Color specifications 3

[dgallup]

We buy a fair number of molded plastic components that need to be molded with different colors. Usually the color requirements are pretty loose but our wonderful offshore "low cost opportunity" suppliers must be pretty color blind. So we end up putting something like this on the drawing:

It makes for a really messy drawing and it's a PITA to sit down with the Pantone color chips and sort through them all to come up with all the possible colors that will be acceptable.

It has been suggested to use something like this:

But I don't know where to get sets of accurate color chip samples for engineering, receiving inspection and suppliers so we can all work to a common standard. Note that neither we nor the suppliers have access to colorimeters, color calibrated displays, color calibrated printers, etc.

I'd love to hear others solutions to this.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[Nescius]

But I don't know where to get sets of accurate color chip samples for engineering, receiving inspection and suppliers so we can all work to a common standard.

In another life, I worked for an outfit that imported consumer goods where color was an important thing to watch...fabric mostly. Using calibrated, well, anything really, was not an option. If I recall correctly, the "solution" was to buy two identical Pantone booklets at the same time and send one to China. This does nothing to define a range of acceptable color, but it was at least a common mode of communication.

Replace Pantone booklets as a pair.

 

[MikeHalloran]

<tangent>
My friend Mitch told a story about making a lab instrument that had a tray recessed in the top.
He ordered a truckload of cafeteria trays, custom colored, with pigment mixed in the resin, to match the instrument's ivory color.
He supplied a color sample with the order, comprising a sheet of posterboard, custom painted by him, with the exact paint and texture used elsewhere on the instrument.
The trays arrived just before scheduled production start, uniformly colored in a hideous shade of green.
... which was an exact match to the green on the back of the posterboard sent as a color sample.

</tangent>

Mike Halloran
Pembroke Pines, FL, USA

 

[dgallup]

Nescius - That's what we do currently (not necessarily the paired replacement). The Pantone color chips are not the problem, it's the ridiculous seemingly random number designations that make it nearly impossible to specify a range of acceptable colors.

I've come across the Munsell color space which looks like it would work well as it uses 3 parameters for hue, value and chroma that describe a 3D color space. The only problem is it depends on using color calibrated displays or colorimeters, neither of which are available.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[drawoh]

dgallup,

Pantone is for printer's ink. I believe this is enough information to mix dyes. I do not know how well plastics manufacturers can match it.

I try to call up RAL[&nbsp;]numbers for paint. There also is Fed-STD-505, again for paint. My links are to online colour charts. Is a computer screen a close enough match for you?

There must be a standard for colouring plastic.

--
JHG

 

[CheckerHater]

Note that neither we nor the suppliers have access to colorimeters, color calibrated displays, color calibrated printers, etc.

How about customer?

I am not trolling. If all three of you are merely eyeballing the color, why not specify just one Pantone number as a goal everybody is trying to reach?

And resort to expensive measures in the event of conflict.

In our place we specify Pantone number for customer logo printed on decals and such. Nobody complained yet.

Colors for plastic parts may vary more, but still work.

Unless YOUR parts are actual color chips...



"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[Nescius]

With the "calibrated" pair of Pantone booklets, we specified a color. Sampling was the next step, with both parties retaining a sample of the approved color. We then had a "calibrated" real item. Continuing production QC was completely subjective, but defining the edge of an acceptable color range would also be completely subjective. Without hard science, we were chasing our shadow, but it worked "good enough".

 

[dgallup]

Perceived color is perceived color, I don't care how it's produced. RAL numbers suffer from the same problem as Pantone numbers, no way to specify a tolerance. We have a set of RAL color chips but never use them as none of our suppliers have them. The Fed standard looks to have the same problem, there are greens all over the chart with completely different colors in between. The problem fundamentally is you can't define color on a simple linear number line. The system has to have at least 2 and preferably 3 dimensions. The RGB hex code would work better as you could give a +/- tolerance to each component. Now just try to explain to someone in a "low cost opportunity" country what you mean by 373D2F +/-5+/-C+/-6(hex)!

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[dgallup]

CheckerHater - I tried just giving a reference Pantone color but when you tell the supplier to make it red and it comes in light pink and we reject the shipment then there has to be a tolerance. We are not trying to get picky but we want them in the ballpark and fairly consistent. The only downstream complaints have been from our own marketing department.

Look at the range of colors I've allowed in the first post, it's not rocket science. However, there is no way you can do this with standard color monitors and color printers, they are all over the place just like our suppliers. This would not be a problem if we were not forced by upper management to use "low cost opportunity" suppliers.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[dgallup]

MikeHalloran - Great story.

MadMango - We get PPAP samples of every part but it can take 3 submissions to get anywhere close and then the next shipment is out in the weeds. Some of these parts have a dozen different colors, some are very low volume, some may be 300,000/yr.

I just want a color system with an easy to specify tolerance. Should be so simple.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[pylfrm]

dgallup,

Specifying a tolerance is relatively easy. Deciding what the nominal color and tolerance should be without a calibrated display might be rather more difficult.

