IET Power Electronics

Research Article

State-space modelling of LLC resonant half- ISSN 1755-4535

Received on 3rd December 2019
Revised 12th January 2020
bridge DC–DC converter Accepted on 3rd February 2020
E-First on 11th March 2020
doi: 10.1049/iet-pel.2019.1503
www.ietdl.org

Sen Li1 , Babak Fahimi1

1Department of Electrical & Computer Engineering, University of Texas at Dallas, Richardson, TX, USA
E-mail: [email protected]

Abstract: Recently, LLC resonant converters have attracted significant research from industry and academia for AC–DC and
DC–DC power conversion with high efficiency and remarkable power density. They are appealing candidates for numerous
vehicular and renewable energy applications including battery chargers for electric vehicles and drivers of LED lights. This study
introduces a mathematical model of LLC resonant half-bridge DC–DC converter, which captures its steady-state behaviours for
both continuous conduction mode and discontinuous conduction mode operations. One major advantage of the proposed model
lies in accurate estimation of the switching frequency of power switches under a wide range of parametric variations. This
benefit is, however, not offered by the prevailing method based on the first harmonic approximation (FHA). The analytical
derivations of the system's state-space model, as well as equations for calculating the switching frequency by FHA, are
discussed in details. For illustration, a 340 W digitally controlled LLC resonant converter is targeted in this study. The simulation
analyses of current and voltage waveforms for light and heavy load conditions are presented. Moreover, the experimental
results, along with the comparison of switching frequency estimation for both methods, are demonstrated and discussed, which
confirms the validity and effectiveness of the proposed model.

1 Introduction series and parallel configurations [7–16]. A brief comparison of

diverse topologies of resonant DC–DC converter is given in
Switched-mode DC–DC converters are essential components for Table 1. In particular, for the LLC resonant converter in its half-
diverse types of the energy conversion system. They are playing a bridge implementation, the two inductors can be realised by
crucial role in industrial applications such as laptop computers, utilising the leakage and magnetising inductance of a single
electric/hybrid vehicles, wireless power transfer devices, wind transformer. This contributes to the conciseness of converter
generators, and photovoltaic power systems. The main design structure and improvement of the overall power density. Some
challenges for power supply circuits include high power density, additional benefits can be found in the LLC resonant converter
high conversion efficiency, and low EMI emission. For the including highly efficient load and line regulation under small
converters implemented with PWM control, the switching devices variation of switching frequency, zero-voltage switching (ZVS)
are usually hard-switched with non-zero voltage and current during during turn-on period, small turn-off current for the power
the switching transient, which leads to high switching loss and switches, low switching loss, and electrical isolation of input and
causes potential thermal issue [1, 2]. On the other hand, the output stages [17–19].
resonant converters are competent candidates for addressing these In most practical applications, the LLC resonant converter
drawbacks since soft-switching can be realised by proper control of operates with a frequency-modulation controller; namely, the
the phase angle difference between the input voltage and current switching frequency serves as an important variable in the closed-
flowing in the resonant tank. With this advantage, the power loop feedback system, and the duty cycle of the power switches is
converter can operate at high switching frequency which, in turn, equally set to 0.5. Due to this unique trait, unlike the case of PWM-
leads to significant size reduction of the passive devices such as controlled power converters, conventional methods such as state-
tank inductor and output filtering capacitor. Consequently, higher space averaging cannot be applied for modelling of LLC resonant
conversion efficiency and better power density of the system can converter since the voltage transfer ratio is directly dependent on
be achieved [3–6]. the switching frequency instead of duty cycle [20–25]. Therefore,
Among numerous available topologies, the series–parallel the switching frequency is closely related to the input–output
resonant converters (SPRCs) including LCC and LLC types are characteristics of the converter and determines the system's
gaining more popularity due to the combined benefits from both dynamical behaviours. Considering the fact that the normal
operating condition is usually in the vicinity of the series
Table 1 Comparison of diverse topologies of the resonant resonance, it can be assumed that the circulating current in the
DC–DC converters resonant tank is purely sinusoidal. The modelling approach based
Parameter Series type Parallel type Series–parallel on such assumption is the so-called first harmonic approximation
type (FHA), which is now widely adopted in the design of resonant
conversion mode step-down step-up/down step-up/down converters [26–28]. Although this method provides a simple way to
power density moderate moderate high develop the input-to-output transfer function, the accuracy of this
approach becomes a major issue as the operating frequency
switching loss low low low
deviates from the series resonance, due to the inclusion of more
conduction loss moderate high moderate frequency components within the current flowing in the resonant
light load efficiency moderate low moderate tank.
heavy load efficiency moderate moderate high In this paper, the general state-space model for the closed-loop
operating frequency high high high LLC resonant half-bridge DC–DC converter, which is controlled
voltage transfer ratio moderate moderate high by the voltage-controlled oscillator (VCO), is presented in details.
The main contribution of the work lies in the introduction of an

IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592 1583

© The Institution of Engineering and Technology 2020
Fig. 1  Diagram of LLC resonant half-bridge DC–DC converter

analytical model which is capable of predicting the steady-state ẋ = An x + Bnu, n = 1, 2, 3, 4 (2)

dynamic behaviours of the converter system under normal stable
operation for both of continuous and discontinuous conduction where An and Bn denote the state and input matrices. Note that
modes with satisfactory precision. Furthermore, detailed depending on the switching frequency of power switches, the LLC
mathematical derivations of the converter's system model are resonant DC–DC converter could operate in either discontinuous
demonstrated for both operating modes, along with the simulation conduction mode (DCM) or continuous conduction mode (CCM).
and experimental analyses of the current and voltage signals. In Considering the completeness of the analysis, the following
addition, the comparisons between the proposed model and the discussion will cover the analytical modelling for both modes of
prevailing FHA method in estimation of the switching frequency of operation.
the power switches are also covered. The main purpose of
conducting such comparisons is to illustrate the accuracy and
2.1 Discontinuous conduction mode
superiority of the proposed model to the FHA method in capturing
the dynamic behaviours of the converter system. Besides, the Fig. 2 demonstrates the ideal time-domain waveforms of ir, im,
precise estimation of the switching frequency can also help output rectifying current id, gate pulses vGS1 and vGS2, and the pole
engineers choose proper passive and active devices for the actual voltage vDS2 for DCM operation. The corresponding equivalent
design of the converter system. circuits for the periods t0 ∼ t1, t1 ∼ t2, t2 ∼ t3, and t3 ∼ t4 are shown in
This paper is organised as follows: the derivation of the state- Fig. 3. During the period t0 ∼ t1, the system model is given by the
space model is demonstrated in Section 2; the analytical equations
following equation:
for calculating the switching frequency by the FHA-based method
are developed in Section 3; Sections 4 and 5 present the results and
ẋ = A1 x + B1u (3)
discussions of the simulation analysis and experimental test of a
mid-power LLC resonant DC–DC converter, along with a
comparison of switching frequency estimation using the FHA where
method and the proposed model; a brief summary and additional
0 e11 0 0 0 0 0
comments of the work are given in Section 6.
e12 0 0 0 0 b11 b12
2 Derivation of the state-space model A1 = e13 0 0 0 0 , B1 = b13 b14 (4)
Fig. 1 demonstrates the system diagram of LLC resonant half- 0 e14 e15 0 0 0 b15
bridge DC–DC converter. Lr and Lm are the leakage and 0 0 0 e16 0 0 b16
magnetising inductances of the primary side of the transformer,
respectively. Ls′ is the referred leakage inductance Ls of the In (4), the expressions for e11 ∼ e16 and b11 ∼ b16 are given in the
secondary winding which is given by Ls′ = n2Ls. Capacitor Cr and Appendix. Considering the fact that the ripple component in output
inductor Lr form the series resonant tank. The half-bridge consists voltage is usually very small, the term vc in the equations for
of two MOSFET switches Q1 and Q2. D1 and D2 are the secondary (dir /dt), (dim /dt), and dvc /dt has been replaced by V ref /k. By
side rectifying diodes. The error signal of the scaled output voltage adopting such modification, the complexity of the system model
serves as the input for the VCO, which outputs the switching can be significantly reduced without compromising the model's
frequency fs as expressed by (1). The gate signals for Q1 and Q2 are accuracy. During the period t1 ∼ t2, the system equation is given by
produced from a dual-channel gate driver. the following equation:

