BU-907: Testing Lithium-based Batteries

With the large number of lithium-ion batteries in use and the applications growing, a functional rapid-testing method is becoming a necessity. Several attempts have been tried, including measuring internal resistance, and the results have been mixed. Additives keep the internal resistance of modern Li-ion low throughout most of the life, making ohmic test unreliable. The internal resistance is measured either by the AC or DC method. Both provide different reading(See BU-802a: How does Rising Internal Resistance affect Performance?)

Electrochemical dynamic response, the method QuickSort™ uses, measures the mobility of ion flow between the electrodes. Based on time domain analysis by applying brief load pulses, the response time on attack and recovery is measured; an algorithm computes the results and compares them against a set of parameters. As seen in Figure 1, a good battery resists the attack and recovers quickly whereas the impact of a weaker battery is larger and the recovery is slower. Figure 1 illustrates the concept of the technology.

The electrochemical dynamic response measures the ion flow between the positive and negative plates. A strong battery recovers quickly from an attack whereas a weaker pack behaves more sluggishly.

Lithium-ion batteries have different diffusion rates. In terms of electrochemical dynamic response, Li-ion polymer with gelled electrolyte is found to be faster than standard Li-ion and needs modified parameters to achieve accuracy. Unique active materials and additives that are kept top secret by battery manufacturers complicate the test procedure.

Cadex devoted much effort to testing small single Li-ion cell in mobile phones. The objective is to also test larger Li-ion in multi-cell configuration, over a broad range of state-of-charge, which involves combining time domain test with frequency domain.

When scanning a battery from kilohertz down to millihertz in frequency domain mode, the high frequency range called migration reveals the resistive qualities of a battery that present a bird’s-eye view of the landscape. However, the unique characteristics of Li-ion lie in the mid frequency range called charge transfer and the low range dubbed diffusion. Batteries with faded capacity suffer from low charge transfer and slow active Li-ion diffusion.

Evaluating batteries at sub one-hertz frequency would require prolonged test times. At one millihertz, for example, one cycle takes 1,000 seconds, or 16 minutes, and several data points are required to complete the analysis. Rapid-tests should only last a few seconds and not longer than 5 minutes. With clever software simulation, the duration can be shortened to fall within the desired short test time.

In Figure 2, a good battery and a faded battery are scanned from 0.1Hz to 1kHz. The difference in impedance (-Imp -Z) is strongest between 1Hz and 10Hz. It should be noted that capturing resistive readings alone has limited value as state-of-charge (SoC) and temperature also affect the signature and muddle SoH references. Furthermore, different Li-ion architectures and how the battery has aged also affect the results. Natural aging produces a different signature than artificial aging and the reason for this discrepancy is not fully understood.

Impedance variances are most visible below 10Hz. The horizontal scale is logarithmic to condense the frequency range.

Test results captured by frequency domain are best represented by the Nyquist plot. Invented by Harry Nyquist (1889–1976) while at Bell Laboratories; a Nyquist plot presents the frequency response of a linear systems displaying both amplitude and phase angle on a single plot using frequency as parameter. The horizontal x axis of a Nyquist plot reveals the real Ohm impedance while the vertical y axis represents the imaginary impedance(Impedance is explained on BU-902: How to Measure Internal Resistance)

Figure 3 divides the scanned battery results as delivered by the Nyquist plot into migration, charge transfer and diffusion. Migration derived at high frequency on the left provides resistive characteristics of a battery; the all-important charge transfer in the middle forms a semi-circle that represents the kinetics of the battery; and the low frequency part on the right represents diffusion.

The mid-frequency semi-circle represents battery characteristics best. Larger batteries require lower frequencies.

A rapid-test should last from a few seconds to no more than 5 minutes, but applying ultra-low frequencies prolongs time. For example, at one millihertz (mHz), one cycle takes 1,000 seconds, or 16 minutes, and several data points are required to complete the analysis. Test durations can often be shortened with clever software simulation.

Li-ion shares similarities with lead acid; the Spectro™ technology that is used to measure the capacity of lead acid batteries will also be able to service Li-ion(See BU-904: How to Measure Capacity)

Summary

No rapid-test can evaluate all battery symptoms and there are always outliers that defy the test protocol. Correct prediction should be 9 out of 10. QuickSort™ (by Cadex) exceeds this requirement with most Li-ion packs for mobile phones, but this technology only tests single-cell packs up to 1,500mAh. New technologies in development promise to test larger Li-ion packs, but this may extend the test to a few minutes to accommodate low frequency sampling.

Capacity is the gate keeper to battery health, and rapid-test technologies with capacity estimation also enhance battery management systems (BMS). Such rapid-test technologies can be included in chargers to evaluate the integrity of the battery with each charge by giving the green ready light only if the set target capacity is met; low capacity batteries are shown the backdoor. This provides quality control without adding an extra layer of overhead.

References

^[1] J. Tinnemeyer and Z. Carlin, "Pulse-discharge battery testing methods and apparatus". US Patent US7622929B2, 25 07 2006.
^[2] Courtesy of Cadex