Don’t get stuck at the light: start-stop technology and battery management
These days, though, gasoline is still King of The Hill, and the DOT [USA’s Department of Transport] has been slowly turning the screws on vehicle fuel economy ever since the first CAFE regulations passed Congress in 1975 following the Arab Oil Embargo.
There are a couple of well-recognised ways to reduce fuel consumption – make smaller or less powerful vehicles. But [American] buyers love their trucks, SUVs and V8 power.
The quest for reduced fuel consumption, while still catering to consumer preferences, has led to the introduction of several new technologies. One of them is the “start-stop” (or stop-start – we’ll leave that discussion to the “chicken or the egg” folks) system; it automatically shuts down and restarts the engine to reduce the amount of time the engine spends idling, thereby reducing fuel consumption and emissions.
This is most advantageous if you a spend significant amount of time waiting at traffic lights or frequently get stuck in traffic jams. This feature is present in hybrid electric vehicles, but has also appeared in otherwise conventionally-powered vehicles, so-called “micro” hybrids. For such vehicles, fuel economy gains from this technology are typically in the range of 5 to 10%.
Turning off the engine at a stop light. What could possibly go wrong?
Predicting battery failure
Battery failure is the single biggest cause of vehicle breakdown, accounting for 52% of failures in one study of 1.95 million vehicles less than 6 years old; in another study, failure rates ranged from 1000ppm to over 10000ppm for batteries older than 3 years. Battery failure is exacerbated by operating conditions that don’t allow the battery to fully charge, such as driving short distances using heavy accessory power. Excessive heat is another contributing factor.
Battery death can occur without warning especially if you inhabit the USA’s Desert Southwest – both [EDN editor] Steve Taranovich and I live in the Phoenix, AZ, area. I once drove 40 miles, turned off the engine for a moment outside a restaurant, and had to be pushed to the nearest parking space. Lucky I wasn’t on te intersate highway during rush hour.
To forestall this possibility, it’s critical to have precise knowledge of the current state of the battery to disengage the start-stop system if there’s a possibility that the vehicle may not start following a deliberate shutdown, as well as alert the driver of impending battery failure.
It might seem reasonable that a measurement of minimum battery voltage during crank would give a good indication of impending doom, but this is not the case: Figure 1 shows a comparison of cranking voltages of a battery that was subject to accelerated ageing. In week 13, a week before eventual failure, both the worst-case droop voltages are essentially the same as a cranking sample from week 9, so a more sophisticated approach is needed.
Figure 1: Comparison of cranking voltages of battery with accelerated ageing (source: Ryan J. Grube)
In commercial battery monitoring systems, there are two primary battery metrics used to forecast incipient failure: State of Charge (SoC) and State of Health (SoH). Related to these, battery State of Function (SoF) provides a yes/no rating as to whether the battery can perform its required function of cranking the vehicle.
The battery SoC represents the current capability of the battery; it’s the actual amount of charge available – its capacity, expressed in Ah – as a percentage of the capacity of a new cell. SoC declines with age, environmental factors, and charge-discharge profiles.
There are two main methods of measuring SoC. Coulomb counting, also known as current integration or ampere-hour balancing, is the best method to track fast changes in the SoC. It is based on integrating the current that is flowing in and out of the battery and changing the calculated SoC of the battery accordingly.
Figure 2: SoC calculation using coulomb counting (source: Freescale)
The coulomb counting technique depends on making highly accurate synchronised current and voltage measurements, in both the charge and discharge phases, at precisely timed intervals. Battery temperature measurements must also be logged. The battery is monitored continually over its operational life including when the vehicle is parked.
The efficiency factor α is derived from Peukert’s Law, which expresses the capacity of a lead-acid battery in terms of the rate at which it is discharged. As the discharge rate increases, the battery’s available capacity decreases. Another parameter which impacts the available capacity is the temperature. At higher temperatures, the available capacity is higher. Both effects are described using α, so there is a 2- dimensional array (temperature vs. discharge rate) of α values required. According to measured temperature and discharge rate, the appropriate value is taken for each integration step individually.
The α- values are highly dependent on the battery design and chemistry and are usually different even for varying models from a single manufacturer. They are obtained from charge and discharge tests during product development.
Although Peukert’s law is only valid for discharging, an efficiency factor similar to α must also be applied for charging cycles. In addition to temperature and charge rate, the current SoC has to be taken into account, since the efficiency factor at high SoCs is smaller than at lower SoCs.
Due to the integration of the current and α values, the cumulative error both from changing battery conditions and current measurement quantisation increases over time. The parameter Q(t0) – the starting point of the current integration – is therefore obtained by the second technique mentioned above, the open circuit voltage (OCV) method, which provides better accuracy. The OCV is the voltage between the battery poles under no-load conditions. The OCV and the SoC of a battery have a nearly linear relationship so the SoC can be determined from a look-up table.
