Backlash is the movement between the gear teeth that allows them to mate without binding. Backlash can cause large torque fluctuations in vehicle powertrains when the input torque changes direction. These fluctuations cause a jerk and shuddering, which negatively affects drive quality. Input torque frequently changes direction in electric vehicles due to regenerative braking. Limiting zero crossings is an option for better drive quality; however, this leads to decreased vehicle efficiency. Because of this, modulating the torque through the backlash region is preferred, yet, if done poorly, it can result in sluggish torque response.
This paper proposes a torque-shaping algorithm for an electric motor and gear/differential system to reduce backlash in electric vehicles. The control algorithm modulates the commanded torque’s rate of change based on the vehicle speed and zero-crossing torque. The torque change is dynamically de-rated over the backlash region, and the gears are gently re-engaged to transmit the torque. The algorithm considers the variability in electric machine designs and e-gear drives and provides calibration tables to change the base rate. This rate is modulated and filtered before applying to commanded torque in both directions. The initial parameters were calculated using model-in-loop simulations. They are later calibrated in closed course testing with specific tip-in and tip-out conditions at various speeds and accelerator pedal positions.
The algorithm was applied on a modified 2019 Chevrolet Blazer with a P0-P4 parallel hybrid architecture using a GVM210-150 Parker Hannifin motor connected to a BorgWarner single-speed gear unit on the rear axle. The algorithm reduced shudder during tip-in and tip-out torque commands from the driver, which yielded a 70% reduction in jerk while maintaining responsive control of the vehicle. The vehicle was subjected to various customer-focused evaluations as part of the DOE/GM EcoCAR program.