The formation process is a critical step in the manufacturing of high-capacity power batteries, especially for electric vehicle battery applications. During formation, cells undergo repeated charge and discharge cycles, which consumes a significant amount of energy. Conventional formation equipment often dissipates the discharge energy as heat in resistors, leading to low efficiency and excessive waste. This thesis focuses on the design of a high-efficiency formation system tailored for large-format electric vehicle battery cells. The work is divided into two main parts: a discharge module and a charging module. The discharge module is designed to recycle the battery energy back to a DC bus, while the charging module provides controlled constant-current/constant-voltage charging. Both modules operate under low-voltage and high-current conditions, which poses significant challenges for efficiency. I propose a novel transformer series-parallel topology for the discharge module and adopt a current-doubler synchronous rectification topology for the charging module. I derive the design equations, select appropriate magnetic components and power devices, and verify the system through prototype experiments. The experimental results confirm that the discharge module achieves an efficiency of 92% at a discharge current of 100 A, and the charging module achieves an efficiency of 91% at a charging current of 100 A. The proposed topologies and design methodologies provide a practical and efficient solution for formation equipment in the electric vehicle battery industry.
1. Introduction
Large-capacity power batteries, particularly lithium-ion batteries, are becoming the dominant energy storage solution for electric vehicles, small electric tools, and stationary storage. Battery formation is an essential industrial process that activates the active materials and stabilizes the cell performance. In a typical production line, each cell must undergo at least three charge cycles and two discharge cycles before being shipped. For a high-capacity electric vehicle battery cell with a capacity of 100 Ah or more, the energy involved in each cycle is substantial. Conventional formation equipment used by many domestic manufacturers still relies on resistive discharge, where the energy stored in the cell is dissipated as heat. This practice leads to enormous electricity waste and raises the production cost. For a large battery plant, the electricity consumption for formation can account for up to 10% of the total production cost. Therefore, there is an urgent need to develop high-efficiency formation equipment that can recycle the discharge energy and minimize losses during charging.
The main difficulty in designing such equipment lies in the low-voltage and high-current characteristics of both the input side of the discharge module and the output side of the charging module. For example, a single electric vehicle battery cell has a terminal voltage in the range of 3.0–4.2 V, while the discharge current can be as high as 100–300 A. Similarly, the charging module must deliver a current of 100 A at a voltage lower than 4.2 V. These specifications make conventional power-conversion topologies inefficient because the forward voltage drop of diodes and the on-resistance of switching devices become significant relative to the output voltage. To overcome these issues, I have analyzed several isolated and non-isolated topologies and selected suitable structures that can operate with high efficiency under these demanding conditions.
This thesis is organized as follows. Section 2 presents the analysis and design of the discharge module, including a new transformer series-parallel topology, the design of high-frequency transformers, and the control circuit. Section 3 describes the charging module, focusing on the current-doubler synchronous rectification topology, the design of the main circuit parameters, and the control/drive circuitry. Section 4 reports the experimental prototype and measured waveforms. Finally, Section 5 summarizes the main conclusions.
2. Discharge Module Analysis and Design
The discharge module is intended to transfer energy from an electric vehicle battery cell to a high-voltage DC bus. The input voltage is the cell voltage, which ranges from 3.0 V to 4.2 V, and the output DC bus voltage is set to 400 V. The discharge current ranges from 20 A to 100 A, with a current ripple less than 5%. The target efficiency at 100 A is 92%. Because the voltage gain is as high as 130 times, a simple boost converter would require an extremely large duty cycle and suffer from high conduction losses. Therefore, an isolated DC-DC converter with a high-frequency transformer is necessary. The transformer not only provides the required voltage step-up but also galvanically isolates the battery from the DC bus.
2.1 Topology Comparison and Selection
Several isolated DC-DC topologies were considered for the discharge module:
| Topology | Advantages | Disadvantages | Power Range | Typical Applications |
|---|---|---|---|---|
| Forward | Simple, low cost, high reliability | Single-ended excitation, low transformer utilization | 100 W – few kW | Small- and medium-power supplies |
| Flyback | Very simple, low cost | Difficult to scale, transformer utilization low | 10 W – <500 W | Low-power consumer electronics |
| Push-pull | Bidirectional excitation, high transformer utilization, simple drive | Potential flux imbalance | 100 W – several kW | Medium-power industrial supplies |
| Half-bridge | Bidirectional excitation, low switch count | Requires isolated drive, DC-bus imbalance issues | 100 W – several kW | Industrial power supplies, computer supplies |
| Full-bridge | Bidirectional excitation, easy to achieve high power | Complex, high cost | kW – hundreds kW | High-power industrial supplies, welding |
The push-pull topology stands out because it uses only two power switches and the transformer is used bidirectionally, which improves the core utilization and reduces the number of primary turns. The main disadvantage of the push-pull converter is the possibility of transformer saturation due to unequal volt-second products. This can be mitigated by using a small series blocking capacitor, by matching the switches, and by employing current-mode control. For our discharge application, the input current is very high (up to 100 A). To further reduce the current stress on each switch and to increase the output voltage level, I propose a transformer series-parallel structure.
