Transient thermal analysis (TTA) with the measurement of the single pulse thermal
impedance (Z(th)(t)) is a standard method to verify the thermal integrity of power
semiconductors modules. For best evaluation of measured data, the signal to noise
ratio (SNR) should be as high as possible. Especially in the early time domain, it
is difficult to achieve high SNR because of the required high bandwidth. Most common
way to increase SNR is averaging over several TTA measurement repetitions. Since the
semiconductor module therefore has to reach thermal equilibrium, this solution is
very time consuming. This paper introduces a new averaging algorithm for TTA, wherein
several short deterministic pulses are applied to the semiconductor before standard
TTA. Over superposition, the influence of the previous pulses is removed from all
short pulses and averaging is possible without reaching thermal equilibrium. Result
is a standard single pulse Z(th)(t) and not a duty cycle form of it. The algorithm
is tested by simulations and experimentally using automotive LEDs to verify feasibility
and demonstrate benefit. Thereby a SNR increase equivalent to 33 repetitions in standard
TTA was reached.
