Assessing the dependence of li-Ion batteries capacity on load current frequency

Lithium-ion batteries are the most common chemistry thanks to their many desirable properties, such as high energy density and long lifetimes; however, they still exhibit several non-ideal properties common to other battery chemistries, such as rated capacity effect, capacity dependence on load variation, and recovery effects. In the last two decades, many works in the EDA community have focused on methods to characterize and model these non-idealities. In particular, for the rated capacity effect and the recovery effect, accurate models have been derived that allow tracking battery behaviourr with a reasonable degree of fidelity. However, one effect that has not been thoroughly explored is how battery capacity is affected by the spectral properties of the load current and its frequency distribution. Some preliminary works analyzed this relation based only on the information provided by datasheets and lacked an experimental validation of the resulting models. We conduct experimental measurements to analyze this dependence between usable battery capacity and the frequency spectrum of load current in this work. Our measurements illustrate that this relation is much more sophisticated than the one extracted from ddatasheet information: it is not monotonic and exhibits a local maximum point that changes over time as the battery ages.