Home
calibration

Calibration


Description of a Coordinate Measuring Machine


A coordinate measuring machine is a class of equipment designed for three dimensional inspection.  These machines provide accurate XYZ point data for any location within their measurement volume.  Each axis of the machine has a precision scale with the three axis held at right angles to each other (orthogonal).  Measurements of length are done by finding the XYZ difference between any pair of points in the machine volume and converting this into a length or 3D distance.
calibration_cs
As computers evolved the ability to expand the measurement capability of a CMM has changed significantly.  Computers allow the user to combine individual point locations into complex geometric features such as planes, spheres, cones, cylinders, and so on.  The geometry allows for additional evaluations such as the angle between the axis of two cylinders or the diameter of a circle.  Computers allow the user to record inspection steps into a part program that can be run over and over allowing faster turn around time than what would be possible using traditional methods such as a surface table and height gauge.

CMM's can be manual or computer controlled.  Manual machines are ideal for single part or simple one time measurements.  Motorized machines are ideal for manufacturing environments where many parts of the same type are produced and sampling is required to monitor the production process.  CMM's come in various shapes and sizes and are capable of using a variety of sensors such as hard probes, switch probes, analogue probes, laser probes, laser scanners, and cameras.

Calibration of a Coordinate Measuring Machine

Calibration, in the context of a CMM, is adjusting the machine in order to reduce the relative error between any two XYZ locations in the machine volume to zero.  The purpose of a CMM is one of measuring length where at least two points are required to produce something meaningful.  Combining two of more points into geometric features is a variation where the relative positions of the input features defines the size, location, and orientation of the output feature.  If there is no relative position error between any two XYZ locations in the machine volume then there is no error contribution from the machine.

It is not enough to simply report the current errors of a CMM to the end user as would be the case when a machine is tested but no updates are performed.  For some equipment such as ring gauges or gauge blocks knowing the actual length of size of the feature is information that is actually usable but this doesn't really work on a CMM as these errors are almost impossible to subtract from measurements performed using the machine.  At best, in this situation, the measurement uncertainty from using the CMM would need to be set higher than necessary in order to contain all the known machine errors.  This is the primary reason that Select Calibration Incorporated does not perform verification only calibrations of CMM's as this is not the expectation of the end user.

Calibration of a coordinate measuring machine is surprisingly complex.  Removing known errors of the machine is something that is needed when the limits of mechanical adjustment has been reached.  Over time almost all manufacturers had adopted a nearly identical process of describing basic geometry errors of the individual axis and using this to apply software correction at a level that is hidden from the user.  Most coordinate measuring machines have a minimum of 21 calibration map parameters that can be adjusted.  Some machines have additional compensation parameters to deal with problems associated with deflection errors for horizontal arm CMM's, axis with two or more scales, hysteresis errors, or even expected changes in shape from the weight of the part placed on the machine.  Machines with parametric temperature compensation can have up to 36 compensation parameters active at any one time (18 static, 18 dynamic based on temperature) which can add a good deal of complexity to something that is already complex.

Calibration Requirements

Calibration of a three axis CMM is far more involved then simply comparing each machine axis to a reference standard.  On a three axis machine a twist or bend in any one axis will directly affect the other axis.  The effects from twists and bends will change as you move throughout the machine volume requiring a valid look-up table of corrections depending on where the machine is located.

calibration_me Manufacturers minimize accuracy related problems as much as possible by using materials and manufacturing processes that produce guide-ways with minimal twisting and bending.  Mechanically producing a perfect machine is either impossible or impractical due to the costs involved.  Residual errors from manufacturing are removed from the machine by software using the data from the compensation map as a basis for correction.  The software uses the known mechanical errors and calculates the real position of any point in the machine volume.

In order to properly calibrate a CMM it is important to inspect and update all geometry errors from each axis.  Some machines are quite stable and will change very little over time while others will change shape from thermal cycling over the course of their lifespan.  All machines require periodic updates to ensure the best measurement performance possible.

Compensation maps are an essential part of most CMM's and allows for adjustments that can be very difficult, if not impossible, to perform by only mechanical means.  Most manufactures of modern CMM's rely on the compensation map to achieve target performance whereas the very best machines use a combination of good mechanical design and software compensation.