2.2 Transformer Series-Parallel Structure
The proposed discharge module consists of three identical push-pull DC-DC converters. The input sides of these converters are connected in parallel, while their output sides are connected in series. A simplified network representation of two converters is shown in Figure 1. The overall transmission matrix of the series-parallel connection is the sum of the individual transmission matrices:
$$ \begin{bmatrix} A & B \\ C & D \end{bmatrix} = \begin{bmatrix} A_1 & B_1 \\ C_1 & D_1 \end{bmatrix} + \begin{bmatrix} A_2 & B_2 \\ C_2 & D_2 \end{bmatrix} $$
If the converters are identical, the input admittance of each sub-converter is half that of the total network. Consequently, the input current of each converter is half of the total battery current. This effectively reduces the current rating of each switch by half for the same power level and input voltage. Similarly, the output voltage of each sub-converter is half of the total output voltage, allowing the use of diodes and capacitors with lower voltage ratings. Because the transformer secondary windings are in series, the same current flows through each secondary winding, forcing the input currents of the primary sides to be balanced automatically. This inherent current-sharing property eliminates the need for complex control circuits. I selected three parallel modules as a compromise between component count and performance.
2.3 Main Circuit of the Discharge Module
The main circuit of the discharge module is shown in Figure 2. Each push-pull converter has its own high-frequency transformer. The primary windings are connected to the battery through two power MOSFETs, which operate with complementary drive signals. The secondary windings are connected to three full-bridge rectifiers, whose outputs are series-connected to form the 400 V DC bus. An output filter inductor and capacitor are placed after the rectifier to reduce voltage ripple.

Figure 2 shows the complete topology. The battery voltage is applied to the center taps of the three transformers. Each push-pull leg includes a low-on-resistance MOSFET. The outputs of the three full-bridge rectifiers are stacked, so the total output voltage is the sum of the three transformer secondary voltages. The switching frequency is set to 20 kHz, and the dead time is adjusted to avoid short-through.
2.4 High-Frequency Transformer Design
The design of the high-frequency transformer is crucial for achieving high efficiency. I used the core geometry coefficient method. The design specifications are summarized below:
| Parameter | Symbol | Value |
|---|---|---|
| Minimum input voltage | \(V_{in,min}\) | 3.0 V |
| Maximum input voltage | \(V_{in,max}\) | 4.2 V |
| Output current (total) | \(I_o\) | 1.0 A |
| Output power (total) | \(P_o\) | 400 W |
| Switching frequency | \(f_s\) | 20 kHz |
| Efficiency target | \(\eta\) | 92% |
| Allowable regulation | \(\alpha\) | 1% |
| Operating flux density | \(B_m\) | 0.15 T |
| Window utilization factor | \(K_u\) | 0.4 |
| Core material | — | Ferrite |
Since the frequency is 20 kHz, the skin depth in copper is given by:
$$ \delta = \frac{66.2}{\sqrt{f_s}} = \frac{66.2}{\sqrt{20000}} \approx 0.43 \text{ mm} $$
Therefore, the maximum diameter of the litz wire should be less than \(2\delta = 0.86\) mm. The bare area \(A_w\) should be smaller than \(\pi (0.43)^2 /4 \approx 0.145 \text{ mm}^2\). I used multiple strands of thin wire in parallel to achieve the required current capacity while mitigating skin effect.
The apparent power of the transformer is:
$$ P_t = P_o \left( \frac{1}{\eta} + \sqrt{2} \right) \approx P_o \times 2.11 $$
For the push-pull converter, the wave factor is \(K_p = 1.0\). The electrical coefficient is:
$$ K_e = K_f^2 \cdot f_s^2 \cdot B_m^2 \times 10^{-4} $$
where \(K_f = 4.0\) for a square wave. Thus:
$$ K_e = 16 \times (20000)^2 \times (0.15)^2 \times 10^{-4} = 1440 $$
The core geometry coefficient is:
$$ K_g = \frac{P_t}{2 K_e \alpha} = \frac{400 \times 2.11}{2 \times 1440 \times 0.01} \approx 29.3 \text{ cm}^5 $$
A suitable core is ETD49, made of ferrite material with a geometry constant of 39.1 cm5. The core area is \(A_e = 2.12 \text{ cm}^2\), and the window area is \(W_a = 3.56 \text{ cm}^2\). The mean length per turn is \(MLT = 8.1\) cm.
