MSA (Measurement Systems Analysis) on Visual Color Check

[Dan M]

Our CB auditor suggested that we should consider doing an MSA study on our Visual color check.

Here is how our visual color check is performed:

The specification indicates the product should be blue.
The QC technician looks at a sample of the product and visually judges what the color is.
The QC technician records the actual color.
(Any result other than blue = Fail)
No reference standard is used for this test.

It is possible to do an MSA study on this type of test? How?

Many Thanks.

 

[trainerbob]

I would suggest that if you feel you need an MSA study on this you would need a blue standard developed by you or other members of your organization. It would not be hard to have a chart that shows a range of blue. Using this you could consider an attribute study with a pass/fail cirteria. It is usually pretty hard to calibrate ones eyes. Remember, the simpler the better!

 

[Jim Wynne]

Our CB auditor suggested that we should consider doing an MSA study on our Visual color check.

Here is how our visual color check is performed:

The specification indicates the product should be blue.
The QC technician looks at a sample of the product and visually judges what the color is.
The QC technician records the actual color.
(Any result other than blue = Fail)
No reference standard is used for this test.

It is possible to do an MSA study on this type of test? How?

Many Thanks.

If it's really as simple as you say, there's no point in even attempting it. Do your best to understand the risks involved, if there are any, and mitigate as well as you can.

 

[JaxQC]

I’m guessing it is not as simple as Blue vs Red. If what the real question is do you have the correct shade of Blue or matching a specific range then you can still do it as an Attribute study or mini-variable (1-3 scale). As with paint or anodizing metal where you want the parts in Lot 5 to match Lot 1 but you don’t want to go to the level of Pantone color identification, easiest is with a color range of samples. Say (3) pieces 1)light as you want 2)target color 3)dark as you want. Can your operators correctly identify where the sample falls into the 1-3 scale to correctly make process adjustments or do they look at the part and adjust the wrong direction?

 

[bobdoering]

Have you ever rejected internally or externally based on color? If so, you have some idea of what the "fail" samples should be. from there I would do a siple attribute MSA. If not, I may have to agree with Jim - what is the point? But, in order to go there it would have to be somewhing worth removing from the control plan and PFMEA as a no occurrence, no severity issue. If there is no need to verify the detection technique, there is no need to check it.

 

[jelly1921]

Our CB auditor suggested that we should consider doing an MSA study on our Visual color check.

Here is how our visual color check is performed:

The specification indicates the product should be blue.
The QC technician looks at a sample of the product and visually judges what the color is.
The QC technician records the actual color.
(Any result other than blue = Fail)
No reference standard is used for this test.

It is possible to do an MSA study on this type of test? How?

Many Thanks.

You shall make a MSA study by hypothesis test analysis cross-tab method, to calculate kappa value, or signal detection approach.
Simply put, there are three areas, which are Area I ( out of a range of blue), Area II in the margin of a range of blue and Area III within a range of blue.
For Area I and III, there is no problem to decide the product is passed or fail.
But for Area II, there could be a problem, because different appraisers may have different decision.

You can read MSA 4th, Chapter 3 Section C attribute measurement systems study.

Jelly

 

[Howard Atkins]

First perform a colour blindness test

 

[Jim Wynne]

First perform a colour blindness test

Depending on the risks involved and the level of accuracy needed, it might also be a good idea to test for color acuity--how well a person is able to discern differences in hue.

 

[tomvehoski]

1. BLUE
2. BLUE
3. BLUE
4. BLUE
5. BLUE
6. BLUE

It all depends on your specifications. If 1-4 above are all within the "Blue" spec, plus you only have three cans of paint in your shop (blue, red, yellow), an MSA study is overkill and you probably don't even need samples.

If 1-3 are acceptable but 4 is not, you might need to establish boundry samples.

If only #1 is acceptable, you may need to look at Pantone standards, light booths or a spectrophotometer for more critical measurement.

 

[Ninja]

First perform a colour blindness test

Good call, Howard.

One of the things I've found necessary (for audit, not for the actual line) is to have an array of color blocks printed on a piece of paper with each operator filling in the color name under each color block.

This not only confirms that they can discern colors from each other, but also serves as a record for an audit. Seemed like a severe waste of time when I first did it, but it came in quite handy when the auditor asked about color blindness.