MSA (Measurement System Analysis) on a Photospectrometer

[mcammaert]

Dear members,

I was wondering if it is possible/useful to conduct a MSA/GR&R on a photospectrometer used for color measurements (L*,a*,b*) of plastic parts.
If yes, could somebody explain me how to conduct this MSA?

Thank you in advance for your help!

Best regards!

 

[Ninja]

Re: MSA on photospectrometer

Hello mcammaert,
Welcome to the Cove

Lots of previous posts on GRR of just about anything...you might want to do a search on the forum.

That said, a photospectrometer is a tool.
You want to do your study on the test method.
There is a huge difference between the two.
How you use the tool likely imparts more variation than the tool itself.

Test your test method the same way as you would test your dimensional tests...(not the tool, the test method).

Have multiple operators test multiple parts multiple times and compare the results.

 

[Ron Rompen]

Re: MSA on photospectrometer

Yes, you can perform a GR&R study on your photospectrometer - this would be done in the same way as any measuring instrument (i.e. 3 operators, 10 parts, 3 measurements per operator per part). HOWEVER because of the nature of the instrument you need to take special steps to ensure that your test series is valuable.
Ensure that there is a clearly-understood written instruction on how to measure the sample parts - particularly regarding what location, and how to use the instrument. Ensure that the lighting conditions for the test series are the same (ideally you would want to do this in the lab, to minimize variation). Between each measurement, the part surface should be cleaned of fingerprints and accumulated dust/dirt (this process also needs to be in the work instruction).
Now you need to determine WHICH value(s) you want to calculate your GR&R on - L a b doesn't really mean much, what you are really looking at is either your deltaE or deltaEcmc. I don't know what your customer accepts as a variance, but that value would be your 'tolerance band' for measurement.

Hope this is of some help. I am also in the painted plastics industry (automotive components) so if I can be of any more assistance, don't hesitate to contact me directly here, and I will do what I can.

 

[Miner]

Be alert to the possibility of within part variation in the results. While this is a legitimate part of measurement variation is is also somewhat controllable by the measurement method.

 

[mcammaert]

Dear cove members,

Thank you for the replies you already provided. They are very helpful.

It is clear for me now that is is useful to conduct a MSA (GRR) on the photospectrometer while following the instructions described in the Work Instructions. It is actually about color measurements of plastic parts for automotive purpose. I would like to ask my supplier to deliver some MSA results (because of some color mismatch problems) but since it seemed like they never conducted a MSA in the past, I was thinking if it was indeed useful to ask.

To go a little bit more into detail, especially for the comments that Ron Rompen had.
I would actually use L*,a* and b* values to calculate my GRR. It seems strange for me to use deltaE or deltaEcmc values because you are including the variance of the reference part and the variance of the actual part you are measuring.
My customer has only tolerances for deltaL*, deltaa* and deltab*.

What are your opinions about this?

Best regards!

 

[Ron Rompen]

Your customer does, as always, have the final say in what they want, however if you are evaluating L, a, and b you are now evaluating 3 separate characteristics of your gauge, and the potential for failure is increased. In our plant, we report the 3 values, (L, a, b) but the final determination of 'good' or 'bad' is based on deltaEcmc.

As for your supplier providing results, I would be very surprised if they have not conducted GR&R studies in the past (presuming they are a manufacturer of the paint you are purchasing, and not just a distributor). We deal directly with one supplier, who is also registered to ISO 17025, and have that type of information readily available.

 

[notadog]

Please define (L, a, b)?

 

[Miner]

See CIELAB in Wikipedia.

 

[notadog]

That doesn't look like anything that I would want any part of. Thank you for your quick response.

 

[Ron Rompen]

Please define (L, a, b)?

L = Lightness or darkness
a = hue or color
b = saturation or shade

It is a standard method of mathematically describing a particular color. Based on a concept devised by Munsell many years ago.

That is the basics of it, as was posted afterwards, you can go on the CIELab website for much more information if you are interested.