I need help. Our customer requires we measure color and gloss and we have added this equipment to our control plan. Because it is in the control plan we now are required to perform a GR&R or some other type of statistical analysis to identify accurracy. A spectrophotometer and gloss meter will not provide an acceptable GR&R. I would be grateful if someone could provide me with an alternative method or a solution to my problem.

Both provide variable data, and you have masters for the measurement base. So if you had 2 operators why can't you do a gage R&R?

Both provide variable data, and you have masters for the measurement base. So if you had 2 operators why can't you do a gage R&R?
We do have color and gloss masters for the spectrophotometer, however because there are so many variables in the measurement process, we can not obtain an acceptable GR&R. By design, our plastic molded parts are not perfectly flat. Chrysler requires SPC for color measurement, yet they indicate in their own standard that visual is the acceptance criteria. To me this indicates they do not themseves trust the spectrophotometer readings.

Probably a conflict they don't know exists. Suggest you contact your SDE and provide Chrysler conflicting info, and ask which is the correct one to follow. Can you make flat panels to test the compound, such as with lab equipment molding?

In general, it shouldn't be necessary to do instrumental color and gloss evaluation in production--it's a matter of process control and visual inspection, using trained evaluators and the proper light source. You do need the spectrophotometer for development purposes, to provide tristimulus data for the Appearance Approval Report, and to provide guidance in achieving an acceptable match to the master. You need to talk to your customer about this, and get direct measurement off of the control plan. You probably won't be able to do an R&R that anyone will be happy with.
Note: if you aren't doing so already, you should be using an FM 100 Hue Test Kit (http://www.gretagmacbeth.com/index/products/products_color-standards/products_color-vision-tests/products_fm-100-hue-test.htm?) to qualify inspectors and operators who are responsible for visual color matching. Use of this test provides good evidence for your customer that inspectors are qualified to do visual color evaluations.

I used to be the QM at a company that made PVC in color, closs, grain for Boeing, Douglas, Sikorski, etc. We had close color controls and used the spectrophotometer for start up approval and monitoring by the Tehs. during production. You are correct about the color /hue testing. Anyone who was making color decisions had to pass. However, our Technicial Director was color blind and had to rely on others to help judge formulation adjustments.
We kept sample during production, and also the print outs from the spectrophotometer readings. We were not required to do any R&R studies at the time.

Probably a conflict they don't know exists. Suggest you contact your SDE and provide Chrysler conflicting info, and ask which is the correct one to follow. Can you make flat panels to test the compound, such as with lab equipment molding?
I plan on contacting Chrysler. I just thought I would put this out there to see what other suppliers are doing. I can't beleive we are the first to run into this conflict in the standard.

In general, it shouldn't be necessary to do instrumental color and gloss evaluation in production--it's a matter of process control and visual inspection, using trained evaluators and the proper light source. You do need the spectrophotometer for development purposes, to provide tristimulus data for the Appearance Approval Report, and to provide guidance in achieving an acceptable match to the master. You need to talk to your customer about this, and get direct measurement off of the control plan. You probably won't be able to do an R&R that anyone will be happy with.
We do conduct color acuity testing using the Farnsworth-Munsell 100. And we use the McBeth Light Booth. In addition to performing visuals, the standard does require mechanical readings for initial process capability studies, incoming materials, 1st off parts, inprocess parts and outgoing parts. It then goes on to say that outgoing SPC data must be compared to 1st off and in process SPC data. To get meaningful data, I need a spectrophotometer that has an acceptable R&R. Thus my problem.

In your original post, you said that the customer requires GR&R "...or some other type of statistical analysis." If you're sure that GR&R will not provide reliable results, how do you know that the measurement system will provide reliable results in production? That's a rhetorical question. The problem seems to be that you've agreed to meet a standard that you can't meet, which means you must negotiate something that makes sense with the customer (not always possible in the automotive world) or make something up. I recommend the former, of course, but I've personally taken refuge in the latter when the customer refused to listen to reason and I was sure that what I was doing made sense.

