Pick up a book on statistical process control... a simple one (stay away
from those textbooks that confuse everything).  These will include the
standard formulas and the Log-Normal distribution analysis.  The beauty of
SPC is it is independent of the statistical model (which means the
Log-Normal is only for the dog-and-pony show).

Every industry has it's own standard for yield.  In my past life I had to
demonstrate the manufacturability of a process used to mount chips onto flat
panel displays.  I used the "Motorola" standard (suggested by an ex-IBMer).
Their standard was using 4.5sigma (which is equivalent to 3 ppm).  A minimum
for my comfort is ~64 samples, but I'd recommend hundreds if you can afford
it.  If you are doing x-sectioning, you can relax this standard.  Basically,
the issue is whether you believe your results represent your sampling
(because anyone can distort SPC with preferential samples).

I see you're in Colorado.  Do you have contacts at HP for advise or the
former NCR/AT&T fab?
Carey

-----Original Message-----
From: LI YUAN <[log in to unmask]>
To: [log in to unmask] <[log in to unmask]>
Date: Wednesday, September 22, 1999 3:59 PM
Subject: [TN] Sample size for board level solder joint test


>I and one of my colleagues in reliability group were arguing the sample
>size and its statistical significance. Can you help us? Our arguments are:
>
>1) What is the sample size industry typically use for BGA solder joint
>reliability test? 32, 40 ...? Why use this size?
>
>2) Is lognormal distribution a good one to fit the test data? If yes, is
>0.1% value of the lognormal fit a good representation of failure free life
>of the whole population?
>
>Appreciate your help in advance,
>
>Dr. Yuan Li
>
>##############################################################
>TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
>##############################################################
>To subscribe/unsubscribe, send a message to [log in to unmask] with following
text in
>the body:
>To subscribe:   SUBSCRIBE TECHNET <your full name>
>To unsubscribe:   SIGNOFF TECHNET
>##############################################################
>Please visit IPC web site (http://www.ipc.org/html/forum.htm) for
additional
>information.
>If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or
>847-509-9700 ext.5365
>##############################################################
>

##############################################################
TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
##############################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in
the body:
To subscribe:   SUBSCRIBE TECHNET <your full name>
To unsubscribe:   SIGNOFF TECHNET
##############################################################
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information.
If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or
847-509-9700 ext.5365
##############################################################