Considerations for achieving reliable and statistically relevant power cycling results for SiC MOSFETs in discrete housings
Year: 2026
Type: article
Abstract: This paper presents initial findings on the number of test specimens required to obtain statistically reliable conclusions about their power cycling capability. The precondition is that the power cycling station operates with a very low thermal condition deviation for each sample. Therefore, 120 discrete SiC MOSFET devices (60 mΩ, 1200 V, TO-247) were selected and tested under nearly the same test conditions until End of Life (EoL). All devices exhibited the same failure mode (bond wire deterioration), which was characterised by an increase in forward voltage drop (V DS,cold – measured immediately after switching on the load current) by 5 % as well as in the failure analysis. The variation in lifetime is very high (up to factor five – normalized values) which is typical for discrete test items [1-3]. Furthermore, a minimum sample size of 35 randomly chosen devices is necessary to make a reliable statement about the power cycling lifetime (63.2 quantile of the Weibull distribution) out of a total sample size of 120 for the investigated device. This resulting sample size ensures that the estimated lifetime falls within an interval of ± 5 % around the target lifetime of the entire population.
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,500
OpenAlex ID: W7223549261