“Lithium-ion (Li-Ion) batteries are a common energy storage method for electric and hybrid vehicles. The energy density that these batteries can provide is very high among all existing battery technologies, but if performance is to be maximized, a battery monitoring system (BMS) must be used. An advanced BMS not only enables you to extract substantial amounts of charge from your battery pack, but it also manages charge and discharge cycles in a safer manner, resulting in longer lifespan. Analog Devices offers a comprehensive portfolio of BMS devices focused on precision and robust operation.
“
Summary
Lithium-ion (Li-Ion) batteries are a common energy storage method for electric and hybrid vehicles. The energy density that these batteries can provide is very high among all existing battery technologies, but if performance is to be maximized, a battery monitoring system (BMS) must be used. An advanced BMS not only enables you to extract substantial amounts of charge from your battery pack, but it also manages charge and discharge cycles in a safer manner, resulting in longer lifespan. Analog Devices offers a comprehensive portfolio of BMS devices focused on precision and robust operation.
Accurately measuring a battery’s state of charge (SOC) can extend battery runtime or reduce weight. Precise and stable devices require no factory calibration after PCB assembly. Long-term stability improves security and avoids warranty issues. Self-diagnostics help achieve the appropriate Automotive Safety Integrity Level (ASIL). The battery pack is an electromagnetic interference (EMI) challenging environment, so special care is needed when designing the data communication link to ensure robust and reliable communication between the measurement chip and the system controller. Cables and connectors are a major cause of battery system failure, so this article describes wireless solutions. The wireless communication design increases reliability and reduces overall system weight, which in turn increases driving range per charge.
Introduction
The energy storage unit must be able to provide a large capacity and release energy in a controlled manner. If not properly controlled, the storage and release of energy can lead to catastrophic battery failure and, ultimately, a fire. Batteries can fail for a variety of reasons, most of which are related to improper use. Failures can arise from mechanical stress or damage, as well as electrical overloads in the form of deep discharge, overcharge, overcurrent and thermal overstress. To maximize efficiency and safety, a battery monitoring system is essential.
The primary function of a BMS is to keep all cells in the pack within their Safe Operating Area (SOA) by monitoring the following physical quantities: pack charge and discharge current, cell voltage, and pack temperature. Based on these values, not only can the battery operate safely, but also SOC and state of health (SOH) calculations can be made.
Another important function provided by BMS is cell balancing. In a battery pack, individual cells can be placed in parallel or in series to achieve the desired capacity and operating voltage (up to 1 kV or more). Battery makers try to supply battery packs with the same cells, but that’s not physically realistic. Even small differences can result in different charge or discharge levels, and the weakest cells in a pack can seriously affect the pack’s overall performance. Precise cell balancing is an important feature of a BMS that ensures that the battery system operates safely at its maximum capacity.
BMS Architecture
Electric vehicle batteries consist of several cells connected in series. A typical battery pack (with 96 cells in series) will produce over 400 V total when charged at 4.2 V. The more cells in the battery pack, the higher the voltage achieved. The charge and discharge currents are the same for all batteries, but the voltage on each cell must be monitored. To accommodate the large number of batteries required by high-power automotive systems, the multi-cell batteries are often divided into several modules and distributed throughout the available space of the vehicle. Typical modules have 10 to 24 cells and can be assembled in different configurations to suit multiple vehicle platforms. The modular design can serve as the basis for larger battery packs. It allows the battery components to be placed in a larger area, allowing for a more efficient use of space.
Analog Devices has developed a family of battery monitors capable of measuring up to 18 batteries connected in series. The AD7284 can measure 8 cells, the LTC6811 can measure 12 cells, and the LTC6813 can measure 18 cells. Figure 1 shows a typical battery pack with 96 cells divided into 8 modules of 12 cells each. In this example, the battery monitor IC is the LTC6811, which measures 12 cells. The IC has a battery measurement range of 0 V to 5 V and is suitable for most battery chemistry applications. Multiple devices can be connected in series for simultaneous monitoring of long high voltage battery packs. The device includes passive balancing for each cell. Data is exchanged across the isolation barrier and compiled by the system controller, which is responsible for calculating SOC, controlling cell balancing, checking SOH, and keeping the overall system within safe limits.
Figure 1. Battery pack architecture with 96 cells using the LTC6811 12-channel measurement IC.
Robust communication systems are essential to support distributed modular topologies in the high EMI environment of EVs/HEVs. Both the isolated CAN bus and ADI’s isoSPI™ provide proven solutions for interconnecting modules in this environment. 1 Although the CAN bus provides a complete network for interconnecting battery modules in automotive applications, it requires many additional components. For example, implementing an isolated CAN bus through the LTC6811’s isoSPI interface requires the addition of a CAN transceiver, a microprocessor, and an isolator. The main disadvantage of the CAN bus is that these extra components add cost and board space. Figure 2 shows a possible CAN-based architecture. In this example, all modules are connected in parallel.
