Electric vehicle power design challenges: High-voltage lithium-ion battery pack management system

At the heart of the Chevrolet Volt sedan lies a sophisticated power pack management system that ensures the safety and reliability of the multi-cell lithium-ion battery pack that powers the Volt's powertrain.

The battery monitoring board within this management system uses two key subsystems to reliably monitor battery health and provide digital results to the main processor, which then coordinates the overall operation of the system. Separating these two subsystems reveals a signal interface that ensures good isolation between the high-voltage battery sensing circuitry and the onboard communication devices.

In this teardown report, we review the challenges associated with high-voltage lithium-ion battery pack management in automotive applications and discuss how the overall architecture of the Chevrolet Volt battery pack management system meets these challenges. In particular, we discuss the requirements for lithium-ion battery monitoring, focusing on the architecture and components used in the battery monitoring subsystem, digital communication subsystem, and isolation interface. We also take a detailed look at the parts selected for this design, including a custom ASIC, Freescale's S9S08DZ32, Avago's ACPL-M43T, and Infineon's TLE6250G. Finally, we discuss the benefits of this specific solution for mission-critical battery pack management and weigh possible alternatives that could meet similar design challenges.

To provide more information on the role of isolation in automotive battery management systems, we also provide a series of three in-depth video interviews.

Part 1: Introduction to the role of isolation in automotive battery management systems;

Part 2: Discussion of some considerations when selecting devices for these applications;

Part 3: Explore the use of isolation devices in the Chevrolet Volt battery management system.

Electric Vehicle Challenges

The Chevrolet Volt is one of the first battery-powered electric vehicles (EVs) to be produced, and it can travel nearly 40 miles on batteries alone. When the battery reaches low limits, the gasoline engine can be started to generate additional electricity, extending the vehicle's range to several hundred miles. At the heart of the Volt sedan is a lithium-ion battery pack that is 1.8 meters long, weighs 181 kilograms, and produces 16 kWh of power, enough to start the drive motors, power passenger devices, and power the complex battery management system. The complexity of this management system is comparable to that of an aircraft system.

Robert LeBlanc, IBM senior vice president, noted that the Volt’s software content, at 10 million lines of code, exceeds the 7.5 million lines of code said to fly the U.S. DOD F-35 Lightning II Joint Strike Fighter—a software size that is itself more than three times the size of current jet fighter code, according to the U.S. Government Accountability Office. While LeBlanc may have picked a less controversial system for comparison, the Volt has certainly generated its own fair share of controversy. Perhaps no other car has received as much attention as the Volt. In fact, when a Volt test vehicle caught fire in a test crash after being parked for weeks, the incident immediately attracted the attention of government agencies and prompted a buyback by GM—even though there had been no battery-related fires following “actual crash events,” according to the National Highway Traffic Safety Administration.

Ultimately, the Volt’s success depends on public acceptance—and its functionality. To that end, while designing the Volt, GM worked with IBM to simulate the performance of the “systems of systems” in the Volt. Using detailed models of key systems, the IBM software not only validated the behavior but even generated key portions of the software code used in the Volt’s systems. This approach to code generation and system modeling is critical to ensuring the performance of the Volt battery management system, as ensuring optimal lithium-ion battery performance and life requires complex algorithms; in fact, optimizing the performance of such batteries remains a research topic of great interest to industry, government, and academia. For the Volt, ensuring battery performance enables the final multi-board design (Figure 1) to integrate the work of multiple embedded systems into a single complete system that meets the range, safety, performance, and extended life requirements of the Volt's lithium-ion battery pack.

Figure 1: The Chevrolet Volt battery management system divides all functions into multiple subsystems implemented on multiple PCBs. The focus of this teardown is the battery interface control module - the red, blue, and green boards in the second column from the right in the figure above. (Courtesy of UBM TechInsights)

Lithium-ion Battery Characteristics

The complex system required to meet the performance, safety, and reliability requirements of the Volt is directly related to the characteristics of lithium-ion batteries. When a lithium-ion battery is discharged, the lithium ion is ionized in the (typically) graphite anode, and the lithium ions enter the electrolyte and pass through the separator to the cathode, thereby generating charge flow. The charging process is the opposite, with lithium ions flowing from the cathode into the electrolyte and through the separator back to the anode.

