start.
Settling

voltage

0.025

0.023

0.018

0.009

In this study, the simplest model of the DC-DC

0.15
0.8
0.2

boost converter is obtained with linearization and using

the ideal L and C equivalent circuits. Although the PI
CCM + DCM

controller’s parameters are calculated by proposed

MBCF formulations using the simplest DC-DC boost
converter model, experimental results are approved by
Mode

DCM
CCM

CCM
CCM
CCM

CCM

the simulation results. Furthermore, the robustness of

the MBCF formulations against the variations such as
electrical parameters (L and C), capacitor equivalent
DC/DC Buck-Boost

DC/DC Buck-Boost

Flyback Controller

series resistance (ESR), non-linearity of pulse width

Boost Converter

Boost Converter
Boost Converter
Boost Converter

modulation (PWM) modulator, quantization effects

System type

and so on are demonstrated by regulating DC-DC

Converter

Converter

boost converter’s output voltage within 61% tolerance

as shown in Figure 11.
Simulation results for all cases mentioned above
were experimentally validated with predefined perfor-
PI + Sliding Mode Control

mance criteria. Proposed MBCF formulations provide

(u): undershoot; CLF: common Lyapunov function.
PI + switching base CLF

easy and accurate calculation of PI controller para-

meters for industrial application engineers and
PID-Sliding Mode

researchers.
VSL + PI–ABC

PID + PID

PI + Fuzzy
Method

Declaration of conflicting interests

Table 5. Benchmark table.

PI + PI

The author(s) declared no potential conflicts of interest

with respect to the research, authorship and/or publica-
tion of this article.
Agorreta et al.24
Proposed study
Salimi et al.39

Gundemir40

Ren et al.37

Funding
Salimi41
Chen1
Study

The author(s) received no financial support for the

research, authorship and/or publication of this article.
10 Proc IMechE Part I: J Systems and Control Engineering 00(0)

References conference, Clemson, SC, 8–11 March 2016, pp.1–6.