ẋ = A2 x + B2u (5)
f s = kp k(vc t − V ref + ki∫ (kv c t − V ref )dt + f o (1)
where
In (1), k is the voltage scaler (k = Ro1 /(Ro1 + Ro2)), kp and ki are the
0 e21 0 0 0 0 0
proportional and integral gains, respectively, vc is the voltage
e22 0 0 0 0 b21 0
across the output capacitor CL, V ref is the reference voltage, f o is
the centre frequency of VCO. Let vcr (voltage across the capacitor A2 = e23 0 0 0 0 , B2 = b22 0 (6)
Cr), ir (current of the resonant tank), im (magnetising current), vc, 0 0 0 0 0 0 b23
and the feedback variable α (α = ki∫ kvc(t) − V ref dt) be the state 0 0 0 e24 0 0 b24
variables, x be the state vector, i.e. x = vcr ir im vc α T, and u be
The expressions for e21 ∼ e24 and b21 ∼ b24 are provided in the
the input vector, i.e. u = V in V ref T, then the system equation of the
Appendix. Further examination of the time-domain waveforms
converter is given by the following equation: reveals that there exists an underlying symmetric relationship

1584 IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592

© The Institution of Engineering and Technology 2020
Fig. 2  Time-domain waveforms for DCM operation

Fig. 3  Equivalent circuits of the converter for DCM mode of operation

(a) t0 ∼ t1, (b) t1 ∼ t2, (c) t2 ∼ t3, (d) t3 ∼ t4

between the period t0 ∼ t2 and t2 ∼ t4. Therefore, the dynamic −1 0 0 0 0

behaviour of the converter can be fully characterised using the 0 −1 0 0 0
model for the period t0 ∼ t2 only. Let t0 = 0 and xt0, xt1, and xt2 be T
W1 = 0 0 −1 0 0 , W2 = V in 0 0 0 0 (10)
the state vectors at t0, t1, and t2, then the mapping relations for xt0 to
0 0 0 1 0
xt1 and xt1 to xt2 are given by the following equation:
0 0 0 0 1
t1
xt1 = e A1t1 xt0 + ∫ 0
e A1(t1 − σ)dσB1u (7) The conditions for the state switching at t1 and t2 are given by the
following equations:
t2 W3 xt1 = W4 xt1
xt2 = e A2(t2 − t1) xt1 + ∫ t1
e A2(t2 − σ)dσB2u (8)
(11)

2t2 ⋅ f o + kp W5 xt2 − V ref + W6 xt2 = 1 (12)

Due to the existing symmetry, the linear relation between xt0 and xt2
can be expressed by the following equation: where

xt2 = W1 xt0 + W2 (9) W3 = 0 1 0 0 0 , W4 = 0 0 1 0 0 (13)

where W5 = 0 0 0 k 0 , W6 = 0 0 0 0 1 (14)

IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592 1585

© The Institution of Engineering and Technology 2020
Fig. 4  Time-domain waveforms for CCM operation

