Switched-mode power supply (SMPS) is an electronic power supply that utilize switching regulator to ensure high efficiency in power conversion. Advancement in semiconductor permits SMPS to operate at high switching frequency. This results in significant reduction of size, weight and volume of SMPS .
SMPS are found in a wide range of applications, from power supply for computers to high power traction battery chargers. These power supplies generally have AC voltage as the input, which eventually converts into a regulated DC voltage output. For the purposes of maximizing the ratio of real power to apparent power, and electrical safety considerations, SMPS are usually incorporated with power factor correction and electrical isolation features .
Selected Power Supply and Economic Issue
In this design project, the designed SMPS served as a 17kW Electric Vehicle (EV) on-board battery charger. This battery charger was targeted to conform to the anticipated SAE J2894 standard. Therefore, it would be the first-to-market battery charger upon the finalization of the standard.
Design Requirements
The charger would be used to charge up a 24kWh, 342V Lithium-ion Manganese Oxide battery pack under the constant current constant voltage charging scheme. The battery pack has a nominal charge voltage of 378V with 4.5V tolerance (1% ripple). The continuous and peak charge rates are 1C (70A) and 3C (210A) respectively.
The EV on-board charger is targeted to be SAE J2894 compliant. Since works on the SAE J2894 is in the progress, the requirements on the charger's input, power factor (PF) and efficiencies would be based on prevailing standards.
In view of the available infrastructure, the charger would be designed to accommodate the two different AC charge levels as defined by SAE J1772; which is expected to be incorporated into the SAE J2894:
Level 1 AC Charging: 120V AC, 15A or 20A, 1.9kW
Level 2 AC Charging: 208V - 240V AC, up to 80A, 19kW
The PF of traction battery chargers are typically less than 0.78. In view of the availability of the Level 2 AC charging, it would be certain that SAE J2894 would require chargers to incorporate power factor corrections. Considering that most utilities would impose levy on customers if the PF falls below 0.9~0.95, the charger would be designed to achieve PF larger than 0.95.
Lastly, the efficiency of the charger should be at least 90%, typical efficiency of traction battery chargers.
The specifications for the EV on-board charger were summarized in the following Table 1:
Input
120V - 240V single-phase AC, max 80A continuous, 19kW
Output
378V DC nominal, ripple < 1.1%, 50A continuous (1.5 hours), 210A peak
Power Factor
At least 0.95, active PFC
Efficiency
At least 90%
Electrical Isolation
Yes
Table 1: Specifications of Battery Charger
Theory of Design
The overall design of the EV battery charger was as shown in Figure 1.
Figure 1: Battery Charger Overall Design
The designs of each of the blocks in Figure 1 were explained in the following subsections.
Load Modeling
The charger adopts the constant current-constant voltage scheme to charge up the lithium-ion battery pack. Under this charging scheme, the given Lithium-ion Manganese Oxide battery pack were initially charged at constant current until battery pack's voltage reached 378V. Thereafter, the charge current was reduced to maintain the battery pack voltage (1.1% voltage ripple). The charging process was complete after the charge current became 0A or negligible.
With the current and voltage controls implemented in the charger, the battery pack behaved like a variable resistive load to the charger during the charging process; minimal resistance at start of the charging process and maximum resistance at the onset of constant voltage charging phase. Thus, the battery pack was modeled as a variable resistive load. The minimum and maximum resistive loads were shown in Table 2:
Charging level
Minimum Load
Maximum Load
1
2
Assuming 90% Efficiency (peak output power, Level 1 = 1.7kW, Level 2 = 17kW)
Table 2: Variable Resistive Load
Full Bridge Converter
The isolated DC-DC converter topology used in the charger was the full bridge converter. Since the full bridge converter could handle output power up to 2kW , a single converter was suffice for Level 1 charging scheme while 10 such converters were cascaded in parallel to deliver the maximum power for Level 2 charging.
The system equations for the full bridge converter (shown in Figure 2) operated in Continuous Conduction Mode (CCM) was as follows:
where D is the duty ratio and f was the switching frequency of the IGBTs.
Figure 2: Full Bridge Converter
The design of the full bridge converter for Level 1 charging was as follows:
Resistance, Ω
50
84
Output Voltage, V
378
Turns ratio = 1, , , f=20kHz
Duty Ratio, D
Minimum Inductance, H
Inductance (30% over-design)=
Minimum Capacitance, F (1% ripple)
Capacitance =
Table 3: Full bridge Converter Sizing
As the magnetizing inductance is not a design parameter , it was omitted in the component sizing and simulation of the converter. This inductance was assumed much larger than that of the other inductance in the converter, approximately 10 times.
Full-wave rectifier and Boost Converter
The first stage of the converter was the full-wave rectifier with Power factor Correction (PFC) (Figure 3).
Figure 3: Full wave Rectifier with PFC
PFC was accomplished by means of a Boost converter. For steady-state operation, the following are the system equations for the Boost converter in CCM:
or
where is the input voltage, D is the duty ratio, f the switching frequency of the IGBT and is the difference between the maximum and minimum inductor current.
