Sie sind auf Seite 1von 8

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO.

6, DECEMBER 2007

2767

New Method for Testing the Dynamic Performance


of CMM Scanning Probes
Adam Wozniak

AbstractIn this paper, a new method that applies a piezoelectric translator to test the dynamic accuracy of scanning probes
for coordinate measuring machines (CMMs) is proposed. The
principle of the method is presented, and this method is used for
the evaluation of measurement uncertainty. The validity of the
method is experimentally confirmed on a bridge Zeiss PRISMO
CMM equipped with a VAST XT scanning probe.
Index TermsCoordinate measuring machines (CMMs),
dimensional metrology, probe calibration, scanning probe (head).

I. I NTRODUCTION

HE CONSTANT progress in device and machine element


manufacturing, along with the necessity to increase the
speed of dimensional and shape error checks, has caused a
continuous increase in the use of coordinate measuring machines (CMMs). These measuring instruments are used in both
laboratories and manufacturing plants. The advantages of these
modern machines are measurement automation, graphic visualization of the results, numerical data archiving in electronic
media, and capability for integration with computer-aided
design/computer-aided manufacturing systems. Therefore, although they represent a substantial cost, these machines are
utilized more and more frequently where speed and precision
of measurements are required.
A. Probing System
One of the fundamental elements that determine the precision
of a CMM is the probe, which locates points on the surface of a
measured part located within the machines measurement volume. New-generation measurement scanning probes are now
being used more frequently. These probes locate sequences of
measured points without being moved away from the tested
surface. This scanning approach now offers new and effective
possibilities for taking measurements. It allows checking not
only dimensions but also positioning and shape. After only
Manuscript received March 9, 2007; revised August 21, 2007. This work
was supported by the Foreign Postdoc Fellowships of the Foundation for Polish
Science for a one-year visit to the Virtual Manufacturing Research Laboratory, Mechanical Engineering Department, cole Polytechnique de Montral,
Montreal, QC, Canada. This ongoing collaboration is made possible through
the HOMING Program of the Foundation as support for the reintegration of
researchers after their stay abroad and the fostering of cooperation with their
former host institutions.
The author is with the Institute of Metrology and Measuring Systems,
Faculty of Mechatronics, Warsaw University of Technology, 02-525 Warsaw,
Poland (e-mail: wozniaka@mchtr.pw.edu.pl).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIM.2007.908154

Fig. 1. CMM scanning probe. (a) Schematic. (b) View of the Zeiss VAST XT
probe. (c) View of the SIP OMNI SIP probe.

one installation of the probe, all the possible dimensions and


tolerances can be determined in a single checking operation,
thus considerably reducing the time required for total part measurement. There is no need for the removal and reinstallation
of the measuring device. This is particularly important for
manufacturing plants that are generally not equipped with onetask measuring instruments (e.g., devices for shape error testing). Regardless of their high purchase price, scanning probes
are used more frequently and will soon be commonplace in
industrial and laboratory applications. As these numbers grow,
it will become more necessary to verify their accuracy.
A scanning probe is in fact a small coordinate measuring
device. It usually consists of three Cartesian length-measuring
systems that are parallel to the axes of the CMM. The scanning
probe schematic is shown in Fig. 1(a). The probes length
transducers measure the probe tip displacement along the x-,
y-, and z-axes. The scanning probe output W p and the CMM
length measurement results W t are added in all the measuring
axes [1][3]. Thus, the probe does not have to be in zero position during probing. If an analog signal of the scanning probe
system is digitized and added to the length measurement value
of the CMM, the result should be the same as that obtained
with the probe in the zero position. There are a few types of
probe measuring transducers. However, the most popular are
inductive and optoelectronic incremental transducers. Fig. 1(b)
shows a scanning probe that has been specifically designed
for high-speed scanning in VAST technology. This type of
probe (made by Zeiss, Germany) is fitted with an inductive
transducer. Fig. 1(c) shows another example, i.e., an OMNI
SIP probe manufactured by SIP Switzerland. It is fitted with
optoelectronic incremental transducers in each of the three axes.
In the active systems of these new-generation probes, the
magnitude and direction of the probing force are controlled by

0018-9456/$25.00 2007 IEEE

2768

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 6, DECEMBER 2007

electronic springs consisting of solenoids in annular magnets.