In your original post, you said that the customer requires GR&R "...or some other type of statistical analysis." If you're sure that GR&R will not provide reliable results, how do you know that the measurement system will provide reliable results in production? That's a rhetorical question. The problem seems to be that you've agreed to meet a standard that you can't meet, which means you must negotiate something that makes sense with the customer (not always possible in the automotive world) or make something up. I recommend the former, of course, but I've personally taken refuge in the latter when the customer refused to listen to reason and I was sure that what I was doing made sense.
You are absolutely right when you say it is difficult to negotiate with the customer. We have found them for the most part to agree verbally. However, we have yet to get anything in writing where the standard normally requires a signed waiver. I was just wondering how other suppliers got around this. I'm tempted to use your "latter" advice, but will of course try to negotiate with the customer first. I can tell from your response that you have been in the automotive arena for awhile. Good advice and you know the game. Thank You for your advice and honesty.

You are absolutely right when you say it is difficult to negotiate with the customer. We have found them for the most part to agree verbally. However, we have yet to get anything in writing where the standard normally requires a signed waiver. I was just wondering how other suppliers got around this. I'm tempted to use your "latter" advice, but will of course try to negotiate with the customer first. I can tell from your response that you have been in the automotive arena for awhile. Good advice and you know the game. Thank You for your advice and honesty.
You're welcome. Unfortunately, the Big 3 spend a great deal of time in inadvertently urging suppliers to follow the path of least resistance. I worked for a time on the supplier end of things, so I do have firsthand experience in that sort of creative writing. I currently work for a vehicle manufacturer and spend most of my time dealing with suppliers, so I've learned how not to do it.

The great irony of this color matching thing is that even when you supply spot-on colorimeter readings with excellent delta-E values, the customer's final judgement is likely to be visual and subjective anyway. Good luck.

(not always possible in the automotive world) or make something up. I recommend the former, of course, but I've personally taken refuge in the latter when the customer refused to listen to reason and I was sure that what I was doing made sense.

Et tu Brute :mg:

Et tu Brute :mg:
You betcha. Never trust anyone who has been in quality any appreciable amount of time and hasn't fudged something. As it is with sculpting, the trick is in knowing when to stop.:D

You stated that the R&R was unacceptable, but did not elaborate on whether the issue was Equipment Repeatability, Operator Reproducibility or both. This would help to narrow the focus of the problem. You also mentioned that the parts are not flat. Can you fixture the part and the test equipment such that the parts are measured in the exact same location and orientation every time? This should help improve both repeatability and reproducibility. It would also eliminate the impact of within-part variation on repeatability.

Note: if you aren't doing so already, you should be using an FM 100 Hue Test Kit (http://www.gretagmacbeth.com/index/products/products_color-standards/products_color-vision-tests/products_fm-100-hue-test.htm?) to qualify inspectors and operators who are responsible for visual color matching. Use of this test provides good evidence for your customer that inspectors are qualified to do visual color evaluations.

Does anybody know the required frequency of this Munsell Hue test? Is once in a lifetime enough or is it annual?

Does anybody know the required frequency of this Munsell Hue test? Is once in a lifetime enough or is it annual?

I don't deal much with color evaluation anymore, but I seem to recall that it was a matter of controversy. In my own experience in administering the test annually to perhaps 10 or 12 individuals over about 3 years, I was struck by how consistent the results were, and didn't see anything that I could identify as a trend in the data. The guy who did the best on the test has only one transposition error (the idea is to put different-colored disks in order according to subtly graduated differences between them), and in three annual tests he transposed the same two pieces!

My suggestion would be (if you're doing automotive work), stick to annual testing unless the results tell you it's safe to do otherwise.

One of my colleagues told me that there are definitely changes to color
perception that result from aging - apparently the receptors that respond
to different hues age at different rates. So certainly re-tests are in order.


My colleague says that this is the reason why the classic expression about
"blue haired old ladies" originated -- because the women themselves had
lost some of their color perception.

Regards,

Brad

PS -- at GM, they have some people at their corporate offices who were
extremely knowledgeable about color. (In the paint section).

I would assume that all of the OEM's
have such staff folks. Perhaps you can engage one of these people and
ask you to recommend the best method for solving your concern; while
this may not constitute an official "waiver", it might help you focus your
concern on real priorities instead of meaningless requirements.