For steady-state operation, the input voltage to the Boost converter is in fact the average output voltage of the full-wave rectifier. Thus:
Similar to the design of the Full Bridge converter, the load of the Boost converter (CCM) was assumed to be a variable resistive load. The design of the PFC was as follows:
Level 1 (120V AC)
Level 2 (240V AC)
Average Input Voltage, V
, f=20kHz
Resistive Load, Ω
Average Inductor Current, A
Duty Ratio, D
Inductance, H
Inductance(10% over-design) =
Minimum Capacitance, F (0.1% ripple)
Capacitance =
Table 4: Boost Converter System Sizing
Modeling, Simulation & Interpretation
The modeling and simulation of the converters and charger were initially performed using Orcad PSpice software. However, due to convergence issues, these tasks were performed using the Matlab Simulink software. In these simulations, components were assumed to be ideal.
Full Bridge Converter
The model of the Full bridge converter simulated in Simulink environment was shown in Figure 4.
Figure 4: Simulink Model of Full Bridge Converter
The results of the converter output based on the two different loads at Level 1 charging were shown in Figure 5 and 6.
Figure 5: Load Voltage and Current, Inductor Current for 50Ω Load
Figure 6: Load Voltage and Current, Inductor Current for 84Ω Load
During steady-state operation, the converter was operating in CCM and the output voltage ripple for both loads were less than 1%. On the other hand, the overshoot in output voltage in both loads during the transient phase were undesirable.
The transient response of the full bridge converter were affected by its inductance and capacitance, as shown in Figure 7. The overshoot in the output voltage were greatly reduced when the inductance was increased with the capacitance decreased by the same factor. On the other hand, increased in capacitance worsen the response time as well as the overshoot in the output voltage. Thus, the overshoot in the output voltage could be minimized, without having the inductance increased, by increasing the switching frequency and reducing the capacitance.
Figure 8 were the results of the converter output, with the switching frequency increased by a factor of 10 and the corresponding reduction of the capacitance by a factor of 100. The overshoot in the output voltage was similar to the case when the inductance was increased by a factor of 10.
Figure 7: Output voltage and Inductor Current with Varying Inductance and Capacitance
Figure 8: Load Voltage and Current, Inductor Current for f=200kHz
Full-wave Rectifier and Boost Converter
The model of the Full-wave converter with PFC, simulated in Simulink environment was shown in Figure 9. The control of the boost converter for PFC was based on the assumption that a controller was available to maintain the output voltage at 600V while at the same time, the current was controlled for PFC. With this assumption, the switching circuit for the boost converter was simplified as follows (Figure 10):
Figure 10: Simplified Boost Converter Switch Control
The above Boost converter control circuit was purely for simulation purpose, it does not represent the actual implementation of the Boost converter control for PFC. The value of the resistor, RControl, was such that the output voltage of the Boost converter was maintained at 600V. The current through RControl thus represent the current (with PF=1) that would flow into the Full-wave rectifier.
Figure 9: Simulink Model of Full-wave Rectifier with PFC
The model in Figure 9 were simulated for the two extreme input operating points of the charger; Level 1 charging when the battery is fully depleted and Level 2 charging at constant voltage phase. The results, input voltage, current, PF and output voltage and inductor current, were shown in Figure 10 and 11 respectively.
Figure 10: Simulation Results of Full-wave rectifier with PFC for Level 1 Charging
Figure 11: Simulation Results of Full-wave rectifier with PFC for Level 2 Charging
The average inductor current for both results were approximately around the values defined in Table 4. In addition, the inductor currents were always larger than zero, which indicate the converter was always operating in the desired CCM. The PF in both cases were around 90%.
Unlike in the results for Level 1 charging, the output voltage ripple for the Level 2 charging case was significant. This could be due to a lower switching frequency administered by the current control mode. As such, the capacitance needs to be increased in order to compensate for the high output voltage ripple. The results of the converter for Level 2 charging with capacitance increased to 30mF were shown in Figure 12. As discussed previously, increasing the capacitance would increase the rise time of the transient response. The increase in capacitance had brought about desirable results for Level 2 charging, however, at the expense of the rise time in Level 1 charging. As shown in Figure 13, the output voltage of the converter with Level 1 charging inputs were significantly increased. Therefore, separate capacitors shall be used for the two charging levels.
Figure 12: Simulation Results of Full-wave rectifier with PFC for Level 2 Charging (Capacitance = 30mF)
Figure 13: Rise Time of Converter for Level 1 Charging with 30mF Capacitance
Charger at Level 1 Charging
The charger at level 1 charging utilized a single Full Bridge converter that was explained in Section 5.3. On the other hand, the charger would utilize ten of the Full Bridge converters connected in parallel to provide the desired output power. Therefore, it would be suffice to simulate the charger operating in Level 1 charging and results for Level 2 charging could be inferred.
The cases for Level 1 charging for the when the battery was fully depleted and constant voltage charging phase were simulated. Figure 14 depict the resulting characteristic of the charger when charging a fully depleted battery pack. In this charging mode, constant current was desired with low current ripple. Since the load was depicted as a constant resistive load, the characteristic of the output current could be inferred from the output voltage plot.
