Task to build PAT rule by JMP

[#1]

Hi Team,

I am asking this is because we are going to implement dynamic PAT (part average testing) to the new product.

Here is the idea temporarily.

1. Load the data table to JMP . 

    Idenity the item we want to show +/- 6 sigma by JSL. Which funciton / plot is suitable from JMP if I would like to rule out coordinate out of +/-6 sigma?

2. One the JMP rule of the coordinate, how can I out the coordinate to the simple file? 

3. All of the above that I would like to automatically like daily task.

Thanks for all assistance, I am sorry that I try to search in the community but I can't find the answer, please teach me how ot handle this .

[#2]

Hi Lichun,

Is this the same issue or a different one than the other you posted? 

Task to build control chart with automatic warning - JMP User Community

We're happy to help, but want to be sure we're not answering the same question twice. :)

[#3]

Hi Jed,

Thanks for asking.

This will be different case.

Another case is for SPC control chart to monitor fab process like WAT. I will use the current rule in JMP like 1S,2S,3S.... 

PAT (part average testing) is only implemented in Sort database. I am sorry to make you confusion.

The idea here is we have like 100 hundred test in wafer sort that we would like to apply average +/-6 sigma for the good unit (each of data point have it's own value, means we can get distribution for each of items).

For automotive product, follow the JDEC spec that we want to rule out the passing die which is outlier in distribution even it pass spec.

Hence, I would like to get support from JMP how I can rule out the die which is out of +/-6 sigma then print out to a file 

Attached one distribution as your reference. the point here is to rule out the data point which is out of mean +/- sigma.

Then print out the file with the coordinate.

I try to attach jrp file but this was rejected by system. I provided you snapshot and sample of raw data. 

[#4]

I'm not certain I understand how you're grouping your data, but I suspect the Analyze...Screening...Explore Outliers platform might be what you're looking for. I attached a JMP table with your data, with a script that:

1. Runs the Explore Outliers platform, assuming you want to look at outliers in the icc_aux... column for each grouping of DIE-Y (I'm likely grouping this wrong for your use case).
2. Specifies to select ± 6 Sigma, Huber method, and launches the Robust Fit Outliers tool
3. Selects all the outlier rows in the original data table.
4. shows a little graph with the outliers selected

[#5]

Hi Lichun, were you able to develop a JSL script that performs DPAT analysis on a set of data? if you did, can you pls share your script? I think it will be very useful. btw, there are also other algorithms which are also useful to detect outliers such as GDBN ( Good Die in Bad Neighborhood) etc..