Easing the complexity of automated tests: EV battery performance testing with the PXB power supply

In order to record measurement data and control test equipment, running software such as LabVIEW or PyVISA/Python for automated tests may necessitate bulky and expensive setups. The complexity of these setups, however, can increase dramatically with EV battery testing. Automotive validation testing can be a large and expensive endeavor with battery cyclers, measurement/recording devices, large environment chambers, and cooling systems. This infrastructure is necessary in order to ensure automotive equipment can operate nominally despite many potential environmental, mechanical, and electrical strains during a vehicle’s lifetime.

EV battery performance testing diverges from traditional battery testing with CC and CV charging profiles. These tests take into account, for instance, the impact of harmonics from power converters on the EV powertrain and ultimately the battery’s charging and discharging. Transient response characteristics of the batteries must also be taken into account during these tests and require more real-world dynamic profiles to charge the battery. During this testing, it is useful to have a power supply that can easily sequence and output these waveforms without too much software complexity. The Kikusui PXB power supply is able to easily perform sequencing and output a test waveform at the click of a button with its built-in web server. An internal data logger can record all measurements so that the user can then analyze battery performance. This article describes various battery cycling tests and how even the complex tests can be easily performed with the PXB power supply.

Lithium ion (Li-ion) batteries are often the battery of choice for EVs and these batteries will suffer from both calendar and cyclic aging. Calendar again mainly depends upon temperature and state of charge (SoC) of the battery. The SoC is the battery’s level of charge relative to itself and is typically expressed as a percentage. Other aging mechanisms occur due to charging and driving operation, the latter requiring a dynamic load profile.

Battery cycling tests require repeatedly charging the battery until it reaches its energy storage potential and discharging it. These reliability tests are costly and energy-intensive, as they can require months of continuous cycling. All discharged energy is often dissipated via a convective-cooled (liquid or forced air) resistive load bank. Various parameters such as SoC, temperature, capacity, and depth of discharge (DoD) are measured after every cycle or at different predetermined points during each test. The charging methods can vary between one of several methods: constant current (CC), constant voltage (CV), constant power (CP), and the hybrid CC-CV charging method. An idea of the various tests performed on different Li-ion technologies and their publicly available datasets can be viewed here. The hybrid CC-CV method often used can be seen in Figure 1. The battery is charged with a constant current (Ich) that allows the battery to reach a preset voltage level (Uch). Then, Uch is kept constant while the charge current gradually decreases. The charging process is then halted when the charge current is below a cut-off level (Iend) that is typically less than 3% of the rated current [1].

The Kikusui PXB power supplies can perform CC, CV, CP, and CC-CV charging to analyze the steady-state behavior and lifetime performances of different battery topologies. It should be noted that overshoots can occur depending upon the relationship between the load’s impedance and the voltage settings divided by the current settings (V/I=R) of the power supply. Overshoots can be prevented by setting CC mode priority when batteries are connected. The required response speed of the power supply can be optimized depending upon the test conditions and load specifications (Figure 2).

The bidirectional supply also offers on-site regenerative capabilities, providing both a source and a sink for battery cycling tests. When power is regenerated to the unit from the battery under test, the load power is converted to reusable power and regenerated to the AC LINE — potentially supplying neighboring racks and nearby test equipment instead of wasting excess energy with load banks.

As stated earlier, battery stress tests can range between simpler CC-CV profiles and more dynamic profiles to, for example, simulate the impact of the harmonics from the powertrain on the DC current on the traction battery [2]. Regenerative braking has also been known to cause recharge periods of a high current rate, which can accelerate battery aging or even damage the Li-ion battery. This might require dynamic profiles to better test battery performance [3].

Synthetic driving cycles, or a vehicle speed-time profile, offer a more dynamic stress test and can be used to validate various battery models (e.g., first-order transient response model, model at fixed SoC, etc.) to ultimately accumulate data on EV battery lifetimes and optimize a traction battery’s life. These cycling tests simulate real-world conditions where a culmination of performance tests over a shortened period of time are used to estimate failure rates. The tests are also used to test the battery’s limits under harsher conditions such as thermal shock, subzero temperatures as well as severe vibrations, and mechanical shock. It is not uncommon to use multiple test cycles for a singular battery to ensure they can meet various performance requirements. These tests will use long and detailed sequences that adjust current, voltage, and power dynamically over time. An example of this can be seen in Figure 3. The Federal Urban Driving Schedule (FUDS) data, for instance, is a test cycle standard pattern that tests EV batteries under conditions similar to driving around in a busy city with irregular bouts of acceleration and regenerative braking. Profiles such as these can consist of hundreds of thousands of data points at very short spacings on the order of milliseconds.

Similar to the more traditional CC-CV battery cycle testing, these setups will require a DC source (programmable DC power supply) and an DC load (electronic load or resistive load bank). However, the DC source must be able to exercise these complex driving cycles — a task that not all power supplies can accomplish. The PXB has the benefit of being able to perform these tests while also including a regenerative capability allowing manufacturers to sink their power with this piece of test equipment. Essentially, the PXB can act as both the source and the load for these specific tests.

The PXB allows preset operations such as these to run continuously. A total of 30 programs can be created and linked for up to 10,000 steps with all of the programs combined. Moreover, if there are repetitive windows of voltage over time, specific programs can be looped (Figure 4). All of these programs can be stored within the PXB’s memory, and any data can also be exported to a USB flash drive from the front panel.

Complex waveforms such as these are easily uploaded to the PXB via its built-in web server with a CSV file, containing the steps a user wants to run. The user simply selects the required sequence and runs it to perform their test (Figure 5).

For added flexibility, the built-in web server allows for the remote test monitoring by providing the same information of the front screen such as status and activity information for each PXB supply (e.g., set current, set voltage, power, slew rate, AC line power, measurements, output settings, memory, system settings, etc.). The utility of the sequence characteristic is not limited to battery testing/validation. This can also be useful for compliance testing automotive power supply units such as inverters and converters and other automotive components as per LV123, LV124, or LV148.

Not only can the PXB automate and remotely monitor a complex battery performance test, but as the test is running, the PXB is able to record data through an integrated data logger (Figure 6).

A recording is easily set up on the PXB and can be downloaded later for analysis. Internal measurements such as time, voltage, current, and power can be obtained from the data logger to extrapolate information such as battery capacity, SoC, etc., without added expense and the need to add more equipment to the setup. This also removes the need for expensive test bench software or the complexity of custom programming. The data is recorded in a CSV format for further analysis.

The PXB series can automate complex battery performance tests by easily uploading dynamic load profiles such as synthetic driving cycles. This is achieved simply by uploading a CSV to its built-in server and running it. The test can be easily monitored remotely to ensure that everything is operating smoothly. All internal PXB data can also be recorded via the PXB’s integrated data logger, which allows time, voltage, current, and power data to be stored and viewed easily at a later time for any required analysis.