Fig. 5  Equivalent circuits of the converter for CCM mode of operation

(a) t0 ∼ t1, (b) t1 ∼ t2, (c) t2 ∼ t3, (d) t3 ∼ t4

By solving (7)–(9), (11), and (12), one can find the steady-state where
solutions of t1, t2, and xt0. Therefore, the switching frequency f s of
the power converter can then be expressed by the following 0 e11 0 0 0 0 0
equation: e12 0 0 0 0 b11 −b12
A1′ = e13 0 0 0 0 , B1′ = b13 −b14 (18)
1
fs = (15) 0 −e14 −e15 0 0 0 b15
2t2
0 0 0 e16 0 0 b16
2.2 Continuous conduction mode
0 e11 0 0 0 0 0
The ideal time-domain waveforms of ir, im, id, vGS1, vGS2, and vDS2 for
e12 0 0 0 0 b11 b12
CCM operation are demonstrated in Fig. 4. Fig. 5 shows the
corresponding equivalent circuits for the periods A2′ = e13 0 0 0 0 , B2′ = b13 b14 (19)
t0 ∼ t1, t1 ∼ t2, t2 ∼ t3, and t3 ∼ t4. Similar to the case of DCM 0 e14 e15 0 0 0 b15
operation, symmetry can still be found between period t0 ∼ t2 and 0 0 0 e16 0 0 b16
t2 ∼ t4. The system equations for periods t0 ∼ t1 and t1 ∼ t2 are given
by the following equation: Then the mapping equations for xt0 to xt1 and xt1 to xt2 are given by
′ the following equation:
Period t0 ∼ t1 : ẋ = A1 x + B1′ u (16)
t1
Period t1 ∼ t2 : ẋ = A2′ x + B2′ u (17)
′
xt1 = e A1t1 xt0 + ∫ 0
′
e A1(t1 − σ)dσB1′ u (20)

1586 IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592

© The Institution of Engineering and Technology 2020
 t2
∫
′ ′ To transform (29) into normalised form, the normalised frequency
xt2 = e A2(t2 − t1) xt1 + e A2(t2 − σ)dσB2′ u (21) f n and inductance Ln are given by the following equation:
t1

The condition equations for state switching at t1 and t2 are identical fs Lm

fn = , Ln = (30)
to those of DCM operation, which are given by (11) and (12). fr Lr
Solving (9), (11), (12), (20), and (21), the steady-state solutions of
t1, t2, and xt0 can be found. The Q factor of the series resonant circuit Qe is

Lr /Cr
3 Switching frequency estimation from FHA- Qe = (31)
Re
based method
The FHA-based method is being widely used in modelling and Consequently, (29) can be rewritten as
design process of LLC resonant converter. To compare with the
proposed model, the analytical equations for calculating the Ln f n2
switching frequency using FHA are derived here. Fig. 6 shows the Ge = (32)
Ln + 1 f n − 1 + j ( f n2 − 1) f nQeLn
2
linear equivalent circuit of LLC resonant converter. The pole
frequency f p and series resonant frequency f r are given by the
From (32), the function of f n can found as
following equations:

1 A1 f n6 + A2 f n4 + A3 f n2 + A4 = 0 (33)
fp = (22)
2π (Lr + Lm)Cr
where
1
fr = (23) A1 = Ge2Ln2 Qe2 (34)
2π LrCr

The RMS value of the input voltage vie is A2 = Ge2 −2Ln2 Qe2 + Ln2 + 2Ln + 1 − Ln2 (35)

2 A3 = Ge2 Ln2 Qe2 − 2Ln − 2 , A4 = Ge2 (36)

V ie, rms = V (24)
π in
Note that Ge can also be approximated as
The RMS values of the output voltage voe and current ioe are given
by the following equations: 2nV out
Ge = (37)
V in
2 2
V oe, rms = nV out (25) Solving (33) by using (34)–(37), one can obtain six roots of f n.
π
Ignoring the negative and complex roots and selecting f n which
π 1 satisfies f n > f p / f r (for ZVS operation), then the switching
Ioe, rms = Iout (26)
2 2n frequency f s can be found using (23) and (30).

