Thanks, Paul, your case study confirms one of the biggest misunderstandings & confusions in MSA, because many people only know about GRR%, which is here "P.Total". The P.Total highly depends on the sample selection (=part-to-part variation) and therefore it is NOT an objective KPI for Meas.system assessment. AIAG even recommends option 4) of calculating GRR%, saying the Tolerance should be used, which gives you the KPI "P.Tolerance". I also like to use the formulas behind in trainings, because it makes it easier to demonstrate & follow: GRR(Total) = P.Total = sigma(MS) / sigma(Total), where sigma(Total) = SQRT( sigma(MS)² + sigma(Part-to-Part)²) GRR(TOL) = P.Tolerance = 6 * sigma(MS) / Tolerance From there you can see that P.Tolerance is independent from part-to-part variation, so indenpendent of sample selection, so a much more objective figure for MSA judgement. And as you mention, bad Process Cp means a big part-to-part variation, big sigma Total, which makes P.Total low and therefore look good. (P.Tolerance and P.Total are similar, if Process Cp is about 1.0, a 3-sigma process.) Or as other consequence: if you have a good process e.g. Cp=2, then you get a bad P.Total, so you would punish your Measuring Process.
Joey - Thanks for the comment, glad you like the approach. I Try to keep it simple, to understand the reason I'm using the technique and what the result is really telling me. And not just apply a set of 'if this then accept' type rules. The rules apply most times, but just occassionally you have ot use your comment sense...
Hi Paul, first I have to say that I'm happily surprised with your fast response!! I really appreciate it. Now my question is mostly about which type of MSA should I perform, due to such small quantity, however they do are expensive devices that we most make sure are their measurements are accurate.
Maria - thanks for the positive comments, I was trained as a draughtsman many years ago, they made me repeat my printed dimensions etc until they were perfect, then some invented computers and i never needed that skill! Maybe you'd be less distracted if you bought my book 'drink tea and read the paper' you can get if from LULU.COM..I think there is a current 10% discount...
Hi Paul, thank you so much for your very informative videos. Could you please advice what type of MSA would you recommend to perform in a service shop where you can only find 2 pieces of the device to be measured on any given day? and no more than 50 devices are serviced per year, do we really need to perform MSA?
Good to hear from you, it really depends on what those measures are for. If you make expensive business decisions or important customer checks that could be costly it would be great to be accurate. If they are not important measurements, sure don't need an MSA...
Hi Paul. What are your thoughts about the main difference between MSA TYPE 1 study and MSA TYPE 3 study? if you are measuring a Diameter of a product with a non contact measuring device, would either be ok to use? any advice would be appreciated
@@paulallen5321 Thanks Again for your swift response, so you wouldn't do a type 1 study based on the operator placing the component onto the measuring device and its a non contact measurement? once its in place the operator has no influence and can only work if its placed where it should be? Also do you thing a type 3 study is feasible, somebody has mentioned this before but this is something i have not come across (only seen type 1 and type 2)
Thanks Paul for the video to clear up my question. As suggested I have repeated the MSA on the same part but different roll of fabric and pretty much got the same results. The values you see are %, and the test was based on moisture reading of the fabric before a wash treatment.
Sorry to be so offtopic but does any of you know a way to get back into an Instagram account?? I stupidly forgot my login password. I would love any help you can give me!
@Sean Vance Thanks so much for your reply. I found the site on google and I'm waiting for the hacking stuff now. Seems to take quite some time so I will reply here later when my account password hopefully is recovered.
In a Gage R&R do they ever measure 3 different parts (Totally different sizes, say a tray with 163 meds, a tray with 108 meds, and 1 tray with 0 meds) or in Gage R&R's do they only measure 3 versions of the same part (3 different trays, each tray with 163 meds)
Hi Anthony - no if you do that you ruin the MSA. It's ratio of the natural variability of the part v natural variability of the measurement system. If you use 3 different parts you inflate one of these values and invalidate the results...
Hi Paul, with automation in production line, the operator may even not have any physical contact (to parts, to equipment, ...) to the entire measurement activity. Can I remove the factor "operator"? And with that, I probably can only use Anova-method to calculate GRR%, because AIAG template/formula won't work anymore. You advice?
Hi Vincent - Firstly, yes the phrase operator really means measurement system, so if this measurement happens automatically replace the word operator with system. Now for the Anova calc to work you have to have 2 systems to compare against each other. If you only have one system then you have to use the X-bar R method. If you have rep-designed spreadsheet for an AIAG calc, simple thing is to paste the same data set from your single measurement system into both systems. You'll get zero reproducablilty error, but is correct as you're only checking one system...
@@paulallen5321 Tx. I tweaked AIAG calc template. Input readings from 1st operator, leave out 2nd, 3rd. Keeping 10 parts, 3 replications. AV (appraiser variance)=0, GRR comes purely from EV (equipment variance). Tx!
If you were doing MSA for SPC process control then this would be an issue as you would be adjusting your process based on flawed data and therefore introducing more product variation.
Wow your videos are GEM. No one has explained MSA so practically
Thanks Vishnu - enjoy the other video's as well
@@paulallen5321 Yes I am and will purchase your "Drink tea and read paper" book as well 😀
Thanks, Paul, your case study confirms one of the biggest misunderstandings & confusions in MSA, because many people only know about GRR%, which is here "P.Total".
The P.Total highly depends on the sample selection (=part-to-part variation) and therefore it is NOT an objective KPI for Meas.system assessment.
