Specialized Battery Emulator for Automotive Electrical Systems Thorsten Baumhöfer, Wladislaw Waag, Dirk Uwe Sauer The complexity of the automotive electrical power system is continuously increasing and thus also the effort for design and verification. In micro hybrids for example one has to find the optimal balance between fuel savings and battery life. But most important, care has to be taken considering voltage quality, to always be able to supply safety-related systems. The battery behavior is dependent on a lot of conditions (state of charge, temperature, age, …), which leads to high testing complexity. It is therefore reasonable to use tools to help with the design and verification process. In this work a novel tool was engineered, a combination of a power electronics device with a simulation model, targeted on the emulation of SLI batteries. Behavior / Model Parameter Stimulation Time effort / Cost Simulation Idealized due to computation restrictions Defined by software, easy to change Possible to change battery technology Synthesized No Randomness Low Emulation Dedicated DSP for the model Advanced algorithms possible Limited by implemented model Real input data Feedback without randomness Medium Battery Test Real behavior Changes only with conditioning Not reproducible Real feedback High Concept Requirements An emulator combines the benefits of a simulation model with the benefits of a real test setup. By controlling a voltage source by a simulation model the emulator provides a virtual battery with adjustable state of charge, age, temperature and so on. The battery emulator gets supplied by two standard lead-acid batteries and provides a virtual battery, adjustable by software. 50 100 150 200 250 -800 -600 -400 -200 current [A] time [ms] ~ 200A/ms Exemplary current profile at the battery during cranking. The requirements for the emulator are derived from the battery behavior in the considered applica-tion. In SLI batteries the roughest conditions apply during cranking or other high power pulses like electrical power steering. Highly dynamic 500 A/ms current slope 5 ms settling time Mains independent, power supplied by standard lead-acid batteries, mobile operation possible. A power electronics device in combination with a simulation model provides the desired terminal behavior. Automotive compliant layout, malfunction monitoring, fail-save mode. Has to be able to carry charge and discharge currents. For precise emulation the voltage error has to be within ±30 mV. 200 A continuous current, 1200 A maximum cranking current (1s). Hardware Bidirectional DC/DC converter. Unconventional combination of low voltages and high currents. Low inductance to reach high dynamics, high switching frequency to reduce ripple current. Multiphase design to reach the desired precision (eight phases, interleaved) and a small package. Thermal heating during cranking gets buffered by the large thermal case capacitance of the transistor. A view inside the prototype hardware. The multiphase design is clearly apparent. The inductors have to be quite large to not saturate at the high cranking currents. DSP Battery- Model Control/Logging Output current Voltage setpoint Phase currents Gate signals Converter control FPGA Fault diagnosis Output voltage Supply voltages Temperature Control and fault diagnosis of the converter are completely integrated into programmable logic. The complete processing power of the signal processor can be used for the model. High control frequency for small response time achieved by programmable logic. Non integral relation between control and PWM frequency to distribute limit-cycling. (222 kHz / 125 kHz) Simple PID-type voltage control, P-type phase current control. Model The size of the Battery Emulator is kept small to be able to use it in cars without the need for large modifications. Above seen together with the supply batteries in the trunk of a small car. Based on a dynamic electrical and thermal model previously developed at ISEA. Impedance based with physical submodels for time based processes like gassing, electrolyte transport and charge acceptance. Parametrization by measurements, but scaleable with capacity and cold cranking current. Easy transfer of model changes due to implementation within MATLAB® Simulink and use of automatic code generation. Exemplary impedance spectrum of a lead-acid battery.