The primary turns for the push-pull configuration are calculated by Faraday’s law:
$$ N_p = \frac{V_{in,min}}{4 f_s B_m A_e} = \frac{3.0}{4 \times 20000 \times 0.15 \times 2.12 \times 10^{-4}} \approx 1.18 $$
Because the primary has a center tap, the total primary turns are \(2N_p\), and I rounded this to 2 turns per half winding. I chose a litz wire with a bare copper area of 0.05 mm2 per strand. The number of parallel strands required is:
$$ n_p = \frac{I_{in,rms}}{J \cdot A_{strand}} $$
The current density is determined from the window utilization factor. More specifically, the primary side uses 10 strands of AWG 22 wire in parallel. The secondary winding is designed to produce an output voltage of about 133 V per module (since three modules are series-connected to obtain 400 V). The secondary turns are:
$$ N_s = N_p \frac{V_o}{V_{in,min}} \approx 2 \times \frac{133}{3.0} \approx 89 $$
With a wire area of 0.05 mm2 and the secondary current of 1 A, the current density is low, and a single strand of AWG 24 might suffice. However, because the voltage is high, the isolation requirement forces a thicker insulation layer. The actual transformer used a copper foil for the primary and a multi-strand secondary with a “sandwich” winding arrangement to reduce leakage inductance and proximity effect.
2.5 Rectifier and Output Filter
The output rectifier uses a full-bridge configuration. Since the output voltage is high (400 V) but the current is low (1 A), the diode reverse voltage stress is approximately 133 V per module. I selected fast-recovery diodes with a rated reverse voltage of 600 V and a forward current of 2 A. The output filter inductance is calculated from the allowable current ripple. For a ripple of less than 5% at \(I_o = 1\) A, the critical inductance is:
$$ L_o = \frac{V_o}{2 f_s \Delta I_o} = \frac{400}{2 \times 20000 \times 0.05 \times 1} = 200 \text{ mH} $$
In practice, because the three modules are series-connected and the rectified waveforms are interleaved, the ripple cancellation allows a smaller inductance. I used an inductor of 50 mH with an air gap. The output capacitor is selected to limit the output voltage ripple. Using the ESR criterion:
$$ C_o = \frac{I_o \cdot T_s}{2 \Delta V_o} $$
For \(\Delta V_o = 0.1\) V, \(C_o \approx 100 \mu\text{F}\). Two electrolytic capacitors of \(220 \mu\text{F}\) each, rated at 450 V, are connected in series to increase the voltage margin.
2.6 Control Circuit of the Discharge Module
The control circuit uses a UC3846 current-mode controller. This chip provides two outputs that are alternately driven by the internal oscillator and a T flip-flop. The oscillator frequency is set by an external resistor and capacitor. The dead time is controlled by an external RC network. The output voltage error amplifier is configured as a voltage follower to allow remote sensing. The battery current is sensed by a Hall-effect current sensor and fed into the current-sense comparator. The control circuit also includes a soft-start capacitor and a shutdown input. The measured waveforms confirm that the gate drive signals have a 50% duty cycle and a small dead time to avoid simultaneous conduction.
3. Charging Module Analysis and Design
The charging module must convert the 400 V DC bus voltage to a low voltage (up to 4.2 V) with a high current of up to 100 A. The output current ripple must be less than 5% of the rated current, and the efficiency at 100 A should be at least 91%. This is a typical low-voltage high-current DC-DC converter. The key to achieving high efficiency lies in the output rectification stage. Conventional Schottky diodes have a forward voltage drop of about 0.4 V. At 100 A, that would cause 40 W of loss per diode, which is unacceptable. Synchronous rectification using MOSFETs can drastically reduce this loss because the on-resistance of a low-voltage MOSFET can be as low as 2 m\(\Omega\). However, synchronous rectifiers require proper gate drive signals. I chose the current-doubler rectifier topology combined with self-driven synchronous rectification.