The performance and reliability of this chemical process depends on the temperature and voltage of the battery. At low temperatures, the chemical reaction is slow, thereby reducing the battery voltage. As the temperature rises, the reaction rate increases until the lithium-ion battery components begin to decompose. At temperatures above 100°C, the electrolyte begins to decompose and release gases, which will cause pressure to accumulate in the battery without a pressure relief mechanism. At sufficiently high temperatures, lithium-ion batteries can experience thermal runaway, with the decomposition of oxides and the release of oxygen, which further accelerates the temperature rise.

Keeping lithium-ion batteries in optimal operating condition is therefore a key requirement for the Volt’s battery management system. The problem for Volt engineers is to ensure reliable data collection and analysis to properly monitor and control the state of lithium-ion batteries in the vehicle—a problem made more severe by the characteristics of lithium-ion batteries themselves. One of

the characteristics of our lithium-ion battery technology is that lithium-ion batteries are able to maintain a nearly flat voltage output in the middle of their capacity range for a given temperature and output current value (Figure 2). While this characteristic enhances the advantages of lithium-ion batteries as an energy source, it also complicates engineers’ attempts to provide users with a means of maintaining battery power or state of charge (SOC) using simple cell voltage measurements. For Volt drivers, accurate SOC measurement is key to accurately estimating the vehicle’s remaining driving range. In fact, in the emerging electric vehicle market, “range anxiety” is a key factor hindering the popularity and sales of electric vehicles, so accurate descriptions of SOC are very important.

Figure 2: Li-ion batteries like the Panasonic CGR18650CG have a nearly flat output voltage in the middle of their discharge range for a given temperature and discharge current. This is an advantage for energy sources, but adds design complexity for engineers who need to accurately measure state of charge (SOC). (Courtesy of Panasonic)

Keeping the SOC within a specific range is also important for extending battery life. A battery with a state of charge that is too low or too high will degrade faster than if it is kept in the middle, which is a specific range that is generally determined by experience. If allowed to discharge completely, the performance of the lithium-ion battery components will begin to deteriorate and cause permanent damage. If a lithium-ion battery is allowed to charge above the recommended upper voltage limit, the battery may overheat or suffer permanent deformation of the structure.

In the Volt, GM engineers established a safe SOC window of 58% to 65%, which can be adjusted based on driving mode. The lower limit can be set to 30% SOC in normal driving mode and a higher limit of 45% in "mountain driving" mode to ensure that there is enough charge to go uphill and extend driving time. When the Volt reaches the appropriate lower SOC limit, the car's gasoline engine will be started, thereby extending the driving range.
Because state-of-charge (SOC) measurements of lithium-ion batteries are not very reliable, engineers are left with SOC estimates, which are generally performed using current-based or voltage-based methods.

Current-based methods provide the most accurate results. Such methods track the change in charge, essentially counting the number of coulombs added to the battery during charging or subtracted during the discharge cycle, and then judging the SOC of the battery relative to the fully charged state. However, self-discharge losses or inefficiencies in the battery itself can make the "coulomb counting" method erroneous. In addition, because continuous monitoring is impractical for many applications, the coulomb counting method requires the use of a sampling method. In automotive applications, this method must be fast enough and can automatically track the rapid discharge associated with acceleration and the rapid charging associated with regenerative braking.

Voltage-based methods use the battery's transient voltage output as the basis for further calculations to estimate SOC, which takes into account changes in battery temperature, aging, current output, and discharge rate. When used with accurate characterization data of a single lithium-ion battery under a variety of operating conditions, the voltage method can provide accurate SOC estimates. For production vehicles like the Volt, maintenance requires precise battery characterization and specific tools and procedures that enable the battery management system to learn the capacity of new battery modules—or relearn the battery capacity when necessary.

Lithium-ion Battery Chemistry

Lithium-ion batteries come in a variety of chemistries, each with different characteristics in terms of energy density, efficiency, durability, and nominal cell voltage. The battery manufactured by LG Chem for the Volt uses the company's manganese spinel cathode lithium-ion chemistry and a proprietary safety-enhanced separator—a ceramic-coated semipermeable membrane. Across the industry, lithium-ion batteries are manufactured in a variety of forms, including the familiar cylindrical form; the flat pack used in mobile phones; and the hard plastic prismatic form. The original LG Chem battery used in the Volt uses a prismatic form.

