Understanding shunt-based current measurement and other BMS trends.
TTI Inc. has submitted this post.
The rechargeable battery industry has experienced significant growth and is expected to continue to grow into the future. Most of this growth is expected to be propelled by next-generation high voltage energy systems for electric vehicles, and marine and home storage applications that use series-connected battery packs. The most popular batteries for these applications are lithium-ion or nickel metal hydride batteries that require battery management systems (BMS) to monitor and maintain the cells in good condition so as to maximize output power. Analyst firm Markets and Markets confirms the huge expected growth, estimating that the battery management system market will grow from 1.98 billion USD in 2015 to 7.25 billion USD by 2022, at a CAGR of 20.5 percent between 2016 and 2022.
Another important function of a BMS is to help enhance the life expectancy of battery cells and protect them from damage. To achieve maximum efficiency and long battery cell life, the BMS needs to determine the state of charge (SOC) to govern the capacity remaining in the battery, and also to control the rate of charging or discharging.
This article reviews the trends in the BMS market and challenges that designers of BMS face. It focuses on the isolation of communications and transient protection challenges, and introduces isolated sigma delta converters with dynamic ranges less than 200 mV. The attractiveness of shunt-based current measurement for BMS is also reviewed.
Overview of Battery Monitoring Circuits
The image below shows a functional block diagram of a BMS for industrial and automotive applications. The left side of the diagram shows the battery cells, current sensing and data acquisition functions. This area of the BMS design is where one integrated Circuit (IC) or separate discrete ICs perform the SOC measurement, temperature and current measurement.
The trend in automotive electric vehicles and marine applications is for the number of cells in stacks to increase beyond 100S (100 cells) and working voltages to approach 1000 V. Such BMS systems will have several cell voltage monitoring modules networked together using two-wire interfaces like CANbus, Ethernet or SPI. The communications line has galvanic isolation using an Ethernet LAN transformer. The isolation barrier prevents the hazardous voltage from jumping onto low voltage lines, although the level of isolation (basic, supplementary, double, and reinforced) is not specified by any standard and is left to each manufacturer’s safety engineer to evaluate the environment and determine the required safety level.
Each cell has its voltage checked by the monitoring system with typical tolerances of 0.1 percent (to the order of millivolts) to determine the state of charge of each cell. This allows those cells that are overcharged to be bled off (passive balancing), which would prevent other cells with lower capacities from being charged. Active balancing redistributes the charge between over and undercharged cells.
Integrated protection from short transients that many times can occur during the handling and wiring of the cells can be provided by Zener or TVS diodes. However, the junctions in these diodes are not typically designed to handle the amount of energy in a transient created by the sudden disconnection of the battery stack busbar. In the event the high current busbar connecting cells together is disconnected, the BMS IC, which is usually connected to the very last cell on the high end of the break, will suddenly go from reading the voltage on a single cell (1S) to reading the entire stack (which could be 100S or more). This energy may puncture through the protection diodes before control loops can react in time, disconnecting the entire pack. What is needed in this scenario is a high-speed overcurrent protector, such as the Bourns Model TBU-CA, placed in series with every A/D line. The below image shows how this can protect the IC from catastrophic damage.
Reliable Current Measurement Using Shunt Resistors
Shunt-based current measurements are well known in the battery industry for monitoring the battery charge and discharge current. One of the drawbacks of shunt-based measurement is power wasted due to ohmic heating by the current being measured. As a rule of thumb, the maximum current with which shunt-based measurement systems work is 50 amperes, although this is now changing with the advent of lower resistance shunts and high dynamic range, high resolution sigma delta modulator technology. Traditionally, Hall Effect sensors measure currents greater than 50 amperes due to the lower power losses incurred. Yet, Hall Effect sensors typically exhibit a wide variation in zero flux output voltage and sensitivity to overtemperature. Temperature compensation circuit solutions are available but can be expensive. By lowering the resistance of the shunt, the power dissipated can also be reduced which allows for higher currents. Shunts have a superior temperature drift characteristic to Hall Effect allowing for higher accuracy with digital sigma delta modulation. Therefore, new developments in signal processing will make the shunt resistor very attractive for future applications that previously used Hall Effect technology.