Temperature Compensation

Temperature compensation is an attempt to minimize the errors of a machine due to changes in temperature.  As temperature changes the material used to construct the CMM will grow or shrink, twist, and bend.  Since the machine structures are complex shapes with welded reinforcements throughout the change in shape is never predictable.  For this reason it is always preferred to have the machine in a proper thermal environment over relying on temperature compensation.

One type of temperature compensation that can be used on a machine is simple scale correction for each axis and the part.  Each axis of the machine has a known expansion coefficient and will grow and shrink by a predictable amount.  The part expansion coefficient can be estimated by knowing the material of the part and referring to established values.  This kind of temperature compensation does not take into account changes in the shape of the machine but only the expected changes in the machine scales and part.

Another type of temperature compensation involves actively changing the amount of software compensation applied to the CMM as the temperature changes.  Each family of machines that use this type of compensation will undergo testing in an environmental chamber in order to produce the required correction coefficients based on where the changes in shape are occurring.  The correction can be either a complex curve or a simple gradient depending on which compensation map parameter the correction is applied to.

For CMM's that use temperature compensation, particularly models with full parametric compensation, it is necessary to take this into account when performing a calibration.  The physical machine error will be the sum of the error described in the compensation map and any active temperature correction being applied.

Data Collection

xdlaserSelect Calibration Incorporated uses a combination of special hardware and software to collect and apply compensation data.  The technology exists to collect all compensation parameters simultaneously and, when combined with custom software to process the raw measurement data, a CMM axis can be completely re-mapped in as little as one measurement cycle.  The compensation parameters collected from the laser are roll, pitch, yaw, horizontal straightness, vertical straightness, and the scale error.

The preferred method for data collection is by direct measurements which is the method used by Select Calibration Incorporated.  Using equipment such as Laser Trackers to map a machine volume is an example of indirect measurements.  Indirect measurement data collection using a Laser Tracker is effective but it does rely on fitting measurement lines from several points of views in the machine volume and then guessing at the source of the error (Monty Carlo method).  A Laser Tracker should not be used in place of traditional lasers as they are not accurate enough unless the Laser Tracker is used in a overlapping data collection pattern, from multiple points in the machine volume, relying primarily on the distance measured by the laser.

Data collection using traditional lasers is good but is also a slow process.  Using a traditional laser or suitable equipment each measurement parameter must be setup and measured independent of every other measurement parameter.  Collecting the standard six sets of measurement data for an axis will require six measurement setups and six measurement cycles of the machine.  If, for example, the setup for each measurement is 20 minutes and the measurement cycle takes 20 minutes to perform then one axis can be re-mapped in 4 hours (6 parameters, 20 + 20 min for each).  When using a laser that can collect all compensation parameters simultaneously the calibration time is reduce to 40 minutes.  The advantage of using a multi-parameter laser becomes obvious in this context.

Completely re-mapping a CMM axis eliminates the need to perform initial investigative measurements to determine which parameters require updating and which ones can be ignored.  Since all parameters are updated the investigative testing is redundant.  Data collection with traditional lasers is quite laborious and to reduce the amount of time to calibrate the machine investigative measurements are usually done in order to plan the measurement strategy to use.

Unattended or automated data collection is used in all cases where possible.  There are several advantages of automated data capture most notably that the data collection is consistent where every data point is measured in the exact same manor as every other data point.  Automated data collection is preferred in less than ideal environments where many measurements can be performed with no additional effort by the technician performing the tests.

map2mapbd The Map2Map Compensation Processor software is at the heart of the ability to remap a CMM axis in a single measurement cycle.  This software calculates the residual errors from changes in the compensation data while taking into account effects from parametric compensation.   Without this unique software single cycle data collection would not be possible.

The most commonly overlooked problem when capturing measurement data during a CMM calibration is the handling of the data.  For a machine that has only 10 measurement positions per axis the results would be 180 data entry values that must be handled.  Manual data entry is often used to transport data between incompatible software when this problem is not considered or has been ignored.  To eliminate data entry errors and reduce the calibration time a variety of tools have been developed to transport and process measurement data directly from the equipment to the compensation map.  With rare exception most equipment software can provide data by either file or through direct communication with the target device.

Calibration Standards And Performance Testing

Testing is necessary in order to prove the coordinate measuring machine is measuring properly.