As described by analysts at UBM TechInsights and Munro & Associates, the entire Chevrolet Volt battery pack consists of 288 prismatic lithium-ion cells, which are packaged into 96-cell groups, ultimately providing the 386.6V DC system voltage measured by the analysts. The battery cell groups are combined with temperature sensors and cooling units to form four main battery modules. Voltage sense wires from each battery group terminate at connectors on the top of each battery module, which are then connected by voltage sense harnesses to the battery interface module located on the top of each battery module. There are four color-coded battery interface modules that operate at different locations in the battery pack, corresponding to the low, medium, and high voltage ranges of the DC voltage offset of the four module groups.

Data from the battery interface module is transmitted upward to the battery energy control module. This control module in turn transmits fault conditions, status, and diagnostic information to the hybrid powertrain control module, which acts as the master controller to complete vehicle-level diagnostics. At any given time, the entire system is running more than 500 diagnostics every 0.1 seconds. 85% of these diagnostics are focused on battery pack safety, and the rest are used for battery performance and life.

Multilayer Circuit Board

The subsequent analysis of battery performance began with a focused teardown of the battery interface control module (Figure 3). This module uses a four-layer PCB with most of the components mounted on the top layer, as well as the orange battery connector and black data communication connector. The top layer has a ground plane and some signal traces, some of which are connected to the layers below through multiple vias. In Layer 2, power and ground planes are laid under the high-voltage areas of the PCB. Layer 3 contains signal traces that pass under these areas. The other side of the PCB, Layer 4, is used for the ground plane and signal traces, and contains a few auxiliary components.

Figure 3: The Chevrolet Volt has four battery interface control module PCBs, each of which integrates multiple sensing circuits and CAN communication circuits, isolated by optocouplers located at the edge of the communication subsystem. (Courtesy of UBM TechInsights) The

black ATLPB-21-2AK PCB-mount connector carries the 5V reference, low-voltage reference, signal ground, CAN bus high-speed serial data, CAN bus low-speed serial data, and high-voltage fault signals. The orange battery connector carries the battery module temperature signal, low-voltage reference, and voltage sense lines from the battery cell group.
At the heart of the battery interface control system is a complex sensing subsystem—a complete embedded system circuit responsible for monitoring the output voltage of each lithium-ion battery pack and the temperature of the battery pack. The battery voltage passes through the battery connector to the L9763, an ASIC jointly developed by STMicroelectronics and LG Chem.

The L9763 ASIC can monitor up to 10 independent lithium-ion battery packs, and can monitor the battery-load-current through the on-chip current-sense amplifier and the battery voltage through the on-chip analog multiplexer and sample-and-hold circuit (Figure 4). The device’s differential inputs ensure millivolt-accurate measurements in the presence of large offset voltages, depending on the location of the battery cell in the battery pack. Additionally, PCB designers can use a combination of trace layout techniques, isolation techniques, and the aforementioned ground planes to ensure signal integrity in this challenging environment.

Figure 4: The L9763 ASIC includes on-chip circuitry for measuring the voltage and current of a Volt battery pack and balancing the charge in those cells through a passive resistor cell balancing technique. (Courtesy of STMicroelectronics)

Based on these measurements, the L9763’s on-chip circuitry switches individual battery packs to external resistor networks to selectively discharge the cells, thereby reducing stress caused by large voltage differences. This simple passive technique provides a simple, low-cost solution for cell balancing, but loses efficiency because energy is lost as heat in the discharge resistors (Figure 5). An alternative cell balancing technique uses an active approach that stores the charge from the highest voltage cell and redistributes it to the lowest cells. This technique requires switching between each cell sequentially and using capacitors, inductors, or transformers to store or redistribute the charge. While active approaches offer the advantage of energy savings over passive approaches, they add system cost and complexity.