The probing force remains largely constant, and it is always
applied perpendicular to the surface to be measured. The scanning probes evaluate measuring points on a work piece surface
through a process of continuous contact of the probe tip with
the measured surface.
B. Well-Known Methods of Checking the Accuracy of
Scanning Probes
The precisions of the CMM and probing unit are closely
interrelated [1][3]. The majority of the test methods consider
a CMM as an integral system, i.e., the geometrical errors
of machine axes and the probing errors are not separately
considered. This is not always the right approach. The main
objective of machine users is to obtain a reliable evaluation
of the integral measuring system. However, in cases of gross
machine functional errors (particularly random ones), information pertaining to the correct operation of its subassemblies,
including the probing unit, are crucial. In spite of this, a
method of testing the accuracy of the probes has not been
determined so far, and this particularly applies to scanning
probes.
Users of CMMs who face the problem of testing the probes
accuracy use intermediate evaluation methods, e.g., checking
simple master artifacts on the machine, which are usually
certified spheres or rings. These tests are recommended by the
International Organization for Standardization (ISO 10360) [4].
In accordance with these recommendations, the scanning probe
error T ij is calculated as an interspace of radial distances from
the sphere center of all the measured points on the master
sphere calculated according to the Gauss method. Scanning
probe parameters are considered as verified as long as the value
of T ij (the calculated error) is not superior to the maximum
permissible scanning probing error M P ET ij. The results of
this check only reflect errors of the probe, assuming that the
shape errors of the masters and the errors of CMM scales and
guides in the machine axes are negligible. In practice, it is
relatively easy to be in possession of a master with shape errors
that are very small, but on the other hand, the errors of the CMM
are difficult to eliminate. The probes characteristics are mostly
erroneous because the machine errors, although they are small,
are comparable in magnitude with the probe errors. Therefore,
despite the fact that the standards [4][6] recommend checking
the machine contact system according to the described method,
the obtained characteristics are to be considered only as a check
to determine whether the coordinate machine together with the
contact system operates correctly.
The second existing method of checking the scanning probe
characteristics involves comparing the readings of the probe
transducer in one axis with a laser interferometer (or another transducer). The precision of the latter is one order of
magnitude higher than the checked probe. Manufacturers of
CMMs use these methods to determine the correction table for
the probes nonlinearity characteristics. However, this method
only provides for the acquisition of the static characteristics
of the probes, which can be considerably different from their
dynamic characteristics. This severely limits the applicability

of this method for obtaining information of scanning probes on


coordinate measurement machines.
Existing methods of the dynamic performance calibration
of CMM scanning probes reflect the errors of the coordinate
machine itself [7]. Information about the errors of the scanning
probe is difficult to evaluate. Further, published investigations
on the performance evaluation of dynamic probes concern the
transducers applied in devices for shape error testing [8], [9].
C. Proposed New Approach for the Calibration of
Scanning Probes
A scanning probe test that is carried out using a CMM to
measure a master artifact can provide satisfactory results only
when performed on highly repeatable machines. In addition,
this kind of calibration yields information about probe inaccuracy, which is limited to the selected shapes. Not enough
information is gained to enable form measurements of any
surface.
Special test setups are required to gain a more precise evaluation of probe linearity and repeatability. These are necessary
to enable the separation of probe inaccuracy and CMM lengthmeasuring systems. The following are the main advantages.
The probe is a part of the exchangeable instrumentation on
the machine. Probe tests performed by the proposed setup
permit an independent study of the probes behavior. This
information is particularly important for certain groups,
such as probe manufacturers and scientists.
Precise knowledge of the probe systematic errors should
allow numerical correction of the most important components of those errors during measurement on CMM.
Modern measuring machines equipped with advanced
computer systems are able to compensate for the inaccuracy of the probes online.
It is therefore necessary to study methods for the precise
determination of not only the static errors of the probes but also
the dynamic errors, which have so far been unknown.
II. N EW M ETHOD OF S CANNING P ROBE T ESTING
A. Objectives of the New Method
The objectives for the development of a new method for
testing the accuracy of scanning probes used on CMMs were
defined as follows.
There must be the capability for probe testing in an arbitrary measurement axis (x, y, z).
There must be the capability for testing the inaccuracy of
the probe while it is installed on the CMM. (The testing
method must include participation of the machine itself
and more precisely participation of its computer. This can
be done because the machine software provides features
that allow nonlinear displacements of probe transducers.)
There must be the capability for generating both static and
dynamic characteristics of the scanning probe under test.
The uncertainty of the method must be at least five times
smaller than the errors of the tested probes.


