CMM Capability Study (MSA Type 1) Excel Needed

[[Ry...]]

Good Day All,

I've been asked by a customer to provide a GR&R Study of our CMM.
I understand that the most common study would be an MSA Type 1, or a repeatability study.
Has anyone done this?
What was your procedures? (What type gauge, What type of Probe, How did you program, What type of strategy and filters used)
Is there an Excel Spreadsheet available?

Thank you for all your help,
Ryan

[[Me...]]

Please sign in to view this quote.

PiWeb has a built-in study for this if you have it. I've only dealt with the CPK results but I'm sure folks on here can point you in the right direction.

[[Ry...]]

Please sign in to view this quote.

Hi Shawn,

Yes, PiWeb does have that, but only if one has the Reporting Plus license, which I do not.
So I'm left with excel data dumping.

Thanks,
Ryan

[[Ri...]]

A Type-1 is only a repeatability check. Some people will run a part 50 times, or only 10. The n amount of times is completely up to you. You are typically only checking repeatability of the machine/software with this test. With this one you typically only focus on the range of the deviation and calculate the percent of tolerance usage. Most likely if you cannot achieve your goal with a Type-1 (say 10%) you are not going to pass a full blown MSA.

The next is a full-blown MSA where you take 10 parts, 3 operators, and measure each part 3 times per operator. In this case you are testing not only the machine and software, but also your fixture. In this case, you will need software to calculate the GR&R values for you. PiWeb has its own built in functionality, but there are Excel plugins, MiniTab, etc that you can use.

As for the measurement strategies, that is completely dependent on the application. There is no one size fits all.

[[Wo...]]

Our gage R&R studies were usually for a specific dimension on the print.
We would use the same Probe/Evaluation etc. as our plan that's checking the part normally and then just plug the data into a standard gage RR form, in that case it was a customer's form.

[[Ry...]]

If I loop the alignment, does that not take the fixture out of the picture? Then in that case, it is just a repeatability check?
Do people rely on the fixtures to accurately hold the base alignment origin from part to part?

I was thinking of using a 1" ring gauge for running the MSA.(Its close to the critical sizes of the part I will be checking)
In that case, what would I use for a tolerance? I'm looking at the calibration report for my Duramax, and it says:
Error for Length measurements is 2.7+L/250um
Probing Error limit 2.4um
Scanning Probing Error 3.8um

Thanks,
Ryan

[[Je...]]

Hi Ryan. Just chiming in. I think the feedback provided above is excellent.

If a customer had requested an R&R study from my company, we would want to meet or exceed their expectations. Although, in most cases our annual Zeiss-supplied calibration per ISO 10360 has been sufficient, even for aerospace customers. However, we are a low-volume, prototyping company. Perhaps R&R would be of more value for high-volume, multi-user situations. Perhaps you could have further discussion with your customer to find specifics of what they are requesting.

As far as your question about base alignment looping, it's important to remember that in Calypso, base alignment is used primarily for navigation and only secondary for evaluation. I always loop the B.A., but most of my measurement data that makes it to the report is constrained in a secondary alignment. Even when my base alignment uses identical features in a part's Datum Reference Frame, I always re-probe these features at a higher scan density and slower speed when capturing data for evaluation. At my previous company, the other inspectors poked fun at my obsession for accuracy and said things like, "Hey, we're not building parts for NASA," but now at my current company, we literally are building parts for space applications, so every micron counts.


Jeff Frodermann
Windings, Inc.
New Ulm, Minnesota

[[Da...]]

Hi Ryan, looping your alignment will help the software fine tune the fit between your actual part and the model however your fixture will still play a huge roll. Ive done hundreds of GR&R studies at my job and based on what I have encountered I can tell you the more robust you can make your holding fixture the better your study will turn out. going back to what Richard said your GRR will test your machine, software, and fixture as well as the operators involved in the study. I sometimes run into issues where we will complete a GRR and 2 of the operators results line up perfectly but the 3rd operators results are different from the first two. You then have you ask yourself what is operator 3 doing differently. is the operator loading the part into the fixture differently? are they applying to much tension to a clamp? is the part sitting up against the stop all the way? did a piece of modular componentry loosen up? these are all variables that will either make or break your GR&R. If you have a fixture already designed that you think will do the trick I would start by doing a Type 1 Study. I personally like todo 2 different Type 1 studies My first study will be fixturing one part and looping the program 10 times without taking the part out of the fixture. ideally your looking for minimal difference between all 10 cycles. Afterwards I will take the same part and run it 10 more times but this time I will refixture the part in between each cycle. Like Richard was saying if your results are not under 10% of the total tolerance then your probably not going to pass the GR&R. switching gears to your program depending on what type of dimensions you are inspecting use fitting algorithm's when you can and watch your scan speeds as well as point density. I hope this post helps in some way shape or form. Good luck on your GR&R.