where V out and Iout are the output voltage and current, respectively.
4 Simulation results
Then the AC equivalent load resistance Re is
To proceed with the simulation analysis, a standalone LLC
V oe, rms 8n2 resonant half-bridge DC–DC converter is utilised as the system
Re = = 2 RL (27) model in this study. The model of this converter has been
Ioe, rms π
constructed in MATLAB–Simulink environment. The design
parameters of the converter are summarised in Table 2. The
The impedances of Cr, Lr, and Lm are given by the following
Simulink model of the converter system is shown in Fig. 7. Both
equation: the heavy load and light load conditions are considered in the
analysis. For the case of heavy load operation (iL = 23 A), Figs. 8
1
XCr = , XLr = ωLr, XLm = ωLm (28) and 9 display the time-domain waveforms of the current flowing in
ωCr
the resonant tank ir, magnetising current im, overall diode current id,
output voltage vout, pole voltage vDS2, and voltage vcr across Cr
Therefore, the input-to-output voltage gain Ge is
when the input voltage V in is set to 350 and 400 V.
XLmRe The same set of figures for light load operation (iL = 6.3A) is
Ge = (29) shown in Figs. 10 and 11, respectively. It is found that both CCM
XLm + XLr − XCr Re + jXLm(XLr − XCr)
and DCM operations of the converter can be expected for different
values of V in. Specifically, when V in is varied close to the lower
limit, the switching frequency of the power switches is regulated
and stays at a low level such that the voltage drop across the
resonant tank reduces accordingly. Therefore, the nominal value of
the output voltage vout can be maintained. Since the current flowing
in the secondary side of the transformer is discontinuous, the
converter operates in DCM. On the other hand, when V in is
changed to the upper limit, the switching frequency is increased by
the feedback controller and, due to this fact, im never reaches the
same level of ir. As a result, the overall diode current at the
secondary side of the transformer remains continuous; hence the
converter operates in CCM.
Fig. 6  Linear equivalent circuit of LLC resonant DC–DC converter

IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592 1587

© The Institution of Engineering and Technology 2020
5 Experimental results the probe amplifier TCPA300 from Tektronix. A 4-channel digital
oscilloscope DPO3014 is used for displaying the signal waveforms.
For verification of the theoretical modelling and simulation Moreover, to eliminate the interference of common-mode noise,
analysis, a 340 W digitally controlled LLC resonant half-bridge
DC–DC converter is utilised as the test subject for the laboratory
examination. The experimental configurations of the converter Table 2 Parameters of the LLC resonant DC–DC converter
system under measurement are demonstrated in Fig. 12. In the Parameter Value or range
prototype circuit, the N-type high power MOSFET input voltage V in, V 350–400
SPW20N60CFD is selected as the half-bridge switches. The diodes leakage inductance Lr, μH 29.8
at the secondary side of the transformer are replaced by the
MOSFET IRLB3036PBF for synchronous rectification. In leakage inductance Ls, μH 0.01
addition, the digital control of the converter is realised by the magnetising inductance Lm, μH 565
digital power supply controller UCD3138A from Texas resonant capacitor Cr, nF 109
Instruments, which incorporates hardware-accelerated PID load capacitor CL, μF 2200
compensators followed by dual-channel digital PWM modules.
The high-side low-side gate driver UCC27714 is selected to maximum load current iL, A 29
produce the gate signals for the power MOSFETs at the primary nominal output voltage V out, V 12
side. For the power switches at the secondary side, UCD7138 transformer turn ratio 16:1:1
synchronous-rectifier drivers are selected to make a close integral gain kint 50,000
behavioural approximation of ideal diodes for the synchronous
rectifiers. To allow for the wide range of load variation, the chassis proportional gain kp 1000
mount resistor TE750B1R0J is adopted and is connected at the centre frequency f o, kHz 90
output port of the converter. The current of the resonant tank is resonant frequency f r, kHz 88.3
measured by the AC–DC current probe TCP312A combined with

Fig. 7  Simulink model of the converter system

Fig. 8  Simulation results when V in is set to 350 V (iL = 23 A, DCM operation)

(a) Current waveforms, (b) Voltage waveforms

1588 IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592

© The Institution of Engineering and Technology 2020
Fig. 9  Simulation results when V in is set to 400 V (iL = 23 A, CCM operation)
(a) Current waveforms, (b) Voltage waveforms

Fig. 10  Simulation results when V in is set to 350 V (iL = 6.3 A, DCM operation)
(a) Current waveforms, (b) Voltage waveforms

