Increasing number of ECU's (Electronic Control Units) being used in modern vehicles have given rise to HIL (hardware in the loop) testing, and model based design approach to design/test ECU's even before the proto-type vehicle is ready. However, it is not uncommon to discover surprising system design lapses during actual vehicle operation reported after vehicle launch. Major cause behind such lapses are found to be the gap between actual field performance/robustness of various vehicle sub-systems interfaced with ECU's and those modeled as ideal cases during HIL testing in the lab. This creates a need to evolve effective vehicle-level validation strategies to expose such performance deficiencies of real life sub-systems provided by the vendors.
This paper describes a new approach to validate ECU in real time. Real time validation approach is based on validation of software and hardware and also analyzes software execution flow through the use of available communication channels i.e. k-line, or CAN communication. ECU software is written with test points as memory bits at the different levels of software - input stage, process stage, and output stage. Using data transfer services of the protocol, status of all the test points are being stored and send to the tester tool developed on labview platform which in turns provides information of the software execution process.
The idea presented is supported with a real life case study involving vehicle validation of “Start-stop controller for micro-hybrid application”. It illustrates how the Start-stop sub-system passed all the performance tests during HIL validation but displayed erratic behavior occasionally during actual vehicle trials. The validation strategy described above was gainfully applied, which quickly helped discover erratic behavior of instrument cluster's vehicle speed output as the root cause. Relevant corrective action could pre-empt costly recalls/field retro-fitments which otherwise would have hampered the vehicle program after the launch.