Designing the circuit to attach a shunt resistor to a sigma delta modulator is straightforward. A simple filter is needed to remove the effects of harmonics from high frequency switching. The equivalent circuit of a shunt resistor consists of a resistor and an inductor in series. At high frequencies, such as harmonics from switching converters or inverters, the inductance changes the response and the voltage across the element will change accordingly. To ensure that the voltage read across the shunt is due to the current itself and not due to inductance, an RC network is added in parallel. The values R3 and C1 as shown in the image below are selected according to the following equation:
Where XL and R1 are the reactance and the DC resistance of the shunt, respectively.
Variation of voltage across shunt resistor across several frequencies. (Image courtesy of Bourns.)The above image shows the response of a shunt with (red) and without (blue) a compensation network over a frequency sweep of 1 MHz. The compensation keeps the voltage stable over different frequencies while the voltage grows significantly as the frequency increases beyond 20 kHz.
High-Speed Protection of Cell Voltages From High Energy
Battery cell monitoring lines in a stack in high voltage systems are vulnerable to hazardous transients and require ultra-fast overcurrent protection to prevent damage to the internal ESD diodes in the event of a hazardous transient. For example, the Bourns Model TBU-CA085-200 high voltage (850 V) MOSFET device behaves like a resistor until the current reaches its threshold (200 mA) at which point the device will trip. To reset, the voltage across the device must fall below its reset voltage (typically, 15 V). The DC load line characteristics for this device are shown below.
Protection Using Galvanic Isolation
Isolation barriers serve to protect equipment and humans from harmful voltage surges. They also enable communication between a transmitter and a receiver, referenced to very different ground potentials. In the case of battery management ICs connected together in a point-to-point configuration as illustrated below, two-wire communications designed to operate over an isolation barrier are used. Signal transformers such as Ethernet LAN magnetics are ideal for providing the isolation needed and provide the value-added feature of a common mode choke inside for reducing common mode noise.
For battery stack applications with voltage levels higher than 400 V, it is common to specify reinforced (or double) insulation with hi-pot testing of 4 kV or higher. Reinforced or double insulation both require the use of three layers of insulation. Reinforced insulation consists of triple insulated wire on either the primary or the secondary. Double insulation requires one side of the transformer to have double insulated wire (supplementary) while the other side requires wire with one layer (single insulation). Often, designers must make compromises to achieve the desired level of safety while maintaining good levels of signal integrity as well as meeting target costs and form factor size goals.
LAN transformers twist the primary and secondary wires together to minimize leakage inductance. However, as the thickness of the insulation increases, it becomes harder to twist the two together and winding machines experience additional challenges to thread the windings through the toroid. It is often more practical to use double insulation rather than reinforced insulation due to the narrower wires involved. The leakage inductance between the primary and the secondary is proportional to the surface area of the insulation between the coils. Therefore, the thicker the insulation of the wire, the higher the leakage inductance. LAN transformers are also bulkier, not just due to wider cores but also due to their creepage and clearance distances. For example, a LAN transformer with a working voltage of 370 V in a pollution degree 2 environment must have a creepage distance of at least 7.7 mm according to IEC 60950. Consequently, isolated communications with high working voltages using Ethernet magnetics require a tradeoff between safety, size, cost and signal integrity.
Battery monitoring ICs require stable DC bias voltages. This is typically accomplished using an isolated topology such as “push-pull” or “flyback.” By putting an isolation barrier around the power supply, the ground potential difference between the supply ground and the battery IC ground is removed and the risk of the ground potential appearing on signal lines as a common mode voltage is essentially eliminated. Power transformers rely on IEC 61558 for determining the required creepage and clearance distances for the relative working voltage and level of insulation.