Fig. 11  Simulation results when V in is set to 400 V (iL = 6.3 A, CCM operation)
(a) Current waveforms, (b) Voltage waveforms

the output voltage is measured by the differential probe frequency f s is conducted and the simplified estimation algorithm
THDP0200. is shown in Fig. 15. As a comparison, the evaluations by FHA-
The obtained time-domain waveforms of vout, vDS2, ir, and vcr are based method are also included. Fig. 16 demonstrates the obtained
displayed in Figs. 13 and 14, respectively, for the light load (iL =  results from FHA and the proposed model along with the
6.3 A) and heavy load (iL = 23 A) conditions. Apparently, the experimental data when V in varies from 350 to 400 V. Tables 3 and
resonant converter can exhibit both DCM and CCM operations 4 summarise the selected results of both methods for light load and
under different values of the input voltage, which is consistent with heavy load operations, respectively. It can be observed that when
the results from simulation analysis. Moreover, to illustrate the V in is in the range of 350–380 V, both FHA method and the
utility of the proposed model, the calculation of switching proposed model can produce reasonably accurate values that are

IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592 1589

© The Institution of Engineering and Technology 2020
Fig. 12  Experimental settings for measurement

Fig. 13  Experimental waveforms for light load condition (iL = 6.3 A, CH1: output voltage vout (10 V/div), CH2: pole voltage vDS2 (100 V/div), CH3: current
of resonant tank ir (1 A/div), CH4: voltage across Cr, vcr (100 V/div))
(a) DCM (Vin = 350 V), (b) CCM (Vin = 400 V)

Fig. 14  Experimental waveforms for heavy load condition (iL = 23 A, CH1: output voltage vout (10 V/div), CH2: pole voltage vDS2 (100 V/div), CH3: current
of resonant tank ir (5 A/div), CH4: voltage across Cr, vcr (100 V/div)
(a) DCM (Vin = 350 V), (b) CCM (Vin = 400 V)

Fig. 15  Simplified estimation algorithm for switching frequency

close to those from test measurement. However, as V in increases values from the proposed model remain in close proximity to the
beyond this range, the evaluated results of the FHA method deviate actual results; hence decent precision can still be retained by using
considerably from targeted values. This tendency is true when the this model. Although the estimation accuracy declines slightly as
converter operates under either heavy load or light load condition, the switching frequency increases, the relative errors are still
as can be seen from Fig. 16. On the other hand, the calculated within 5% for both load conditions, as illustrated in Tables 3 and 4.

1590 IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592

© The Institution of Engineering and Technology 2020
Fig. 16  Switching frequency of power MOSFET when V in varies from 350 to 400 V
(a) Heavy load (iL = 23 A), (b) Light load (iL = 6.3 A)

Table 3 Comparison of switching frequency estimations for the light load operation
Input voltage, V Experimental results, kHz FHA method, kHz Proposed model, kHz
350 56.95 53.76 57.04
360 62.08 59.58 62.19
370 69.36 67.81 69.27
380 79.04 80.68 79.72
390 93.37 105.1 96.19
400 125.91 177.75 121.41

Table 4 Comparison of switching frequency estimations for the heavy load operation
Input voltage, V Experimental results, kHz FHA method, kHz Proposed model, kHz
350 51.38 49.68 50.40
360 56.14 55.58 55.62
370 62.18 63.83 62.22
380 69.72 76.25 70.90
390 79.88 96.46 82.99
400 94.67 128.83 96.47

