Fuzzy Self Tuning PID Control of MC LLC Resonant LED Drivers
Fuzzy Self Tuning PID Control of MC LLC Resonant LED Drivers
https://doi.org/10.1007/s43236-021-00233-9
ORIGINAL ARTICLE
Received: 24 November 2020 / Revised: 22 February 2021 / Accepted: 25 February 2021 / Published online: 10 March 2021
© The Korean Institute of Power Electronics 2021
Abstract
An MC3 LLC resonant light-emitting diode (LED) driver is a multi-channel constant current ( MC3) output solution based
on LLC resonance. Through an effective combination of multiple transformers and a DC block capacitor, this driver allows
constant current outputs in multiple channels when the LED channels are unbalanced. This study considers a fuzzy self-
tuning proportional integral derivative (PID) control strategy that allows for the self-tuning of PID parameters according to
fuzzy inference rules designed using the relationship between the current gain and the switching frequency. Thus, when LED
channels are unbalanced, the system allows a constant current output for each of the channels and controls the input voltage
disturbances well. Simulation and experimental verifications are carried out, and a 20 W 4-channel constant current power
supply design based on LLC resonance is obtained for the LED driver. This realizes soft-switching of the switching device,
which reduces system loss and improves system efficiency. Moreover, the adjustment time of the proposed method is half
that of fixed-parameter control methods, and the overshoot of the proposed method is smaller than that of fixed-parameter
control methods. When compared with PID controllers with fixed parameters, the proposed method is more robust. When a
system is disturbed, it recovers faster and has a smaller range of current fluctuation with the proposed method.
Keywords LLC resonant converter · Fuzzy PID control · Multiple output · LED driver
13
Vol:.(1234567890)
C3 LLC resonant LED drivers
Fuzzy self‑tuning PID control of M 783
of output channels. In [9], the authors proposed a modular An LLC resonant MC3 LED driver is a complex high-order
multi-output LED driver based on LCL-T constant current nonlinear system, and the LED is itself a nonlinear load that is
characteristics. An LCL-T rectifier is connected to the same difficult to model, which leads to difficulties in designing its
voltage bus and works in the same way as a current source. controller [18, 19]. Conventionally, the control parameters are
The authors of [6–9] adopted a complex three-stage solution, only set at an operating point with a specific voltage gain [19].
where each stage requires one or more controllers, especially This resulting in unstable control performance and susceptibil-
the third stage. In the third stage, each of the LED strings ity to changes in parameters, such as the load and switching
needs a separate controller to control the output current. frequency. Since it is difficult to meet the needs of driving
Unlike the three-stage solutions, the authors of [10–14] a constant current with unbalanced multi-channel LEDs, an
adopted an isolated single-stage multi-channel constant optimal control strategy that allows a system to obtain stable
current source to drive multiple LED channels. In [10], a and excellent control performance has become a popular topic
simple rectifying circuit for an isolated two-channel LED of research. However, it remains a difficult problem to solve.
driver without active current-sharing control was proposed. When compared with conventional converters, resonant
Although the circuit has a simple structure, good current- converters contain more switching elements, which makes
sharing performance, and high efficiency, it can only achieve them complex nonlinear systems. The modeling of LLC reso-
current-sharing with two channels. In [11], an MC3 LLC nant converters mainly includes extended describing function
resonant LED driver was proposed based on a half-bridge (EDF) and generalized state-space averaging (GSSA) meth-
LLC resonant converter. The driver achieves constant cur- ods [20–23]. The mathematical models of these methods have
rent outputs in multiple channels through an effective com- a high order and low precision, which makes it difficult for
bination of multiple transformers and a DC block capacitor, small-signal modeling by analytical methods. In addition,
even when the LEDs are unbalanced. The authors of [12] small-signal characteristics are directly affected by factors
introduced a simple and effective M C3 LED drive circuit, such as the switching frequency and load. Conventional con-
and explained its control principle and working principle. trol methods rarely meet the requirements of control precision
A general mathematical model was provided to describe for MC3 LEDs. It is especially difficult to obtain good results
the current cross regulation (CCR) between multi-channel when the multi-channel LED is unbalanced. Since an LLC
outputs. A general model of current cross-regulation was resonant LED driver control system is a time-varying nonlin-
derived, which can easily locate and analyze the factors that ear system, it is difficult for a fixed-parameter proportional-
affect cross-regulation. Based on [11], buck-boost PFCs and integral-derivative (PID) controller to achieve good control
LLC resonant DC-DC converters are integrated to form a performance.
