Rechargeable batteries are increasingly being used to deliver higher voltages and more power in applications such as electric vehicles (EVs) and hybrid-electric vehicles (HEVs), power tools, lawn equipment, and uninterruptible power supplies. While it’s well known that chemistries of all kinds need careful monitoring and management to ensure effective, reliable, and safe operation, the series-connected stacks of many tens of cells or more that are required to meet the power demands of these devices require more attention from designers, particularly as the number of cells per battery increases.
Monitoring and measuring a single cell or a small battery pack with just a few cells is a modest challenge and is far simpler than doing the same for cells in a multicell series string. Designers of stacked, multi-cell implementations need to consider issues such as performing measurements despite high common-mode voltage, presence of hazardous voltages, implications of single-cell failure, multiplexing across a large number of cells, cell mismatch and balancing, and battery-stack temperature differentials, to cite just a few. These require advanced battery management ICs (BMICs) and battery management systems (BMS) to perform parametric measurement and control, and some engineering know-how in order to use them correctly.
This article discusses the basics and challenges of battery management in general, and multi-cell batteries in particular. It then introduces and shows how to apply BMICs from Analog Devices, Renesas Electronics Corp., and Texas Instruments that are specifically designed for the unique issues of managing series-connected strings of cells.
Battery series strings pose unique challenges
Typical battery monitoring involves measuring current flow into and out of the battery (fuel gauging), monitoring terminal voltage, assessing battery capacity, monitoring cell temperatures, and managing charge/discharge cycles to optimize energy storage and maximize the number of such cycles over a battery’s lifetime. Widely used BMICs or BMS’ provide these functions for small battery packs consisting of just one or two cells with single-digit voltages. The BMIC or BMS acts as a data acquisition front-end, with its data reported to a cell management controller (CMC); in more complex systems, the CMC connects to a higher-order function called the battery management controller (BMC).
For the purposes of this article, a “cell” is an individual energy storage unit, while a “battery” is the entire power pack, comprising multiple cells in a series/parallel combination. While an individual cell produces only a few volts, a battery pack can be built up of dozens or more cells and deliver many tens of volts, and combinations of battery packs go even higher.
For effective management, the critical cell parameters to be measured are terminal voltage, charge/discharge current, and temperature. The measurement performance needed for modern battery packs is fairly high: each cell must be measured to within a few millivolts (mV) and milliamps (mA), and to about a degree centigrade (°C). The reasons for such close cell monitoring include:
- Determining battery pack state-of-charge (SOC) and state-of-health (SOH) in order to provide accurate predictions of remaining battery pack capacity (run time) and overall life expectancy.
- Providing the data needed to implement cell balancing, which equalizes the voltage of charged cells with respect to each other, despite their internal differences, as well as different locations, temperatures, and aging. Failure to perform cell balancing results in reduction in battery pack performance at best, and cell failure at worst. Balancing can be accomplished using passive or active techniques; the latter provides somewhat better results but is more costly and complex.
- Preventing many conditions that can damage the battery and lead to safety concerns for the user (such as a vehicle and its occupants). These include undesired scenarios such as:
- Overvoltage or charging at excessive currents, which can lead to thermal runaway.
- Undervoltage: a single over-discharge won’t cause catastrophic failure, but it may start to dissolve the anode conductor. Subsequent repeated over-discharge cycles can lead to lithium plating in the recharging cell and, again, potential thermal runaway.
- Overtemperature affects the cell electrolyte material, reducing the SOC; this can also increase solid-electrolyte interphase (SEI) formation, resulting in increased and non-uniform resistivity and power loss.
- Under temperature is also a problem as it can cause deposition of lithium, which also results in capacity loss.
- Overcurrent, and resultant internal heating due to uneven internal impedance and eventual thermal runaway; this can increase the SEI layers in the battery and increase resistivity.
There’s a conundrum here as, for example, it is fairly straightforward to accurately measure the voltage of an individual cell at the test bench or other benign setting. A designer just needs to connect a floating (non-grounded) or battery-powered digital voltmeter (DVM) across the cell of interest (Figure 1).
Figure 1: Measuring the voltage across any single cell of a series string is simple in concept, requiring only a floating digital voltmeter. (Image source: Bill Schweber)
However, it is far more difficult for many reasons to do so with confidence and safety in an electrically and environmentally harsh situation such as in an EV or HEV. This is made clear by a representative EV power pack example comprising 6720 Li+ cells, managed by eight control modules (Figure 2).