1. Chen Z. Double loop control of buck-boost converters New York: IEEE.
for wide range of load resistance and reference voltage. 19. Ibrahim IN and Al Akkad MA. Exploiting an intelligent
IET Control Theory A 2012; 6(7): 90. fuzzy-PID system in nonlinear aircraft pitch control. In:
2. Sanjeevikumar P and Rajambal K. Extra high voltage 2016 international Siberian conference on control and com-
DC-DC boost converters topology with simple control munications, Moscow, Russia, 12–14 May 2016, pp.1–7.
New York: IEEE.
strategy. Model Simul Eng 2008; 2008: 593042.
20. Stefanutti W, Mattavelli P, Saggini S, et al. A PID auto-
3. Hägglund T and Åström KJ. Revisiting the Ziegler-
tuning method for digitally controlled DC-DC boost con-
Nichols tuning rules for PI control. Part II: the frequency
verters. In: European conference on power electronics and
response method. Asian J Control 2008; 6(4): 469–482.
applications, Dresden, 11–14 September 2005, pp.10–17.
4. Kalaviani R and Lakshami P. Biogeography-based opti-
New York: IEEE.
mization of PID tuning parameters for the vibration con-
21. Guo L. Implementation of digital PID controllers for
trol of active suspension system. J Control Eng Appl
DC-DC converters using digital signal processors. In:
Inform 2014; 16(1): 31–39.
IEEE international conference on electro/information tech-
5. Nasri M, Nezamabadi-pour H and Maghfoori M. A
nology, Chicago, IL, 17–20 May 2007, pp.306–311. New
PSO-based optimum design of PID controller for a linear
York: IEEE.
brushless DC motor. J World Acad Sci Eng Technol
22. Arulselvi S, Uma G and Chimdambaram M. Design of
2007; 2: 211–215.
PID controller for boost converter with RHS zero. In: 4th
6. De Moura Oliveira PB. Modern heuristics review for
international power electronics and motion control confer-
PID control systems optimization: a teaching experiment.
ence, Xi’an, China, 14–16 August 2004, pp.532–537. New
In: 2005 international conference on control and automa-
York: IEEE.
tion, Budapest, 26–29 June 2005, pp.828–833. New York:
23. Chen Z, Gao W, Hu J, et al. Closed-loop analysis and
IEEE.
cascade control of a nonminimum phase boost converter.
7. Yachen Z and Yueming H. On PID controllers based on
IEEE T Power Electr 2011; 26(4): 1237–1252.
simulated annealing algorithm. In: 2008 27th Chinese
24. Agorreta JL, Reinaldos L, Gonzalez R, et al. Fuzzy
control conference, Kunming, China, 16–18 July 2008,
switching technique applied to PWM boost converter
pp.225–228. New York: IEEE.
operating in mixed conduction mode for PV systems.
8. Bryant B and Kazimierczuk MK. Open-loop power-stage
IEEE T Ind Electron 2009; 56(11): 4363–4373.
transfer functions relevant to current-mode control of
25. Lee SW. Practical feedback loop analysis for current-mode
boost PWM converter operating in CCM. IEEE T Cir-
boost converter. Application Report, March 2014. Dallas,
cuits I 2005; 52(10): 2158–2164.
TX: Texas Instruments.
9. Guo L, Hung JY and Nelms RM. Evaluation of DSP-
26. Hegarty T. Current-mode control stability analysis for
based PID and fuzzy controllers for DC–DC converters.
DC-DC converters (part 1). How2Power Today. June
IEEE T Ind Electron 2009; 56(6): 2237–2248.
2014; 1–7.
10. Shen Z, Yan N and Min H. A multimode digitally con-
27. Moradi MH. New techniques for PID controller design.
trolled boost converter with PID autotuning and constant
In: IEEE conference on control applications, Istanbul, 25–
frequency/constant off-time hybrid PWM control. IEEE
25 June 2003. New York: IEEE.
T Power Electr 2011; 26(9): 2588–2598. 28. Pai N-S, Chang S-C and Huang C-T. Tuning PI/PID
11. Maksimovic D and Zane R. Small-signal discrete-time
controllers for integrating processes with deadtime and
modeling of digitally controlled PWM converters. IEEE inverse response by simple calculations. J Process Contr
T Power Electr 2007; 22: 6: 2552–2556. 2010; 20(6): 726–733.
12. Guo L, Hung JY and Nelms RM. Comparative evalua- 29. Ho WK, Hang CC and Zhou J. Self-tuning PID control
tion of sliding mode fuzzy controller and PID controller of a plant with under-damped response with specifica-
for a boost converter. Electr Pow Syst Res 2011; 81(1): tions on gain and phase margins. IEEE T Contr Syst T
99–106. 1997; 5(4): 446–452.
13. Hua C, Lin J and Shen C. Implementation of a DSP-con- 30. Ho W, Hang C and Cao L. Tuning of PID controllers
trolled photovoltaic system with peak power tracking. based on gain and phase margin specifications. Automa-
IEEE T Ind Electron 1998; 45(1): 99–107. tica 1995; 31(3): 3–8.
14. Trejos A, Gonzalez D and Ramos-Paja CA. Modeling of 31. Ozdemir A and Erdem Z. A new approach for calcula-
step-up grid-connected photovoltaic systems for control tion of PID parameters with model based compact form
purposes. J Energ 2012; 5(12): 1900–1926. formulations. Elektronika Ir Elektrotechnika 2014;
15. Seo KW and Choi H-H. Simple fuzzy PID controllers 20(3): 3–10.
for DC-DC converters. J Electr Eng Technol 2012; 7(5): 32. Ntogramatzidis L and Ferrante A. Exact tuning of PID
724–729. controllers in control feedback design. IET Control The-
16. Volosencu C, Danica R and Pasgianos GD. Introduction ory Appl 2011; 5(4): 565.
to PID controllers: theory, tuning and application to fron- 33. Eltamaly AM, El-Tamaly HH and Enjeti P. An improved
tier areas. (ed. Panda RC). London: InTech, 2012. maximum power point tracker for photovoltaic energy
17. Luoren L and Jinling L. Research of PID control algo- systems. In: The 2nd Minia international conference for
rithm based on neural network. Enrgy Proced 2011; 13: advanced trends in engineering, El-minia, 7–9 April 2002.
6988–6993. Egypt: Elminia University.
18. Agee JT, Davidson IE and Kombani LT. Intelligent 34. Hua C and Chihming S. Study of maximum power track-
proportional integral control of a polar axis solar ing techniques and control of DC/DC converters for
tracker. In: 2016 Clemson University power systems photovoltaic power system. In: 29th annual IEEE power
Özdemir and Erdem 11