multi-channel LED output circuit in [13]. This structure Fuzzy control is widely used in the design of LLC resonant
reduces the number of switches and controllers and simpli- converter controllers since it does not need to establish a small-
fies the circuit structure. Based on [11] and [12], the authors signal model of the system [24–30]. The fact that the current
of [14] considered the impact of LED dimming on a system gain of an LLC resonant circuit has a clear relationship with
and adopted optimal trajectory control to achieve a multi- the switching frequency makes it suitable for designing fuzzy
channel LED constant current output. control rules. Hence, this study combines PID and fuzzy con-
The authors of [15–17] proposed or improved control trol schemes while taking into consideration their respective
methods of LLC LED drivers. In [15], a time-domain analy- advantages and limitations. A fuzzy algorithm is applied to
sis and a design method were proposed for LLC LED driv- PID parameter tuning, and the PID parameters are optimized
ers based on estimated efficiency. When compared with the in real time according to fuzzy rules, which results in fuzzy
first harmonic approximation (FHA), the control precision self-tuning PID control.
was improved here. It was oriented by the weighted-average- The remainder of this article is organized as follows. Sec-
efficiency concept and constraints that ensure zero voltage tion 2 introduces the working principle of an MC3 LLC res-
switching, enough power gain, and a practical switching onant LED driver. Section 3 provides the design method of
frequency range over a wide operating window. In [16], a fuzzy self-tuning PID control. The fuzzy self-tuning PID con-
primary-side regulation (PSR) control circuit for an LLC trol and fixed-parameter PID control are compared and verified
LED driver based on output current estimation was stud- by simulation and experimental results in Sect. 4. Finally, the
ied. This circuit utilized the “magnetizing current cancella- conclusion is given in Sect. 5.
tion method” to estimate the output current and to achieve
good control over the output. In [17], a single-stage AC/
DC converter based on an interleaved buck-boost circuit and
an LLC resonant converter was proposed. The interleaved
circuit allows the system to achieve soft switching under
high-voltage inputs with a high power factor.
13
784 D. Li et al.
2 Working principle of the MC3 LLC resonant conduction loss, the magnetizing inductance needs to be
LED driver as large as possible. On the other hand, in order to ensure
the minimum switching loss, the magnetizing inductance
The MC3 LLC resonant LED driver adopts multiple trans- needs to be small enough to achieve zero-voltage switch-
formers (T1, T2, …, Tk) and DC block capacitors (Cdc1, Ccd2, ing and large enough to have a small turn-off current [31].
…, Ccdk) to provide multi-channel outputs of the current The DC block capacitor are connected in series with the
sources, which achieves current sharing between the differ- transformer windings on the secondary side to balance the
ent channels of LEDs [11]. A system schematic is shown in forward voltage differences between pairs of LED chan-
Fig. 1. In this figure Cr is the resonant capacitor; Lr is the nels. Taking the first transformer circuit on the secondary
resonant inductance; Lm1, Lm2, …, Lmk are the magnetizing side as an example, based on the charge-balancing prin-
inductances of each of the transformers; isec1, isec2, …, iseck ciple of capacitors, the charge of the DC block capaci-
and vsec1, vsec2, …, vseck are the currents and voltages of each tor is balanced in a switching period Ts, and the charge
of the channels on the secondary side; and Io11, Io12, …, Iok1, quantities Q1 and Q2 of each half period must be equal.
Iok2 and Vo11, Vo12, …, Vok1, Vok2 are the output currents and Therefore, the average DC currents in the 2-channels,
voltages of each of the channels. Io11 = Q1/Ts and Io12 = Q2/Ts, are forced to be identical. The
To provide multiple current source outputs, the M C3 average voltage VCdc1 of the DC block capacitor Cdc1 can
LLC resonant LED driver uses multiple independent trans- be obtained according to the volt-second balance of the
formers with identical designs to achieve good current transformer [10] as follows:
sharing. The primary side of the transformer is connected
Vo11 − Vo12
in series, and the secondary side is connected in parallel. VCdc1 = (1)
2
Since the primary side of the transformer is connected
in series, the currents in all of the transformers on the The LLC resonant converter has two resonant frequen-
primary side are identical. When the turn ratios are also cies, fr and fm:
identical, the currents in all of the transformers on the
1
secondary side are also identical. When the magnetizing fr = √ (2)
inductance of the transformer is sufficiently large, cur- 2𝜋 Lr Cr
rent sharing can be achieved through a multi-transformer
setup even when the number of LEDs is inconsistent [11]. 1
fm =
The choice of a magnetizing inductance needs to consider �
�� �
requirements such as the minimum conduction loss and the
� k
∑ (3)
2𝜋 � Lr + Lmi Cr
minimum switching loss. On the one hand, to minimize the i=1
- D12
LED12
on the primary side achieves ZVS, and the current on the
S2
D21
LED21 secondary side of the transformer isec1 is continuous, as
T2
isec2
+ shown in Fig. 2b.
