The Cove Business Standards Discussion Forums
Go Back   The Elsmar Cove Business Systems and Standards Discussion Forums > >
Forum Username

Elsmar Cove Forum Visitor Notice(s)

Wooden Line

Capability for CMM Scanned Profiles on a feature with a Bilateral Profile Tolerance - Page 3

Monitor the Elsmar Forum
Courtesy Quick Links


Links Elsmar Cove visitors will find useful in the quest for knowledge and support:

Jennifer Kirley's
Conway Business Services


Howard's
International Quality Services


Marcelo Antunes'
SQR Consulting, and
Medical Devices Expert Forum


Bob Doering
Bob Doering's Blogs and,
Correct SPC - Precision Machining


Ajit Basrur
Claritas Consulting, LLC



International Standards Bodies - World Wide Standards Bodies

AIAG - Automotive Industry Action Group

ASQ - American Society for Quality

International Organization for Standardization - ISO Standards and Information

NIST's Engineering Statistics Handbook

IRCA - International Register of Certified Auditors

SAE - Society of Automotive Engineers

Quality Digest

IEST - Institute of Environmental Sciences and Technology


Some Related Topic Tags
capability study (part dimensions from a process or machine), cmm (coordinate measuring machine)
Reply
 
Thread Tools Search this Thread Rate Thread Content Display Modes
  Post Number #17  
Old 30th November 2012, 12:06 PM
Rand T

 
 
Total Posts: 54
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Quote:
In Reply to Parent Post by JAltmann View Post

For what your trying to do you'll really want to use single vectored points, and not a line scan as is currently being perfomed. This will allow you to compare the exact same points for trends/fliers ect.
I agree that this makes the analysis of a specific point easier, but from a standpoint of determining the profile of the entire part, the line scans are much quicker and check the entirety of the part.

Either way, I need to develop a strategy for capability using what data I have.

Sponsored Links
  Post Number #18  
Old 30th November 2012, 12:09 PM
bobdoering's Avatar
bobdoering

 
 
Total Posts: 4,097
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Quote:
In Reply to Parent Post by Rand T View Post

Here is part of a spreadsheet with the data for feature #12
Are these several parts, in time order of production?
  Post Number #19  
Old 30th November 2012, 12:16 PM
Rand T

 
 
Total Posts: 54
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Quote:
In Reply to Parent Post by bobdoering View Post

Are these several parts, in time order of production?
Yes, 34 pcs.

The data is the min profile, max profile, and the "measured" value which is merely the difference between the two.

Feature has 0.08 mm profile tolerance
  Post Number #20  
Old 30th November 2012, 12:43 PM
bobdoering's Avatar
bobdoering

 
 
Total Posts: 4,097
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Here is the analysis I made based on your data. If you plot the difference from min to max, you are always going to have a positive value. Your lower limit is zero. I used the .08 as your tolerance, since it is the width of your profile zone, and 0 as your minimum (also identified as a boundary, since you can not go below it.) So you compare your zone usage to your total zone tolerance for capability - and if you do that, you get the attached results. Note that it is a non-normal distribution (used weibull as the distribution) since it is physically unilateral. Also, being non-normal and unilateral, Cpk and Ppk really don't apply - but are estimated for academic discussion.

This may not be my "final answer", but I provide it as a starting point for discussion.
Attached Files: 1. Scan for viruses before opening, 2. Please report any 'bad' files by Reporting this post, 3. Use at your Own Risk.
File Type: docx scan capability.docx (159.3 KB, 52 views)
Thanks to bobdoering for your informative Post and/or Attachment!
  Post Number #21  
Old 30th November 2012, 01:00 PM
bobdoering's Avatar
bobdoering

 
 
Total Posts: 4,097
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Quote:
In Reply to Parent Post by JAltmann View Post

For what your trying to do you'll really want to use single vectored points, and not a line scan as is currently being perfomed. This will allow you to compare the exact same points for trends/fliers ect.

One down side with scanning like this is that you end up with "fliers" in the data. Sometimes dues to controller errors, they are minimal and could possibly be eliminated or greatly reduced with filters.

Any issue that needs to be understood is how repeatable the alignment system being used is. This can be checked easily by loading, running, and unloading the same part thru multiple runs and looking at this repeatability. If you have high repeatability values this will cause other errors with the SPC study in the way of poor results.
I agree - if the data you presented was for parts in the order that they were manufactured, and only one tool machines the parts and all parts were made with that tool (no tool changes) I think there is some measurement error masking your process variation.
  Post Number #22  
Old 30th November 2012, 01:16 PM
Rand T

 
 
Total Posts: 54
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Thanks, Bob

I did a similar calculation (using minitab normal capability; see attached).

I just wasn't sure if that was correct in using the difference between min and max. At first glance I thought it might overstate the variance from nominal.

If I have a part where the profile is +.03 and - .01 mm, it seemed more logical to calculate the +.03 against the + bilateral tolerance of +.04 rather than the min /max difference of .04 against a unilateral tolerance of .08.

I then decided to select whichever of the min or max had the greatest magnitude, but then found that I was bouncing back & forth between + and - which also made the range much higher than it really is.