WOZNIAK:
NEW METHOD FOR TESTING THE DYNAMIC PERFORMANCE OF CMM SCANNING PROBES

Fig. 2.

New device for scanning probe testing. (a) Schematic. (b) View.

The measurement range must correspond to the probes


scanning range.
It must be easily adaptable to automation and
computerization.
There must be a short duration for measurements.
The method must be flexible (i.e., the method should be
able to check the accuracy of the various probes used on
different CMMs).
The location of the testing stand must be flexible (it must
be possible to check the accuracy of probes operating on
remote test stands or in remote manufacturing plants).
B. Principles of the New Method
The new concept is based on double scanning measurements
on a flat reference surface. The two scans follow the same measuring path. The first scanning check is made on a stationary
master surface. During the second check, the reference surface
affects a controlled movement, which causes perpendicular
deflections of the test probe in relation to the measured surface,
i.e., in the direction of the test probe axis. The difference of the
results of these measurements, i.e., one without and the other
with surface displacement compared to the master reference,
allows the determination of the characteristics of probe errors.
C. Implementation of the Method
The scanning probe testing device is shown in Fig. 2(a).
To actuate a displacement of the reference surface, a Physic
Instrument piezoelectric translator of 90-m measuring range

2769

was used. The other device elements are a piezo translator


controller, a signal generator, a holder to mechanically fix
and position the element on the coordinate machine table, and
a computer. An optic glass flat plate has been used as the
reference surface. The shape errors of the plate do not exceed
a value of /3 (approximately 0.2 m). These existing shape
errors of the reference surface have no effect on the method
accuracy because they are compensated for due to the procedure
of subtracting the two scanning results carried out on the same
measuring path. The signal generator has been incorporated
and set up to emit a repeatable signal that controls the displacement of the reference surface via the piezo translator.
The piezo translator controller is a high-precision amplifier that
adapts the signal from the generator to the corresponding level
of the translators output signal. The piezo translator that actuates the reference surface must be rigidly fixed to the measuring
machine table. Height adjustment is not necessary because
the machine itself may displace the probe together with the
machine head into any position within the machine measuring
range. On the other hand, angular adjustment is necessary to
position the reference surface in accordance with the direction
of the scanning probe. A special device that can make this
angular adjustment of the measuring axis was, therefore, built.
The computer has been incorporated into the test stand
to perform the following tasks: analysis, visualization, and
activation of the measurement results. The characteristics of
the displacement of the reference surface as affected by the
piezo translator are compared with the probe response in
the memory of the computer. The computer can also control
the measurements. A view of the prototype scanning probe
testing device built according to the described concept can be
seen in Fig. 2(b).
The reference surface deflection has to be carefully selected
to ensure that it does not exceed the measuring range of
the tested scanning probe and to avoid changing the machine
head displacement (together with probe) without displacing the
reference surface. Based on the results of tests on a Pismo
coordinate machine equipped with a Zeiss VAST XT head,
along with published Renishaw data (another manufacturer of
probes), we can state that the effective range of the scanning
probe does not usually exceed a value of 100 m. The
active portion of the probe range depends on the surface being
scanned. Any remaining probe tip deflection is nonactive or
idle, although necessary for safety reasons. In fact, during the
scanning measurements, only a small portion of the probe range
is used, i.e., a few tenths of a micrometer on either side of
the probes neutral position. The inaccuracy of probe deflection
within this measurement subrange determines the accuracy of
measurements affected on the CMM.
D. Calibration of the Device
The precision of the proposed method for testing the scanning probe accuracy depends on the characteristics of the
reference surface and the repeatability of the described device.
A Zeiss ZLM 500 laser interferometer was used to calibrate the
system made up of the piezo translator, controller, and signal
generator. During the tests, a vibration-measuring configuration

