When faced with a measuring problem, the first task is to reach for a catalog of measuring instruments. This leads down a path of sorting out characteristics of several instruments that appear to be suited for the application. Too often, a key detail is overlooked and only after the instrument has been put in service, the importance of the missing detail comes to light.

When looking for an instrument to measure a new or special feature, the following needs to be known to create a shopping list:

• What exactly is the feature to be measured? This may sound obvious but it is not. For example, it's a 1-inch or 25-millimeter outside diameter and it's long enough to use a micrometer. But if it runs up to a shoulder, your regular micrometer may not be able to detect a taper condition. The actual size and tolerance of the feature must be known and understood, as well as any related limitations such as roundness or taper.

• What is the process that will produce this feature? This is important so that you know what the instrument must be able to detect-negative byproducts of the machining process. This includes roundness errors on centerless ground parts or rolled threads, out-of-round conditions on thin-walled components held in three- or four-jawed chucks.

• What time and skill level is available for the use of the instrument? The perfect instrument may take too long or a machine minder is expected to work like a skilled metrologist. Often central planning expects miracles from nothing. They want such a short cycle that the person doing it may revolt and push the data transfer button twice for each part that is measured to get some breathing time.

• What masters are required to set and check the instrument? The best will duplicate the shape of the part feature being measured. Who will calibrate the masters?

• What are the operating conditions? Is the component being measured in a laboratory setting or hot off the machine? If it is measured while in the machine, will there be enough room to manipulate the instrument? How much cleaning is needed before measuring?

• What accuracy is needed? As you've probably guessed, this is not an easily answered question. What you need to know is what will the measurement uncertainty be for this measuring process using this instrument? This requires preparation of an uncertainty budget to determine if the process will meet the company's needs. As a worst-case acceptable ratio, the expanded uncertainty at K=2 should be less than 25% of the component tolerance.

• Is output required for statistical process control (SPC)? Finally, a simple question! If the answer is yes, in which case you need to check that its output is suitable for the system being used for SPC.

The answer to these questions will help you determine whether an off-the-shelf instrument-the most economical choice-is suitable for your application. However, sometimes what appears to be the perfect instrument is just not good enough.

Being the clever type that you are, you'll take a look at your uncertainty budget for the measurement to see what can be done. But when it comes to the instrument, not much can be done. However, other items in the budget may be sufficient, and changing one or more item could bring the overall uncertainty where needed.

For example, your budget may show a very high value for temperature variation. This could be reduced by adjusting the limits on the component to offset for the dimensional effects of this variation.

Perhaps repeatability is too high. This might be improved by the use of a backstop or centering device. Part geometry such as roundness that has been too high could be reduced with a change in machine setup, chucking or fixturing.

A benefit of preparing an uncertainty budget for such measurements is seeing the elements that make up the uncertainty number and meaningful changes that enable using an instrument that otherwise might be avoided.

You can't beat watching the operator doing the job before you start the instrument selection process. By doing so, you won't be in a position in which the ideal instrument is not used because the operator is left-handed.