Hi Trisha,

I'm having the same problem with R&R gages that I was performed on my glossmeter. The %G R&R always is greater than % 50, and ndc value is lower than 2. I think that reason for bad results is due to the tight tolerance range for this process (0,5 U.G) and the low equipment resolution (0,1 U.G). Somebody can help us to explain clearly to our costumer that it is impossible to get good results with these process conditions.

You guys have to post the data for any help.

My first thought is that you are not controlling the sample placement well enough. You really need to be able to measure exactly in the same spot. I would measure without moving the unit a few times to see if the unit repetes at all, if it is not repeteing you may have a unit issue. If it is repeteing according to normal specs, then make sure your samples have normal laydown variation.

You may have to cut up the samples so you get a uniform sample. Remember you need to have pieces that reflect the normal process spread.

Oh and post the data......

Hi,

I attached my R&R study data in order to be checked better and somebody can explain me what could be the cause for these bad results.

Thanks

Your gage's resolution is inadequate for use as an inspection device or as a process control device.

There are also indications that there may be issues with operator reproducibility, but this is not definite due to the inadequate resolution.

You need a gage with another decimal place of resolution to improve your results.

Your gage's resolution is inadequate for use as an inspection device or as a process control device.

There are also indications that there may be issues with operator reproducibility, but this is not definite due to the inadequate resolution.

You need a gage with another decimal place of resolution to improve your results.

Do the samples represent the full process spread? The samples measured only have a total spread of 0.4, and the equipment resolution is 0.1.
If your actual process spread is greater than 0.4, you may get much improved results by using a more representative sample.
If not, I agree with Miner that the equipment resolution is not adequate.

Hi Trisha,

Can you obtain good CgK results? As long as the master samples can be located / mounted in a suitable manner which is consistant it should'nt be a problem - so run a CgK study first if that does not work then GR&R is a waste of time.

Best Regards,

Mr Gunz

Hi Miner,

Thanks for your comment. I knew that I should to get an glossmeter with better resolution, but the problem here is that in market there are not any equipment with resolution lower than 0,1. So, What should be the next step to follow?

I have had a look at the manufacturers data sheets for several gloss meters and every one I've found has a resolution of 0.1 gloss units. They typically quote repeatability of 0.2 gloss units and reproducability of 0.5 gloss units.
As your tolerance is 0.5 GU and the spread from your sample is 0.4 GU, you will have no chance of acheiving acceptable R&R using this equipment.
Explain to your customer that you are using the best equipment economically available (Industry standard equipment). Ask your customer to advise what equipment they use to measure gloss and suggest that they talk to their own paint experts on how they validate their measurement process.
I will be very surprised if they have acceptable R&R studies!!!

Thanks prototyper,
I'll negotiate with my costumer in accordance with your proposals!!

Hi, I attached my R&R study data in order to be checked better and somebody can explain me what could be the cause for these bad results. Thanks

Is this your own spreadsheet? If so, nicely done!

The great irony of this color matching thing is that even when you supply spot-on colorimeter readings with excellent delta-E values, the customer's final judgment is likely to be visual and subjective anyway. Good luck.

Truer/sadder words have never been typed. It doesn't matter one bit what the approved master looks like or what the color meter says, once a Customer Representative announces "I don't like the color" tons of non value added work will begin.

Our paint supplier tells a story about a guy who would spray out many color plaques at PPAP time and keep them in his desk. After a customer flap about color he would let a few days go by, and send the "new" plaque along with a suitable tale about process changes, new paint lots, re training, and so on. The customer ended up happy and production went on....

Hi Miner,

Thanks for your comment. I knew that I should to get an glossmeter with better resolution, but the problem here is that in market there are not any equipment with resolution lower than 0,1. So, What should be the next step to follow?
Based on this and prototyper's comments, I recommend the option of last resort, namely averaging multiple readings.

If you take n readings and average them, you will reduce the Repeatability component by the SQRT(n). That is, averaging 4 measurements will reduce the repeatability by the SQRT(4) or 2.

Repeat your MSA using the average as an individual reading. Your resolution and Repeatability should improve and expose Reproducibility as the next issue to address.

Averaging multiple readings is a good idea. We have used the gloss meter from BYK Gardner and had good results with averaging. Very accurate.

Hi Djlance,

Averaging multiple reading sounds like a good idea, but what can I'll spect from the results of [B]ndc value... it'll still being low???