Decreasing Test Time to Determine SOH for Lithium Chemistry
Abstract
Warranty return testing can take more time than a customer is willing to accept. Decreasing test time over the standard charge-discharge-charge routine needs to be found so that customer satisfaction with the process is increased. Decreasing warranty return testing time should not compromise results beyond what is acceptable and also be safe and repeatable.
In this paper we will discuss and test procedures to decrease the test time of lithium based cells/batteries using shorter charge and discharge times.
Introduction
Warranty testing of lithium based cells and batteries is currently based upon a coulomb counting routine of fully charging the unit under test, then fully discharging and finally fully recharging the battery before a result is given. The result is given as a percentage of the cells nominal capacity.
For example, a 3.6V 1500mAh cell will be charged to 4.2V and a charge cut off current of 0.02C (30mA in this case). At this point the cell is said to be at 100% state of charge (SOC). This cell is then discharged to 3.0V at 1C (1500mA). If a result of 1275mA is given by the test equipment then the battery is said to have an 85% State of Health (SOH). The cell is then fully recharged and returned to the customer.
This test can take up to 360 minutes if the test is done at 1C, see Table 2. At less than 1C tests can take far longer.
There is no definitive definition of SOH, the above test is just one of many methods to determine it. Other methods may be based on internal resistance, self-discharge, cell heating, charge efficiency, charge acceptance, power delivery or a combination of these. Testing a single parameter may not reveal another parameter that is failing. Training and understanding is an essential part of battery testing.
Eliminating the Constant Voltage Stage
A general rule is that a lithium cell reaches 85% SOC when the charge moves from the constant current (CC) stage to the constant voltage (CV) stage. We will test this idea as one method of reducing test times.
Looking at the charge profile of a UR18650FK 3.6V Lithium cell it can be seen that it takes approximately 150 minutes to fully charge with 1C at 20 °C. Most of this time is spent in the CV stage.
Figure 1 UR18650FK charge characteristics
The current profile in Figure 1 shows that the CC stage with 1C charge lasts approximately 50 minutes. Lithium chemistry has 98-100% coulometric efficiency and a quick calculation (50/60*100) tells us we should have approximately 85% SOC.
This raises two points; a 45 minute charge reaching 85% SOC is less than a third of the charge time to achieve 100% SOC, and, is it possible to report SOH by the measuring the time it takes to reach the CV charge state from 0% SOC?
As a cell ages its internal resistances rises, this demonstrates itself by increases in voltage sag during discharge and voltage rise during charge. During a charge cycle this increased resistance will cause the voltage to increase and reach the CV stage sooner. If an earlier charge cut off means less charge accepted, will charge time give an accurate value for SOH?
As an aside, Figure 1 shows that the temperature of the cell will determine cell voltage at various SOC levels during charge, therefore we must use “room-temperature” as a test requirement if using transition from CC to CV as a charge cut-off point.
Tests
A set of fourteen 7.2V NP-W126 lithium batteries of various capacities, manufacturers and calendar life were chosen as a sample to test if cutting off charge at the transition of CC to CV was a good indication of battery SOH. Some of these batteries have seen a long life in professional photography equipment while others were new from stock. The NP-W126S version of battery have a lower internal resistance to cater for the high power requirements of latter model Fujifilm cameras.
Test equipment used were programmable power supply (Rigol DP-811A) and programmable load (Rigol DL-3031). These devices were set up with sense wires for Kelvin connections eliminating errors introduced by voltage drops on current carrying wires.
A 0.5C charge and discharge rate was chosen as this is close to the actual charge and discharge rate of the equipment.
The first test was to get a baseline SOH using the standard CC/CV 0.5C charge with 0.02C charge termination. Battery capacities varied from 79.3% to 107.8% of nominal value.
The second test involved charging the batteries to the CV stage, as soon as the voltage reached 8.4V the charge was terminated. Results showed 62.6% to 92.8% with normalisation showing 73.6% to 109.1%.
The shortened tests showed a maximum of 8.8% variation with the standard test.
The third test was using the time to charge from 0% to CV stage as a criteria of SOH. This showed a correlation with actual SOH using the full charge to discharge routine with a maximum 10.5% difference in result.
