CellPac PLUS FUEL GAUGING

Power Pack Solutions
Understanding the latest techniques for
implementing high value-added battery
fuel-gauging
Users today expect to find a battery status indicator in the display screen of sophisticated
portable electronic devices. The implementation of battery ‘fuel-gauging’ has been
popularised through its use in mobile phones and laptop computers.
Fuel-gauging offers more than convenience for the consumer.
In some mission-critical applications in, for instance, the
industrial and medical markets, data integrity – or even
human health – can be compromised by unexpected
shutdowns that occur when the battery pack runs out of
power. Reliable, accurate fuel-gauging enables the device
designer to implement safe shutdown processes. It also
allows the user to switch to a power-saving mode to
extend operation until the battery can be recharged.
cell voltage between 3.9V and 3.6V, but this would be
expensive, and is not merited since the voltage-monitoring
technique suffers from another fundamental drawback: the
correlation between cell voltage and DoD changes over
time and over temperature (see Figures 2 and 3). So even
if a precise value for cell voltage could be measured, the
DoD value it produced would still not be reliable.
Voltage monitoring, then, is cheap and simple to
implement, but also highly inaccurate and unreliable.
Older implementations of fuel-gauging in consumer
devices with rechargeable lithium-ion (Li-ion) or lithiumpolymer (Li-Po) batteries prioritised low cost over accuracy
and robustness, using cell voltage measurement to
estimate the state of charge. This method, however, is
both imprecise and inaccurate. Across 70% of the
discharge cycle of a lithium battery the cell voltage is
between 3.9V and 3.6V (see Figure 1). This means that a
large rise in the Depth of Discharge (DoD) value is
associated with a minute fall in the monitored voltage.
The cell voltage then falls steeply from 3.6V as the DoD
value rises from around 80% to 100%. Voltage monitoring
schemes therefore typically sacrifice accuracy across the
middle portion of the discharge cycle. In theory, it would be
possible to provide very high-resolution sampling of the
Technical Article
Fig. 1: typical discharge curves of a Li-ion (Prismatic) cell
Fig. 2: typical temperature profile of a Li-ion cell, showing that the
discharge curve is strongly dependent on temperature, and that capacity
is different at different temperatures
Fig. 3: voltage-monitoring as a fuel-gauging technique becomes less
accurate as the cell ages. Source: Texas Instruments
Accurate fuel-gauging requires a different approach
This article examines the options for system developers
who wish to implement a fuel-gauging circuit that remains
accurate for the lifetime of the battery pack.
The typical accuracy of a coulomb-counting circuit in a
new battery is 2-3%, but over time, as the cell’s charge
capacity shrinks, this can rise as high as 25% after 500
cycles. To counter this effect, fuel-gauging IC
manufacturers such as Texas Instruments and Maxim have
sought to implement algorithms that compensate for the
effects of ageing. Devices such as the bq27200/bq27210
from TI, for instance, have the ability to ‘relearn’ the charge
capacity of a cell, based on the amount of charge drawn
over a full discharge cycle (ie between the end of a full
charging event and the point at which the cell voltage falls
to a pre-programmed minimum level).
A technique widely adopted in mobile phones, known as
‘coulomb counting’, goes some way towards addressing
the problem of maintaining accuracy as the physical
characteristics of a cell change over time. A coulombcounting circuit requires a current sense resistor and a
fuel-gauging IC. (Such devices are readily available on the
merchant IC market.) Over an initial full charge/discharge
cycle, the IC ‘learns’ the actual charge capacity of the
battery. (The nominal capacity for common battery types is
stored in the device’s memory.) By monitoring the voltage
across the current sense resistor, the device can then
measure the amount of charge added to or drawn from
the battery.
Since the fuel-gauging IC knows the amount of charge the
battery holds when fully charged, and the amount of
charge that has been drawn at any given time, it can easily
derive dynamic DoD, remaining capacity and remaining
time values.
Except… the capacity of a lithium battery shrinks with
every charge/discharge cycle, so over time the system’s
stored value for charge capacity, and the actual charge
capacity, steadily move further apart. This in turn makes
the fuel gauge steadily less accurate.
Take as an example a new battery with a 1,000mAh
capacity. Starting from a fully-charged state, the host
device draws charge equivalent to 700mAh; the fuelgauge IC therefore calculates that the SoC is 30%
(300mAh/1,000mAh).
After a given number of charge/discharge cycles, assume
the battery’s charge capacity falls to 900mAh. Now the
same device usage, drawing 700mAh of charge, produces
an actual SoC of 22.2% (200mAh/900mAh). A basic
coulomb-counting fuel-gauge, however, will still register
the SoC as 30%, because the charge capacity value
stored in its memory is 1,000mAh.