- Co21 -
Vo21 Io21
+
Lm2
vsec2 Co22
√
+ Io22 Lr ∕Cr
Vo22
Cdc2 - Qe = � (4)
- D22 (8N 2 𝜋 2 )Rd
LED22
LEDk1
Tk Dk1
iseck
Cok1
+
-
Vok1 Iok1 When the number of LEDs in any channel changes,
+
Lmk
vseck Cok2 + Iok2 the quality factor, Qe, is no longer a constant value. It is
Vok2
Cdck - obtained using (4), which results in a different current gain
- Dk2
LEDk2 curve at the operating point of the converter. In [11], the
normalized LLC current gain with an LED load is given
C3 LLC resonant LED driver
Fig. 1 Topology of an M as:
13
C3 LLC resonant LED drivers
Fuzzy self‑tuning PID control of M 785
2
vab iLr vab iLr
Qe
Qe/2
Qe/3
t
isec1
t
Qe/4
vsec1
vsec1
isec1 Qe/5
MI=Io/In
Q1
Q1 Q1 1
t
Q2 t Q2
(a) (b)
0
Fig. 2 Typical waveforms of the primary and secondary sides of a 0.5 fs/frmin 1 1.5
transformer: a fm < fs < fr; b fs ≥ fr fn=fs/fr
(5) S1
Cr Lr LED11
where In is the LED current at the resonance point, Rd is T1
isec1
D11
-
D12
)]2
)]4 S2 LED12
LED21
[ [ ( (
1 1 1 1 T2 D21
A=− 1+ 1− 2 ,B = 1 + 1− 2 isec2
+
H fn H fn Co21 Vo21 Io21
+ -
Lm2
vsec2
[[ )]2 )2 ] Io22
( ( +
1 1 1 Co22 Vo22
C = 1+ 1− 2 + Q2e fn − - Cdc2 -
H fn fn - D22
LED22 ILED
ei -
[[ )] fs
)]2 (
Vin ∕2N 2 Drive
(
1 1 PFM PID controller ec de dt + Iref
D= 1+ 1− 2 − module i
H fn Vth KT
Fuzzy
control rules
The LLC current gain curves for different quality fac-
tors are shown in Fig. 3, where fn = 1 is at the resonant
Fig. 4 Fuzzy self-tuning PID control system schematic
point of the LLC.
ei KT
Input Fuzzy logic
ec Fuzzification Unblur
quantization reasoning
3 Fuzzy self‑tuning PID control based
Fuzzy rule
on deviation and deviation increment base
In this study, a fuzzy self-tuning PID controller is designed Fig. 5 Fuzzy control schematic
for a 4-channel MC3 resonant LED driver. A block dia-
gram of the control system is shown in Fig. 4. The main
goal of controlling the LED driver is to ensure that the ( )
1
output current is adjusted to its expected value even in the Gc (s) = Kp 1 + + Td s (6)
Ti s
presence of disturbances in the input voltage. Therefore, a
fuzzy self-tuning PID current-loop controller is adopted.
In the figure, the deviation ei = I ref–I LED and the devia- Fuzzy control makes online corrections to the PID
tion increment e c show the rate of change in the current parameters, and the process consists of four parts: fuzzi-
deviation ei. The transfer function of the PID controller fication, fuzzy reasoning, defuzzification, and fuzzy rule
is given as: base, as shown in Fig. 5. First, the inputs ei and ec undergo
13
786 D. Li et al.
fuzzification. After that, the inference is carried out accord- the output current decreases, both the deviation ei and
ing to the fuzzy rule base. Then, the tuning factor KT for the the deviation increment ec increase. The tuning factor KT
proportional gain is obtained by defuzzification, and the new decreases more than in Rule (1), and the switching fre-
PID proportional gain Kp is obtained using (7): quency decreases. When fm < fs < fr, as the output current
increases, both the deviation ei and the deviation increment
Kp = KT Kpo (7) ec decrease. The tuning factor KT and the switching fre-
where Kpo is the fixed part of the proportional gain, which quency also decrease. On the other hand, when the output
is set when the switching frequency is equal to the resonant current decreases, the deviation ei increases while the devia-
frequency in series with the rated load. tion increment ec decreases. The tuning factor KT and the
Fuzzification is first carried out for ei and ec. Fuzzy terms, switching frequency both increase.
