What good is an automobile without its battery?

Reducing the number of failures requires precise sensing of the battery's voltage, current and temperature; preprocessing of results; calculation of state of charge and state of health; communication of results to the ECU; and control of the charging function.

The modern automobile was born at the dawn of the 20th century. The first car was started by hand, which required strength and entailed a large degree of risk, with many deaths attributed to the automobile's "hand crank." In 1902, the first battery and starter motor were introduced, and by 1920, all cars had electric starters.

Initially a dry-cell battery was used, but it had to be replaced once depleted. A wet battery-specifically, the venerable lead-acid battery-soon replaced the dry cell. The lead-acid battery had the benefit that it could be recharged by the automobile engine itself, once started.

Over the last century, the lead-acid battery has changed very little. The last major advance was the introduction of the sealed lead-acid battery. What has changed are the demands placed upon it. At the outset, it was used only to start the car, sound the horn and power the lights. Today, it is used to power all electrical systems in an automobile, prior to ignition.

The proliferation of new electronic devices is not just limited to consumer electronics such as GPS and DVD players. Body electronics such as engine control units (ECUs), electric windows and power seats are now standard in many basic models. This exponentially increasing load has taken its toll, as is evident from the increasing number of breakdowns attributed to electrical-system failures. According to ADAC and RAC statistics, almost 36 percent of all car breakdowns are due to electrical failures. If this number is analyzed, more than 50 percent of all failures are due to one component, the lead-acid battery.

Assessing a battery's health
A lead-acid battery has two key attributes that tell how "healthy" it is:

1) State of charge (SoC): The SoC, which indicates how much charge can be delivered, is expressed as a percentage of a battery's rated capacity (that is, the SoC of a new battery).

2) State of health (SoH): The SoH indicates how much charge the battery can store.

State of charge
State-of-charge indication may be thought of as the battery's "fuel gauge." Many methods exist for the calculation of SoC, but the two most common methods are open-circuit voltage measurement and coulometric measurement (also called coulomb counting).

1) Open-circuit voltage (VOC) : A linear relationship exists between a battery's open-circuit voltage, where no load is applied, and its state of charge. This method of calculation has two fundamental limitations: To calculate the SoC, the battery must be open-circuited, with no loads connected; and this measurement is only accurate after a considerable stabilization period.

These limitations make the VOC method unsuitable for in-line calculation of SoC. It is regularly used in auto shops, where the battery is removed and a voltmeter is used to measure the voltage across the positive and negative terminals of the battery.

2) Coulometric measurement: This approach to SoC determination uses coulomb counting to integrate current over time. That makes it possible to use this measurement for real-time calculation of SoC, even when the battery is under load conditions. However, the coulometric-measurement method is prone to cumulative errors over time.

A combination of both open-circuit voltage and coulomb counting is often used to calculate a battery's state of charge.

State of health
The state of health reflects the general state of the battery and its ability to store charge, compared with a new battery. SoH calculations are, by their very nature, extremely complex and dependent on the understanding of a battery's chemistry and environment. A battery's SoH is influenced by numerous factors, including charge acceptance, internal resistance, voltage, self-discharge and temperature.

These factors are generally considered difficult to measure in real-time, such as in the automotive environment. The best indicator of SoH occurs during the cranking phase (starting of the engine), when the battery is under the most strain.

The actual algorithms that leading automotive-battery sensor developers, such as Bosch, Hella and others, use for SoC and SoH calculations are closely guarded secrets that are often covered by patents. As the vendors' intellectual property, these calculations are frequently developed in close conjunction with battery manufacturers such as Varta and Moll.

Figure 1 shows a discrete circuit commonly used for battery sensing.

Figure 1: Discrete battery-sensing solution
(Click to Enlarge Image)

The circuit can be broken into three separate regions:

1) Battery sensing: The battery voltage is sensed through a resistive attenuator tapped directly off the positive terminal of the battery. For sensing current flow, a sense resistor (normally 100 microohms for 12-V applications) is placed between the negative terminal of the battery and ground. In this configuration, the metal chassis of the car is generally ground, and the sense resistor is placed in the return path of current to the battery. In an alternate configuration, the negative terminal of the battery is ground. For SoH calculations, the temperature of the battery must also be sensed.

2) Microcontroller: The microcontroller, or MCU, performs two primary tasks. The first is processing the A/D converter results. This task may be as simple as performing basic filtering or as complex as calculating the SoC and SoH. The actual function is dependent on the processing power of the MCU and the demands of the automotive manufacturers. The second task is to transmit the processed data to the ECU via the communications interface.

3) Communications interface: At present, the local interconnect network (LIN) interface is the most commonly used communications interface between battery sensors and the ECU. LIN is a single-wire, low-cost alternative to the familiar CAN protocol.

This is the simplest configuration available for battery sensing. However, most precision battery-sensing algorithms require simultaneous sampling of battery voltage and current, or battery voltage, current and temperature. This simultaneous-sampling requirement demands the addition of up to two A/Ds. Also, an added complexity is that the A/D and MCU require regulated power supplies to operate correctly. This has been addressed by the integration of a regulated power supply by LIN transceiver manufacturers.

The next step forward in automotive precision battery sensing is an integrated device incorporating A/D, MCU and LIN transceiver. One example is the ADuC703x family of precision analog microcontrollers from Analog Devices Inc. The ADuC703x features a choice of two or three 8-ksample/second, 16-bit sigma-delta A/Ds, a 20.48-MHz ARM7TDMI MCU and an integrated LIN v2.0-compliant transceiver. The ADuC703x family is powered directly from the lead-acid battery through an on-chip low-dropout regulator.

To address the needs of automotive battery sensing, the front end includes a voltage attenuator for battery-voltage monitoring; a programmable-gain amplifier facilitating measurement of currents over full-scale ranges from less than 1 A to 1,500 A, when used in conjunction with a 100-microohm resistor; an accumulator function that allows coulomb counting without software supervision; and an on-chip temperature sensor.

An example solution using such an integrated device is shown in Figure 2.

Figure 1: Discrete battery-sensing solution
(Click to Enlarge Image)

A few years ago, only high-end cars had battery sensors. Today, increasing numbers of mid- to low-end vehicles are beginning to feature electronic gadgets and gizmos seen only in high-end models 10 years ago. This has led to an increasing number of failures due to lead-acid batteries. In a few short years, every car manufactured will come with a battery sensor, reducing the risk of a breakdown from the ever increasing amount of electronics.

About the author
David McKenna (david.mckenna@analog.com) is an applications engineer for the industrial and automotive converters product line at Analog Devices Inc. He graduated from University College Cork in Ireland with a bachelor of electrical and electronic engineering degree.