AIAG even recommends option 4) of calculating GRR%, saying the Tolerance should be used, which gives you the KPI "P.Tolerance".
I also like to use the formulas behind in trainings, because it makes it easier to demonstrate & follow:
GRR(Total) = P.Total = sigma(MS) / sigma(Total), where sigma(Total) = SQRT( sigma(MS)² + sigma(Part-to-Part)²)
GRR(TOL) = P.Tolerance = 6 * sigma(MS) / Tolerance
From there you can see that P.Tolerance is independent from part-to-part variation, so indenpendent of sample selection, so a much more objective figure for MSA judgement.
And as you mention, bad Process Cp means a big part-to-part variation, big sigma Total, which makes P.Total low and therefore look good.
(P.Tolerance and P.Total are similar, if Process Cp is about 1.0, a 3-sigma process.)
Or as other consequence: if you have a good process e.g. Cp=2, then you get a bad P.Total, so you would punish your Measuring Process.
Joey - Thanks for the comment, glad you like the approach. I Try to keep it simple, to understand the reason I'm using the technique and what the result is really telling me. And not just apply a set of 'if this then accept' type rules. The rules apply most times, but just occassionally you have ot use your comment sense...
Hi Paul, first I have to say that I'm happily surprised with your fast response!! I really appreciate it. Now my question is mostly about which type of MSA should I perform, due to such small quantity, however they do are expensive devices that we most make sure are their measurements are
accurate.
drop me a message on paul.allen@allenp.co.uk and i'll see if I can help...
I really tried to focus on your story, but I was more impressed with your handwriting and drawing skills!
Maria - thanks for the positive comments, I was trained as a draughtsman many years ago, they made me repeat my printed dimensions etc until they were perfect, then some invented computers and i never needed that skill! Maybe you'd be less distracted if you bought my book 'drink tea and read the paper' you can get if from LULU.COM..I think there is a current 10% discount...
Hi Paul, thank you so much for your very informative videos.
Could you please advice what type of MSA would you recommend to perform in a service shop where you can only find 2 pieces of the device to be measured on any given day? and no more than 50 devices are serviced per year, do we really need to perform MSA?
Good to hear from you, it really depends on what those measures are for. If you make expensive business decisions or important customer checks that could be costly it would be great to be accurate. If they are not important measurements, sure don't need an MSA...
Hi Paul. What are your thoughts about the main difference between MSA TYPE 1 study and MSA TYPE 3 study? if you are measuring a Diameter of a product with a non contact measuring device, would either be ok to use? any advice would be appreciated
Type 1 only works where the process is all machine based. You'll get bias if you try type 1 with operators. Otherwise type 2 is always best to do.
@@paulallen5321 Thanks Again for your swift response, so you wouldn't do a type 1 study based on the operator placing the component onto the measuring device and its a non contact measurement? once its in place the operator has no influence and can only work if its placed where it should be? Also do you thing a type 3 study is feasible, somebody has mentioned this before but this is something i have not come across (only seen type 1 and type 2)
Thanks Paul for the video to clear up my question. As suggested I have repeated the MSA on the same part but different roll of fabric and pretty much got the same results. The values you see are %, and the test was based on moisture reading of the fabric before a wash treatment.
Sorry to be so offtopic but does any of you know a way to get back into an Instagram account??
I stupidly forgot my login password. I would love any help you can give me!
@Justus Pierce Instablaster :)
@Sean Vance Thanks so much for your reply. I found the site on google and I'm waiting for the hacking stuff now.
Seems to take quite some time so I will reply here later when my account password hopefully is recovered.
@Sean Vance It did the trick and I finally got access to my account again. I am so happy!
Thank you so much, you really help me out !
@Justus Pierce No problem =)
In a Gage R&R do they ever measure 3 different parts (Totally different sizes, say a tray with 163 meds, a tray with 108 meds, and 1 tray with 0 meds) or in Gage R&R's do they only measure 3 versions of the same part (3 different trays, each tray with 163 meds)
Hi Anthony - no if you do that you ruin the MSA. It's ratio of the natural variability of the part v natural variability of the measurement system. If you use 3 different parts you inflate one of these values and invalidate the results...
Thanks a lot. You are the best
Cheers David....
Hi Paul, with automation in production line, the operator may even not have any physical contact (to parts, to equipment, ...) to the entire measurement activity. Can I remove the factor "operator"? And with that, I probably can only use Anova-method to calculate GRR%, because AIAG template/formula won't work anymore. You advice?
Hi Vincent - Firstly, yes the phrase operator really means measurement system, so if this measurement happens automatically replace the word operator with system. Now for the Anova calc to work you have to have 2 systems to compare against each other. If you only have one system then you have to use the X-bar R method. If you have rep-designed spreadsheet for an AIAG calc, simple thing is to paste the same data set from your single measurement system into both systems. You'll get zero reproducablilty error, but is correct as you're only checking one system...
@@paulallen5321 Tx. I tweaked AIAG calc template. Input readings from 1st operator, leave out 2nd, 3rd. Keeping 10 parts, 3 replications. AV (appraiser variance)=0, GRR comes purely from EV (equipment variance). Tx!
If you were doing MSA for SPC process control then this would be an issue as you would be adjusting your process based on flawed data and therefore introducing more product variation.
Spot on Martin - I've seen good machines be adjusted over and over when they hadn't really moved...
Thanks, Pauls for the Video. SO could you give me a file excel to practice?
Trong - Please contact me on paul.allen@allenp.co.uk and I'll send it...
@@paulallen5321 hi. PAul,. I already sent mail for you. Thanks