including negative big (NB), negative medium (NM), nega- (4) If the quality factor Qe decreases, the current gain
tive small (NS), zero (ZO), positive small (PS), positive curve moves upward, the output current gain increases, and
medium (PM), and positive big (PB), are used to represent the change rate of the current gain increases. When fs ≥ fr,
the fuzzy subsets of ei and ec. The fuzzy subsets of both ei as the output current increases, both the deviation ei and the
and ec are defined as {NB, NM, NS, ZO, PS, PM, PB} using deviation increment ec decrease. The tuning factor KT and
the triangular membership function. the switching frequency both increase. On the other hand,
The following tuning rules can be obtained according to when the output current decreases, the deviation ei increases
the current gain curves. while the deviation increment ec decreases. The tuning fac-
(1) At a constant quality factor Qe, when fs ≥ fr, as the tor KT and the switching frequency also decrease. When
output current increases, both the deviation ei and the devia- fm < fs < fr, as the output current increases, the deviation ei
tion increment ec decrease. The tuning factor KT is greater decreases while the deviation increment ec increases. The
than or equal to 1, and the switching frequency increases. tuning factor KT and the switching frequency both decrease.
On the other hand, when the output current decreases, both On the other hand, when the output current decreases, both
the deviation ei and the deviation increment ec increase. The the deviation ei and the deviation increment ec increase. The
tuning factor KT is less than 1, and the switching frequency tuning factor KT and the switching frequency also increase.
decreases. The above fuzzy rules are summarized in Table 1.
(2) At a constant quality factor Qe, when fm < fs < fr, as the To reduce the calculation time of the DSP program and to
output current increases, the deviation ei decreases while the expedite the system response, a fuzzy lookup table, shown in
deviation increment ec increases. The tuning factor KT and Table 2, is created according to the triangular membership
the switching frequency also decrease. On the other hand, function, fuzzy rules, and defuzzification methods, using the
when the output current decreases, both the deviation ei and Fuzzy Logic Toolbox provided in MATLAB. When the DSP
the deviation increment ec increase. The tuning factor KT and is running, the deviation ei and the deviation increment ec
the switching frequency also increase. are calculated in real time. They are then used to obtain the
(3) If the quality factor Qe increases, the current gain
curve moves downward, the output current gain decreases,
Table 2 Fuzzy lookup table ei1 ei2 ein
and the change rate of the current gain decreases. When …
f s ≥ f r, as the output current increases, the deviation e i ec1 KT11 KT21 … KTn1
decreases while the deviation increment ec increases. The ec2 KT12 KT22 … KTn2
tuning factor KT increases more than in Rule (1), and the ⋮ ⋮ ⋮ ⋮ ⋮
switching frequency increases. On the other hand, when ecn KTn1 KTn2 … KTnn
NB NB NB NM ZO PB PB PB
NM NB NB NM ZO PM PB PB
NS NB NB NS ZO PM PB PB
ZO NB NB NS ZO PS PS PS
PS NM NM PM ZO PS PS PS
PM NM PM PM ZO ZO ZO ZO
PB PM PM PM ZO ZO ZO ZO
13
C3 LLC resonant LED drivers
Fuzzy self‑tuning PID control of M 787
tuning factor KT from the fuzzy lookup table, and the new 16
Vo12=Vo21=Vo22=16.67V
proportional gain Kp is calculated using (7).
