Inspection Machine Capability Analysis
- March 3, 2008 at 4:58 pm #49497
ScottMember@Rick Include @Rick in your post and this person will
be notified via email.
I am trying to quality a camera-based optical inspection machine. The machine is used to inspect print circuit boards (PCB). The data that I have collected is the number (count) of defects per PCB. I have 10 sample PCBs with known defects. I have run these 10 PCBs through the machine a total of 4 times.
My question is, how do I perform a capability analysis on an attribute machine? I know how to perform one using variables data, but attribute data is trickier.
Any and all help will be appreciated.
Rick0March 13, 2008 at 7:29 am #169635
TriveniMember@Triveni Include @Triveni in your post and this person will
be notified via email.
For attribute data first caluculate the DPU,DPO,DPMO
DPU( Defects Per Unit ) =No.of defects / (Divided by)Total No.of Observations
DPO(Defects Per Unit)= DPU/ (Divided by)No.of Opprtunities
DPMO(Defects Per Million Opportunities) =DPO * 1000000
From the DPMO We Can Caluculate Sigma Value
Ex:- DPMO = 6209
Sigma Level or Process Capability is 40March 24, 2008 at 8:45 am #170003
As I understand you want to guage the Quality of an PCB inspection machine. Now since the primary function of the Optical Inspection M/C is to identify if the PCB is defective or not. So the primary purpose of the M/s is inspection, which means measurement. So in situations where M/s is not doing a productive activity or a process step and is rather doing an inspection test, then we need to consider this M/s as a part of inspection system.
So in this case your problem is to guage the quality of this measurement system of identifying / counting defects using this Optical Inspection M/c. To guage the Quality of the Measurement system, the techniques used is called as Measurement System Analysis (MSA). MSA will help in answering your question of quality of the M/c.
Since your data is a discrete (attribute type) data i.e count of defects, so you need to do an Attribute Type MSA. You can use software tools like Minitab and perform this Measurement System Analysis and choose ‘Attribute type’ option while doing MSA. This attribute type MSA will give you a Kappa statistic. This will tell you how good / bad is the quality of the the measurement system or in this Inspection M/C.
As a benchmark, kappa statistic of >=0.9 is considered significantly good and the Measurement system is considered to be OK. Kappa statistic value of 0.8 to 0.9 is also at times accepted when the Measurements are not that critical. And Kappa value of less than 0.8 indicates that the Measurement System (M/c) does not have a good Quality and need to be corrected before actually using for making measurements. In your case I would suggest that Kappa value of more than 0.9 should be must as the M/C is an fully autometic M/c and the measurements are also critical.
Hope this helps and answers your query.
The forum ‘General’ is closed to new topics and replies.