ADI’s innovative two-wire isoSPI interface is an alternative to the CAN bus interface. The 1isoSPI interface is integrated into each LTC6811, using a simple transformer and a simple twisted pair instead of the four required for CAN bus. The isoSPI interface provides a noise-immune interface (for high-level RF signals) that allows modules to be daisy-chained over long cables and operate at data rates up to 1 Mbps. Figure 3 shows an architecture based on isoSPI and using a CAN module as a gateway.
Both architectures shown in Figures 2 and 3 have pros and cons. CAN modules are standardized modules that can share the same bus operation with other CAN subsystems; the isoSPI interface is a proprietary interface that can only communicate with devices of the same type. On the other hand, the isoSPI module does not require additional transceivers and MCUs to handle the software stack, making the solution more compact and easier to use. Both architectures require wired connections, which has a distinct disadvantage in modern BMSs, where routing wires to different modules can be a tricky problem, adding weight and complexity. Wires also tend to absorb noise, requiring additional filtering.
Wireless BMS
Wireless BMS is a novel architecture that eliminates communication wiring. 1 In wireless BMS, the interconnection of each module is realized by wireless connection. The advantages of wireless connection of large multi-cell battery packs are:
Less complex wiring
lighter weight
lower cost
lower cost
Greater safety and reliability
Wireless communication is a challenge due to harsh EMI environments and signal propagation barriers posed by RF shielding metals.
ADI’s SmartMesh® embedded wireless network is field-proven in Industrial Internet of Things (IoT) applications and provides redundancy in excess of 99.999% in industrial, automotive and other harsh environments through the use of path and frequency diversity for redundancy Connection.
In addition to improving reliability by creating multiple redundant connection points, wireless mesh networks extend the capabilities of the BMS. SmartMesh wireless network enables flexible placement of battery modules and improves battery SOC and SOH calculations. This is because more data can be collected from sensors installed in places that were previously unsuitable for wiring. SmartMesh also provides time-correlated measurements from each node, enabling more precise data collection. Figure 4 shows a comparison of wired and wirelessly interconnected battery modules.
ADI demonstrated the industry’s first wireless automotive BMS concept, incorporating the LTC6811 battery pack monitor and ADI SmartMesh networking technology in a BMW i3.2 model. This is a major breakthrough that promises to improve the reliability of large multi-cell battery packs for EVs/HEVs and reduce cost, weight and wiring complexity.
Figure 2. Independent CAN modules in parallel.
Figure 3. Series connection of modules with CAN gateway.
The importance of accurate measurements
Accuracy is an important characteristic of BMS and is crucial for LiFePO4 batteries. 3,4 To understand the importance of this feature, we consider the example in Figure 5. To prevent overcharging and discharging, battery cells should be kept between 10% and 90% of full capacity. In the 85 kWh battery, the capacity available for normal driving is just 67.4 kWh. If the measurement error is 5%, the battery capacity must be maintained between 15% and 85% in order to continue safe battery operation. Total usable capacity has been reduced from 80% to 70%. If the accuracy is increased to 1% (for LiFePO4 cells, a measurement error of 1 mV is equivalent to a 1% SOC error), the cells can now operate between 11% and 89% of full capacity, an 8% increase. With the same battery and a more accurate BMS, the car’s range per charge can be increased.
Circuit designers estimate the accuracy of battery measurement circuits based on the specifications in the data sheet. Other real-world effects often dominate the measurement error. Factors that affect measurement accuracy include:
Initial tolerance
temperature drift
long-term drift
humidity
PCB Assembly Stress
noise suppression
Figure 4. Comparison of battery monitoring interconnects.
Figure 5. Battery charge limits.
A well-established technology must take all of these factors into consideration in order to provide very good performance. The measurement accuracy of the IC is mainly limited by the reference voltage. The reference voltage is sensitive to mechanical stress. Thermal cycling during PCB soldering creates silicon stress. Humidity is another cause of silicon stress because the package absorbs moisture. Silicon stress relaxes over time, causing long-term drift in the reference voltage.
Battery measurement ICs use either bandgap or zener references. IC designers use the NPN emitter-base junction at reverse breakdown as a Zener diode reference. Breakdown occurs at the chip surface, where contamination and oxide charge are most effective. These junctions are noisy and have unpredictable short- and long-term drift. Buried Zener diodes place the junction below the silicon surface, away from the effects of contaminants and oxide layers. The result is a Zener diode with excellent long-term stability, low noise, and relatively accurate initial tolerances. Therefore, Zener diode references excel in mitigating time-varying real-world effects.
The LTC68xx family uses a laboratory-grade Zener diode reference, a technology that ADI has perfected over 30 years. Figure 6 shows the battery measurement IC error drift with temperature for five typical cells. Drift is less than 1 mV over the entire automotive temperature range of -40°C to +125°C.