Vo(V)
14
12
Vo11=10V
10
0
VCdc2=0V
4 Simulation and experimental results
-1
VCdc(V)
-2
-3 VCdc1=-3.335V
Io(A)
0
Io11=Io12=Io21=Io22=0.3A
-0.5
[11, 12], and shown in Table 3. -1
0.0305 0.031 0.0315 0.032 0.0325 0.033 0.0335 0.034 0.0345
Vo(V)
12
Vo22=10V
and 0, respectively. 10
VCdc2=3.335V
3
4.1 Simulation results
VCdc(V)
2
1
VCdc1=0V
0
Io(A)
0
Io11=Io12=Io21=Io22=0.3A
-0.5
simulations in Simulink when the numbers of LEDs in each -1
0.0415 0.042 0.0425 0.043 0.0435 0.044 0.0445 0.045 0.0455
of the channels are not equal. The numbers of LEDs in each t(s)
of the channels are listed in Table 4. It can be seen from the (b)
figure that the voltage differences between channel 1 and
channel 2 as well as that between channel 3 and channel Fig. 6 Steady-state response waveforms of a system when the load of
4 are shared by the capacitors C dc1 and Cdc2, respectively. each channel is unbalanced: a unbalanced state 1; b unbalanced state
2
As a result, the current in the four channels is maintained
at 300 mA. When the numbers of LEDs are different in the
four channels, the designed driver can still output a constant 27.2 V), the output current is stabilized to a set value after
current, which shows that the current sharing effect among a short adjustment. The fuzzy PID controller is observed
the channels is relatively good. to have a faster recovery, a smaller current overshoot, and
When the input voltage (V in) changes, the dynamic stronger robustness than the conventional PID controller.
responses of the output currents of the fuzzy and conven-
tional PID controllers are shown in Fig. 7. It can be seen 4.2 Experimental setup
from this figure that both of the controllers are capable of
adjusting the output current to its required value. When Vin An experimental setup was established to verify the control
decreases (i.e., at t = 8 ms when Vin changes from 24 to 22 V, strategy proposed in this paper. It includes a digital signal
and at t = 20 ms when Vin changes from 27.2 V to 24 V) and processor (DSP28335), an MC3 LLC resonant LED driver
increases (i.e., at t = 16 ms when Vin changes from 22 V to circuit, four LEDs, and a sampling network, as shown in
Fig. 8. The main circuit parameters are selected according
to the specifications in Table 3.
Table 3 Main circuit parameters of an LLC resonant converter Voltage waveforms of the MOSFET are shown in Fig. 9.
Parameter Value It can be seen from this figure that the turn-on signal vgs is
13
788 D. Li et al.
0.2 Vo11(10V/div)
0.305
0.15
0.3 VCdc1(5V/div) Io11(500mA/div)
0.1
0.05 0.295 Io11(500mA/div)
VCdc1(2V/div)
0 7.5 8 8.5 9 9.5 10 10.5 Io12(500mA/div) Io12(500mA/div)
×10-3
Vo22(10V/div)
24 Vo21(10V/div)
23
VCdc2(2V/div) Io21(500mA/div)
22
5ms/div 5ms/div
Fig. 7 Dynamic response waveforms of a system under changes in
the input voltage (c) (d)
Io21(500mA/div) Io21(500mA/div)
2.5µs/div 5ms/div
4-channel Sampling
LED T2 network (a) (b)
13
C3 LLC resonant LED drivers
Fuzzy self‑tuning PID control of M 789
36ms
200mA
250mA
34ms
50ms/div 50ms/div
(a) (b)
DC block capacitor voltage (V) VCdc1 VCdc2 A fuzzy self-tuning PID controller was designed for a reso-
0 3.35 nant LED driver. The designed controller has good dynamic
Output current of each channel (mA) Io11 Io12 Io21 Io22 performance and robustness. When there are disturbances in
300 300 300 300 the input voltage or unbalanced LEDs, the driver can output
currents in multiple channels with relatively good current
sharing effects. In addition, soft switching is achieved, which
reduces the power loss. The proposed fuzzy control method
Vin (20V/div) Vin (20V/div) does not depend on a precise model of the control target.
Io11 (500mA/div)
17ms Moreover, the fuzzy control rules are simple and effec-
tive, and the calculation load is small. Hence, the proposed
120mA
18ms
Io11 (500mA/div)
models for such systems.
50ms/div 50ms/div
(a) (b) Acknowledgements This work was supported by the Key Program of
Natural Science Foundation of Guangxi Province of China under Grant
Fig. 13 Fuzzy PID control system dynamic response waveforms: a 2018GXNSFDA281037 and the National Natural Science Foundation
input voltage increases; b input voltage decreases of China under Grant 51667005.