Variation with the results of the second set of tests showed a 5.1% difference (battery B2). This 5.1% may be an outlier with a bad result as all other differences were at a maximum of 2.0%.
See Table 1 for a full set of results from these 14 batteries.
It is interesting to note that the batteries with the most deviation from their nominal value also had the greatest difference between the timed charge test (T3) and the full standard test (T1).
Note that results using the timed charge method (T3) showed increased departure from the T1 tests in the direction of difference from nominal value. That is, if T1 showed greater than nominal capacity, then T3 test showed that increase in a larger deviation. Figure 2 shows the x-axis as deviation from nominal value of the battery and the y-axis as deviation from the standard T1 test result.
Fourteen batteries is a small sample set, even so, we could surmise that the increased internal resistance of an aging battery is resulting in a reduced charge acceptance time and therefore an exaggerated departure from nominal capacity.
Figure 2. Reduction in capacity and increase in internal resistance showing as large departure from nominal capacity.
Batteries B2 and B12 were both slightly swollen (16.1mm instead of 15.5mm standard) and were no longer an easy fit in the camera body. As far as SOH is concerned, battery dimensions may well be another parameter to use for prismatic cells.
The Final Charge
Once a result has been determined, what recharge should be given to the cell or battery under test? A customer may expect a 100% charge, this may demonstrate good will on the testers part and also begin a conversation about the charger and charge routine the customer has. But it adds considerable time as we have seen. 85% may be a sufficient SOC and at most should take 48 minutes with a 1C charge.
If the cell or battery requires transportation back to a customer, 30% is the maximum recommended SOC by IATA and is the accepted industry standard.
A cell or battery that has failed a warranty test neither needs to be returned to a customer nor recharged. Replacement is the expected policy.
Hidden Problems
Results may be skewed in battery packs with multiple cells. A single cell in a string of cells may not show its problem of lower capacity in a test that only covers 85% of nominal capacity. If a cell in a 2 cell battery has a low capacity or is not well balanced in SOC in that battery, its voltage change is prominent. But in a battery pack of 10 cells a 0.2V change within a pack of up to 42V may not be recognised.
Further tests in larger packs may be required before full confidence is attained in shorter test routines.
Test Time Reductions
Batteries are returned for testing in various states of charge, anywhere from 0% to 100%. Data on received SOC of batteries may be drawn from a history of data from test machines in a lab, this data was not available at the time of this paper.
Of the 3 tests described, is there a test that is clearly the quickest every time? Eliminating a test step will speed up results no matter what capacity a battery is returned at.
Table 2 to Table 5 show the test routines from various SOC ranging from 0% to 100%:
1. Charge to 100%. Discharge to 0%. Charge to 100%. C100/D0/C100
2. Charge to 100%. Discharge to 0%. Charge to 85%. C100/D0/C85
3. Charge to 85%. Discharge to 0%. Charge to 85%. C85/D0/C85
4. Discharge to 0%. Charge to 85%. D0/C85
Method 4, discharge to 0% and then timing the charge to the constant voltage state is faster in nearly every situation. If a customer brings the battery for testing in a depleted state test times may be as low as 50 minutes. The battery after the test, if in good condition, is at 85% charge and ready to use.
Conclusion
Decreasing test time by modifying the charge termination and normalising the discharge capacity reduces test time with a less than 10% difference in results for this limited data set of tests.
A further reduction in test time can be found by timing the 0% SOC to transition of Constant Current to Constant Voltage and calculating charge over time. In as little as 50 minutes a customer may have a result and an 85% charged battery.
Battery testing will be a contentious subject, future methods may will decrease test time further from what is demonstrated here. It is clear that charging to 30% or 85% is faster than charging to 100% SOC. Final charge levels can be a determined depending upon customer/shipping and test result. We can reduce test time further by modifying the charge-discharge-charge routine to a 2 step test.
References
Association, I. A. (2019). Transport of Lithium Metal and. IATA. Retrieved from https://www.iata.org/whatwedo/cargo/dgr/Documents/lithium-battery-shipping-guidelines.pdf
Buchmann, I. (2016). Batteries in a Portable World (4th ed.).
Plett, G. (2015). Battery Management Systems, Volume I: Battery Modeling.
Plett, G. (2015). Battery Management Systems, Volume II: Equivalent-Circuit Methods.