2770

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 6, DECEMBER 2007

Fig. 3. Special setup for the calibration of the device using a Zeiss laser
interferometer.

of the interferometer was used. The sampling frequency was


equal to f = 25 kHz. At a selected sampling time of 0.6 s, this
resulted in 16 932 measurement points on the reference surface.
The measurement uncertainty using the ZLM 500 interferometer was 5 nm [10]. Tests of stability over time of the ZLM 500
interferometer were conducted [11]. Additionally, the effect of
measurement time on the average reading and repeatability was
examined. Statistical tests confirmed that the average value does
not depend on the measurement time, providing the latter with
less than 10 s. In the case of a measurement time less than than
1 s, the uncertainty of reading evaluated in 2 s does not exceed
10 nm.
Using the configuration shown in Fig. 3, the reference
characteristics of the setup for scanning probe testing were
determined using a ZLM 500 laser interferometer. At the same
time, the uncertainty of these reference characteristics can be
evaluated. It does not exceed 2 s = 0.3 m at a 95% confidence
level.
E. Procedure for Testing the Scanning Probe and Analysis of
Measurement Results
The scanning probe is tested orthogonal to the reference
surface and the scanning direction. Therefore, the reference
surface must be positioned parallel to the probe axis direction
at a high level of accuracy. This is to ensure that the piezo
translator axis movement coincides with the tested direction of
the scanning probe (Fig. 4). The CMM has control options that
enable this positioning.
The measurement of the probe characteristics can be determined in two stages. The first stage involves scanning a motionless reference surface (orthogonal scanning measurement of
the surface carried out over a length of a few millimeters). This
scanning result reflects the straightness of the surface including
the superimposed errors of the coordinate machine axes. The
second stage is a repetition of the previous scanning at the
same parameters but with the displacements of the reference
surface orthogonal to the scanning direction. The second series
of measurements causes perpendicular movements of the tip
of the tested probe. The difference in measurement results,
carried out with and without reference surface displacement,

compared to the master displacement separately obtained by the


interferometer, allows the determination of the characteristics
of probe errors.
The use of a double scanning check, i.e., with and without
reference surface displacement, and a successive subtraction of
the recorded results enables the elimination of shape errors of
the measured surface. More importantly, however, systematic
errors of the CMM guides and scales are eliminated. The
repeated scanning check allows the quantification of the
characteristics of the tested probe in relation to tip deflection
and allows the determination of the repeatability of these
characteristics.
The length of the scanning measurement path should be
selected to achieve a maximum number of measurement points.
Of course this must be done while considering the scanning
speed and the size of each scanning step as set on the CMM.
In the case of the VAST XT probe and PRISMO machine, at
a constant scanning step equal to 1 m, the relation between
the number of measured points and the measurement speed is
shown in Table I. Additional tests revealed that the number of
measurement points under identical conditions is not constant
and can differ by some percentage. In the tests run so far, a
scanning distance of 8 mm was chosen, which is symmetrically
located in relation to the reference surface axis and the piezo
translator.
Three series of recorded characteristics are sent to the
computer. Two of these are straightness scanning tests
affected without reference surface displacement (static
characteristics). The third is the master displacement characteristic (reference characteristic). The method for measuring
this latter series is described in Section II-D. An additional set
of characteristics is made using a master displacement of the
scanning probe actuated by the translator (test characteristics), as created in the CMM computer. The test characteristics
include information on the probe operating errors superimposed
on the controlled displacement of the reference surface. The test
characteristic also describes the straightness of the reference
surface and errors of the coordinate machine itself. During
the scanning operation, the machine moves the head together
with the probe along the measurement axis. Any imperfection
of the guides of the coordinate machine (i.e., straightness
and perpendicularity errors) and its measuring scales may be
a further source of the errors that appear in the probe test
characteristics.
F. Parameters Proposed for the Evaluation of Scanning
Probe Accuracy
Dispersion of the values of the resulting characteristics can
be a natural measure of test probe errors. The M AE statistical parameter (i.e., average absolute error [14], [15]) may be
defined to describe this dispersion as
1
|ei |
n i=1
n

M AE =

(1)

where ei is the deviation of observation i from the reference


characteristics. SSE is a similar parameter and is defined in


WOZNIAK:
NEW METHOD FOR TESTING THE DYNAMIC PERFORMANCE OF CMM SCANNING PROBES

Fig. 4.