3.1 Topology Selection
For the primary side, a half-bridge converter is suitable for input voltages around 400 V. The half-bridge topology uses two power MOSFETs, two input capacitors, and a blocking capacitor. It naturally limits the voltage stress across each switch to the input voltage. For the secondary side, I compared three approaches:
| Topology | Advantages | Disadvantages |
|---|---|---|
| Forward secondary | Simple, fewest components | High filter inductance, higher rectifier loss |
| Center-tapped secondary | Low inductance, low loss | Two secondary windings, complex layout |
| Current-doubler secondary | Single secondary winding, interleaved inductors, low ripple, low loss | Two output inductors |
The current-doubler rectifier is a well-known technique for low-voltage high-current applications. It uses two inductors and two transformers on the secondary side, as shown in Figure 3. The two inductors carry half of the output current each, and their ripple currents are 180° apart, so the output current ripple is substantially reduced. The current-doubler also eliminates the center-tap of the transformer, which simplifies the magnetic design and reduces the AC winding loss.
3.2 Operation of Current-Doubler Synchronous Rectifier
In the current-doubler topology, the transformer secondary voltage is a symmetrical square wave. When the secondary voltage is positive, the upper synchronous rectifier (SR1) is turned on, and the load current flows through SR1 and inductor L1. At the same time, inductor L2 is freewheeling through the body diode of SR2. When the secondary voltage is negative, SR2 is turned on and L2 carries the load current while L1 freewheels. The two output inductor currents are phase-shifted by 180°, so the ripple components tend to cancel at the output.
The steady-state voltage gain of the half-bridge current-doubler converter is given by:
$$ \frac{V_o}{V_{in}} = \frac{2 D}{n} $$
where \(D\) is the duty cycle of each primary switch, and \(n\) is the turns ratio of the transformer (primary-to-secondary). For \(V_{in}=400\) V and \(V_o=4.2\) V, with \(D=0.4\), the required turns ratio is:
$$ n = \frac{2 D V_{in}}{V_o} = \frac{2 \times 0.4 \times 400}{4.2} \approx 76.2 $$
I selected a turns ratio of \(n=75:1\) for the transformer.
3.3 Main Circuit Components
The main circuit of the charging module is shown in Figure 4. The primary side consists of two power MOSFETs (Q1 and Q2), each rated at 500 V and 20 A. The two resistors \(C_{dc1}\) and \(C_{dc2}\) split the input voltage. A blocking capacitor \(C_b\) is placed in series with the primary winding to prevent transformer saturation due to asymmetry. The secondary side uses two synchronous rectifiers (SR1 and SR2) in parallel to lower the total on-resistance. Each SR is a low-voltage MOSFET with a rating of 30 V and 100 A.
The transformer was designed using the same core geometry method. The core chosen is ETD49 as well, but with an air gap to accommodate the half-bridge operation. The primary winding has 75 turns of AWG 28 wire, and the secondary winding has 1 turn of copper foil. Because the secondary carries 100 A, using a single turn of copper foil minimizes the winding resistance and the conduction loss.
3.4 Output Filter Design
In the current-doubler rectifier, the two output inductors are identical. The inductance value is determined by the allowed current ripple and the switching frequency. The ripple current through each inductor is:
$$ \Delta i_{L1} = \frac{V_o (1-2D)}{2 L_1 f_s} $$
For \(D=0.4\), \(V_o=4.2\) V, \(f_s=100\) kHz, and a desired total output ripple of \(\Delta i_o = 2\) A, the individual inductor ripple should be about 1 A. Thus:
$$ L_1 = \frac{V_o (1-2D)}{2 \Delta i_{L1} f_s} = \frac{4.2 \times 0.2}{2 \times 1 \times 100000} = 4.2 \mu\text{H} $$
I used \(L_1 = L_2 = 5 \mu\text{H}\). The output capacitor is selected to suppress voltage spikes. The required capacitance is:
$$ C_o = \frac{\Delta i_{L1}}{8 f_s \Delta V_o} $$
For \(\Delta V_o = 20\) mV, \(C_o \approx 12.5 \mu\text{F}\). I used two 100 \(\mu\text{F}\) ceramic capacitors and four 470 \(\mu\text{F}\) electrolytic capacitors in parallel to reduce the ESR.
3.5 Control and Drive Circuit
The charging module uses a voltage and current dual-loop control. The outer voltage loop regulates the battery voltage, while the inner current loop regulates the charging current. To achieve constant-current/constant-voltage (CC/CV) charging, I implemented a hardware circuit that selects the lower of the two regulator outputs. The control diagram is shown in Figure 5. The error amplifiers are PI controllers. The output of the voltage regulator saturates when the battery voltage is below the preset value, so the current regulator determines the duty cycle. When the battery voltage reaches the preset value, the voltage regulator exits saturation and takes over, gradually reducing the charging current.
The gate drive signals for the primary switches are generated by a PWM controller (UC3825) and delivered through two pulse transformers. The synchronous rectifiers are self-driven from the transformer secondary voltage, as shown in Figure 6. When the secondary voltage is positive, the upper SR is turned on and the lower SR is off. To avoid cross-conduction, the gate signals are shaped with a small delay. The dead time is set by the RC network of the controller.
