22 November 2011
EV battery management essential for desired lifetime and charge cycles
For hybrid (HEV) and full electric vehicles (EVs), Li-Ion batteries offer the best trade off of power, energy density, efficiency and environmental impact. But Li-Ion batteries can be delicate and dangerous, while automobiles can be cruel and unforgiving. The challenge is to bridge the gap between the two environments.
Operating large Li-Ion automotive battery packs safely and reliably is not simple. The capacity of Li-Ion cells will degrade when operated to a full state of charge or discharge, while charge cycling, lot to lot differences and different environmental conditions mean battery cell capaci¬ties diminish and diverge over time. Thus, to meet a 15 year, 5000 charge cycle goal for a battery pack, each cell must be kept within a limited operating range. By controlling each cell's state of charge (SOC), a battery pack's capacity can be maximised and degradation minimised. The battery management system (BMS) ensures the safe and efficient use of the vehicle's battery pack.
The BMS tracks and controls the SOC of each cell – discharging cells or moving charge such that cell voltages are matched and within a specific range. Measurement accuracy is critical; it determines how close each cell can be operated to the edge of its reliable SOC range. The ability to maximise usable capacity determines the number of battery cells needed, with cost and weight implications.
Measuring the voltage of every cell accurately is difficult because battery pack cells are subjected to high common mode voltages and high frequency noise. To understand, consider that EV/HEV battery packs typically comprise a group of 100 to 200 cells connected in series. These packs must deliver charge and discharge currents that can exceed 200A, with voltage transients potentially exceeding 100V at the top of the stack.
Driven by cost and reliability concerns, automotive electronics is moving to higher levels of integration and reduced component count. This is especially true with BMS electronics, where highly integrated devices are emerging as the key data acquisition component.
The battery monitor's primary function is the direct measurement of series connected battery potentials, typically 12 channels per IC. Other functions include cell balancing control and additional measurements, such as temperature. To handle a high voltage stack, these devices typically communicate via a daisy chained serial interface. One element of the BMS, typically not integrated, is embedded software. The SOC algorithm is highly guarded, specific to the chemistry, size, form factor, operating conditions and application. Figure 1 illustrates the basic configuration of an arbitrary cell count battery module, where the BMS algorithm is software coded and controlled by the developer.
A critical consideration is how battery montoring ICs handle noise. For example, many battery monitors use a fast successive approximation register (SAR) converter for cell digitisation. While this might seem advantageous in a data acquisition system with more than 100 channels, the noise environment requires significant filtering and this determines effective throughput, not sampling rate. For this reason, a Delta-Sigma (DS) ADC offers better performance. For a given amount of 10kHz noise rejection, a 1ksample/s DS a/d converter provides a throughput equivalent to a 1Msample/s SAR a/d converter.
For example, the LTC6802 uses a 1ksample/s DS a/d converter that can sequence through 10 input channels in 10ms. A built in linear phase digital filter provides 36dB of rejection to 10kHz switching noise. By contrast, a 1Msample/s SAR converter requires a single pole RC filter on each cell, with a corner frequency of 160Hz, to get the same noise rejection at 10kHz. The 12bit settling time of the RC filter is 8.4ms so, even though a SAR can sequence through 10 channels in 10ms, scanning more than once every 8.4ms is pointless because of the filter's response.
Given the long chain of battery monitoring ICs, the serial interface is also important and Linear offers two options. One option, supported by most battery monitoring ICs, is the daisy chained interface in which each IC in the chain communicates to its neighbour without optocouplers or isolators: only the bottom devices interface to a microprocessor or control unit.
The second option features individually addressable serial interfaces. Here, a microcontroller communicates via isolation with multiple devices in parallel. This topology offers the inherently more reliable 'star configuration': loss of communication with one device does not eliminate communication with any others. Addressable devices can also be used in a modified daisy chain topology, where relatively expensive isolators are replaced with a less expensive 'transistor¬ised' SPI bus configuration. The result is a serial interface with a wide compliance range.
After two years of production, Linear has introduced a second generation device: the LT6803. A comparison of the first and second generation devices offers insight into the direction of future high voltage battery systems. One of the goals of the LTC6803 is to ensure error free communications, even under the most extreme noise. Packet error detection is implemented for all commands and data to ensure communication integrity. The LTC6803 family continues to support daisy chained and individually addressable serial communications and the LTC6803 daisy chain can be subjected to more than 20V of ac noise and 30V of fast switching spikes with no errors.
Its independent supply input can be disconnected while leaving other connections intact. In this hardware shutdown condition, the LTC6803 draws only a few nanoamps; important for long term battery pack storage, since the current consumed by the integrated BMS can potentially unbalance the cells in a battery pack. The LTC6803 can also be operated from an independent power supply, allowing the supply current to be drawn from a separate source, instead of the battery pack. With a separate power supply, the LTC6803 can continue to monitor a stack of cells, even when all cell voltages have collapsed.
Increasing electronic content in cars is driving new standards in automotive electronic quality and reliability, hence, the emergence of automotive electronic standards such as AEC Q100 and ISO 26262. This translates to extensive qualification and internal capability to ensure that system safety requirements can be satisfied. The LTC6803, designed for ISO 26262 compliant systems, includes an open wire check, digital filter check, multiplexer decoder check, watchdog timer and a redundant voltage reference for full self test capability (see fig 2).
A number of other improvements are included to address demands outside of the standard automotive design. For example, the part has an extended measurement range – from -300mV to 5V – to support supercapacitors and NiMH batteries. It is specified for operation from -40 to 125°C and has been designed to withstand supply voltages up to 75V.
The automobile is a tough environment for electronics, but increased automotive electrification is not up for debate. Lithium-Ion battery systems in EVs and HEVs will soon become mainstream and sophisticated measurement devices, such as the LTC6803, are essential to their success. Not only must these devices be accurate, they must also operate reliably operate for long periods under difficult conditions.
Greg Zimmer is senior product marketing engineer, signal conditioning products, for Linear Technology.