2771

Location of scanning and probe testing directions during the probe accuracy test along the (a) x-axis, (b) y-axis, and (c) z-axis.

TABLE I
EFFECT OF SCANNING SPEED ON THE NUMBER OF MEASUREMENT
POINTS. PRISMO MACHINE WITH VAST XT PROBE

statistical analysis [14], [15] as the sum of distance squares of


observation i in relation to reference characteristics, i.e.,
SSE =

n


e2i .

(2)

i=1

The maximum error is another parameter that describes the


accuracy of the scanning probe and is defined as the maximum
distance of reference characteristics from the probe response
(resulting characteristics), i.e.,
emax = max |ei |.

(3)

The proposed parameters are calculated by the research


software used for the analysis of probe accuracy test results.

G. Evaluation of Method Uncertainty


The uncertainty of evaluation of the actual characteristics
of the scanning probe errors is obtained by superimposing
component errors during the dynamic operation of the CMM.
The principal source of errors is the repeatability of reference
surface displacement. It is also essential to include the uncertainty of the calibrating method. Some additional errors may
result from the repeatability of static characteristics related to
the measurement systems of the machine. The total uncertainty
of the method and test stand can be written as

 3

u =  (ui )2

(4)

i=1

where ui is the repeatability of reference characteristics, u2 is


the error of reference characteristics determination, and u3 is
the repeatability of static characteristics.

We assume in the proposed method that the errors related to


the surface straightness and measurement systems of the machine are mostly repeatable. Therefore, they can be eliminated
by subtracting the static characteristics from the test characteristics. To evaluate the repeatability of static characteristics,
a scanning measurement of reference surface straightness was
performed five times on a PRISMO CMM equipped with a
VAST XT scanning probe. In the tested case, the maximum
standard deviation was equal to 0.15 m, and the average was
0.06 m. It is to be noted at this point that the static characteristics include errors that should be considered along with those
of the test probe mentioned above. However, due to the small
shape errors of the tested surface, the probe operates in a range
close to zero. Comparatively speaking, these probe errors are
very small and are close to the probe reading repeatability [13].
To minimize the range of the scanning probe for static
characteristic measurements, a more precise surface must be
used, and a method of precise location of the reference surface
parallel to the scanning direction must be developed. In an earlier research [12], [13], repeatability tests of the scanning probe
readings used on CMMs have been performed. These were done
using a displacement transducer with a reticular interferometer.
Chi-square and ShapiroWilks statistical tests have confirmed a
quantitative conformity of the results distribution with a normal
distribution. The probability level of the Chi-square statistical
test was 0.404941, whereas that of the ShapiroWilks test
was 0.988301. One can conclude at a confidence level of at
least 95% that the test results follow a normal distribution.
Therefore, the limit error at a 95% confidence level is equal to
2 s = 0.16 m.
The repeatability of static characteristics is related to the
repeatability of the tested scanning probe and to the systematic
shape errors of the scanned reference surface, the straightness
and perpendicularity errors of machine guides, and the reading
errors of the machine scales. Repeatability tests of the scanning
probe readings reveal that the instability of static characteristics
and the repeatability of probe readings are approximately the
same (having similar values of dispersion measured by standard
deviation). To eliminate the systematic errors not related to the
tested scanning probe, the static characteristics must be subtracted from the test characteristics. Consequently, a resulting
characteristic is acquired, and it is the response of the tested
probe, which is actuated by the piezoelectric translator, to its
reference deflection.
The analysis of the reaction of the CMM scanning probe
during dynamic operation is done by comparing the resulting

2772

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 6, DECEMBER 2007

Fig. 5. VAST XT scanning probe test results in the (a) x-axis, (b) z-axis, and (c) y-axis, and (d)(f) additional repetition in the y-axis.

