ValeryD wrote:
1/20 wave? Yes? Did I understood you correctly?

Anybody else believe in such a statement?


No, but this doesn't necessarily mean I believe the manufacturer is being
intentionally dishonest.

As Dan McShane pointed out, stating a surface accuracy without the details
of the test procedure isn't particularly useful. Aside from the obvious
issues like testing wavelength, rms vs. p-v, etc., you need to know how the
interferometric data reduction is performed. Take the Zygo GPI software for
example. There are user inputs like "trimmed", "noise", "filter", and
"aperture percentage". These variables have legitimate purposes, like
eliminating unused edge areas from the test, but if set wrong they can give
overly optimistic numbers. Also, some software relies on the user to mouse
click on the fringe pattern to define the shape. This is reasonably
accurate and repeatable up to a point, but not for 1/20th. I can take the
same fringe pattern and make it read wildly different numbers depending on
where, and how many, points I define.

Your point is well taken, though, Valery. 1/20th is thrown around far too
commonly.

Sincerely,
Bryan Greer