Sponsored By
Bill Tobin

May 24, 2016

6 Min Read
Tolerances: Are you hitting the target or the bull’s-eye?

"Why will only being within tolerance make a good part?" the consultant asked.

A good analogy is target shooting. Hitting the target gets you points; missing it gets you nothing. Specifying a target that's too small to hit ruins the competition, because now it's down to sheer luck and not accuracy!

Consider the concepts of Cp and CpK, which originated in the precision machining business. Cp looks strictly at process variation. In the machining business, if your drill wobbles in the press, the accuracy and precision of the hole will be no better than the amount of wobble. Going back to our analogy, Cp defines the size of the target. CpK looks at how much you varied from the specified dimension (the bull’s-eye). This was developed in the machine tool industry, because the equipment running the cutting tool only knows the dead center of the cutter, but the size of the hole is where the cutting teeth are. CpK was the tool developed to look at the wear of the cutter.

A CpK of 1.0 means the entire target. But a CpK of 1.3 has essentially cut the target size by 50%! In plastics terms, if the tolerance is ±0.002, the designer has essentially required you to hold ±0.001. 

Going back to our analogy: If your shots are all over the target, you've just defined your CP (assuming a stable process). If you have a tight group of shots in the upper right hand corner of the target, the precision of your shooting is very repeatable; with a little tweaking of your sites, style or something else, you can move this tight group into the bull’s-eye, thereby improving your CpK of target shooting. 

Now, let's look at the real world, and switch analogies. When an area has a spring flood, there's always a lively debate that boils down to raising the dam or lowering the water. Raising the dam is a massive capital expense for which nobody wants to pay. Lowering the water means you create the flood earlier on in the hope that the real flood can be contained.

In terms of design requirements, raising the dam equates to buying/refurbishing equipment with the expectation that the newly acquired precision will meet the requirements of the design. Buyers routinely make these demands of their customers because they don't care what it costs them.  

Lowering the water is where I come in. It's a wonderful debate when I ask the designer: "Explain to me . . . " usually in the context of why his specified ±0.0005-inch tolerance is the difference between a successful product and either shortened product life or a catastrophic product failure.  

It's interesting to have this discussion. When I ask: "What happens if the part is ±0.002 inches? Will it still assemble? Will it fail?” More importantly, “perhaps we should be looking at CNC machining as either a secondary operation or replacing injection molding altogether and machine this part.” This is really a design for manufacturing (DFM) discussion. DFM says you design it to be made, not simply designed. This assumes the designer has a working knowledge of the technology used to manufacture the part. In many cases, he or she does not.  A design can be made more robust almost always by opening the tolerances. But you can only go so far. 

The robust machine 

First, you need to know the “natural” variation of your equipment. This is the combination of all the variables of the process: 

  • What's the “slop” in your screw and barrel? 

  • How tight is the shutoff in your non-return valve? 

  • How accurate are your heaters? 

  • Do the techs hook up the waterlines the same way every time? 

  • Are the machine heats accurate? 

  • While the machine settings will be different, can you duplicate what the plastic sees when you run the mold in an 8-ounce machine this week and a 12-ounce machine next week? More importantly, do you know how to do this?

  • Lastly—the ship sinker—is your equipment adequately maintained to be classified as robust, meaning it can mold parts within the variation the machine manufacturer specified? 

If you see cat litter absorbing leaking oil, your screw not loading up in the same time, your operators trimming flash that mysteriously comes and goes, or your techs telling you “that mold doesn't like to run in machine xyz,” then your equipment isn't robust and you have no hope maintaining any version of Cp or CpK, unless the tolerances are specified in whole inches, not thousandths of an inch. 

RJG Associates, which provides injection molding training, came up with an interesting test you can do on your machines to determine their robustness. This screening test separates the good from the not-so-good machines by giving you a score in percentages.

You take the reading for 10 shots: The time it took from a full shot to when you switched over to packing and its associated pressure. Then you make a small false sprue bushing that, instead of injecting the plastic into the mold, directs the plastic on the machine, much like a purging. You repeat the same measurements for these air shots. In theory, a new or robust machine will show only a difference in pressure between the air shots and production shots. Further, the variation in times and pressures will be minimal. RJG further proved the lower the percentage, the less controllable the process and, therefore, the more variation in the parts, i.e. there was a point in time when the machine would no longer be considered robust for precision molding.  

Once you've separated the good from the bad machines, there's a different test that takes a lot more work, which involves looking at the variability of each parameter in your machine's computer memory. That will usually point to worn component(s) that require maintenance or replacement.

In most cases you can convert a non-robust machine to a high-precision robust machine for a small fraction of what it would cost to replace it.  

Challenging design tolerances so that you can make acceptable parts is good business for both you and your customer. Improving your equipment's precision is inexpensive compared to low yields and the cost of resolving customer rejects. Try hiring a summer student to show you the profit you can make by raising your yields by 1%, or what a rejected shipment will actually cost you. It might make you a true believer when you see the profits you've lost. 

On request, the spreadsheets for these two experiments can be provided. 

It's your choice.

Bill Tobin is a consultant and owner of WJT Associates. He has authored several books and articles and teaches courses on plastics processing. Books, additional articles and the current public training seminar schedule are available at www.wjtassociates.com, or contact the author directly at [email protected].

Sign up for the PlasticsToday NewsFeed newsletter.

You May Also Like