Figure 2: A real-world battery pack is an array of series and parallel connected cells in modules, with a significant amount of stored energy; these are factors that greatly complicate the task of measuring cell voltages. (Image source: Analog Devices)
Each cell has a capacity of 3.54 ampere-hours (Ah), resulting in a total nominal energy storage of 100 kilowatt-hours (kWh) (3.54 Ah x 4.2 volts x 6720 cells). Each of the 96 series-connected rows is made up of 70 cells in parallel, for a battery voltage of 403.2 volts (96 rows × 4.2 volts), with a capacity of 248 Ah (100 kWh/403.2 volts or 3.54 Ah × 70 columns).
Among the issues are:
- It is a challenge to provide the needed resolution and accuracy when measuring a low, single-digit voltage to get meaningful precision at several millivolts due to the presence of a high common-mode voltage (CMV), which can overload the measurement system or affect reading validity. This CMV is the sum of the voltages of all the series-connected cells, up to the one being measured, with respect to system common (also referred to as “ground” although that is a misnomer). Note that in an EV, there can be up to as many as 96 or even 128 battery cells in series, yielding a CMV in the hundreds of volts.
- Due to the high CMV, it is necessary to galvanically isolate the cells from the rest of the system for both electrical integrity and user/system safety since neither should potentially be exposed to the full CMV.
- Electrical noise and surges can easily corrupt the millivolt-range reading.
- The multiple cells must be measured nearly simultaneously within a few milliseconds to create an accurate overall picture of the cells and battery pack status. Otherwise, time skew between the cell measurements can result in misleading conclusions and resultant actions.
- The large number of cells means that some sort of multiplexing arrangement is needed between the cells and the rest of the data-acquisition subsystem or else the size, weight, and cost of the interconnection wiring becomes prohibitive.
Finally, there are significant and mandatory considerations related to safety, redundancy, and error reporting that must be satisfied. The standards differ from industry to industry; industrial and power tools are very different than autos, and those for the latter are the most stringent. In mission-critical automotive systems such as those related to battery management, a loss of functionality must not lead to a hazardous situation. In the case of a malfunction within the system, the “safe” state requires that the electronics be switched off and the vehicle driver must be alerted via a dashboard light or other indicator.
For some systems, however, a malfunction or the loss of functionality can potentially lead to a hazardous event and cannot simply be switched off, so safety goals may include a defined “safety-related availability” requirement. In such cases, tolerance for some types of faults in the system may be required to avoid hazardous events.
Such safety-related availability requires the provision of basic functionality or a defined “exit” path for a specified time period—despite the defined fault conditions—and the safety system must tolerate a fault for that time period. This fault-tolerance enables the system to continue functioning longer with an acceptable level of safety. Key sections of ISO 26262 “Functional Safety for Road vehicles” provide guidance for system developers regarding safety-related availability requirements.
ICs step up to provide solutions
Vendors have developed BMS ICs that are designed to solve the problem of reading a single cell in a series string with accuracy—despite high CMV and the harsh electrical environment. These ICs not only provide the basic readings but also address multiplexing, isolation, and timing skew technical issues. They meet the relevant safety standards and, if appropriate, are rated for ASIL-D approval for automotive applications, which is the highest and most stringent level.
Automotive Safety Integrity Level (ASIL) is a risk classification scheme defined by the ISO 26262 – Functional Safety for Road Vehicles standard. This is an adaptation of the Safety Integrity Level (SIL) used in IEC 61508 for the automotive industry.
Although the “broad-brush” functions of these BMS devices are similar, they differ to some extent in architecture, number of cells they can handle, scan speed, resolution, unique features, and interconnection approach:
•The isolated CAN architecture is based on a star configuration and is robust, as a break in the communications wire in the isolated CAN architecture disrupts only one IC, while the rest of the battery pack remains safe. However, the CAN architecture requires a microprocessor and CAN for each IC, making this approach more costly, while providing relatively slow communication speeds.
•The daisy-chain architecture is generally more cost-effective, as its universal asynchronous receiver/transmitter (UART)-based daisy chain can deliver reliable and fast communication without the complexity of CAN. It most often uses capacitive isolation, but may also support transformer-based isolation. However, a wire break in the daisy-chain architecture can disrupt communication, so some such daisy-chain systems offer “workarounds” and support some operation during the wire break.