6 Conclusion Bandgap Power Devices and Applications, Blacksburg, USA, November

2015, pp. 257–262
Resonant DC–DC power converters can exhibit wide varieties of [3] Bonache-Samaniego, R., Olalla, C., Martínez-Salamero, L., et al.: ‘Design of
self-oscillating resonant converters based on a variable structure systems
nonlinear dynamic behaviours. This paper has proposed the general approach’, IET Power Electron., 2016, 9, (1), pp. 111–119
state-space model of the LLC resonant half-bridge DC–DC [4] Lee, S.H., Cho, Y.W., Cha, W.J., et al.: ‘High efficient series resonant
converter. The analytical derivations of the system equations for converter using direct power conversion’, IET Power Electron., 2014, 7, (12),
DCM and CCM operations along with the simulation analysis and pp. 3045–3051
[5] Ghahderijani, M.M., Castilla, M., Momeneh, A., et al.: ‘Robust and fast
experimental results have been demonstrated and discussed. sliding-mode control for a DC–DC current-source parallel-resonant
Furthermore, a comparison between the proposed model and the converter’, IET Power Electron., 2017, 11, (2), pp. 262–271
FHA approach has also been included in this study. It has been [6] Wang, Y., Han, F., Yang, L., et al.: ‘Dual-transformer soft-switching DC–DC
discovered that the proposed model is capable of evaluating steady- resonant converter with multiple resonant elements’, IET Power Electron.,
2018, 11, (15), pp. 2538–2544
state behaviours of the converter for both CCM and DCM [7] Yang, B., Lee, F.C., Zhang, A.J., et al.: ‘LLC resonant converter for front end
operations under light load and heavy load conditions. In addition, DC/DC conversion’. Proc. IEEE Applied Power Electronics Conf. &
compared with the FHA-based method, it provides a more accurate Exposition, Dallas, USA, March 2006, pp. 1108–1112
estimation for the switching frequency over a wide variation range [8] Hua, C.C., Fang, Y.H., Lin, C.W.: ‘LLC resonant converter for electric vehicle
battery chargers’, IET Power Electron., 2016, 9, (12), pp. 2369–2376
of the selected system parameter. In view of these advantages, the [9] Ku, B., Cai, W., Fahimi, B.: ‘Low-power LLC resonant AC-DC converter for
proposed model can provide one with better insight into the phone charging applications’. Proc. IEEE Dallas Circuits and Systems Conf.,
dynamical properties of the converter operations. Overall, this Dallas, USA, October 2016, pp. 1–4
model can serve as a useful tool for the researchers and engineers [10] Sun, W., Wu, H., Hu, H., et al.: ‘Modified LLC resonant converter with
secondary paralleled bidirectional switch for applications with hold-up time
who are involved in the design of series–parallel resonant DC–DC requirement’, IET Power Electron., 2017, 10, (3), pp. 398–404
converters. [11] Tan, X., Ruan, X.: ‘Optimal design of DCM LCC resonant converter with
inductive filter based on mode boundary map’, IEEE Trans. Power Electron.,
2015, 30, (8), pp. 4144–4155
7 References [12] Yang, R., Ding, H., Xu, Y., et al.: ‘An analytical steady-state model of LCC
[1] Aydemir, M.T., Bendre, A., Venkataramanan, G.: ‘A critical evaluation of type series–parallel resonant converter with capacitive output filter’, IEEE
high power hard and soft switched isolated DC–DC converters’. Proc. IEEE Trans. Power Electron., 2013, 29, (1), pp. 328–338
Ind. Appl. Conf., Pittsburgh, USA, October 2002, pp. 1338–1345 [13] Huang, D., Ji, S., Lee, F.C.: ‘LLC resonant converter with matrix
[2] Li, X., Zhang, L., Guo, S., et al.: ‘Understanding switching losses in SiC transformer’, IEEE Trans. Power Electron., 2014, 29, (8), pp. 4339–4347
MOSFET: toward lossless switching’. Proc. IEEE Workshop on Wide [14] Lin, L., Xu, J., Chen, Y., et al.: ‘Asymmetrical hybrid-controlled half-bridge
LCC resonant converter with low conduction loss and wide ZVS operation
range’, IET Electron. Lett., 2017, 53, (21), pp. 1422–1424

IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592 1591

© The Institution of Engineering and Technology 2020
[15] Li, Z., Wu, T., Zhang, G., et al.: ‘Hybrid modulation method combining e16 = kik (42)
variable frequency and double phase-shift for a 10 kW LLC resonant
converter’, IET Power Electron., 2018, 11, (13), pp. 2161–2169
[16] Tang, X., Wu, H., Zhao, J., et al.: ‘Family of half-bridge LLC resonant 1 Ls′
b11 = − (43)
converters with auxiliary switches for hold-up operation’, IET Power Lr Lr2k3
Electron., 2019, 12, (6), pp. 1376–1384
[17] Choi, H.: ‘Analysis and design of LLC resonant converter with integrated
transformer’. Proc. IEEE Applied Power Electronics Conf. & Exposition, k1 + k2 Ls′
Anaheim, USA, 2007, pp. 1630–1635 b12 = − k1 + ⋅ (44)
[18] Jang, J., Joung, M., Choi, B., et al.: ‘Dynamic analysis and control design of
k3 Lr
optocoupler-isolated LLC series resonant converters with wide input and load
variations’, IET Power Electron., 2012, 5, (6), pp. 755–764 Ls′
[19] Zhao, Z., Xu, Q., Dai, Y.: ‘Minimum resonant capacitor design of high-power b13 = (45)
LLC resonant converter for comprehensive efficiency improvement in battery
LmLrk3
charging application’, IET Power Electron., 2018, 11, (11), pp. 1866–1874
[20] Kang, S.W., Kim, H.J., Cho, B.H.: ‘Adaptive voltage-controlled oscillator for k1 + k2 Ls′
improved dynamic performance in LLC resonant converter’, IEEE Trans. Ind. b14 = k2 − ⋅ (46)
Appl., 2016, 52, (2), pp. 1652–1659
k3 Lm
[21] Li, S., Fahimi, B.: ‘On the period-doubling bifurcation in PWM controlled
buck converter’. Proc. IEEE Transportation Electrification Conf. & Expo, 1
Long Beach, USA, June 2018, pp. 589–594 b15 = − (47)
[22] Li, S., Fahimi, B., Kiani, M.: ‘Bifurcation analysis and experimentation of
RLCLk
peak current-controlled boost converter’. Proc. IEEE IECON, Lisbon,
Portugal, October 2019, pp. 1633–1638 b16 = − ki (48)
[23] Hsu, W.C., Chen, J.F., Hsieh, Y.P., et al.: ‘Design and steady-state analysis of
parallel resonant DC–DC converter for high-voltage power generator’, IEEE
Trans. Power Electron., 2017, 32, (2), pp. 957–966 where
[24] Fang, X., Hu, H., Shen, Z.J., et al.: ‘Operation mode analysis and peak gain
approximation of the LLC resonant converter’, IEEE Trans. Power Electron., n 1
2012, 27, (4), pp. 1985–1995 k1 = + kdi (49)
[25] Tian, S., Lee, F.C., Li, Q.: ‘A simplified equivalent circuit model of series Lr k
resonant converter’, IEEE Trans. Power Electron., 2016, 31, (5), pp. 3922–
3931 n 1
[26] Huang, H.: ‘Designing an LLC resonant half-bridge power converter’. TI k2 = + kdi (50)
Literature No. SLUP263, 2010, vol. 3, pp. 2010–2011
Lm k
[27] Zeng, H., Peng, F.Z., Karki, U.: ‘First order frequency-domain analytical
model for resonant converters in CCM’. Proc. IEEE Energy Conversion Conf. Ls′ Ls′
& Exposition, Portland, USA, September 2018, pp. 1997–2002 k3 = 1 + + (51)
[28] Ivensky, G., Bronshtein, S., Abramovitz, A.: ‘Approximate analysis of
Lm Lr
resonant LLC DC-DC converter’, IEEE Trans. Power Electron., 2011, 26,
(11), pp. 3274–3284 In (49) and (50), kdi is the scalar of the diode voltage drop to the
output reference voltage (kdi = vD /V ref ).
8 Appendix
1
  e21 = (52)
Cr
The analytical expressions for e11 ∼ e16 and b11 ∼ b16 are given by
the following equation: 1
e22 = e23 = − (53)
Lr + Lm
1
e11 = (38)
Cr e24 = kik (54)

1 Ls′ 1
e12 = − + (39) b21 = b22 = (55)
Lr Lr2k3 Lr + Lm

Ls′ 1
e13 = − (40) b23 = − (56)
LmLrk3 RLCLk

n b24 = − ki (57)
e14 = − e15 = (41)
CL

1592 IET Power Electron., 2020, Vol. 13 Iss. 8, pp. 1583-1592

© The Institution of Engineering and Technology 2020