Tools Six Sigma for my Automated Machine

THEERANART

Starting to get Involved
My company (kind of computer HDD parts) has a machine to measure distance from center point to edge of parts . Below are my information.

1. within 1 sec able to measure distance 3 parts (that mean 1 part per 1/3 second)
2.There are 2 side to measure left , right
3.I have data of 1 hour around 9000 row data point both of left and right.

I would like to compare between left and right side .

What tools can do to compare ? .
1. Can I use one way Anova to compare left and right ?
2. Can I use average of 3 parts as per one second to represent time.?
3.If I want to use GRR , Which GRR you would to recommence , Nest or Crossed
4. if I want to analyze Capability Process (Cp) , Can do?
Thank you in advance for your help , I am new in Six Sigma..
I have attacked example and specification .
 

Attachments

  • minitab_test.xlsx
    31 KB · Views: 166
S

supreecha

:cool: In attached file for my concept (some data my key mistaken)
What tools can do to compare ? .
1. Can I use one way Anova to compare left and right ? Ans: Yes you can use :T-est/ F-Test , Regreesion Model etc.
2. Can I use average of 3 parts as per one second to represent time.? Ans: Yes ,by Implemented: X-Bar R Chart (n=3)
3.If I want to use GRR , Which GRR you would to recommence , Nest or Crossed Ans : If implement 3 Operators , 3 trials , 10 samples used Crossed , If 1 Operator , 1 machine used Nest .
4. if I want to analyze Capability Process (Cp) , Can do? Ans: Cp=20.08 ,Cpk =16.68 (left
) , Cp=15.30 , Cpk =15.19 (right)
 

Attachments

  • minitab_test.xlsx
    560.2 KB · Views: 197
Last edited by a moderator:

THEERANART

Starting to get Involved
Thank you so much for your concept and for your time for making sample for me.

Dear Expert,
I have one more question about using Nested GRR , Using minitab there was some error "
Gage R&R Study - Nested ANOVA
* ERROR * Design is not balanced; execution aborted.

I have attached details for 3 required column , I think my data is obtained by automated measure instrument (not people bias)
Could you help to recommend how to change data field or change system .
 

Attachments

  • Minitab_NEST.MPJ
    27 KB · Views: 107

Bev D

Heretical Statistician
Leader
Super Moderator
First - Six Sigma isn't about throwing a bunch of statistical tools at a pile of data and seeing what happens. It is a thoughtful process of trying to understand your process and improve it.

Begin by plotting your data in time series to understand what is happening: you have a drift to the high side on both holes...

...you could do a nested GR&R but a crossed study is probably better unless it is very difficult to make repeat measures.

I wouldn't use a control chart using the average of the two holes as your subgroup. this is not a rational subgroup as the between hole differences is NOT representative of the hole to hole variation over time which is whatyou really want to plot.

You could determine capability, but as this process is not stable (upward drift) the results would be meaningless. additionally the time period is extremely short. you have not included setup to setup, raw material, lot to lot, operator to operator, etc.

yes you could do an ANOVA on hole 1 to hole 2, or any number of other statistical analyses, but what are trying to understand? that is what tells you the best statistical analysis to use.

However, the biggest gap is: what are you trying to do?
 
Last edited:

Miner

Forum Moderator
Leader
Admin
Some additional observations that may or may not be relevant, but may help understand your process.

  • There is a very strong correlation between the right and left measurements (see scatterplot)
  • However, there is one apparent outlier and another factor that causes a shift in this relationship (see the tri-modal histogram of the delta between right and left measurements)
 

Attachments

  • Scatterplot.png
    Scatterplot.png
    7.4 KB · Views: 207
  • Histogram.png
    Histogram.png
    5.3 KB · Views: 152

THEERANART

Starting to get Involved
First - Six Sigma isn't about throwing a bunch of statistical tools at a pile of data and seeing what happens. It is a thoughtful process of trying to understand your process and improve it.

Begin by plotting your data in time series to understand what is happening: you have a drift to the high side on both holes...

...you could do a nested GR&R but a crossed study is probably better unless it is very difficult to make repeat measures.

I wouldn't use a control chart using the average of the two holes as your subgroup. this is not a rational subgroup as the between hole differences is NOT representative of the hole to hole variation over time which is whatyou really want to plot.

You could determine capability, but as this process is not stable (upward drift) the results would be meaningless. additionally the time period is extremely short. you have not included setup to setup, raw material, lot to lot, operator to operator, etc.

yes you could do an ANOVA on hole 1 to hole 2, or any number of other statistical analyses, but what are trying to understand? that is what tells you the best statistical analysis to use.

However, the biggest gap is: what are you trying to do?

Thank you,
My purpose is to compare left and right head m/c ,
"Inline camera inspection. Camera measure each product parts in a coil .100% inspection.
Camera measure parts edge (left, right) to pilot hole distance .Delta is the difference between left and right measurement.
Camera can inspect 3 parts in a second of a moving coil."

So, In ideal , left and right should be equal .
 

Bev D

Heretical Statistician
Leader
Super Moderator
Ok - so as Miner's analysis shows ther is a difference. now what?
 
Top Bottom