The principal disadvantage of the OCV method is that to attain no-load conditions in a practical application, the vehicle must be parked with the maximum number of devices turned off. Even after the load has been removed, the battery OCV will slowly rise over several hours as the electrolyte concentration equalises across the cell, a process known as relaxation.
In practice, the coulomb counting method runs continuously; the OCV method is used to recalibrate the system whenever conditions permit. This combination provides a good calculation of SoC and can be further refined by correcting the SoC with the battery self-discharge rate during long parking periods.
Battery SoH depends both on battery SoC and temperature. It describes the decrease in maximum battery capacity due to ageing, expressed as a percentage of the nominal (new) capacity.
It can be evaluated in several ways: looking at the maximum SoC reached after consecutive full charge cycles; counting the number of charge/discharge cycles and their depth and comparing it to manufacturer data, after correcting for differences in temperature and SoC; or measuring another electrical parameter that is well correlated with SoH as determined during manufacturer testing.
The precise correction factors and discharge data are found after extensive testing; they are highly proprietary and vary according to manufacturer and battery model.
s mentioned earlier, the State Of Function is an indication to the start-stop system of whether the battery will be able to crank the vehicle after the next shutdown.
One good way of obtaining the SoF is to analyse the most recent engine cranks, the remaining battery charge (as a function of SoC and SoH), and the battery temperature. During cranking, the internal resistance (Ri) of the battery (which is calculated from voltage drop and current) is recorded. Ri is usually quite constant over battery lifetime and rises significantly just before end of life, so its average value needs to be below a certain threshold to guarantee safe cranking.
Another effect of aged batteries is that during the cranking phase, Ri is linear for new batteries; for batteries about to fail, the calculated Ri values from the voltage and current samples tend to be nonlinear, i.e. there are different current values for equal voltage samples.
Several manufacturers offer integrated battery monitoring solutions which are part of a battery management system. These systems are usually mounted directly at the negative pole of the battery as shown in Figure 3. The battery current is measured with an external precision shunt resistor of a very low value – around 100 μΩ.
Since it can vary over a range of 100 μA – ±2000A, a high-resolution ADC with a PGA front end is needed. Battery voltage is measured via a series resistor at the positive pole, measured concurrently with the battery current.
Other features include an internal temperature sensor which serves as a proxy for battery temperature, although an interface for an external temperature is typically also provided. Multiple low-power and wake-up modes are also a requirement: the device must wake up when it detects a crank; take multiple measurements at up to 1 kHz during the event; go into sleep mode to minimise the drain on the battery when the vehicle is parked for long periods; and wake up to take periodic measurements as part of the OCV calculation.
Figure 3: Typical mounting of a battery monitoring IC (source: Freescale)
Freescale’s MM9Z1_638 is an intelligent battery sensor that can handle battery monitoring for both 12V lead-acid and multi-cell Li-ion batteries. It includes an embedded 16-bit S12Z microcontroller with 96 or 128 kB flash, 8.0 kB RAM, 4.0 kB EEPROM. Other features include: a LIN 2.2 protocol and PHY, and an MSCAN protocol controller; synchronised voltage and current measurements; an integral PGA; up to five direct voltage measurements; and up to five external temperature sensors.
Analog Devices’ AduC7039 includes a dual channel, simultaneous sampling, 16-bit Σ-Δ ADC with programmable throughput from 10 Hz to 1 kHz, a PGA, internal or external temperature sensors, LIN 2.1, an on-chip 5ppm/ºC VREF. The core is an ARM7TDMI-S, which is a 16-/32-bit RISC architecture with 64 kB of flash, 4 kB SRAM, JTAG support, and a 20.48 MHz PLL on-chip oscillator.
Figure 4: ADuC7039 (source: Analog Devices)
A start-stop system is a small step on the road to re-architecting vehicle functions in the quest for better fuel economy – after all, a micro hybrid is really just a standard vehicle with a twist. The next stage is the “mild” hybrid, which incorporates a combination starter-alternator to stop (and, hopefully, later restart) the engine whenever the vehicle is coasting, braking or stopped. A mild hybrid may also use the starter-alternator to provide power assist to the engine or implement regenerative braking.
About the author
Paul Pickering is a consultant who has over 35 years of engineering and marketing experience in the electronics industry, including time spent in automotive electronics, precision analogue, power semiconductors, flight simulation and robotics. Originally from the North-East of England, he has lived and worked in Europe, the US, and Japan. He has hands-on experience in both digital and analogue circuit design, embedded software, and Web technologies. He has a B.Sc. (Hons) in Physics & Electronics from Royal Holloway College, University of London, and has done graduate work at Tulsa University. In his spare time he plays and teaches the guitar in the Phoenix, Arizona area.