Analog fault simulation in automotive radar 77 Ghz circuit for safety requirements

Abstract

Nowadays, many electronics are being embedded in road vehicles to assist drivers while they are conducting, and potentially, to prevent accidents. One of the most popular technologies onboarded in vehicles is radar, which detects targets with good precision. However, all systems embedded in cars must go through safety checks, because during the device’s life cycle, it can present random failures. Thus, to reduce and mitigate the risks of electronic failures, some methods must be applied following safety guidelines, such as fault simulation. This is important because, due to safety reasons, it is assumed all faults are dangerous, and will generate failures in the system. However, by simulating, it can be seen which type of failures the device presents and the precise distribution of “dangerous” and “safe” faults. Hence, in the radar device, inside the transceiver block, “bridges” and “open” faults were injected in both voltage regulator and frequency doubler blocks. To model both “bridge” and “open”, a 10Ω and a 1GΩ resistor were used. Each fault was injected to interact with each terminal of all component elements inside the studied blocks. Legato, a Cadence simulation tool, was used to execute those simulations. For the voltage regulator, 2981 faults were systematically simulated, and 3 types of failures were identified, such as undervoltage, overvoltage, and oscillation outside the safety voltage range. Besides that, 78.87% of faults were “safe” and did not affect the main operation of the voltage regulator. For the frequency doubler, the simulation software presented an error to execute the high-frequency analysis. However, considering this block demands high computational power for each simulation, and in total is expected to simulate 4607 faults, the testbench had to be optimized. After the optimization, the single run time decreased from 45 minutes to 20 minutes. Then, once a new software version is available, fault simulations can be performed to identify the failures in the frequency doubler block.