Among the representative BMS ICs are:
• MAX17843 BMS from Analog Devices: The MAX17843 is a programmable, 12-channel battery-monitoring data-acquisition interface with extensive safety features (Figure 3). It is optimized for use with batteries for automotive systems, HEV battery packs, EVs, and any system that stacks long series strings of secondary metal batteries up to 48 volts.
Figure 3: The MAX17843 12-channel battery-monitoring data-acquisition interface incorporates multiple safety features, making it suitable for automotive applications and mandates. (Image source: Analog Devices)
The MAX17843 incorporates a high-speed differential UART bus for robust daisy-chained serial communication, supporting up to 32 ICs connected in a single daisy-chain (Figure 4). The UART uses capacitive isolation which not only reduces the bill of materials (BOM) cost but also improves failure in time (FIT) rates.
Figure 4: The 12-channel MAX17843 uses capacitive galvanic isolation in its daisy-chain UART configuration, supporting up to 32 devices in a single chain. (Image source: Analog Devices)
The analog front-end combines a 12-channel voltage-measurement data-acquisition system with a high-voltage switch-bank input. All measurements are done differentially across each cell. The full-scale measurement range is from 0 to 5.0 volts, with a usable range of 0.2 to 4.8 volts. A high-speed successive approximation (SAR) analog-to-digital converter (ADC) is used to digitize the cell voltages at 14-bit resolution with oversampling. All twelve cells can be measured in under 142 microseconds (μs).
The MAX17843 uses a two-scan approach for collecting cell measurements and correcting them for errors, which yields excellent accuracy over the operating temperature range. Accuracy of the cell differential measurement is specified at ±2 millivolts (mV) at +25°C and 3.6 volts. To facilitate design-in with this IC, Analog Devices offers the MAX17843EVKIT# evaluation kit with a PC-based graphical user interface (GUI) for set-up, configuration, and assessment.
• ISL78714ANZ-T from Renesas: The ISL78714 Li-ion BMS IC supervises up to 14 series-connected cells and provides accurate cell voltage and temperature monitoring, cell balancing, and extensive system diagnostics. In a typical configuration, a master ISL78714 communicates to a host microcontroller through a serial peripheral interface (SPI) port, and up to 29 additional ISL78714 devices connected together by a robust, proprietary two-wire daisy chain (Figure 5). This communication system is highly flexible and can use capacitor isolation, transformer isolation, or a combination of both at up to 1 megabits per second (Mbits/s).
Figure 5: The ISL78714 uses an SPI port to link multiple devices in a two-wire daisy chain that can use either capacitive or transformer-based isolation. (Image source: Renesas Electronics Corp.)
Initial voltage measurement accuracy is ±2 mV with 14-bit resolution over a range of 1.65 to 4.28 volts from 20°C to +85°C; post-board assembly device accuracy is a tight ±2.5 mV over a cell input range of ±5.0 volts (the negative voltage range is often needed for bus bars).
This BMS includes three cell-balancing modes: manual-balance mode, timed-balance mode, and auto-balance mode. Auto-balance mode terminates balancing after a host-specified amount of charge has been removed from every cell. Among the integrated system diagnostics for all key functions is a watchdog shutdown device if communication is lost.
• BQ76PL455APFCR (and BQ79616PAPRQ1) from Texas Instruments: The bq76PL455A is an integrated 16-cell battery monitoring and protection device designed for high-reliability, high-voltage industrial applications. The integrated high-speed, differential, capacitor-isolated interface supports up to sixteen bq76PL455A devices, communicating with a host through a single high-speed UART interface via a daisy-chain with twisted-pair cabling at up to 1 Mbits/s (Figure 6).
Figure 6: The bq76PL455A 16-cell battery management IC targets industrial applications, using capacitive isolation to link up to 16 devices with twisted-pair cabling communicating at up to 1 Mbits/s via a daisy-chain arrangement. (Image source: Texas Instruments)
The 14-bit ADC uses an internal reference with all cell outputs converted in 2.4 milliseconds (ms). The bq76PL455A monitors and detects several different fault conditions including overvoltage, undervoltage, overtemperature, and communication faults. It supports passive cell balancing with external n-FETs, as well as active balancing via external switch-matrix gate drivers.