Figure 5: Passive cell balancing (left) switches the high-voltage cell to a discharge resistor; active cell balancing can sequentially accumulate charge on a capacitor (right) or inductor, or use a transformer to distribute the charge to the lower-voltage cells. (Courtesy of STMicroelectronics)

To charge or discharge a multi-cell Li-ion battery pack, designs typically use a constant-current or constant-voltage approach, where the charging system uses a pair of MOSFETs to reduce the charge current when the desired charge voltage is reached, or increase the current during discharge operations. The L9763 provides a charge pump to drive the power MOSFET devices. The L9763 transmits the measured data of the monitored Li-ion cells to Freescale's S9S08DZ32 MCU via an SPI interface. The L9763 also provides a 5V LDO output to the MCU. For overall battery management functions, the individual L9763 devices are linked via an on-chip interface and are individually addressed by the main control unit via vertical daisy-chain communication.

Sensing Circuit MCU

As mentioned above, SOC estimation of Li-ion batteries is a complex task that requires sufficient processing power. In this design, each sensing subsystem has an L9763 ASIC and a Freescale S9S08DZ32 40-MHz HCS08 MCU with 32kB flash, 2kB RAM, and 1kB E2PROM. An external 4MHz oscillator provides the reference frequency for the MCU clock operation.

In the GM-LG Chem design, the MCU is required to perform the operations required to estimate the SOC based on the voltage and current measurement data provided by the L9763. Although the SOC algorithms are proprietary, the hardware configuration and maintenance procedures suggest that these estimation algorithms can combine voltage-driven estimates using stored battery characterization data with more direct charge measurements for temporary recalibration during the charging process. The use of a detailed system modeling environment described by IBM provides an ideal platform to help find the right data set for optimizing SOC calculations and to help validate the method under a wide range of sampled operating conditions.

Safety features of the HCS08, such as the computer health watchdog timer, help ensure reliable operation and automatically generate a reset signal in the event of an unrecoverable application software failure. Of particular importance in this application is that the S9S08DZ32 has a sophisticated on-chip CAN controller that can be selectively powered down or put into sleep mode when not in use (Figure 6). To help ensure predictable real-time performance, the on-chip controller integrates five receive buffers organized into a FIFO buffer, and three transmit buffers that allow outgoing messages to be prioritized.

Figure 6: The on-chip CAN controller was a key factor in choosing Freescale's S9S08DZ32 MCU to build the battery interface control module detection subsystem. (Courtesy of Fairchild Semiconductor)
In the Chevrolet Volt's system of systems, communication and control are fundamental to the operation of the vehicle, and the Volt provides multiple networks to isolate and protect the various subsystems. The complex algorithms mentioned above are required to manage each lithium-ion battery pack and monitor the battery pack within each detection subsystem on a specific battery interface control module. However, the key data required for the final overall battery management is contained in the CAN bus signal interface and high-voltage fault signals. At the same time, system safety and reliability depend on the safe isolation of the CAN bus network from the high-voltage detection circuit. Although isolation can be achieved using a variety of methods and components, harsh environments and multiple safety regulations make optocouplers the preferred solution for this application.

Optocouplers provide high common-mode noise rejection and are essentially immune to the EMC and EMI associated with electrically noisy environments such as automobiles. In addition, these devices have thick multi-layer insulation, which is very useful when faced with long-term DC voltage stress from the battery pack and fast high-voltage transients that may occur during testing, charger connection/disconnection, and DC/DC conversion.

When selecting these important components, key requirements for automotive applications include the right package and operating voltage specifications. While performance specifications such as speed, data rate, and power consumption remain important, EMI considerations from fast switching times and large transient currents often limit the need for very high-speed devices, and instead increase the need for more flexibility in adjusting slew rate and performance to further limit EMI.

Automotive-Grade Optocouplers

The ACPL-M43T optocoupler from Avago Technologies provides isolation in the Volt vehicle battery interface control module PCB. The M43T, a member of Avago's R2Coupler family, is an automotive-grade, single-channel digital optocoupler in a 5-pin SO-5 Jedec surface-mount package. To enhance isolation, Avago's R2Coupler devices such as the M43T use dual wire bonds to enhance key functional pads (Figure 7). In addition, the use of sealed automotive-grade LEDs demonstrates extended reliability and a wide temperature range, which is much higher than optocouplers based on consumer-grade LEDs. Avago devices targeted for automotive applications are manufactured under ISO/TS16949 quality systems and are certified to AEC-Q100 specifications.