Tolerance zones are often given as ellipsoids in a 3-dimensional color space. If you insist on not using a colorimeter, then you could obtain sample sets that show the nominal color and two limits for each axis. This turns your tolerance zone into more of a rectangular cuboid, but I imagine you can live with that. Here's an example source for such sample sets:

http://munsell.com/color-products/custom-color-services/munsell-color-tolerance-sets/

A web search for "Delta E" will turn up lots of info, including the Wikipedia article on color difference. It even has a section on tolerances:

https://en.wikipedia.org/wiki/Color_difference#Tolerance

Keep in mind that these tolerances are for the reflected/emitted light, which is dependent on the illuminant.

There are some ASTM specifications available as well, but I don't know anything about them.



pylfrm

 

[3DDave]

Welcome to color space problems.

Problem 1: You need to specify the color components of the lighting source. Daylight vs tungsten vs LED vs fluorescent vs mercury vapor vs an endless number of other options will shift the reflected colors.

Problem 2: Color is subjective

Problem 3: Color can change depending on surface texture. A glossy finish will look darker for the same material than a matte surface due to scattering.

Solution 1: Mix a really big batch of a color you like and hope it doesn't run out or that the pigments can be carefully managed to mix future batches. The latter is how the paint stores manage to match their own color samples - they know how much of each tint to add to reproduce the color chip.

Solution 2: Unfortunately solution 2 requires buying or renting color matching hardware. I did this on the cheap with a digital camera and Photoshop to match a FED-STD chip under direct sunlight to some molded samples. Photoshop can give the RGB coordinates from the image the camera captures. I suppose one can, in Photoshop then vary the RGB values to see how far the values can shift before they are unacceptable.

Solution 3: Be really specific about the surface finish.

My final suggestion is to contact a local printer and pour out your heart to them about color matching. They will probably have had the same situation to deal with. My minor experience is with a photo lab that did commercial photography. For at least one customer, the tolerance was none. If it didn't match, it didn't get paid for.

(edit: Keyboard typed what keys I pressed.)

 

[3DDave]

I just remembered solution 4: do what GM did. They originally wanted to mold all the Fiero parts in finished color, but couldn't get the processes to match, so they went with paint.

(edit: Keyboard typed what keys I pressed.)

 

[CheckerHater]

@dgallup:

I understand your situation. Well, another way is to set up your own standard. Any kind of procedure that you, customer and supplier can agree upon. Then refer to the document on your drawing.

I know the company that was not much into GD&T, but would buy or make gauges and send them to the supplier. Drawing would say "check with the gage ABC123"

Make your own samples, have them approved by the customer, prepare the document describing how to compare parts to samples. Lot of work, I know :-(


"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[dgallup]

pylfrm - I like the Munsell color scale. However, I don't want to have to get multiple sets of color control chips for each color specified on each part. Too many different sets to keep track of. I want something universal like the Pantone or RAL chip sets where every available color is in one set, then just specify the acceptable range. It's just that Pantone and RAL are discontinuous numbering systems making it impossible to specify a tolerance. Look at my example, we are not trying to match automotive paint samples, we are just trying to hit the broad side of a barn. I don't care about the little details like what kind of light source and surface texture. You would cry if you saw some of the initial samples received for the "orange" above. Came in more like brown, nobody would look at them and say they were any shade of orange under any light source.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[CheckerHater]

dgallup, you asked how to specify color on the drawing, while you should ask how to measure it.

You mentioned that you don't have a color-measuring device. You can buy one at Wal-Mart:

https://www.walmart.com/ip/Pantone-...43685&wl11=online&wl12=29222579&wl13=&veh=sem

Once you measure actual Pantone number of your part, you can look at your chart and see how far away from the desired color you are.

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[3DDave]

http://sensing.konicaminolta.us/2014/05/defining-color-tolerances/

 

[pylfrm]

I want something universal like the Pantone or RAL chip sets where every available color is in one set, then just specify the acceptable range.

If you really want a universal set with broad coverage of a 3-dimensional color space, I wouldn't be too surprised if you need upward of 1000 chips. Consider the second image in you original post for example. The main section has 21 rows and 37 columns, for 777 chips total. The red box is drawn around 16 of them, so perhaps you'd say this chart has roughly double the resolution you need in each of its two axes. Maybe you can get away with 200 chips for these two axes, but there is still the third axis to consider. Only 5 gradations there and you are up to 1000.

I imagine the same people that make the color tolerance cards could make universal sets for you, and they may also be able to help with design choices such as color space, resolution, range of values, etc.


pylfrm

 

[dgallup]

That's a good link 3DDave, it nicely defines the problem. Unfortunately, their solution revolves around using a colorimeter or spectrophotometer. Maybe we will have to go down that route in the end. I'd be happier to get something like the second image in my first post printed on stable sturdy stock with stable inks commercially available that we could tell the supplier row 0 +/-3, column 24 +/-3. If we have to replace them every year, so be it.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[pmarc]

dgallup,
If you are okay with square or rectangular area of acceptable color range (just like you showed it on the second image in your first post), you could define it using "coordinates" of 2 diagonal corners. Of course first you would have to somehow clarify exact meaning of that custom drawing specification, but by doing this you would get away from discontinuous numbering system.