This should reduce the extent to which a coulombcounting implementation loses accuracy over time. But the
effect of this workaround can be limited. First, this
relearning process can only take place if the user allows
the battery to become fully discharged. A mobile phone
user who always recharges the battery with two or more
bars showing on the fuel gauge will therefore never allow
the system to relearn the battery’s capacity.
Second, even if the battery does go through a full
charge/discharge cycle, a wide range of conditions –
including cold temperatures, light load, a fast voltage drop,
excessive charging and excessive self-discharge –
invalidate the relearning process.
In summary, then, a coulomb-counting circuit’s accuracy is
only as good as its most recent charge capacity reading –
and in many cases this might be the first reading taken
when the battery was new.
A new technique to maintain accuracy over the
lifetime of the battery
While coulomb-counting can in certain circumstances
achieve high accuracy over the lifetime of a battery, it
cannot guarantee this lifetime performance. For
applications in which accurate SoC data is mission-critical
– such as certain portable medical devices – a more
robust technique is required.
A new approach called Impedance Tracking™, using
technology patented by TI, promises to offer a guarantee
of accurate fuel-gauging over time. Impedance Tracking
uses coulomb-counting in its operation, but also
implements other techniques to negate the limitations of
coulomb-counting.
Impedance Tracking is based on the fact that a cell’s Open
Circuit Voltage (OCV – when the battery is being neither
charged nor discharged) can be correlated to the DoD.
But this correlation changes over time: the more
charge/discharge cycles a cell undergoes, the lower its
OCV for any given value of DoD.
To counter this effect, the Impedance Tracking technique
tracks changes in the cell’s internal resistance over time,
because the change in the ratio OCV:DoD itself correlates
to the change in this internal battery resistance. At the
same time, an Impedance Tracking implementation will
constantly track the reduction in the cell’s charge capacity
(Qmax) over time.
In normal operation, an Impedance Tracking system will
measure the OCV whenever possible. It will then offset this
voltage value to take account of the voltage drop caused
by the battery’s internal resistance. It can then read off a
DoD value from data stored in memory.
When the device is being charged or discharged, it is not
in an OCV condition, so to track the DoD it then counts
coulombs in or out until the next OCV event, when it resets
the DoD value using an updated reading for the internal
battery resistance.
For applications in which high accuracy is required, or a
certain level of accuracy must be guaranteed, developers
should implement Impedance Tracking. In such
applications, the additional bill of materials cost is justified:
Impedance Tracking ICs are available prices ranging from
around $1.40 up to around $4.25; ICs that implement
coulomb-counting typically cost in the region $1.00 $1.25.
When SoC information is not mission-critical, the coulombcounting technique will sometimes be accurate enough,
although the designer must bear in mind that the level of
accuracy is not guaranteed, and might decline over time if
the fuel-gauging IC is not able to ‘relearn’ a cell’s
maximum charge capacity. And while nearly all
microcontroller-based applications will support the I2C
interface used by coulomb-counting systems, some might
not have the SMBus interface required for Impedance
Tracking implementations.
Whichever technique is chosen, developers who do not
wish to master the details of fuel-gauging can have VARTA
Microbattery design it for them, through its CellPac PLUS
service. VARTA Microbattery will both design a customised
battery pack including Impedance Tracking or coulombcounting technology, and manufacture it in volume on its
customer’s behalf.
Because the system is also dynamically monitoring Qmax,
it is also able to derive from the accurate DoD value an
accurate value for remaining charge and remaining time.
Since it is based on impedance data that track changes in
the battery’s physical condition over time, this technique is
able to achieve constant accuracy over the whole lifetime
of a battery of better than 1%.
Fuel-gauging ICs from Texas Instruments that implement
Impedance Tracking, such as the bq20zxx family, provide
straightforward outputs via SMBus, providing values for
SoC, remaining charge, Qmax and so on using the
standard Smart Battery System (SBS) 1.1 protocol. They
also integrate:
• protection features that eliminate the need for certain
external components
• a data-logging function. This provides battery-condition
data that are useful to service and repair technicians
How to choose the appropriate fuel-gauging technology
In the view of this author, the only justification for
implementing basic voltage-monitoring as a fuel-gauging
technique is when the actual readings are unimportant, and
the information is presented to the user for cosmetic purposes,
rather than as an actual guide to SoC. Voltage monitoring
is too inaccurate for users to rely on its SoC readings.
If you are interested in learning more about the latest
techniques for implementing high value-added battery
fuel-gauging, please contact your nearest VARTA
Microbattery sales office. Details can be found at
www.varta-microbattery.com.
Press Contact:
VARTA Microbattery GmbH
Sonja Peitl-Steinert – Corporate Communications
Daimlerstrasse 1
74379 Ellwangen
Germany
Telephone +49 7961 921-526
E-mail: [email protected]