This BMS easily handles strings with fewer than the maximum of 16 cells. The only restriction when doing so is that the inputs must be used in ascending order, with all unused inputs connected together with the input to the highest used VSENSE_ input. For example, in a 13-cell design, inputs VSENSE14, VSENSE15, and VSENSE16 are not used (Figure 7).
Figure 7: The bq76PL455A can be used with fewer than 16 cells; in such cases, the unused cell inputs must be the highest ones in the chain. (Image source: Texas Instruments)
Other ICs, such as the Texas Instruments bq79616PAPRQ1, include support for ring configuration and bidirectional communication, enabling the system to continue monitoring the state of health and safety of the battery pack (Figure 8).
Figure 8: The bq79616PAPRQ1 supports a bidirectional ring topology for an additional link connectivity path in the case of a wire break or node failure. (Image source: Texas Instruments)
If there is a fault, open, or short between two of the battery monitoring ASICs in this configuration, the control processor will be able to continue communicating with all of the battery monitoring ASICs by switching the direction of messaging backward and forward. Thus, if normal communication encounters a fault, the system can maintain availability using the fault tolerance of the ring communication feature, and do so with no loss of voltage and temperature information from the battery modules. For designers looking to experiment with the bq79616PAPRQ1, Texas Instruments provides the BQ79616EVM evaluation board.
• LTC6813-1 from Analog Devices, Inc.: The LTC6813-1 is an automotive qualified, multicell battery-stack monitor that measures up to 18 series-connected battery cells, with a total measurement error of less than 2.2 mV via its 16-bit delta-sigma ADC with programmable noise filter (Figure 9). Note that this is a higher number of cells than some of the other ICs can support directly. All 18 cells can be measured in under 290 microseconds (μs), and lower data acquisition rates can be selected for higher noise reduction.
Figure 9: The LTC6813-1 supports the highest number of cells (18) and uses a 16-bit ADC to achieve 2.2 mV accuracy and high-speed cell scanning. (Image source: Analog Devices, Inc.)
Multiple LTC6813-1 devices can be connected in series, thus permitting simultaneous cell monitoring of long, high-voltage battery strings. The LTC6813-1 supports two types of serial ports: a standard four-wire SPI and a 2-wire isolated interface (isoSPI). The non-isolated four-wire port is suitable for shorter distance links and some non-automotive applications (Figure 10).
Figure 10: The LTC6813-1 supports a standard four-wire SPI interconnection for shorter distance links and some non-automotive applications. (Image source: Analog Devices, Inc.)
The 1 Mbit/s isolated serial communications port uses a single twisted pair for distances of up to 100 meters (m) with low electromagnetic interference (EMI) susceptibility and emissions, as the interface is designed for low packet error rates even when the cabling is subjected to high RF fields. This daisy chain’s bidirectional capability ensures communication integrity even in the event of a fault, such as a broken wire along the communication path.
In its two-wire configuration mode, isolation is achieved through an external transformer, with standard SPI signals encoded into differential pulses. The strength of the transmission pulse and the threshold level of the receiver are set by two external resistors, RB1 and RB2 (Figure 11). The values of the resistors are chosen by the designer to allow a trade-off between power dissipation and noise immunity.
Figure 11: The LTC6813-1 also offers a 2-wire, 1 Mbit/s, transformer-isolated serial communications port via a single twisted pair for distances of up to 100 m, with both low EMI susceptibility and emissions. (Image source: Analog Devices, Inc.)
The LTC6813-1 can be powered directly from the battery stack it is monitoring or from a separate isolated supply. It also includes passive balancing for each cell, along with individual duty-cycle control using pulse-width modulation (PWM).
Accurate measurement of the voltage, current, and temperature of a single cell or small battery pack with only a few cells is a modest technical challenge. However, accurately measuring these same parameters on individual cells in a series string—and doing so in harsh automotive and industrial settings with negligible cell-to-cell time skew—is a challenge due to the large number of cells, high CMV, electrical noise, regulatory mandates, and other issues.
As shown, designers can turn to ICs specifically designed for these applications. They support the required galvanic isolation, precision, and fast scan time to address the problems. As a result, they deliver accurate, actionable results that enable critical, high-level battery management decisions.