Figure 14: Charger Characteristics with Fully Depleted Battery Pack as Load
Results of shown in Figure 14 were very different from that shown for the individual converters, particularly the output voltage ripple and PF. The high charger output voltage ripple was mainly due to a much larger output voltage ripple at the output of the Boost converter. Increasing the Boost converter capacitance would improve the overall output voltage ripple, however, the rise time would be badly affected. On the other hand, the output ripple voltage could be improved without affecting the rise time by increasing the inductance of the Full Bridge converted. This approach was adopted, ignoring the EMI issues that might arise due to the higher inductance (5mH). The results were shown in Figure 15.
Figure 15: Charger Characteristics with Increased Inductance and Fully Depleted Battery Pack as Load
With the new inductance, the output voltage ripple was significantly reduced, without affecting the rise time. This translates to a relatively constant output current.
Figure 16 depict the resulting characteristic of the charger when charging a 90% charged battery pack. In this charging mode, constant voltage was desired with voltage ripple around 1.1%.
Figure 16: Charger Characteristics with 90% Charged Battery Pack as Load
Similar to the results for charging a fully depleted battery pack, both the voltage ripple and the PF differ from the results shown demonstrated by the individual converters. Unlike in the previous case, increasing the inductance of the Full Bridge converter does not have significant effect on the charger's output voltage ripple. This was due to a large voltage ripple at the output of the Boost converter. As such, the capacitance of the Boost converter was increased to 6mF. The results were shown in Figure 17.
Figure 17: Charger Characteristics with 90% Battery Pack as Load
The output voltage ripple of the converter with the new capacitance was around 1.1%.
Practical Design
The list of components for practical implementation of the charger was shown in Table 5.
Specifications
Quantity
Source
Unit Price
Full-Wave Rectifier:
Diode Rectifier Bridge
IdAV = 100A,
VRRM = 600V, trr = 35ns
1
IXYS: VBE 100-06NO7
Boost Converter:
Inductor
L= 15mH
I = 80A
1
-
-
Inductor (Level 1 Charging)
L= 15mH
I = 20A
1
Digi-key: L2015-ND
$182.38
Inductor (Level 2 Charging)
L = 2.5mH
I = 100A
1
Digi-key: 195E100-ND
$608.89
Diode
IF = 100A,
VRRM = 600V, VF = 1.65V
1
Infeneon: IDW100E60
-
Capacitor (Level 1 Charging)
C= 6mF
V = 300V
4
Digi-key: CGS602T300X8L-ND
$319.32
Capacitor (Level 2 Charging)
C= 15mF
V = 300V
8
Digi-key: CGS153T300X8L-ND
$354.74
IGBT
VCES = 600V,
IC = 75A, Vge = 20V
1
Infineon: IGW75N60T
-
Full Bridge Converter
IGBT
VCES = 600V,
IC = 30A, Vge = 20V
4
Infineon: IGW30N100T
-
Diode
IF = 6A,
VRRM = 600V, VF = 1.5V
2x10
Infineon:
Idd06e60G
-
Capacitor
C = 10uF
V = 400V
1x10
Digi-Key: PCE3590TR-ND
$1.11
Inductor
L = 2.5mH
I = 10A
2x10
Digi-Key: DKIL-0346-1002-ND
$28.51
Transformer
Turns ratio = 1,
L = 5mHx10
1x10
-
-
Table 5: Component List
Realistic Constraints and Implementation Issues
The design of the charger were based on the assumption that the battery pack at a given voltage and current could be modeled as a constant resistive load, where the energy dissipated by the resistor is equivalent to the energy transferred to the battery pack and losses. In actual implementation, the voltage of the battery pack is constantly changing, particularly during Level 2 charging when the charging current is high. Therefore, the rise time or response time of the charger has to be fast in order for the above assumption to be valid. In other words, apart from the PFC controller for the Boost converter, another controller is required at the Full Bridge converter to control the current and voltage at the charger output. This controller shall be able to control the charger in such a way that the response time of the charger when delivering peak charge current is at least as much as that of the rise time of the battery pack.
Judging from the component list as shown in Table 5, the cost of the inductance and capacitance for the Boost converter were considerably high. Therefore, Alternative topologies shall be considered that would be able to bring down the cost of these components. While increasing the switching frequency would bring down the cost of these components, the cost benefit of using very high frequency switches shall be analyzed, in addition to the EMI issue that might arise.
Conclusion
An on-board EV battery charger was designed to allow charging off domestic 120Vac and specialized 240Vac power points. The charger consists of a Full-wave rectifier with a Boost converter for PFC, cascaded to ten Full Bridge converters that were connected in parallel. For Level 1 charging, only one of the ten Full bridge converters was active. Apart from the PF, the output voltage, current and voltage ripple were according to requirement, as was demonstrated in the simulation results. The results of the simulation validate the correctness of the converter design. In order for the controller to be fully functional, more work has to be done on the controller design, EMI studies and thermal management of the charger in order for the charger.
