Contents I. Introduction and motivation
Future automotive developments are moving towards autonomous driving. For this, elevated requirements - e.g. fail-operational vehicular power system and proper operation of safety-critical components - must be met according to Automotive Safety Integrity Level (ASIL) [1]. A fail-operational vehicular power system requires that in case of a fault (e.g. short circuit or thermal wire overload), only the affected parts of the power system are switched off. Thus, only the fuse may trigger immediately located before the fault [2] and not an upstream or parallel one (selectivity). In future vehicular power systems melting fuses are being replaced by electronic fuses (eFuses) [3]. For wire protection, eFuses trigger according to a software-implemented algorithm. Various algorithms can be used, e.g., the current is measured via a shunt resistor and, based on the current, an algorithm determines the wire temperature using an electrothermal wire model [4]–[8] or the amount of charge flowed through the wire. The eFuse triggers if a certain threshold is reached. The current time diagram can be used to evaluate the selectivity of cascaded fuses by comparing the characteristics of the fuses. The diagram indicates the triggering time when a certain current flows. Fig. 1 shows schematically a static constant current time diagram. For the dash-dotted wire protection area applies: The higher the current, the shorter the triggering time. In the dashed overcurrent area, the fuse triggers instantly without an intended delay. If the characteristics of cascaded fuses do not overlap, they are considered to switch selectively in case of an excessive current. [9] However, selectivity evaluation based on the static constant current time diagram is only correct for constant currents. For a dynamic current profile selectivity violations can occur, even if, there is selectivity according to the static constant current time diagram. This is due to the different calculation methods of different possible algorithms. In this contribution, a methodology for analytic triggering time determination of two easy-to-implement eFuse algorithms on the basis of a generic periodic current pulse waveform is proposed. Moreover, the selectivity of two cascaded eFuses with different implemented algorithms in case of the dynamic current profile is examined to identify possible selectivity violations despite selective behavior according to the static constant current time diagram.
