SPC on TP

[Jan d.]

I was asked to do SPC on a TP call-out. I have a 0.1873" hole with a +/-0.0002" size tolerance and the following TP call-out:

0.0005 at MMC to A (a datum plane), B at MMC (a datum cylinder with a virtual condition) and C at MMC (a datum cylinder with a virtual condition).

I believe that the type of a call-out represents a functional gauge and it either fits, or it does not. So I think it is a go/no-go type analysis and SPC will not be meaningful, if not outright impossible. But of course management would like to know how the can calculate Cp and CpK for this TP.

Anybody any words of wisdom???


Jan.

[kbotta]

With the type of tolerance you specify, and the datum shift needed to apply, and CMM's inherent in-capability of accuratly 'picking up' a constrained oriented feature (or read as constrained actual mating envelope), and the constrained and located AME of your tertiary...I would absoulutly treat this as pass/fail.
The #'s would be meaningless. You simply cannot back up a 'good part' vs a 'bad part'.
Gotta be a hard gage.
my 0.02
Kev

[kbotta]

Oh, and that is a cool avatar. Nice...

[John Kingston]

This has been discussed on the other site

SPC and TP don't mix. You could do SPC on the basics, but you still don't have spec. limits for your chart. There are others here who may have more input than me.

(Like your avatar too!)

[craiger_ny]

The most you could report is either variance or standard deviation of the readins to indicate a process spread but this would hardly be considered SPC in most peoples minds. Anything else would require control limits and charting to first check for an "in control" condition and how you do that with the material condition modifiers on the datums I have no idea. I'd first explain the need to see if the process is in control and ask whomever is asking you for an "SPC" study to explain how to do that. I'd try to baffle the requestor with his/her own request if that is what it takes.

[Paul Sarrach]

Correct SPC and TP does not mix. I am still fighting 3 weeks later fellows with that engineer ( Using that word loosely ) Who wants a 30 pc cap study on all his holes TP. And Radius, and angles of loc of rotion dimples. aggggg.

And Jerry Sienfield would love your avatar too.

[Matthew D. Hoedeman]

TP is NOT an acceptable variable for SPC, with good reason, it does NOT take into account DIRECTION of variation. Example:

TP of 2.0mm for a hole located at X0 Y0 (for ease of example)
Deviations as follows:
Part #1 X-1.0 Y 0.0 TP = 2.0
Part #2 X 1.0 Y 0.0 TP = 2.0
Part #3 X 0.0 Y 1.0 TP = 2.0
Part #4 X 0.0 Y-1.0 TP = 2.0

These 4 parts will ALL show a TP of 2.0 and will show a range of ZERO, thus giving a 'perfectly' repeatable part, but in reallity, they is 2mm of variation in EACH axis. So, will TP work for SPC? NO WAY! With no variation, SPC will not even be able to calculate, but for a simple example, I ran them to the limits. BUT, in any case, TP DOES NOT WORK FOR SPC!

[kbotta]

And Matt is da' man...man...
Sorry, a little cosigning can go a long way

I would agree with all of the above.

kev

[craiger_ny]

I have a customer that it never fails everytime we ship them something new they want a 30 piece Cpk study done on. It is a product we design and we send them an outline drawing of it with mounting dimensions and spline data referenced so they know what to build the mating part to. I do the same thing everytime, I call and ask "Just what feature would you like a Cpk study done on?". One time the guy told me the spline. I said "ok them which part of the spline specification?" the whole time I'm thinking oh god please don't say circular tooth thickness, we didn't get past that part luckily. This is what you get when people drill 30 piece capablility studies into engineers heads in "Six sigma" classes and then unleash them on us. The only one that made out on that one was the guy with the add in the back of Quality Progress magazine saying "I'll come in to you'r company and make you all black belts in one week........"

[kibbe]

When I am directed to use a TP for capability, I will break out the 2 axis deviations and apply the tolerance needed to make the TP work.
Not perfect, but works when I have no other choice.
Explaining how I get the ± deviations is another subject.

[Paul Sarrach]

When I am directed to use a TP for capability, I will break out the 2 axis deviations and apply the tolerance needed to make the TP work.
Not perfect, but works when I have no other choice.
Explaining how I get the ± deviations is another subject.

Been there done that. I just cant get over the ignorance of so many with a degree and title and not have a clue as to what thier job is.

[Johnson]

THAT IS EXCACTLY WHAT PROJECT I AM WORKING ON NOW!! I preached about not using TP for SPC and they still want it. The engineer even had agreed with me of using linear callouts v.s TP but someone changed his mind. So I am running 60 parts as of right now. Hence I had time to catch up on the board.

[kibbe]

The example Matt gives is how I convinced the powers(once) that TP will not work for SPC.
Now, we do have this come up quite often, but I can usually convince them to use the axis deviations, and if it will pass with the tighter tolerance we should be good to go.

[Clay_Man]

I have a really neat excel spead sheet to do a cap study on TP if you want I can e-mail it to you.

[B. Jacobs]

I would like to see it Clay Man.
Can you send it to me?