Figure 7 compares the long-term drift of a bandgap reference IC and a buried Zener diode reference IC. The error of the initial measurement is calibrated to 0 mV. The ten-year measurement drift was predicted by the drift after 3000 hours at 30°C. The picture clearly shows that the Zener diode reference has better stability over time, at least 5 times better than the bandgap reference. Similar humidity and PCB assembly stress tests show that buried Zener diodes outperform bandgap references.
Figure 6. LTC6811 measurement error versus temperature.
Figure 7. Long-term drift comparison between a buried Zener diode and a bandgap reference.
Figure 8. Programmable range and frequency response of the ADC filter.
Another limiting factor for accuracy is noise. Automotive batteries are a very harsh environment for electronics due to the electromagnetic interference generated by electric motors, power inverters, DC-DC converters and other high current switching systems in EV/HEV. The BMS needs to be able to provide a high level of noise rejection in order to maintain accuracy. Filtering is a classic method for reducing unwanted noise, but it requires a trade-off between noise reduction and conversion speed. Since the battery voltage that needs to be converted and transferred is high, the conversion time cannot be too long. A SAR converter might be ideal, but in a multiplexed system, the speed is limited by the settling time of the multiplexed signal. At this point, the sigma-delta converter becomes a valid alternative.
ADI’s measurement ICs use sigma-delta analog-to-digital converters (ADCs). With a sigma-delta ADC, the input is sampled multiple times during the conversion process and then averaged. The result is built-in low-pass filtering, which removes noise as a source of measurement error; the cutoff frequency is determined by the sampling rate. The LTC6811 employs a third-order sigma-delta ADC with programmable sampling rate and eight selectable cutoff frequencies. Figure 8 shows the filter response for eight programmable cutoff frequencies. Excellent noise reduction is achieved by quickly completing measurements in 290 µs for all 12 cells. The high-current injection test, which coupled 100 mA of RF noise into the wires connecting the battery to the IC, showed a measurement error of less than 3 mV.
Cell balancing to optimize battery capacity
Even when batteries can be manufactured and selected with precision, they can show subtle differences. Any capacity mismatch between cells will result in a reduction in the overall capacity of the battery pack.
To better understand this, let’s consider an example where the cells are held between 10% and 90% of full capacity. Deep discharge or overcharge can greatly reduce the effective life of the battery. Therefore, BMS provides undervoltage protection (UVP) and overvoltage protection (OVP) circuits to help prevent these conditions. The charging process stops when the battery with the lowest capacity reaches the OVP threshold. In this case, the other batteries are not fully charged and the battery energy storage does not reach the maximum allowable capacity. Likewise, the system stops working when the lowest-charged battery reaches the UVP limit. Also, there is still energy in the battery pack to power the system, but for safety reasons the battery pack cannot continue to be used.
Obviously, the weakest cell in the pack dominates the overall pack performance. Cell balancing is a technique that helps overcome this problem by equalizing the voltage and SOC between cells when they are fully charged. 5 There are two cell balancing techniques: passive and active.
With passive balancing, if one cell is overcharged, the excess charge is dissipated into the resistor. Typically, a shunt circuit is used, which consists of a resistor and a power MOSFET that acts as a switch. When the battery is overcharged, the MOSFET turns off, dissipating the excess energy into the resistor. The LTC6811 uses a built-in MOSFET to control the charge current of each cell to balance each cell being monitored. Built-in MOSFETs allow for compact designs and the ability to meet current requirements of 60 mA. For higher charge currents, an external MOSFET can be used. The device also provides a timer to adjust the equilibration time.
The advantages of dissipative techniques are low cost and low complexity. Disadvantages are high energy losses and more complex thermal design. Active balancing, on the other hand, redistributes excess energy among the other cells of the module. In this way, energy can be recovered and the heat generated is lower. The disadvantage of this technique is that the hardware design is more complex.
Figure 9. Battery pack module with actively balanced 12 cells.
Figure 9 shows active balancing using the LT8584. This architecture solves the problem of passive shunt balancers by actively shunting the charging current and returning the energy to the battery pack. The energy is not lost in the form of heat, but is reused to charge the remaining cells in the battery pack. The device’s architecture also addresses the issue of reduced run time when one or more cells in the pack reach a lower safe voltage threshold before the entire pack capacity is exhausted. Only active balancing can redistribute charge from strong cells to weak cells. This allows the weak battery to continue powering the load, allowing a higher percentage of energy to be extracted from the battery pack. The flyback topology allows charge to travel back and forth between any two points within the battery pack. Most applications return charge to battery modules (12 cells or more), others return charge to the entire battery pack, and still others return charge to auxiliary power rails.
in conclusion
The key to low-emission vehicles is electrification, but also intelligent management of energy (lithium-ion batteries). If not managed properly, battery packs can become unreliable, greatly reducing the safety of the car. High precision helps improve battery performance and lifespan. Active and passive cell balancing enables safe and efficient battery management. Distributed battery modules are easy to support, and robust data delivery to the BMS controller (whether wired or wireless) enables reliable SOC and SOH calculations.