electronics specialists conference, Fukuoka, Japan, 22–22 when the Ki parameter is substituted in equation (22),
May 1998. New York: IEEE. Kp is obtained as given in equation (25)
35. Koutroulis E, Kalaitzakis K and Voulgaris NC. Develop-
ment of a microcontroller-based, photovoltaic maximum cos c sin c jz1 j  cos b
Kp =   +  ð25Þ
power point tracking control system. IEEE T Power Gp (z1 ) Gp (z1 ) sin b
Electr 2001; 16(1): 46–54.
36. Mutoh N, Masahiro O and Takayoshi I. A method for Further information about MBCF formulations could
MPPT control while searching for parameters corre- be found in Ozdemir and Erdem.31
sponding to weather conditions for PV generation sys-
tems. IEEE T Ind Electron 2006; 4(53): 1055–1065.
37. Ren H, Xin G and Zi Y. Double loop control of boost Appendix 2
converter based current switching controller and voltage Derivation of inner- and outer-loop transfer functions
compensator. In: International conference on electronics, from state equations42 is given below
computers and artificial intelligence, Bucharest, 25–27
June 2015. New York: IEEE. B1 B2
" #   zfflfflffl}|fflfflffl{
 Vo zffl}|ffl{
1
38. Ogata K. Discrete time control systems. 2nd ed. Upper d~iL (t) (1D) ~iL
0
Saddle River, NJ: Prentice-Hall, 1995, pp.205. dt = L + L d~ + L v~g
39. Salimi M, Soltani J, Zakipour A, et al. Two-loop adaptive
vo (t)
d~ 1D
C  RC 1
v~o  ICL 0
dt |fflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflffl} |fflffl{zfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
and nonlinear control of the DC-DC boost converter in A x B:u
discontinuous conduction mode. In: 4th power electronics, ð26Þ
drive systems and technologies conference, Tehran, 13–14 
~
February 2013, pp.164–169. New York: IEEE. ~iL = ½ 1 0  iL ð27Þ
40. Gundemir H. Sliding mode control of DC-DC boost con- |fflfflffl{zfflfflffl} v~o
C1
verter. J Appl Sci 2005; 5(3): 588–592. 
41. Salimi M. Two-loop control of the DC-DC flyback ~i
v~0 = ½ 0 1  L ð28Þ
converter. In: 3rd international conference on advance- |fflfflffl{zfflfflffl} v~o
ments in electronics and power engineering, Kuala Lum- C2
pur, Malaysia, 8–9 January 2013, pp.230–232. Planetary
Scientific Research Centre (PSRC). From the state equations (26), (27) and (28),
42. Smitha K, Aryar PG, Bijlwan R, et al. Steady state Gn (s) = Ci (sI  A)1 B1 (n = 0, 1, 2 and i = 1, 2) is used
~ open-loop trans-
to obtain G1 (s) and G0 (s) = v~o (s)=d(s)
analysis of PID controlled boost converter using state
space averaging technique. (Special issue published by fer functions
Multidisciplinary Journal of Research in Engineering and
Technology) In: National Conference, 2015, pp.100–110.
~iL (s)
G1 (s) = = C1 (sI  A)1 B1 ð29Þ
Pune, India: Green Publications. ~
d(s)
State equation coefficients C1 , A and B1 are replaced in
Appendix 1 equation (29) and can be rewritten as
Where Euler’s formula expressions are substituted into
 " (1D)
#!1 
Vo

equation (11) and rearranged as follows s 0 0 L L
G1 (s) = ½ 1 0   (1D)
0 s  RC1  ICL
jz1 j(cos b + j sin b) cos c + j sin c C
Kp + Ki =  
jz1 j(cos b + j sin b)  1 Gp (z1 ) ð30Þ
ð21Þ Rewriting equation (30), G1 (s) is obtained as follows
where Kp and Ki parameters are obtained by re- ~iL (s) Vo Cs + 2(1  D)IL
arranging them as the real part and imaginary part as G1 (s) = = G1 (s) =
~
d(s) LCs2 + RL s + (1  D)2
given in equations (22) and (23), respectively
Real part ð31Þ
Where small derivations in v~0 is assumed as zero
jz1 j2  jz1 jcos b cos c
Kp =  Ki    ð22Þ
2
jz1 j  2jz1 jcos b + 1 Gp (z1 ) v~0 = 0 !
V0
= ð1  DÞIL ð32Þ
R
Imaginary part
Outer-loop transfer function G2 (s) = v~o (s)=~
iL (s) is
jz1 jsin b sin c obtained from
Ki 2
=  
 ð23Þ
jz1 j  2jz1 jcos b + 1 Gp (z1 ) v~o (s)
G0 (s) ~
d(s) v~o (s)
where Ki parameter is obtained as follows G2 (s) = = = ð33Þ
G1 (s) ~iL (s) ~iL (s)
~
d(s)
sin c jz1 j2  2jz1 jcos b + 1 v~o (s)
Ki =    ð24Þ G0 (s) = = C2 (sI  A)1 B1 ð34Þ
Gp (z1 ) jz1 jsin b ~
d(s)
12 Proc IMechE Part I: J Systems and Control Engineering 00(0)

For the derivation of G0 (s) state equation, coefficients v~o (s) (1  D)Vo  LIL s
G0 (s) = = ð36Þ
C2 , A and B1 are replaced in equation (34) and rewrite ~
d(s) LCs2 + RL s + (1  D)2
 " (1D)
#!1 
Vo
G (s) outer-loop transfer function is derived from equa-
s 0 0 2
G0 (s) = ½ 0 1   (1D) L L
0 s 1  IL tion (33) and obtained as follows
C  RC C

ð35Þ G0 (s) v~o (s) (1  D)Vo  LIL s

G2 (s) = = = ð37Þ
G1 (s) ~
iL (s) Vo Cs + 2(1  D)IL
G0 (s) is derived from equation (35) and obtained as
follows

View publication stats