Standards exist so that comparison of values is possible between different individuals or organizations.  Without standards, or at least a full understanding of how a particular value was derived, measurement results would be meaningless and not comparable.  It would be the same as stating something like “the measurand from Test A was 0.0034 mm and the measurand from Test B was 0.0058 + 0.0034L mm” but without knowledge of what Test A and Test B are the results have no meaning and could not be reproduced by anyone other than the organization who originally performed those tests.

Select Calibration Incorporated follows the ASME B89.4.10360 or ISO/IEC 10360 family of standards for performance evaluation of coordinate measuring machines.  The 10360 family of standards is titled “Acceptance and Reverification Tests for Coordinate Measuring Machines” and contains the following sub sections:

10360-1 Vocabulary
10360-2 CMMs used for measuring linear dimensions
10360-3 CMMs with the axis of a rotary table as the fourth axis
10360-4 CMMs used in scanning measuring mode
10360-5 CMMs using single and multiple stylus contacting probing systems
10360-6 Estimation of errors in computing Gaussian associated features
10360-7 CMMs equipped with imaging probing systems

The predecessor to the ASME B89.4.10360 standard is ASME B89.4.1 (ball bar).  This is an obsolete standard and has been replaced by ASME B89.4.10360:2008.  It was decided not to offer calibration performance testing following the ball bar standard simply because investing resources necessary to meet the requirements of this standard does not make sense.

The following sections list some of the performances tests from the 10360 family of standards.

ISO/IEC 10360-2:2009 Maximum Permissible Measuring Error E0,MPE

E0 This test is the measurement of 105 lengths performed in seven specified directions with each measurement direction consisting of five lengths measured three times.  Each individual length from the 105 results is compared to the specification (no averages are used).  The probe used to perform this test should have a zero or minimal probe offset perpendicular to the ram axis of the CMM.

Criteria for strict acceptance is that all of the 105 individual lengths must have a deviation from nominal that is less than the manufacturer specification reduced by the expanded uncertainty.

Results are usually displayed graphically where the length deviations can be visually compared to the specification.  The specification is length dependent and forms a funnel shape around the nominal.

ISO/IEC 10360-2:2009 Maximum Permissible Measuring Error E150,MPE

E150 This test is the measurement of 30 lengths performed in either the YZ or ZX measurement planes.  A single direction (not shown in image) can be used if performed by diametrically opposing probes otherwise two directions are required.  Each direction consists of five lengths measured three times.  Each individual length from the 30 results is compared to the specification (no averages are used).  The probe used to perform this test must have an offset of 150 mm perpendicular to the ram axis of the CMM.

Criteria for strict acceptance is that all of the 30 individual lengths must have a deviation from nominal that is less than the manufacturer specification reduced by the expanded uncertainty.

Results are usually displayed graphically where the length deviations can be visually compared to the specification.  The specification is length dependent and forms a funnel shape around the nominal.

ISO/IEC 10360-2:2009 Maximum Permissible Repeatability Limit R0,MPL

This test of repeatability is not a separate test but a derived result from the E0 measurements.  For each of the 35 sets of three repeated length measurements the range of the length results are found.  Criteria for strict acceptance is that all 35 sets of repeatability ranges must be less than the manufacturers specification reduced by the expanded uncertainty.

ISO/IEC 10360-5:2010 Maximum Permissible Probing Error PFTU,MPE

Pftu This test is a measurement consisting of 25 defined points on a precision sphere.  The least squares center of the sphere is found from the 25 points and, for each of the points, the radial distance is calculated from the center of the sphere.  The probing error, PFTU, is the difference between the maximum and minimum radial distance.

Criteria for strict acceptance is that the range from the maximum to minimum radial distance must be less less than the manufacturer specification reduced by the expanded uncertainty.

Note: This test was part of ISO/IEC 10360-2:2001 standard but was removed from the 2009 version and is now part of ISO/IEC 10360-5:2010.

Artifacts That Represent A Calibrated Test Length

Several types of artifacts can be used for performance testing as a measurement of length.  These includes step gauges, multi-ball bar bars, ball plates, and laser interferometry. 

Bidirectional measurements are required in order to represent a calibrated test length as defined by the performance standard.  Unidirectional measurements can be used only if supplemented with a short gauge block bidirectional measurement.  Measurements that are considered unidirectional include laser interferometry and ball plates as an example.