I have been in this argument before.  I believe my opponent and I
reached a stalemate.  I bought the pizza, he bought the beer.

However,

I typically use:

        [(max-min)/(max+min)]*100               (1)

I would guess you are arguing for or against this formula versus:

        [(max-min)/(2*avg)]*100                 (2)

If the population is mean centered, max + min will be equal to twice the
average value -- in this case you have mathematical equivalency.

Eq. 1 will give you a case which is insensitive to the distance of the
max and min values from the average value.

Eq. 2 is obviously sensitive to the mean value of the data set and
outliers have a smaller effect.

In the end, if you have a representative data sample, either method is
satisfactory and should only differ in the third significant figure or
so.  At that point, you need to consider the capability of your
mesaurement system and probably even check your statistical process
capability values.

I have most likely jumped the shark by now.

Best Regards,
Garrett Oakes

EV Group
invent * innovate * implement
Regional Sales Manager North America - Direct: +1 (480) 305 2443, Main:
+1 (480) 305 2400 Fax: +1 (480) 305 2401
Cell: +1 (480) 516 6724
E-Mail: [email protected], Web: www.EVGroup.com


-----Original Message-----
From: GQ Chu [mailto:[email protected]]=20
Sent: Wednesday, January 28, 2009 10:32 AM
To: General MEMS discussion
Subject: [mems-talk] PR Thickness Uniformity Calculation

Hello:

I need your help. What is the formula you use to calculate your PR thickness
uniformity across the wafer? Where is the reference or source for this, is
it ANSI standard calculation or SEMI standard calculation? Can you please
help me site the source? I'm having an argument with few people how to
calculate it. What is the industry standard and where can I find that
source to calcualte it.

Thank You Very Much,
GQ Chu