13
790 D. Li et al.
2. Bender, V.C., Marchesan, T.B., Alonso, J.M.: Solid-state light- 20. Lu, Y., Huang, X., Huang, Y., Liu, D.: Sigmoid Function Model
ing: A concise review of the state of the art on LED and OLED for a PFM Power Electronic Converter. IEEE Trans. Power Elec-
modeling. IEEE Ind. Electron. Mag. 9(2), 6–16 (2015) tron. 35(4), 4233–4241 (2020)
3. Hong, S.S., Lee, S.H., Cho, S.H., Roh, C.W., Han, S.K.: A new 21. Tian, S., Lee, F.C., Li, Q.: Equivalent Circuit Modeling of LLC
cost-effective current-balancing multi-channel LED driver for a Resonant Converter. IEEE Trans. Power Electron. 35(8), 8833–
large screen LCD backlight units. Journal of Power electronics 8845 (2020)
10(4), 351–356 (2010) 22. Chang, C.H., Chang, E.C., Cheng, C.A., Cheng, H.L., Lin, S.C.:
4. Hui, S.Y., Qin, Y.X.: A general photo-electro-thermal theory for Small signal modeling of LLC resonant converters based on
light emitting diode (LED) systems. IEEE Trans. Power Elec- extended describing function. In: 2012 International Symposium
tron. 24(8), 1967–1976 (2009) on Computer, Consumer and Control 365–368 (2012)
5. Wu, X., Wang, Z., Zhang, J.: Design considerations for dual- 23. Cheng, B., Musavi, F., Dunford, W.G.: Novel small signal mod-
output quasi-resonant flyback LED driver with current-sharing eling and control of an LLC resonant converter. In: 2014 IEEE
transformer. IEEE Trans. Power Electron. 28(10), 4820–4830 Applied Power Electronics Conference and Exposition-APEC
(2012) 2014, 2828–2834 (2014)
6. He, Q., Luo, Q., Cao, C.: A modular open-loop multi-channel 24. Wang D., Shao R., Dong F.: Research on LLC resonant converter
resonant constant-current LED driver. Proc.CSEE 39(22), 6688 based on fuzzy PI control. Chinese Journal of Electron Devices,
(2019) 19–20 (2019)
7. Luo, Q., Zou, C., Zhi, S., Yan, H., Zhou, L.: A multi-channel 25. Buccella, C., Cecati, C., Latafat, H., Razi, K.: Comparative tran-
LED driver based on passive resonant constant current net- sient response analysis of LLC resonant converter controlled by
works. Proc. CSEE 33(18), 73–79 (2013) adaptive PID and fuzzy logic controllers. In: IECON 2012–38th
8. He, Q., Luo, Q., Cao, C., Sun, P., Zhou, L., Wei, Y.: September. Annual Conference on IEEE Industrial Electronics Society, 4729–
A modular multi-channel constant-current LED driver with high 4734 (2012)
frequency AC square voltage bus. In 2018 IEEE energy conver- 26. Tian, Z., Fang, Y., Guo, N., Zhou, X., Cao, S.: Fuzzy frequency-
sion congress and exposition (ECCE), 4712–4716 (2018) selecting sliding mode controller for LLC resonant converter. J.
9. He, Q., Luo, Q., Huang, J., Cao, C., Sun, P., Du, X.: LCL-T Eng 2019(15), 571–575 (2019)
resonant network-based modular multi-channel constant-cur- 27. Madheswaran, M., Nagarajan, C.: DSP based fuzzy controller
rent LED driver analysis and design. J. Power Electron. 20(6), for series parallel resonant converter. Front. Electr. Electron. Eng
1616–1628 (2020) 7(4), 438–446 (2012)
10. Wu, X., Zhang, J., Qian, Z.: A simple two-channel LED driver 28. Chen, T., Chen, Q., Lei, L., Zhang, L. and Quan, S.: Research on
with automatic precise current sharing. IEEE Trans. Industr. full bridge LLC resonant converter based on fuzzy self-adaptive
Electron. 58(10), 4783–4788 (2011) PI control. In: 2017 32nd Youth Academic Annual Conference of
11. Wu, H., Ji, S., Lee, F.C., Wu, X.: Multi-channel constant cur- Chinese Association of Automation (YAC), 132–137 (2017)
rent (MC3) LLC resonant LED driver. In 2011 IEEE Energy 29. Liang, G., Du, G., Liu, Y.: Fuzzy PID control strategy for LLC
Conversion Congress and Exposition, 2568–2575 (2011) resonant converter with wide operating range. J. Power Supp.
