Optical Measurements (Profiler), Trust & Traceability

[Conrad Hoffman]

The profiler in question is basically a collimated light source and a camera system. You hold something in the beam and the camera end sees and measures what is essentially the shadow. If the object is a round pin, I can be highly certain that the light is being cut off tangent to the round surface along a line. IOW, I know exactly what surface is controlling the measurement. The situation is about the same as using an optical comparator or measuring the diameter with a micrometer, and I'd expect them to all be in very close agreement.

Being so fast and easy, somebody decides to measure a hole. If the hole is in something thin, say a photo-etched part, where alignment isn't a critical issue, you get the hole diameter based on the limiting aperture formed by burrs and taper. Probably an accurate result and I wouldn't expect a major argument in getting it accepted.

My problem starts when somebody wants to measure a bore with any depth. The alignment becomes critical and it has to be done in two planes to have any hope of a usable result. Even then, I can't say with certainty what edges are controlling the measurement. IMO, and I think my logic is sound, if you can't say what edges/surfaces are being measured, you can't properly say whether a part meets the print or not. The results may be interesting and useful for process control or whatever, but they can never be used as an argument for compliance to a print. (same issue with any flat part, say a gauge block)

I've never seen any "official" comments on this from NIST or any other authority I can find, but it seems fundamental that you have to know what surfaces/edges control a measurement for it to be a deciding factor, legally or otherwise, in compliance.

Or am I just being too nitpicky? Any comments on this?

 

[TGTool]

I'd agree with you completely.

 

[Milling man]

Or am I just being too nitpicky? Any comments on this?

Probably, it is impossible to deceive physics very much. If we need to measure a hole in a thick part, we cannot do it with transmitted light, only with reflected light.

 

[jccaclimber]

You are being picky, but all methods have their faults.
On a comparator, or any other single side facing optical measurement method you need to be squared up, and you're getting the smallest value, with error for some fuzz depending on how well collimated your source is and how telecentric your optics are. When measuring with calipers, you need to make sure they are squared up, and that you have selected the largest/narrowest point.
With a (no-go) gage pin in a hole you only check the smallest point, and if it's a blind hole you're only checking for straight or shrinking holes. Barrel or widening holes (more common than you might think on say a 0.005 mm tolerance hole) aren't detected, nor are holes with an OOT large lobe wider than the uncut portion of a diamond pin.
Anything with a CMM you're checking a small number of points on a surface, and assuming the rest of the surface is within those.
Air gaging you're assuming your surface roughness doesn't change much as that will move measured diameter a tenth, sometimes more.

I could go on for a while, but I think you get the point. For something like a comparator I like to stay a bit farther from the accuracy limits of the machine than say a pair of calipers or a micrometer.

If you want to drive this point home to someone, measure a class X gage pin or ring gage in the axial direction on an optical system. It's not unusual to get a substantially incorrect value in one axis due to not being perfectly square.

 

[Conrad Hoffman]

Appreciate the input. The manufacturers of these systems are very good at pointing out the strengths, but don't really want to say any more than necessary about the weaknesses. The part that got me thinking about this is a tube with a cross drilled hole. The spec is for the distance from the end of the tube to the center of the holes. The measurement system is very good at finding the center of what it sees, and can give an accurate number if both sides are in line. It occurred to me that if one hole is out of spec high and the other hole (other side) is out of spec low, by the same amounts, the machine will locate the center and tell you the part is exactly to spec, when it's anything but.

 

[ttrager]

Would something like a Novacam make more sense relating to your question? They make laser based measuring systems specifically designed to drop into deep bores and take "accurate" measurements (don't recall the accuracy rating). Their videos on youtube show a variety of scanning scenarios from the rifled ID's of rifle barrels using one of their long probes, to complex parts both OD dimensions as well as multiple ID's and holes.

Their systems basically have long probes of different diameters lengths, you pick which is needed for the measurement. The beam actually comes out at 90-deg to the end, sideways, and the probe spins. A CMM program is written that moves the probe around the outside of a part taking measurements and building a 3D model, and if it needs to get into a bore, does the same thing: Drops into the bore with the probe spinning, the laser sweeping the ID on the way down taking data points.

I've never used one. I know nothing other than what I've seen on marketing videos.

Just thought it fit with the question you were asking about using light-based measuring systems on bores, particularly deep ones.