Using Q-YIELD™(Page 2 of 4)
There are 1,830 possible scatter plots of two variables.
There are 35,990 possible plots of 3 variables, (assuming
you can find a good way to visualize them). Supposing that
you only take 15 seconds to create and view each plot, and
that you ignore plots of three or more variables,
that is about eight hours of solid work.
And that is assuming that you don't miss anything.
If you start work now and aren't interrupted, you might
just finish before the start of your next shift.Time to
look for a better solution.
You start up your copy of Q-YIELD and import the above
data set. As a first step, you examine the distribution
of failures per wafer.
(Click image to enlarge)
No obvious clues here. But it does look like there might
be a Poisson distribution of failures (which is what you
were expecting) with a secondary distribution imposed on
the tail of the Poisson. Consequently, you set up Q-YIELD
to ask the question: under what conditions are there
more than 200 failures on a wafer?
Previous page 
Next page 
|
