Hi Steve:
     
     Here are my replies to your questions:
     
     1.  What would be considered a significant variation in the 
     cleanliness level?
     a.  What variation should I see board to board of the same part number
     coming off the same process?
     b.  What variation should I see board to board of different part
     numbers, but using similar parameters in the soldering process ?
     c.  Has anybody studied their flux and soldering parameters to
     determine when they see large increases in residues due to adverse 
     affects on the cleanibility of the flux caused by excessive heat and 
     dwell ???
     
     The significant variation in cleanliness level will become evident as 
     part of your statistical analysis. Have you done or do you have access 
     to a statistical terrorist, for the setting up of a design of 
     experiment (DOE). The use of a balanced DOE with controllable factors 
     will tell a great deal about your process. You will need to "fix" some 
     of the process variables - not necessarily lock up the process 
     variables you are not investigating at one level but understand how 
     these "other" process variables change while you conduct your DOE. 
     Things such as cleaning system load, type of pwb materials that are 
     being processed, operator induced changes, etc., need to be 
     understood. You may want to break up your process variables of 
     interest into smaller chunks and do sequential DOEs to keep your 
     sample sizes small.
     
     2.  What test methodologies would be best used to determine 
     cleanliness level for this effort?
     
     A DOE is only as good as the response measure used. Many of the 
     studies I am familiar with have used the Omegameter, SIR coupons, 
     HPLC, spectrophotometry, or some combination of these tests. I would 
     recommend using the Omegameter to get an understanding of how your DOE 
     results relate to an every day process control check, and HPLC to get 
     a very detailed view of what chemicals are not being removed in the 
     cleaning cycle. The combination of the science and the practical 
     should give you a good idea of which set of process parameters give 
     you the best cleaning. It will also give your statistical terrorist 
     variables data to analyze and cut down the number of samples you need 
     to process.
     
     3.   Has anyone done a gage R & R on their test methods?

        The gauge R&R on HPLC is very good and the same can probably be said 
for most analytical tests but I don't know if the Omegameter falls into that 
category (I suspect not). 

The IPC specification PC-90 on SPC has some good information on DOE 
methodology. Also email Doug Pauls at CSL ([log in to unmask]) for a lively 
discussion on possible test methods.


Dave Hillman
Rockwell Collins
[log in to unmask]



______________________________ Reply Separator 
_________________________________
Subject: Assy:SPC of cleaning process
Author:  [log in to unmask] at ccmgw1
Date:    6/2/96 8:57 


We are reviewing our alternative cleaning process for application of SPC 
methodologies.  The first system to be subjected to the microscope will be 
an in-line water cleaner using a water miscible solvent in the wash.  During 
the course of our evaluation we will attempt to determine the variation in 
the temperature, spray pressure, bath concentration, bath loading, belt 
speed, and incoming water quality.  We also intend to attempt to determine 
the impact of the variation of each process element as well as determine any 
significant interactions.  We have the following questions for comment:
     
1.  What would be considered a significant variation in the cleanliness level?
     a.  What variation should I see board to board of the same part number
coming off the same process?
     b.  What variation should I see board to board of different part
numbers, but using similiar parameters in the soldering process ?
     c.  Has anybody studied their flux and soldering parameters to
determine when they see large increases in residues due to adverse affects 
on the cleanibility of the flux caused by excessive heat and dwell ???
     
2.  What test methodologies would be best used to determine cleanliness 
level for this effort?
     
3.   Has anyone done a gage R & R on their test methods?
     
I would also appreciate hearing from anyone who has conducted such a study, 
with information on the engineered limits you established.
     
Your responses are appreciated
Steve Mikell
SCI Systems, Huntsville, ALA
[log in to unmask]
Steve Mikell, Process Engineer, Soldering & cleaning processes 
SCI Systems Plant 13
[log in to unmask]