[[Er...]]

Personally, I would get feedback directly from the person requesting this to be done. A phone call or email could clarify what's actually needed before doing a lot of work they may not want. It may be needed for a certain dimension for someone down the line. And they may have a form for you to fill out. Asking if they have a preference can't hurt?

I'd send them the Zeiss calibration report and see if that is enough. They may not know what they are asking for too. Or it gets mistranslated between people. Had that happen more than once.

[[Ry...]]

Thank you for all the info so far.

My fixture if very robust and should not come into play at all. Its a basic 3-2-1 setup in that the part sits on 3pts, and is stopped in X by 2pts and Y by 1pt. Its a 5lb part, being measured with 2mm & 3mm probes. So the part doesn't rock and wont move by probe force. And I'm the only operator.

So I tried a MSA Type 1 study, of 25pts and 60pts. Using a couple different excel sheets that I found on the web. I have no idea how accurate these sheets are, and they are very limited and sheet protected. So I can't even manipulate or modify for my use.
Using my 1" ring gauge, I gave a tolerance of +/-0.0001". On the 25pt study, it came up with 3.8% and on the 60pt study it came up with 5%. (Maximum deviation between measurements was 0.000009")
Seeing as the tightest tolerance I'm measuring is +/-.0005", I'm well within a repeatable acceptance. (Less than 1% if using part tolerance.)

So I think I should be good in terms of what I'm measuring for them. We're just a small shop running low volume parts. Our customer is a multi-million dollar facility with tons of people to analyze data. I think this is just a general cookie cutter PPAP request they send to all vendors no matter the quantity or part. Its unfortunate that they can't tailor the PPAP depending on the vendor/part. And trying to get a hold of them for answers is next to impossible.

Dakota, are using excel sheets for all your GR&R's and MSA's? Are you able to share any?

Thanks again for everyone's help,
Ryan

[[Da...]]

Hi Ryan We use Minitab for pretty much everything at my work. it may be a good investment for your shop to get a seat of minitab. alot of big companies are starting to use minitab more and more each year. Ive been working in some sort of inspection/metrology role for going on 12 years now and when I started at my current job 7 years ago only a handful of our customers used minitab for analytics. Now more than 80% of our customer basis use mintab and the majority of them require MSA of some sort. HTH

[[Ry...]]

Thank you for the info Dakota,

Now I guess the million dollar question is, does the upgraded version of PiWeb offer the same capability studies and graphs that Minitab does? The price between each package is similar, so its a matter of whats the bestbang for the buck.

Thanks,
Ryan,

[[Ow...]]

It's been a long time since I've used mini-tab and I've never used the advanced version of PiWeb.
If you're the only one using it, PiWeb might be just fine.
However, the big issue often is if somebody else like the customer, external quality personnel or process capability engineers will have to have the same software to evaluate or, for a lack of a better term, manipulate the results.

[[Ch...]]

You can get QI Macros (an excel add-on) for $299 bucks and it will do type 1 and type II gage R&R's.

[[Ro...]]

I am almost certain that we also use minitab, i know ive heard that word kicked around in meetings that i was trying to escape from.
One thing that our QE's would almost always try to do is 3 to 5 of the 10 parts to be out of tol, or at least be very high low in the tolerance range.
I don't know if this was any kind of requirement but i know that simply grabbing 10 parts that are pretty much identical could cause problems mathematically for the study.

As for what dims to use, we would use Critical dims for whatever part we would study using whatever tips we normally use.

[[La...]]

Please sign in to view this quote.

Yes, you are supposed to use parts that span the tolerance. It reflects in the # of Distinct Categories value. AIAG requires 5, i believe.

[[Da...]]

Hello. I use Calypso 5.4 at the moment. Any chance to do study type 1 in an easy way ? If not does anyone have excel sheet I could use ? How do you normally transfer 10 measurements into one excel with results div % to show managers ?