July 2008 Edition

shop talk

Simple steps to avoid GD&T headaches

‘Entities should not be multiplied beyond necessity.’ — Ockham’s razor

By Richard Clark

If your facility is anything like many of the industrial facilities I have worked with in the "real world," then common GD&T problems might occasionally turn into a "free-for-all."

A machinist looks at a part drawing a certain way because of how he or she sets up the machine in order to make the part. The engineer views the part in a perfect CAD world. And the quality inspector is trying to re-create the drawing to confirm that each feature is correct, when measured exactly as the drawing "calls" out each feature.

The challenge for your facility is that individuals of many different GD&T knowledge levels must not only co-exist, but work toward a common goal without losing their minds. I suggest a simple, yet highly strategic Five-Step process to follow.

• Define it
• Find it
• Plan it
• Test it
• Measure it

To begin, I would create a brainstorming group of four to five people from as many job functions as you have in your equation. A nice blend would be an engineer, a machinist and two inspectors.

Did I forget anyone? Yes. Always include someone from your management team. It will save a lot of explaining later.

The first step of the investigation is to define the GD&T characteristic. There are many sources (or references) you can obtain if you don’t already have a GD&T reference. According to my colleagues at Technical Consultants Inc., parallelism is defined as the condition of a surface, line, or axis, which is equidistant at all points from a datum plane or axis.

What is really helpful for the team to look for within the definition is the "this on the drawing means this on a part" explanation.

Once the definition is clear, the team should be able to find the characteristic on one of your part drawings. The natural response now would be to measure your part but I don’t recommend this just yet. If you measure now and your reading doesn’t come out like you think it should, you’re right back where you started: Is the part bad, is the proposed measuring method incorrect, or is our understanding of the GD&T not correct? (…and nobody trusts the CMM).

This is why I recommend the next step be to plan it. You may have several measurement equipment options so it’s important to choose the best option for your situation. To determine this, I suggest testing a part of a known value. As an example, if you need to confirm parallelism of two faces using a CMM, you could wring two gage blocks together (slightly offset).

You could test your measurement method by measuring the top face of one block and defining it as the datum and then measuring the top face of the second block and "asking" the CMM to calculate the parallelism. These should be nearly perfectly parallel.

If this were the result, I would say the measurement method passes the test. This will also give an idea of what the best result you’ll get might be. If you found these gage blocks to be out of parallel by 0.001mm (just under 50 millionths of an inch) at a reference length of 25mm, that lets you know about what your part might read, even if it were perfect.

Once this cycle is completed, you’re set. When you measure the characteristic in the future and someone questions your method, you have a test ready to use.

Richard Clark is a quality systems consultant from Portland, IN. To order a copy of his book "Exposing the Myths of Industrial Precision Measurement" or to offer feedback, e-mail rcmetrology@yahoo.com

What do you think?
Will the information in this article increase efficiency or save time, money, or effort? Let us know by e-mail from our website at www.ToolingandProduction.com or e-mail the editor at dseeds@nelsonpub.com.