FRIDAY, SEPTEMBER 22, 2017
Font Size
Topic 1.67 Cpk

1.67 Cpk

Home Forums General Forums Implementation 1.67 Cpk

This topic contains 10 replies, has 7 voices, and was last updated by  Strayer 2 months, 1 week ago.

Viewing 11 posts - 1 through 11 (of 11 total)
  • Author
    Posts
  • #647521 Reply

    Hello All,

    New to having to implement a 1.67CPK to a component, and was told that we could provide full print tolerance, but not the 1.67CPK attached to it. There are several dimensions with fairly close tolerances that we will not hold due to the CPK.

    In question is a 19mm hole that is 19.00mm/19.016mm tolerance.

    I do not have any issue holding that tolerance, but by my calculations the 1.67CPK turns it into a + or – .0016mm tolerance.
    Not sure who can hold that in a typical CNC machining center, though I have see much closer tolerances held in machines using single crystal diamond. (Moore Nanotechnology & Precitech)

    I’m not saying its impossible, but I’ve never been given a “CPK at any automotive plant I’ve ever entered, but testing was nearly always fruitful.

    Please explain how any machine, is capable of this? Even the customer said they dont hold that tight tolerance of .0016mm.

    Thank you all for your help!!!

    #648181 Reply

    Hi @jpm9871,

    Your question generates even more questions. Maybe a step or two back to consider other things first.

    It may not be as simple as converting the specification down to a tolerance of +/-0.0016mm. I can prove this by creating a random data set. The range on my created data is 0.0074mm with a Cpk of 1.67 (more than double your suggested range of +/-0.0016mm).

    How have you collected the data? Is it in time order? What does your control chart tell you? It would help if you could upload a control chart here. Is the process stable?

    19.000 +0.016 / -0.000mm is a tight tolerance. In machining, I agree you probably wouldn’t be able to hold 1.67 on that. Even Wire EDM may struggle to hold such a tolerance to 1.67. You may want to find out what you can hold the process to and present that to your customer with a caveat of; “we can meet your specification 99.XX% of the time”.

    What do the manufacturers of the machine say about this? How accurate and precise do they quote the machine to be?

    Can you repeatably measure with this sort of variation / discriminate between results?

    James

    #648344 Reply

    With attachment

    Attachments:
    #650328 Reply

    @jpm9871 You haven’t given enough information to really say much but lets look at what we have, If you have a total tolerance3 of 0.0018 and a Cpk requirement of 1.67 (5 std dev) then if you have that process dead center you can have the largest std dev allowed to meat the requirements i.e +/- 5 std dev. Given that your process has to be able to hold a std dev of 0.00018.

    Is that possible? Probably not but let’s say it is. To make that measurement I need to be out 1 more decimal point and my calibration needs to be one more decimal point beyond that and calibration standard one more beyond that.

    Even if you can achieve it you probably can’t make it with any confidence.

    As far as the 1.67 requirement it is generally given to you in your quality system requirements or in an ISIR document. I am sure someone has put it on a print but it wouldn’t make much sense unless you like rev’ing all of your prints or writing a deviation for every print.

    Just my opinion

    #653108 Reply

    Albert Viljoen
    Participant
    Reputation - 94
    Rank - Aluminum

    Not sure if the technology exist to achieve that AQL with a Cpk of 1.67 at 5 sigma level wit such a small tolerance.

    You can automate and inspect for this AQL and give the customer that sigma quality level but you will need to observe your actual Cpk USL and LSL and calculate your defect rate and factor that into your quotation for ROI target achievement.

    I agree with Mike.

    #653160 Reply

    One of the requirements for standard calculation of Cpk is the assumption of data normality. Your histogram suggests that, at best, you have a truncated normal distribution – which isn’t normal. Before doing anything else I’d recommend using the methods for Cpk calculation that deal with the situation when the data isn’t normal to see what kind of a Cpk you get. Chapter 8 of Bothe’s book Measuring Process Capability has the details. It is my understanding that the methods described in that chapter have been incorportated in Minitab.

    #653210 Reply

    @rbutler I bet one could have time disproving normality with those 30 sample points. ;)

    Your point is understood…just teasing you.

    @jpm9871 This may seem pedantic but one should always check precision and accuracy of a gage before a capability study. You might have great variation control but be off because of poor accuracy.

    #653215 Reply

    Thank you everyone for your input, it has been very helpful.
    I have discussed this with several people in different departments, and no one wants any part of it. The person who previously put together the proposal, only provided the 1.67 CPK on tolerances that were more open/realistic to achieve, now I know why.

    Even after discussing this with out Quality Assurance department, they explained how some deviation is possible, and an average over a large amount of parts, depending on the amount of deviation could still put me in the 1.67 CPK…..except for that deviation being so small…..not much help.

    Customer has a high scrap rate too, I received two numbers 12% and 18%. Contact at customer is on vacation, and I’m waiting on replies to some of my questions, but I have a new one.
    Is the part scrap when it doesn’t meet the CPK or when it is out of print tolerance?

    I have come to the conclusion that it doesn’t matter what machine type, brand or cutting tool used to make this part, that CPK is not going to be achieved. My brothers company has a makino wire that he’s run .003″ inch wire on a part, and in his words incredibly and consistently, extremely accurate, but after explaining CPK, he said that is not possible. Lucky for them they do onesy, twosy amounts, never a CPK issue.

    Holding print dimension would be a walk in the park.

    I am comfortable with saying no, and moving on now.

    What makes me laugh is after being to so many automotive companies in the past, the CPK of that block, head, rod, crank or cam shaft was never mentioned, here’s the print can you do it?

    Again, Thank You all for your time and input on this, I appreciate it greatly, it has been very helpful.

    John

    #655228 Reply

    @jpm9871 I have occasionally seen the Cpk requirement on a print but not often. We would see it at PPAP and then we would have to calculate it (dpending on the OEM) from the initial production runs. If we couldn’t hold it then we weren’t allowed to turn on production.

    #658701 Reply

    Hello Everyone,

    Thank you all for your insight to my CPK dilemma, I appreciate your time.
    Although I believe holding print tolerances would be no problem whatsoever, the addition of the CPK makes this from what I’ve read unattainable with any machining center. I know the customer is getting good parts, but with a high scrap rate. I have not gotten a reply from him asking about his fixturing and tooling, because I have to assume he is not having the same issue with two different machines producing these parts. I have after talking to several people here, thrown in the towel. Plus or minus a couple .0001 or a few micron would be no problem, but to hold the tolerances necessary for this is unobtainable from what I’ve read here.

    #659162 Reply

    It’s only unobtainable under current circumstances. You can change the circumstances by reducing variation in the inputs and the process, and you may need better measurement system R&R. Whether that’s worth doing is a different question.

Viewing 11 posts - 1 through 11 (of 11 total)

Register Now

  • Stop this in-your-face notice
  • Reserve your username
  • Follow people you like, learn from
  • Extend your profile
  • Gain reputation for your contributions
  • No annoying captchas across site
And much more! C'mon, register now.

Reply To: 1.67 Cpk
Your information:






5S and Lean eBooks
GAGEpack for Quality Assurance
Six Sigma Statistical and Graphical Analysis with SigmaXL
Six Sigma Online Certification: White, Yellow, Green and Black Belt
Lean and Six Sigma Project Examples
Six Sigma Online Certification: White, Yellow, Green and Black Belt

Find the Perfect Six Sigma Job

Login Form