# Spec Limit Rounding

Six Sigma – iSixSigma Forums Old Forums General Spec Limit Rounding

Viewing 20 posts - 1 through 20 (of 20 total)
• Author
Posts
• #53031

JBro
Participant

I am calculating Cpk based on the customer spec which is subject to standard rounding rules.  Should I use the spec, or am I allowed to incorporate the rounding rules.  Example:  Acceptable range from the customer is 0.3 – 0.5  When calculating Cpk, can I use 0.25 and 0.54 as my LSL and USL?  This dramatically improves my Cpk, so I hope the answer is yes.

0
#187518

GB
Participant

I would not round. Tools such as minitab allow you to use decimal points.Rounding can be a slippery slope, subject to stack up issues.my \$0.02

0
#187521

Taylor
Participant

Jbro
I would not round. Main reason is that it changes your process center.

0
#187522

Mikel
Member

No. What a stupid question.

0
#187524

Kluttz
Member

Whoah, I don’t think you’re really asking a question about rounding.  It sounds to me as if you’re asking “is it ok to artificially inflate my Cpk expand by expanding my operating limits via data manipulation?”.
What’s the purpose of your Cpk calculation?  If it’s actually to understand your inherent capability to meet customer expectations, then your capability is what your capability is.  You have the customer specification in one hand and your performance in the other.  It is what it is.
Based on your “rounding rules” it would be just as acceptable for your customer to reject product because it doesn’t meet a rounded 0.349-0.450 specification.

0
#187525

Mikel
Member

You are tell us you can only get 3 values for your measurement?0.3, 0.4, or 0.5?

0
#187529

Bower Chiel
Participant

Hi JBro
It seems to me that you need to think about the measurement systems involved.  Suppose an item has a “true” value of 0.273 and you the supplier have a system that can give you a measurement of the true value correct to 2D i.e. that returns the value 0.27.  On the other hand suppose that your customer has a less sophisticated system that can give him a measurement of the true value correct to 1D i.e. that returns the value 0.3.  He would therefore accept that part.  Thus if you are indeed in this sort of scenario I believe that you should “extend” the specification range to match the accuracy of your measurement system.
Much wiser heads than mine have considered this.  You might find the Donald Wheeler article at http://www.spcpress.com/pdf/Manufacturing_Specification.pdf of interest.
Best Wishes
Bower Chiel

0
#187530

zippo
Member

There is no such thing as a stupid question.  Only stupid answers by stupid people.  Stan can you please inform this forum as to the resoning behind your arrogant answer.  I stayed away from this website for 3 months hoping to see more maturity from the likes of you, only to return and see nothing has changed.  Guess I need to stay away longer.

0
#187532

Mikel
Member

Zippy,Stay away as long as you like. I’ve seen no contribution from you
ever, so who cares?The question is stupid on many levels, I’ll give you a few -1) The guy doesn’t care if his process is capable, he just wants a
good number.
2) From his rounding example, he leads us to believe he is only
checking to one significant digit in which case his measurement
system is not appropriate. If he can check to two significant digits
his rounding would be at 2.95 and 5.04.
3) Read the link to the Wheeler article – he should be bringing his
specs in, not out, to assure the result to his customer.And there are definitely stupid questions, especially when a person
is just being lazy instead of learning.

0
#187533

Taylor
Participant

So Zippy, did you wake up this morning and decide you were going to come back to the forum after 3 months and pick a fight with Stan or is this purely coinencidental? The problem here is that while most of us think it was pretty,       well, not so thought out question, Stan merely speaks his mind, and has more than experience to state the obvious if he wants to. If your skin is so thin that you cannot take a little redicule and learn from then maybe you should go back to where it is you have been hiding and stay there.

Just my opinion, really up to you how you procede from here. If you have some knowledge to contribute, then by all means go for it. If you just want to come here and give motherly talks to people who some how offend you by speaking their mind, then you are in the wrong room.

0
#187537

zippo
Member

0
#187538

zippo
Member

actually I have contributed but in your case no contribution would be worth a lot more

0
#187541

GB
Participant

Bye Zippo…Safe journey.

0
#187542

Dharma Bum
Participant

Robert,
Come on now. This unnatural focus on Stan is really getting weird now. Be careful, or the pounding you’ve taken due to your unashamed hucksterism may just follow over to LinkedIn.

0
#187555

Darth
Participant

There are many posters who dislike Stan including myself, HeeBee, Carnell and others so blaming Robert S aka Zippy aka Freddy may be jumping to conclusions. In fact, we could take a poll to see how many of our thousands of posters also dislike Stan.

0
#187560

Mikel
Member

Count me in – I don’t like him

0
#187573

GB
Participant

Darth,
This has nothing to do w/ like/dislike. Heck, I’m sure RS is a pleasant fellow.So, Freddy is called out as being our friend, RS due to the similarity of his posts. The Freddy attack-postings stopped cold. (The same thing happened w/ “Les”). Now Zippo pops up and his attack-posts are very similar to Freddy, Les and RS. Zippo has been called out and the attack-posts have stopped…at least for a bit.It reminds me of DOG SXXT/ATI/Marlon/et al… maybe he is onto something…I should give it a whirl…

0
#187636

JBro
Participant

Wow guys – thanks a lot for all the feedback – constructive and some maybe not so constructive.  Part of me does want to just improve a value to satisfy the customer (rather than actually making a change to become more capable), but I am absolutely not a statistical expert, so I was really just after whether this was an appropriate approach.  The customer spec is between 0.3 and 0.5, and according to rounding rules they will accept 0.25 – 0.54 as good product, so why shouldn’t I use those values as my LSL and USL when calculating Ppk?  I can measure to 3 decimal places, and I am confident in the values.
Call me a glutton for punishment, but I appreciate your thoughts and comments.  Thanks.

0
#187637

Mikel
Member

If you can measure to three places, your rounding is 0.295 – 0.504.Do you know your MSA?Stop trying to cheat.

0
#187647

Taylor
Participant

Jbro
There are many reasons not use rounding as your LSL and USL. The most obvious is measurement error. My advice, and you can take it for what its worth (Which usually is not much). Perform an MSA on your process. Make sure you are capable of measuring the product to meet the customer spec. Center the process and monitor it with control charts to make sure you stay in control. If for some reason you identify product which does not meet customer specification, then you may want to assert “rounding rules” with your “Customer Permission” to send them product that doesn’t meet specification. If you can prove your process capable of measuring at the .25-.54 range, then you may want to push your customer to change their specifications. However, if you have a robust system and your measurement sysem is capable then you have no reason to change anything.

0
Viewing 20 posts - 1 through 20 (of 20 total)

The forum ‘General’ is closed to new topics and replies.