haha 2:52 kinda funny but good point. What if supplier gives you CP,CPK and standard dev? How can you determine whether the data they used resulted in a "normal curve". Should you graph the bell curve if you get the raw data to verify? In terms of the 6 sigma, why would you want that? You'd increase your tolerances to expand the USL and LSL. Or, your bell mean (center of bell curve) is moving left or right. Doesn't make sense if an engineer want's to keep a tighter tolerance band.
Perhaps I've misunderstood your point but the goal is not to achieve 6 sigma performance by widening tolerances (the customer might have something to say about that) but to reduce special cause variation within your process to shrink the SD such that your process mean sits 6 SD from the closest customer spec, upper or lower. This achieves very high process yields while allowing "wiggle" room and time to respond before more defects are produced
John's reply is great on your second question (thanks for contributing, John). As to your first question - you could ask your supplier to also show a Q:Q plot: this plots the observed measurements in a distribution graph that will show a straight diagonal line if the distribution closely resembles a normal distribution, or points deviating from that diagonal if certain values are over or under-represented.I would always first check the bell curve on it's normal scale as this might show all kinds of oddities, but the Q:Q plot is much better at specifically checking for normality. Do keep in mind that your goal is to get a stable supply of parts for your production process, not to get all possible statistics from suppliers. I would use such a step only if you have reason to doubt the validity of your supplier's Cpk statement. Maybe you get half a batch of out of spec (or very close to spec) products from time to time, while the supplier claims a Cpk of >1.33. I would always advise you to go talk with your supplier when you've got processing problems with products involving their parts (don't immediately share all your trade secrets 🤐) - they may know of changes in their process or materials, how their other customers solved similar problems, double-check their production process, etc. As soon as you've opened up dialogue with them, looking at their Cpk data together can be a logical next step (which they'll be much more happy to do when they understand that you're trouble-shooting and not just 'giving them trouble').
As the standard deviation of a proces ................... . the sigma level cabability of the process...........,...................... A. Increase, decrease B.increace, increase c.decrease, increase D.decreas, decrease
Glad you liked it, nice to hear from you. If there are any other topics you’d like some idea or clarification on, please drop a comment, I’ll happily look into to it.
@@tuanpham9600 I did already do such a type of video on Cpk (why and when to use it), which also goes into when to use an SPC control chart a bit: ruclips.net/video/TvcxraxR5m8/видео.html so you can get a first idea. I do also like your suggestion, so you can expect a why and when to SPC video coming up within a few weeks 😉
Can you do a video to explain how to convert different sigma level. Example: Cpk>=1, X = 25um @ 3 sigma is equal to Cpk>=1.66, X = 60um @ 5 signma. I saw a conversion table but don't know what is the formular for this. Thank you in advance.
Hi Lucas, I’ve been thinking about your question for a couple of days now, but honestly, I don’t really get it - I can give more detailed examples of various sigma levels (with sample data), but that wouldn’t make a 3 sigma process equal to a 5 sigma process. Did you mean that by centering a process (variation/sigma stays the same, but we move the centerline of our process farther or closer to the specification limits) the Cpk and sigma level do change (even though sigma itself doesn’t)? About that, I can make a video 😉
@@TomMentink I read an article which has this conversion table. Now can't find back the link. I'm also confuse to this. I can share this to you via email since I printed out a pdf version to read. Thank you in advance.
haha 2:52 kinda funny but good point. What if supplier gives you CP,CPK and standard dev? How can you determine whether the data they used resulted in a "normal curve".
Should you graph the bell curve if you get the raw data to verify?
In terms of the 6 sigma, why would you want that? You'd increase your tolerances to expand the USL and LSL. Or, your bell mean (center of bell curve) is moving left or right. Doesn't make sense if an engineer want's to keep a tighter tolerance band.
Perhaps I've misunderstood your point but the goal is not to achieve 6 sigma performance by widening tolerances (the customer might have something to say about that) but to reduce special cause variation within your process to shrink the SD such that your process mean sits 6 SD from the closest customer spec, upper or lower. This achieves very high process yields while allowing "wiggle" room and time to respond before more defects are produced
John's reply is great on your second question (thanks for contributing, John).
As to your first question - you could ask your supplier to also show a Q:Q plot: this plots the observed measurements in a distribution graph that will show a straight diagonal line if the distribution closely resembles a normal distribution, or points deviating from that diagonal if certain values are over or under-represented.I would always first check the bell curve on it's normal scale as this might show all kinds of oddities, but the Q:Q plot is much better at specifically checking for normality.
Do keep in mind that your goal is to get a stable supply of parts for your production process, not to get all possible statistics from suppliers. I would use such a step only if you have reason to doubt the validity of your supplier's Cpk statement. Maybe you get half a batch of out of spec (or very close to spec) products from time to time, while the supplier claims a Cpk of >1.33.
I would always advise you to go talk with your supplier when you've got processing problems with products involving their parts (don't immediately share all your trade secrets 🤐) - they may know of changes in their process or materials, how their other customers solved similar problems, double-check their production process, etc. As soon as you've opened up dialogue with them, looking at their Cpk data together can be a logical next step (which they'll be much more happy to do when they understand that you're trouble-shooting and not just 'giving them trouble').
Very nice video, very straight forward. Thank you
Thanks for the compliment, glad you liked it
Amazing Explanation.
Thanks Abdullah, glad to hear that my video brought value for you
Wow! I love your video! Thank you so much!
Thanks for sharing such kind words. I’m glad you enjoyed my video.
As the standard deviation of a proces ................... . the sigma level cabability of the process...........,......................
A. Increase, decrease
B.increace, increase
c.decrease, increase
D.decreas, decrease
I hope the answer to this one becomes clear in the video 😉
There are 2 correct answers between these options, by the way.
@@TomMentink
Yes there are two correct answer but my doctor all exam need brain storming as he said
Am still a student 😄
Thanks so much for sharing
Glad you liked it, nice to hear from you.
If there are any other topics you’d like some idea or clarification on, please drop a comment, I’ll happily look into to it.
@@TomMentink yep if you are ok kindly help to talk about Control chart and the function of it, thanks so much :)
@@tuanpham9600 I did already do such a type of video on Cpk (why and when to use it), which also goes into when to use an SPC control chart a bit: ruclips.net/video/TvcxraxR5m8/видео.html so you can get a first idea. I do also like your suggestion, so you can expect a why and when to SPC video coming up within a few weeks 😉
Can you do a video to explain how to convert different sigma level. Example: Cpk>=1, X = 25um @ 3 sigma is equal to Cpk>=1.66, X = 60um @ 5 signma. I saw a conversion table but don't know what is the formular for this. Thank you in advance.
Hi Lucas, I’ve been thinking about your question for a couple of days now, but honestly, I don’t really get it - I can give more detailed examples of various sigma levels (with sample data), but that wouldn’t make a 3 sigma process equal to a 5 sigma process.
Did you mean that by centering a process (variation/sigma stays the same, but we move the centerline of our process farther or closer to the specification limits) the Cpk and sigma level do change (even though sigma itself doesn’t)? About that, I can make a video 😉
@@TomMentink I read an article which has this conversion table. Now can't find back the link. I'm also confuse to this. I can share this to you via email since I printed out a pdf version to read. Thank you in advance.
@@lucasnq3224 please do - you can view my email address in the about section of my channel page (in a browser)
Plz do share with me it will be great favor