CMM Industry Errors Out

Xspect White Papers

Originally ALL Coordinate Measuring Machines were manufactured to a high very precision and were mechanically accurate. The quest for accuracy demanded the selective assembly of mating parts and the production of component parts to very high tolerances. Highly trained and skilled craftsmen undertook the assembly of CMMs in the manufacturing plant, and technicians with similar skills were necessary to install the machine and replicate CMM accuracy at the customer's plant. It was not uncommon for manufacturers to offer 2 different grades of CMM accuracy for the same model; the higher accuracy being attained by dedicating more care and attention in the final assembly and calibration process.

The first error mapping techniques were introduced in the early '80's to aid the linear accuracy calibration of individual CMM axes since scales are non-linear; prior to the introduction of linear error correction software scales were 'tweaked' by the use of scale tensioners. This error mapping technique is perfectly acceptable since it allows scale accuracy to be equated to the Laser Interferometer readout, thereby improving CMM accuracy WITHOUT compromising the intrinsic base accuracy of the CMM structure.

Within any 3 axis device there are 21 parameters of potential error. Straightness (2) per axis, Pitch, Yaw, Roll and linearity for each axis. Squareness of each axis to each other complete the 21 parameter count.

In order to offer large volume style CMMs with accuracies that had comparable accuracies to bridge style CMMs manufacturers developed full error compensation techniques. The machines were built as accurate as possible and data collected during the calibration process was input as errors from nominal into the error correction software to enhance the overall CMM accuracy. This error correction technique eventually found its way onto traditional small and medium measuring volume CMMs. Manufacturers used various names to promote this technique. MEA was used by Sheffield, which stood for Microprocessor Enhanced Accuracy. The use of error correction techniques DID NOT compromise build quality and merely allowed manufactures to offer more accuracy to end-users. The magnitude of errors introduced into the map was very small, generally limited to scale error and squareness error, indicating the intrinsic accuracy in-built into the CMM.

In the late '80's CMM manufacturers facing severe price pressure looked to lower manufacturing costs. One solution followed was the building of CMMs from aluminum; an initiative that started a trend that has subsequently become the norm. Rather than bloated CMM companies reducing their organizations and attempting to become more efficient they concluded to engineer profit back into their products. The only issue they faced was how to make aluminum CMMs accurate. Answer: throw the large errors prevalent in an aluminum structure into the error map and force the map to make the CMM structure accurate. Maps were designed to only meet the accuracy standard currently in vogue and assist the manufacturer in passing the "test". One by one manufacturers went to aluminum structures attracted by much reduced manufacturing costs. Manufacturers became dependant upon error correction software; their confidence in this fundamental technique increased with time and subsequently relaxed manufacturing tolerances further as they improved with experience error collection and mapping procedures. Today's aluminum CMMs are typically built by unskilled personnel and have no accuracy whatsoever until error mapped into specification.

CMM error maps have undergone similar evolution. Early maps only managed scale and later squareness errors. They were not full 21 parameter correction tools and worked well since the machine errors being corrected were "static" and easily reproducible. As you might expect CMM maps were designed to assist the manufacturer in passing the chosen accuracy "test". Today, with CMM mapping so prevalent, one legacy problem remains: no Certification of Map integrity by an Independent Body! How come? Most CMM measuring software's are evaluated through either the PTB or NIST algorithm performance test. Unfortunately, the same is not true for mapping algorithms or mapping models. The situation is made worse by the shear number of maps considering each manufacturer and variants by machine type. Is there any wonder why two machines, measuring the same part, yield different results! With no industry or government oversight the CMM user is at the mercy of the manufacturer and current CMM accuracy testing.

Aluminum CMMs offer no benefit, other than price, to the end users. They were introduced to the industry for purely selfish reasoning. Manufacturer of aluminum CMMs 'claim' aluminum is the perfect material for CMM build, (of course they do!!) how so since its coefficient of expansion is almost 4 times that of granite and yet its specific weight is only 1% less than granite. The reality is aluminum allowed for mass production techniques to be introduced into an industry that was inefficient and losing money. Aluminum IS NOT a Metrological Material. Granite IS; that why all CMMs using granite tables I guess.

The Advantage of using Metrological Material in CMM build 

A few CMM companies have retained the original manufacturing techniques for CMMs and yet have remained competitive and profitable. One such company Wenzel of Germany has grown dramatically by bucking the aluminum and error map trend and has subsequently become the 4th largest CMM manufacture in the world as a consequence. Wenzel is delivering CMM structures that are manufactured to exacting tolerances, built by skilled tradesmen and achieve their competitive accuracies without the use of 21-parameter error mapping techniques. The key to Wenzel's success is a vertically integrated manufacturing facility where the investment has been in people, manufacturing processes and t horoughbred engineering not compromises.

Most manufacturers of aluminum CMMs also "lock-up" their error maps to prevent customer access. Maybe they are embarrassed by the magnitude of errors they are compensating for or maybe they mandate the calibration dollars ($$$$'s) spent annually by CMM users belong to them as they refuse to allow others to adjust 'their maps' to correct for CMM geometry adjustments that occur with aluminum structures over time.

In addition the error map data is collected during calibration with the CMM structure static and yet the CMM is compensated for dynamically during the measurement process. Probe calibration occurs an integral part of CMM usage since static probe diameter differs from that obtained during dynamic probe calibration. How then can static error maps be used dynamically with accuracy.

Homing the CMM is required on boot-up of the measuring system. The CMM triggering end of stroke limit switches on each axis typically is used for homing. The repeatability of machine home is approximately +/- 5mm and yet it's this dynamically created home position that triggers the error map zero position. An error map with large errors 5mm out of position can have a significant impact on compensated measured data.

Most CMM software's have measurement algorithms that have been certified by a national accreditation body such as PTB; no such accreditation of vendor error map algorithms exists. Why not? Today's aluminum CMM relies totally on the validity of these complex mathematical formulae to provide volumetric 3D CMM accuracy. Would you fly on an airplane with invalidated software managing the autopilot; would the airplane even be allowed to leave the ground?

Error mapping is certainly a 'black art' and a subject matter the industry prefers not to discuss. Why not? Does it have something to hide?

CMMs built mechanically accurate in the '70's and '80's remain in regular use today; what length of life can you expect from today's aluminum CMM devices?

Wenzel are proud to be able to quote CMM accuracies of its products with and without error maps. The difference is minimal indicating that Wenzel structures are intrinsically accurate. Maybe the CMM industry should be asked what the accuracy of its product is with the error map switched out. Prepare to be shocked!!!!

Voice: 248.295.4300
Toll Free: 866.4.Xspect (866.497.7328)
Fax: 248.295.4301
inquiries@xspectsolutions.com

Headquarters
Xspect Solutions, Inc.
Wenzel Building
47000 Liberty Drive
Wixom, MI 48393

Click driving directions for interactive mapping and directions.

White Papers Navigation
CMM Industry Errors Out
Basics of ISO 10360-2
Sizing The CMM
CMM History