12. Ji, S., Wu, H., Ren, X., Lee, F.C.: Multi-channel constant cur- 18(2), 138–144 (2020)
rent (MC3) LED driver. In 2011 Twenty-Sixth Annual IEEE 30. Liang G.: The research on fuzzy PID control strategy for antidrift
Applied Power Electronics Conference and Exposition (APEC), of LLC resonant converter. Master’s thesis, South China Univer-
718–722 (2011) sity of Technology (2019).
13. Chang, C.H., Cheng, C.A., Jinno, M., Cheng, H.L.: An inter- 31. Lu, B., Liu, W., Liang, Y., Lee, F.C. and Van Wyk, J.D., 2006,
leaved single-stage LLC resonant converter used for multi- March. Optimal design methodology for LLC resonant converter.
channel LED driving. In 2014 International Power Electronics In: Twenty-First Annual IEEE Applied Power Electronics Confer-
Conference (IPEC-Hiroshima 2014-ECCE ASIA), 3333–3340 ence and Exposition, 6–6 (2006)
(2014).
14. Feng, W., Lee, F.C., Mattavelli, P.: Optimal trajectory control of
LLC resonant converters for LED PWM dimming. IEEE Trans. Delong Li was born in Guiping,
Power Electron. 29(2), 979–987 (2013) China, in 1996. He received his
15. Menke, M.F., Duranti, J.P., Roggia, L., Bisogno, F.E., Tambara, B.S. degree in Light Sources and
R.V., Seidel, Á.R.: Analysis and Design of the LLC LED Driver Lighting from Tiangong Univer-
Based on State-Space Representation Direct Time-Domain Solu- sity, Tianjin, China, in 2018. He
tion. IEEE Trans. Power Electron. 35(12), 12686–12701 (2020) is presently working towards his
16. Dong, H., Xie, X., Mao, F., Zhang, L., He, Y.: A Novel Primary- M.S. degree at Guangxi Univer-
Side Regulation Control Scheme for CCM and DCM LLC LED sity, Nanning, China. His current
Driver Based on “Magnetizing Current Cancellation Method.” research interests include DC/
IEEE Trans. Power Electron. 35(11), 12223–12237 (2020) DC resonant converter control
17. Wang, Y., Guan, Y., Huang, J., Wang, W., Xu, D.: A single-stage and its applications.
LED driver based on interleaved buck–boost circuit and LLC
resonant converter. IEEE J. Emerg. Select. Topics Power Electron
3(3), 732–741 (2015)
18. Menke, M.F., Seidel, Á.R., Tambara, R.V.: LLC LED driver
small-signal modeling and digital control design for active rip-
ple compensation. IEEE Trans. Industr. Electron. 66(1), 387–396
(2018)
19. Zhao, Z., Zong, J.: A LLC Resonant Full-bridge Converter with
Fractional Order PID Controller. In: 2019 International Confer-
ence on Intelligent Informatics and Biomedical Sciences (ICI-
IBMS), 250–255 (2019)
13
C3 LLC resonant LED drivers
Fuzzy self‑tuning PID control of M 791
Yimin Lu received her B.S. degree ENG, UK. From 2014 to 2015, she was a Visiting Research Scholar in
in Measurement Technology and the Department of Mechanical Engineering and WEMPEC, University
Instrumentation from Southeast of Wisconsin-Madison, Madison, WI, USA. Her current research inter-
University, Nanjing, China, in ests include control theory and its applications in power electronics.
1992; her M.S. degree in Control
Theory and Control Engineering Xin Ge was born in Zhuzhou,
from Guangxi University, Nan- China, in 1996. He received his
ning, China, in 2000; and her B.S. degree in Electrical Engi-
Ph.D. degree in Control Theory neering and Automation from
and Control Engineering from the Hunan University of Science
the South China University of and Technology, Xiangtan,
Technology, Guangzhou, China, China, in 2018. He is presently
in 2004. In 1992, she became a working towards his M.S. degree
Teaching Assistant in the School at Guangxi University, Nanning,
of Electrical Engineering, China. His current research
Guangxi University, where she is interest includes wireless power
presently working as a Professor of Electrical Engineering. From 2007 transmission control and its
to 2008, she was a Postdoctoral Researcher in the Department of Elec- applications.
tronic and Electrical Engineering, University of Sheffield, Sheffield,
13