Two Cases of Gage-Improvement Failure

Today’s indicators are approaching bench amplifier levels of performance at a fraction of the price. So why not go out and get a digital indicator with a higher resolution and place it on the gage to improve repeatability? It’s not always so simple. Sometimes the gage design does not allow for this kind of upgrade. All it might do is let you see errors that were once invisible.

Consider the case of a portable height gage that had been used for years to check the height of small buttons inserted into the surface of the part. The buttons had smooth, rounded heads, and the gage moved around the part to check the height of the buttons relative to the surface. Like most height gages, it employed a flat anvil that could be placed over each button.

A simple gaging concept for improving the resolution of the readout from 50 microinches per digit to 10 microinches per digit should have been easy. However, the operators soon lost confidence in the gage because it was not as repeatable as it used to be. What once was one flick of the digit was now eight or nine counts. This seemed terrible to the user.

Two things were happening here. First, the operator could now see something that was previously hidden. The one count of ±50 microinches was now 10 counts to the operator. This was not a change in magnitude but rather a change in the ability to actually see inherent gaging inaccuracies. In this case, the flat anvils were not designed with a tight enough parallel specification to the base of the height gage for the higher resolution. Any out-of-flatness or parallelism of the anvil could now be read by the indicator and seen by the user. What once was invisible was now visible.

In another case, an inside diameter had been measured for years with an internal tri-contact bore gage. This had always been successful and produced a lot of good parts. Then the tolerance tightened. Now, gages with a measuring capability of 0.00015 inch had to measure parts with tolerances of ±0.0003 inch. That was clearly not going to work.

To measure these types of tolerances on the shop floor, a mechanical or air fixed-plug gage is generally the best tool. These high-performance gages virtually eliminate operator influence.

However, once again, things did not work out as planned. Soon, there were no good parts. The operators did not trust the results of the gage. In fact, they could not get repeatable readings with the gage at all. Rotating the plug or moving in or out to explore the bore generated a range of readings. The gages did not work as expected.

In this case, a lot of things were going on in the parts that were not detected by either the previous digital indicators or the gages themselves. To resolve gaging problems such as this, one has to start by looking at all the components of the measuring process. We started by taking a close look at the parts to see what the diameters looked like, knowing that when the tolerances get very tight, form and surface finish can take up a large portion of the tolerance.

A quick run on a form machine clearly showed a four-point, out-of-round condition that was, in fact, greater than the part tolerance. With an even number of lobes, the two-point, fixed-plug gage was ideal for showing the minimum and maximum values of the lobing. Therefore, the gages did not have a repeatability problem. They were displaying what they saw: The part out of roundness. No matter where the gages were placed in the part, a different number was seen, and with the high resolution, the readings were a jumble of numbers to the users.

So why did they never see bad parts with the tri-bore gages? This all comes back to the idea that an odd number of gage contacts works best when there are an odd number of lobes in the part. Because this part had an even number of lobes, the tri-bore gage was always reading an average of the points that the contacts were measuring. No variation was seen, because no matter where the gage was placed, it always read the same results: good.

Even though the operators did not like the results of the new gage, they understood for the first time what the part shape was really like. The new gage also revealed a process problem that had not been evident previously.

So in both cases, steps toward improved gaging did not work out as planned. However, these steps uncovered either gage or part issues in both cases, and that in itself could be termed a success.