The main purpose for my question wasn't specific to this part, but a strategy on how to handle capability on profile tolerances where the only info I get on line scans is min and max.

Either way, it seems like I am overstating the variation, but I guess I may have to live with the best approach of the two. My last (and maybe best) alternative was calculating the capability of the min and max individually and then ignoring the one with the best Cpk
Attached Files: 1. Scan for viruses before opening, 2. Please report any 'bad' files by Reporting this post, 3. Use at your Own Risk.
File Type: pdf 20121130115546.pdf (348.2 KB, 54 views)

Last edited by Rand T; 30th November 2012 at 02:30 PM.
  Post Number #23  
Old 30th November 2012, 02:58 PM
bobdoering's Avatar
bobdoering

 
 
Total Posts: 4,097
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Quote:
In Reply to Parent Post by Rand T View Post

Thanks, Bob

I did a similar calculation (using minitab normal capability; see attached).

I just wasn't sure if that was correct in using the difference between min and max. At first glance I thought it might overstate the variance from nominal.
For sure, using the normal calculation would not be applicable using the difference - unless your variation was really bad. With a physical border (0), you need non-normal. If the normal model goes below zero, that is a good clue that it is a bad fit. Also remember, your best value for the difference is not the "center" 0 to .08 (or .04) - it is 0. Plugging in normal assumes the center is best or nominal. The better you get, the closer to zero - not .04 - you should get.

Quote:
In Reply to Parent Post by Rand T View Post

If I have a part where the profile is +.03 and - .01 mm, it seemed more logical to calculate the +.03 against the + bilateral tolerance of +.04 rather than the min /max difference of .04 against a unilateral tolerance of .08.
I don't think so. As in any measurement, you are looking at how much of the tolerance you are using up. You have a zone where the dimensions are acceptable, and the difference is how much of that zone was consumed by the part. Remember, you are looking at an array of data, no one linear measurement. You thinking is closer to what you can do with one linear measurement.

But, again, it is going to be what logic will best be convincing to the customer.

Quote:
In Reply to Parent Post by Rand T View Post

I then decided to select whichever of the min or max had the greatest magnitude, but then found that I was bouncing back & forth between + and - which also made the range much higher than it really is.

The main purpose for my question wasn't specific to this part, but a strategy on how to handle capability on profile tolerances where the only info I get on line scans is min and max.

Either way, it seems like I am overstating the variation, but I guess I may have to live with the best approach of the two. My last (and maybe best) alternative was calculating the capability of the min and max individually and then ignoring the one with the best Cpk
In really, really broad brush stroke, the difference from nominal (if it was truly nominal, and not a calculated one) would be closer to one of the Cpk calculations (Cpu or Cpl). Looking at the entire variation as I did is closer to Cp. But, I emphasize close - since it is nowhere near what is really calculated with those indices. More of a conceptual comparison.
  Post Number #24  
Old 22nd March 2013, 08:06 AM
Mr.Happy's Avatar
Mr.Happy

 
 
Total Posts: 150
Re: Capability for CMM scanned profiles on a feature with a bilateral profile toleran

Quote:
In Reply to Parent Post by Rand T View Post

Here are 2 attachments

1) Section of print
2) Sample CMM page
Hi Rand T,
Just for my understanding,
11) why did you used profile and location for the .25 profile and tolerance ?0.0
12) why did you used only profile for the .08 profile and tolerance ?0.04

Why don't you use for the .08 profile Nominal 0.08 MAX 0.0 and MIN -0.08 ?

On your sketch there are no datums (A B C) ?
Reply

Lower Navigation Bar
Go Back   The Elsmar Cove Business Systems and Standards Discussion Forums > >

Bookmarks



Visitors Currently Viewing this Thread: 1 (0 Registered Visitors (Members) and 1 Unregistered Guest Visitors)
 
Thread Tools Search this Thread
Search this Thread:

Advanced Forum Search
Display Modes Rate Thread Content
Rate Thread Content:

Forum Posting Settings
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Emoticons are On
[IMG] code is On
HTML code is Off


Similar Discussion Threads
Discussion Thread Title Thread Starter Forum Replies Last Post or Poll Vote
Capability & GDT Features - Surface Profiles, Line Profiles & Symmetries Peter A Capability, Accuracy and Stability - Processes, Machines, etc. 3 13th July 2012 07:20 AM
Converting 1-Sided Design Tolerance to a Bilateral Tolerance abhipatel Statistical Analysis Tools, Techniques and SPC 8 11th June 2012 02:13 PM
Surface Profile Callouts - How the profiles are being dimensioned brianq727 Inspection, Prints (Drawings), Testing, Sampling and Related Topics 5 14th March 2012 07:49 PM
Unequal Bilateral Tolerance spreadsheet wanted rnsvasan Statistical Analysis Tools, Techniques and SPC 7 6th April 2010 08:30 AM
Profiles - User Profile Information / Details Marc Legacy vBulletin Forum Software Instructions (OBSOLETE) 0 31st July 2008 09:43 PM



The time now is 04:30 AM. All times are GMT -4.
Your time zone can be changed in your UserCP --> Options.



Misc. Internal Links


NOTE: This forum uses "Cookies"