Inside view

11 October 2012

Christoph Unterrieder (left), Michael Lunglmayr (centre) and Professor Mario Huemer (right) from the University of Klagenfurt collaborated with Stefano Marsili from  Infineon Technologies Austria to develop their battery state-of-charge estimation technique.

Why are battery monitors important?

With an ever increasing variety of functionalities in portable electronic applications, the demands on the battery market are growing much faster than battery technology is advancing. It has become increasingly important to improve the use of the available cell capacity, to extend the battery lifetime and the battery runtime, and to improve the cell balancing and the battery charging process. It is therefore crucial to be able to accurately and reliably determine the state-of-charge (SoC).

Nearly all battery-powered electronic devices use methods for indicating the battery status - reaching from simply using light-emitting diodes which are roughly indicating the internal battery state to the use of sophisticated integrated battery identification circuits: so-called battery fuel gauges. For a performance increase in terms of power efficiency, cell capacity use and battery runtime as well as a decrease in battery costs, the use of battery fuel gauges is indispensable. In general, the battery fuel gauge IC market is increasing – also affected by the steadily increasing battery management IC market. The portable electronics and electric vehicle markets in particular constitute target markets with a high expected demand on integrated battery management solutions.

How do these circuits work?

State-of-the-art implementations of integrated battery identification circuits are mainly based on two techniques: counting the amount of charge during charge/discharge phases (Coulomb counting) and correcting the SoC estimation by using open-circuit-voltage (OCV) based methods during periods of rest. Furthermore, actual implementations on the market are often enhanced by capacity and impedance re-learning methods. In general, integrated battery management solutions can extend the battery runtime by more than 20%, and they are also indispensable for optimising the use of the available battery cell capacity and for reducing an application’s battery costs.

The problem is that OCV-measurement based approach can only be applied if the considered cell is in a well-relaxed state, but it can take between 20 minutes and 3 hours to reach this state depending on parameters like the SoC, the temperature or the age of the battery. Frequent use of today’s electronic devices means that this state is rarely reached and the determination of SoC is continuously afflicted by an error introduced through the Coulomb counting method.   

What’s new about the monitoring method that you have presented in your Electronics Letters paper?

Our OCV extrapolation methodology offers the possibility to determine the OCV/SoC of a cell that is not in a well-relaxed state. To achieve this we proposed the successive application of individual, voltage model based, sequential least squares estimation approaches. The OCV relaxation process after a charge/discharge pulse is split into several sections. For each of these sections, our approach provides two degrees of freedom: the time period for which the OCV is estimated and the time distance between the values used for the coefficient calculation for the next section. Also, the performance is influenced by the chosen voltage model and the applied sampling time for the least squares curve fitting process. The predicted OCV can then be used for the estimation of the corresponding SoC.

Our approach reduces the period of time that is needed until the SoC of a cell can be accurately determined based on its OCV. An OCV-based correction of the Coulomb counting method can be applied more often which not only improves the estimation of the actual SoC, it also enhances the capacity re-learning capability of the battery fuel gauge algorithm. For a real world user this means having a faster and more reliable battery status display.  

What’s next?

Basically, our model can be used for any battery type and technology. As the model provides some degrees of freedom, its parameters have to be adjusted according to the considered cell. But also cell-to-cell variations, aging aspects and other parameter dependencies can be included in the modelling process. Next steps will include the investigation of the proposed methodology with respect to even to influence factors like the temperature or the applied charge or discharge current rate.  

A related field that our group is working on is the power management of mobile devices. Here, digital, analogue as well as hybrid control strategies for DC-DC-converters are investigated. In co-operation with an industrial partner a number of different control concepts have been implemented and verified on test chips. Further activities are in the fields of digital compensation of analogue non-idealities in RF (radio frequency) frontends, and in signal processing for sensor and communication applications.  

The Letter presenting the results on which this article is based can be found on the IET Digital Library.

For further reading ‘Battery state-of-charge estimation using polynomial enhanced prediction’, C. Unterrieder, M. Lunglmayr, S. Marsili and M. Huemer

Journal content

Cover of Electronics Letters, Volume 49, Issue 25

View all content

Browse or search all papers in the latest or past issues of Electronics Letters on the IET Digital Library.