Figure 7: In automotive-grade R2Coupler devices such as the ACPL-M43T optocoupler, Avago reinforces key functional pads (shown in the highlighted area) with dual wire bonds. (Courtesy of Avago Technologies)

The ACPL-M43T is well suited to meet the requirements of the Chevrolet Volt battery pack, including 567V continuous operating voltage, 6000V maximum transient overvoltage, 5mm creepage distance and 5mm spacing. The device's logic high or logic low output has a common-mode transient suppression performance of 30 kV/μsec at a forward input current of 10mA, which can reduce the possibility of transient signals from other automotive subsystems entering the CAN transmission network. The

ACPL-M43T optocoupler's 1M baud rate is sufficient for such designs. In addition, the device uses an open-drain output, which allows designers to adjust the output slew rate to reduce electromagnetic radiation caused by fast switching in downstream components. Downstream fast switching components include CAN transceivers, although the EMI inherent in the CAN physical layer transmission protocol is relatively low.

In the battery interface module PCB, the M43T device is located at the edge of the communication section, isolating the communication section from the high-voltage detection subsystem, which is further shielded by the ground plane in the deeper PCB layer. The isolation interface provides three independent M43T optocouplers for the three lines from each detection circuit - the Freescale S9S08DZ32 CAN Tx output pin, the MCU CAN Rx input pin, and the high-voltage fault signal from the MCU. For example, the output signal of the MCU CAN Tx pin will reach the pin 1 anode of the M43T device through the shielded signal layer in the PCB to power the embedded LED and cause the pin 5 Vo to change state (Figure 8). The isolated signal is then transmitted to the communication output stage circuit of the battery interface module.

Figure 8: Avago ACPL-M43T optocouplers are used to isolate signals between the Freescale S9S08DZ32 MCU and the Infineon CAN transceiver. (Courtesy of Avago Technologies)

CAN Physical Signals

At the end of the communication signal chain, the Infineon TLE6250G CAN transceiver is an AEC-certified IC that provides the CAN physical layer signals between the physical cable and the CAN protocol handler—in this case, the S9S08DZ32 MCU (isolated via optocouplers). Rated for 1M baud CAN transmission, the device handles the conversion between the CAN_H and CAN_L signals on the differential signal lines, as well as the CAN dominant and recessive bits sent and received by the S9S08DZ32. The

8-pin TLE6250G includes Tx, Rx, Vcc, GND, CAN_H, and CAN_L pins, as well as two mode control pins: INH and RM. When the TLE6250G detects that the signal on the Rx pin changes from the CAN idle state to the CAN occupied state, the device swaps CAN_H high and CAN_L low (Figure 9).

This symmetrical change of state can effectively reduce EMI because the electromagnetic radiation caused by the rise of CAN_H is balanced by the opposite direction change of CAN_L.

Figure 9: Symmetrical changes in CAN_H and CAN_L in the CAN physical layer help reduce EMI. (Courtesy of Infineon Technologies)

The TLE6250G device supports three operating modes: normal, standby, and receive-only. When the RM pin is set low, the device operates in receive-only mode, which is useful for diagnostics. When the INH pin is set high, the device enters a low-power standby mode with both transmit and receive turned off.

Next-generation systems

The Chevrolet Volt is certainly one of the most complex distributed embedded system applications to be put into production in the commercial market, and its design is leading the way in many areas. Among the most important systems that will impact the success of the Volt and the adoption of the electric vehicle market, the car's lithium-ion battery and associated battery management system demonstrate the increasing importance of software and circuitry in automotive applications. According to a recently released McKinsey market research report, by 2025, emerging lithium-ion technologies have the potential to increase battery capacity by 80% to 110%, while prices will drop accordingly, making the total cost of ownership of electric vehicles competitive with traditional vehicles powered by internal combustion engines. For engineers, the challenge remains to unlock the full potential of emerging lithium-ion battery systems in the face of higher DC voltages, battery capacities, data rates and consumer expectations.