It is often said that metrology is a method of communication and when you think about it, that’s a pretty accurate way to explain its purpose. But, too often, there are underlying assumptions that can mess up the communication to such an extent that the most precise measurement can be of little functional value.
This problem can arise in all measurement situations but probably occurs more often when features on machined parts are involved. One reason for this is part drawings that imply certain characteristics but don’t specify them so they are not top of mind when it comes to determining the procedure to be followed and, of course, the hardware to be used.
Continuing with the cylindrical object such as a pin, the drawing will usually stipulate a diameter for it and a tolerance. The assumption is that the diameter of the pin will be consistent anywhere along it. In real life that may not be the case. It may be tapered to such an extent that a diameter measurement taken in one location may show quite a different reading than in another. These measurements are occasionally referred to as ‘local diameters’ since they do not necessarily reflect the condition of the overall surface.
There is also the possibility that you could take a number of measurements that show little or no variation in the reading of diameter but the part won’t assemble because it is not round. A number of readings taken along and around the pin may show ovality or two-point lobing but not three or five lobed roundness variations. Straightness variations can remain undetected unless they are specifically tested for but I think I have beaten up on diameter measurements enough so I won’t go there at this time. These hidden factors can be the reasons parts don’t work even though basic measurements indicate they are within tolerance.
The center distance between two holes in a flat component may appear to be within tolerance according to a coordinate measuring machine (CMM) but won’t assemble. Questioning the results from a costly CMM can be a frustrating situation especially when the diameter of the holes has been verified in a couple of ways. The hidden factor causing the problem may be that the holes are not square to the datum surface and the CMM operator did not check for this condition.
Threaded components can have all the problems noted earlier for a plain diameter and have a number of others unique to them that can take what appears to be a simple problem and turn it into a nightmare. If the fixed limit gages being used on threaded parts reject them, it seems the blame falls on the gages first and foremost as any gage maker will attest, which is not usually the case. But what’s happening if the gages accept the parts and they still won’t assemble? It’s usually one of those hidden factors at work.
An example of this is where a company is making mating parts, one with an internal thread, the other with an external thread, each of which are standard dimensions and fits. At assembly, after a few turns, the threads lock up preventing assembly. The threads may be standard but longer than normal, so a pitch (lead) error was not picked up by the standard length gages.
It is often believed that the humble fixed limit gage may not be good enough for the task at hand and indicating instruments are considered a better choice for controlling a part size. The down side to this is that the indicating device is usually limited in the area it covers and relies on the operator testing the part at various locations. The fixed limit gage is an accurate but simple ‘master’ against which the component is tested but will usually pick up many of the problems I’ve noted when the effects of those problems will interfere with assembly or other functions.
A thorough knowledge of how the parts are produced and what the limitations are of the equipment used to qualify them is the key to preventing hidden factors creating mayhem in controlling dimensions.
In the sixth paragraph of my March column, the words ‘former’ and ‘latter’ were transposed. It should have noted that the accredited calibration was more expensive than the non-accredited calibration. I apologize for any confusion this error may have caused.