Color measurements R&R: what's wrong with white?

[lexaclev]

Dear members,

This is my first post so first of all thank you all for valuable and helpful threads and advices in MSA topic. Most of my questions are already answered here, but I still need you collective wisdom in the field of instrumental color measurements.

The problem we ran into is GR&R for instrumental (spectrophotometer) measurements of white colored painted plastic auto parts. GRR for pure whites is almost twice as bad as for other colors and unacceptable for practical purposes. Measured values are dL, da, db and resulting deltaE. The troublesome component is dL with most between-appraisers variation. Same parts, same appraisers, same experimental plan as for other colors is used. GRR for all other colors is acceptable.

What might be wrong with white colors? Is instrumental control unreliable for them and visual inspection should be used instead? (BTW, customer requires instrumental control)

Thank you in advance for your help!

 

[RCH2016]

Are you certain that the wavelength and geometry are identical for each instrument?

 

[lexaclev]

Are you certain that the wavelength and geometry are identical for each instrument?

Yes, the instrument was also the same without settings changes, ambient light as well. The whole experiment was well controlled and was repeated second time for another set of parts.

GR&R was calculated as %Tolerance and we use same tolerances for all colors: deltaE < 1,7

 

[Ron Rompen]

You are measuring the reflected color of the 'white light' which is in fact, made up of all other colors. This will skew your observations, and unless you measure the exact same spot, at exactly the same orientation, with the same temp and humidity (etc etc ad naseum) then you will get different results each time. The degree of difference may vary, but it will always be there.
If this is a gloss white (vs flat or eggshell) then the effect is even worse.

 

[julsbear]

For whites almost all your variation is going to be in L and it will also vary quite a bit more than in the other achromatic colors (grays and black) because the absolute value is so high.

Placement of part on the port may also be key. Are the parts opaque? Even some light due to lack of opacity could have a big affect especially if the operator keeps their hand on the plunger (if a desktop unit) or where they stand (shadow).

 

[Hershal]

If you want to do GR&R, ignore what may be called light, and go solely by wavelength. No other approach is valid.

The reason is because if with sphere, is the blue coming on stronger? Chips inside? If on a Goneo, then distance?

Wavelength is the only constant as both instruments run through a Spectrophotometer.

Hope this helps.