The step-up method is an experimental technique
for determining the impulse dielectric breakdown voltage of an insulator
which is not reusable after breakdown. This method often does not
yield reliable data because it depends on the applied voltage difference
d between two consecutive steps and the number m; of repeat tests
at a voltage step. This paper proposes a method of estimating d-
and m-independent breakdown voltage distribution, and also an improved
step-up method for optimum testing conditions. This estimation assumes
that breakdown voltage will follow the Weibull distribution with
parameters of scale ~, shape fi, and location y. If all these parameters
are unknown, the estimation error will be too large. If either n
or y is known, the 40, where 02 is the variance of the optimum condition
is that m = I 0.6 ~; d/a ~~ 1 4 andN ' = ' ' ' nlm Weibull distribution
and Nmin is the smallest number of specimens when the maximum likelihood
estimation is valid. Ifflis known the optimum condition is that
m = I d/a イ 1 4 and N ' = 20. ' ' ' ' mln These conditions are similar
to those one obtains when assuming a normal distribution pf breakdown
voltage. This rn value is markedly differentfrom that (rn = 3) in
conventional step-up tests.
|