ANOVA MSA GR&R in ATE replacing operators (Error Design is not balanced)

Rumbero65

Starting to get Involved
hello, I am trying to perform a Gage R&R on a ATE (Automated Test Equipment) without operator contribution to variance, but I's like to replace the "operator" variance source with "socket" (A tester can test several devices of the same type loaded on different "sockets" and ideally all sockets should return the same value).
I am using:
N.10 UUT
N.2 Trial/test
N.4 Sockets
In Minitab I use the Crossed Anova and use Sockets data column for "operator" field,

Minitab returns "Error * Design is not balanced"

Could you please help on this?

regards
 

Miner

Forum Moderator
Leader
Admin
Per your description, you should have 80 data points structured as shown in the attached file. If you attach the worksheet, I can take a look.
 

Attachments

  • MSA with ATE.xlsx
    11.3 KB · Views: 181

Rumbero65

Starting to get Involved
Thanks Miner, I have double checked with "Cross tabulation and Chi square" analysis and actually there was some discrepancy on collected data.
Thanks a lot for your help!
 

Rumbero65

Starting to get Involved
Even if the "error" issue has been solved may I ask another question here?
I am using MINITAB --> Gage R&R Study (Crossed) and then use the ANOVA method assigning the SOCKET data column to OPERATOR for the analysis. Is this approach correct?
I keep on getting %Study Var close to 95 while the customer calculates a %R&R= 7.30 so i assume I am doing a big mistake somewhere...
In attachment the Minitab results.
Thanks a lot
 

Attachments

  • ANOVA MSA GR&R in ATE replacing operators (Error Design is not balanced)
    AnovaGRR_Results.png
    124.8 KB · Views: 37

Miner

Forum Moderator
Leader
Admin
Your approach is correct. I see that you have a lot of socket-to-socket variation that you will need to correct. Hopefully that is a simple calibration issue.

Regarding the customer's results, do you know how they ran the study? Was it the same way, or did they run a type 1 study on a single socket? Are they also using %SV or are they using %Tolerance?

If you attach your worksheet, I can double-check your results. If you cannot attach a Minitab file, copy it into Excel and upload that.
 

Rumbero65

Starting to get Involved
Hello Miner, Here I attach the screenshot I sent to the customer. You can see his results and my notes in red and green.
I Have had a call with him and it seems that they use TOLERANCE VARIATION instead of PART Variation (probably it's hard for them to collect parts that represent the overall manufacturing variation). So I am going to integrate our tool with the possibility to replace (as option) the PART VAriation with TOLERANCE Variation.
IN your opinion does this "workaround" make sense for ARM method too?
I really thank you Miner!
 

Attachments

  • ANOVA MSA GR&R in ATE replacing operators (Error Design is not balanced)
    Customer.png
    44.5 KB · Views: 32

Miner

Forum Moderator
Leader
Admin
No. You should be comparing your customer's 7.3% Tolerance against your 8.5% Tolerance results. No need to change your spreadsheet, just change which metric you are using.
 
Top Bottom