3.6 Snubber Circuit
To suppress voltage spikes caused by leakage inductance and diode reverse recovery, I added a simple RC snubber across the primary winding. The snubber capacitor is sized by:
$$ C_s = \frac{I_p t_f}{V_{ce}} $$
where \(t_f\) is the fall time of the primary current and \(V_{ce}\) is the desired clamp voltage. With \(I_p = 0.5\) A, \(t_f = 100\) ns, and \(V_{ce} = 400\) V, \(C_s = 125\) pF. A 220 pF capacitor and a 20 k\(\Omega\) resistor were selected. The snubber dissipates about 0.2 W at 100 kHz, which is acceptable.
4. System Implementation and Experimental Results
I built two prototypes: a discharge module and a charging module. Both were tested in the laboratory. The discharge module was loaded with a resistive load (since the final grid-connection inverter was not part of this thesis), while the charging module was used to charge a real electric vehicle battery cell with a nominal capacity of 100 Ah.
4.1 Discharge Module Prototype Tests
The discharge module prototype is shown in Figure 7. The switching frequency was set to 20 kHz. The battery voltage was 3.6 V. The three push-pull converter outputs were series-connected to produce a total output voltage of 400 V. The measured gate drive waveforms, transformer primary voltage, transformer secondary voltage, and battery current are shown in Figure 8. The gate-drive voltages were 12 V, sufficient to fully turn on the MOSFETs. The transformer primary voltage was a square wave with a magnitude equal to the battery voltage, as expected. Some voltage spikes were observed, which were attributed to the leakage inductance. The snubber network reduced these spikes to acceptable levels.
The discharge current was set to 20 A, 50 A, and 100 A. The current waveforms were flat, confirming the constant-current control. The current ripple was measured as less than 2% of the set value. The efficiency was measured by dividing the output power by the input power. At 100 A, the input power was 360 W, and the output power was 331 W, resulting in an efficiency of 92%. At lower currents, the efficiency was slightly lower, mainly because the switching and drive losses become relatively more significant.
4.2 Charging Module Prototype Tests
The charging module prototype is shown in Figure 9. The input was connected to a 400 V DC supply. The output was connected to an electric vehicle battery cell. The charging current was set to 100 A in constant-current mode. The measured waveforms are shown in Figure 10. The transformer primary voltage and secondary voltage correspond to the expected half-bridge operation. The output charging current was stable at 100 A with a ripple of about 1.5 A, which is well below the 5% limit. The efficiency at 100 A was measured as 91%.
The charging module was also tested in constant-voltage mode. When the battery voltage reached 4.2 V, the control circuit automatically reduced the charging current, preventing overcharging. The transition from CC to CV was smooth, and the battery was charged to its full capacity without excessive temperature rise.
4.3 Discussion
The experimental results validate the design approach. The transformer series-parallel topology for the discharge module effectively reduces the current stress on the power switches and allows the use of standard MOSFETs. The current-doubler synchronous rectification topology for the charging module achieves high efficiency by minimizing the conduction loss of the secondary rectifier. Both modules meet the initial design specifications. The residual issues are the voltage spikes caused by leakage inductance, which could be further suppressed by using more advanced snubber circuits or by optimizing the transformer winding layout.
5. Conclusion
In this thesis, I have presented a comprehensive study on the design of high-efficiency formation equipment for high-capacity electric vehicle battery cells. The main contributions are:
- For the discharge module, I proposed a transformer series-parallel topology based on push-pull converters. This topology halves the input current of each converter, reduces the output voltage stress, and provides automatic current sharing without additional control.
- For the charging module, I selected a half-bridge primary and a current-doubler synchronous rectifier secondary. This topology is highly suitable for low-voltage high-current applications because the secondary is a single winding, the two inductors reduce output ripple, and synchronous rectification minimizes losses.
- I provided detailed design procedures for the high-frequency transformers, output filters, power devices, and control circuits. The designs were verified by prototyping and testing.
The discharge module achieves an efficiency of 92% at 100 A, and the charging module achieves 91% at 100 A. These values are significantly higher than those of conventional resistive-formation equipment. The proposed topologies and design methodologies can be directly applied in industrial electric vehicle battery formation systems, resulting in substantial energy savings and reduced production costs.
Future work should focus on integrating the discharge module with a grid-connected inverter to feed the recovered energy back to the AC grid, thereby maximizing energy savings. In addition, soft-switching techniques could be introduced to further improve the efficiency and reduce electromagnetic interference.
