Program verification

[[Cl...]]

Verifying a CMM (master) program by measuring the same part with another (alternative) method, i.e., comparator, Smartscope, etc. Results between both methods must be within 10% of the tolerance. For example, tolerance is .005". I get bonus tolerance as well. Let's say another .005". Is my 10% now .001?

[[Ri...]]

Great question.

I think I'd treat it the same way as you treat capability analysis where you convert the results to a percentage of tolerance usage instead of the raw result.

[[Ch...]]

If you need statistics, then consider the following:

Use Cg or possibly Cgk as your statistic. Cg is like Cp; and Cgk is like Cpk; but for gages. You want both to have similar Cg and Cgk numbers. A good measurement is above 1.33. FYI, this is just a type I gage study.

https://support.minitab.com/en-us/minit ... age-study/

Then I highly recommend F and T test.
F test to check if variance is same
Then T test to check if means are the same

Output box and whisker charts to visualize if needed.

[[Cl...]]

Please sign in to view this quote.

I ran 15 replicants through minitab and the cg was 0.35

[[Ch...]]

This is 15 parts measured from the same instrument? If so, then that is saying your measurement is not very repeatable. Even one outlier will throw off results.

If you have a tight tolerance, then Cg and Cgk can be hard to get.

[[Ch...]]

Here is an excel sheet that has Cg on it. Tables from Calypso and many other SPC softwares copy and paste easy into it.

As noted in the sheet, Excel Cg and Cpk will never match Minitab perfectly.

!BLANK_Lab_SPC_Worksheet.xlsx

[[Cl...]]

Please sign in to view this quote.

Thanks Chad. This will come in handy. I ran the 15 samples through
your sheet and the cg results were pretty close but not identical.

[[Ja...]]

Please sign in to view this quote.

Chad, my dude!

While we do have actual SPC software (MeasurLink by Mitutoyo) I haven't been required to use it yet by any customers and it can be a bit wonky to set up and time consuming to run during production. I have been trying to wrap my head around getting SPC data to work in Excel with the right formulas in the right places for a quick and dirty overview and your sheet is perfect for that application. Thanks!

Man, I love the learned minds on this forum.

[[Ch...]]

Please sign in to view this quote.

Please sign in to view this quote.

Please sign in to view this quote.

With the interest; I made a major improvement. I added an If statement to automatically select correct formula based on tolerance. This is important since it is different between one sided and two sided tolerances. New sheet is attached.

Note: All excel Cpk formulas are actually Ppk. Therefore, they will match Ppk more closely in Minitab. This is a minor difference. Search Cpk vs Ppk on Elsmar Cove if you want to go down rabbit hole.

[[Ch...]]

Please sign in to view this quote.

Please sign in to view this quote.

Please sign in to view this quote.

Calypso forum bugged out and wouldn't attach the file. Here it is.

!BLANK_Lab_SPC_Worksheet 4.19.24.xlsx

[[Ri...]]

Please sign in to view this quote.

I've been out of the stat realm since I left manufacturing, but I had gone down a rabbit hole in the past about Cpk vs Ppk when we had first gotten PiWeb all the way back in 2013.

I didn't realize at the time that Minitab didn't really follow the standard for Cpk analysis when the subgroup was 1. They used xbar for the standard deviation calculation which is not what you would normally do. When the subgroup size is 1, Cpk and Ppk should be the same results, but with Minitab it won't be. This is because Minitab hangs on to the idea that Cp/Cpk should be about potential capability, so they use x-bar to try and calculate an in-between subgroup standard deviation.

Anyways, I don't miss statistical analysis at all. Lol.