characteristics with the reference characteristics. A special


software has been developed to calculate the characteristics of
the tested scanning probe errors. The calculation is carried out
based on loaded files containing data recorded during scanning
measurements.
Research on the static characteristics carried out on a ZLM
500 interferometer reveals that u1 = 0.3 m at a confidence
level of 95%. Based on the tests presented in [11], the error
of the reference characteristics measured using a ZLM 500
interferometer does not exceed u2 = 0.01 m at a confidence
level of 95%. The repeatability of the static characteristics of
the VAST XT tested probe does not exceed u3 = 0.16 m, also
at a confidence level of 95%.
The total quncertainty of the presented test stand for the
dynamic testing of CMM scanning probes was obtained by calculating the geometrical sum and does not exceed u = 0.35 m
at a 95% confidence level.
III. R ESULTS OF S CANNING P ROBE T ESTS
To demonstrate the feasibility of the proposed new method
for testing the CMM scanning probe dynamic performance,
experiments were carried out on a bridge Zeiss PRISMO CMM

(Germany) equipped with a VAST XT scanning probe. As per


standard ISO 10360 [17], the machine maximum permissible
error of indication for size measurements is MPEE = 1.7 +
L/300 m, where L is in meters and represents the measured
length. The maximum permissible scanning probe error is
MPET ij = 4.1 m. A Master probe tip set was used for the
test, i.e., a stylus of 42-mm length and a 2-mm-diameter probe
stylus tip. The measuring speed and the sampling step size were
set to 8 mm/s and 13 m, respectively.
The effective measuring radius of the measuring tip was
4.0002 mm. A set of measurements was each performed for
the x-, y-, and z-axes. Fig. 5(a)(c) shows the results of these
tests in the x-, z-, and y-axes, respectively. The dashed line and
triangular points represent the pushing direction of the probe
tip movement from its neutral position. The continuous line
and square points represent the return direction of the probe
tip movement. Additionally, the manufacturers specification of
the probe MPET ij error is indicated on these graphs using two
horizontal parallel straight lines.
To check the repeatability of the method and the VAST XT
probe error characteristics, a y-axis test was repeatedly performed. Fig. 5(d)(f) shows the results of these repeated tests in
the y-axis. As can be seen, the characteristics obtained by the


WOZNIAK:
NEW METHOD FOR TESTING THE DYNAMIC PERFORMANCE OF CMM SCANNING PROBES

2773

TABLE II
RESULTS OF KRUSKALWALLIS TEST

same y-axis look similar. The errors are of the order of several
micrometers appearing at maximum deflection of the probe
tip. Existing differences of the characteristics are caused by
accidental probe errors. Hysteresis is revealed as well. During
the y-axis testing of the VAST probe, its value reaches several
micrometers. A major value of hysteresis appears in the z-axis
[shown in Fig. 5(b)].

IV. E XPERIMENTAL V ERIFICATION OF THE M ETHOD


To experimentally verify the proposed method, the Zeiss
VAST XT probe was tested using a special master called a
microroller master artifact. The microroller is made of a small
diameter wire and small shape errors glued to the reference flat
surface (flatness error less than 0.3 m). The method concept
is to select the wire diameter so that the tip displacement
in the direction perpendicular to the master surface does not
cause any reaction in the servo systems of the machine. The
test results show that the roller diameter should be in the
range of 10100 m; however, the flat surface of the assembly
has to be accurately aligned to avoid CMM movement in the
static axes. A 49.7-m diameter wire was used for the roller
master. The uncertainty of its diameter measurement (together
with the roundness error of the microroller) does not exceed
2 s = 0.2 m at a 95% confidence level.
Based on the test results of the master scan, the inaccuracy
of the probe can be determined when measuring the highest
point of the master. This corresponds to a measurement of the
diameter of the wire used in the master. The difference between
the size of the master and a dimension measured by the probe
provides a value equal to the probe performance error.
Measurements of the master artifact were performed three
times at the same configuration of the probe tip and at three
measuring speeds, including 1, 8, and 16 mm/s.
To verify whether there are significant differences between
the two proposed methods of testing the scanning probe accuracy, statistical analyses were performed on the test results.
The original assumptions of the variance analysis were not met,
and for this reason, the results of the nonparametric testing of
KruskalWallis were used [14], [16].
The KruskalWallis test evaluates the null hypothesis that
the medians of error are the same for each of the two
methods. Since the P-value = 0.226117 is greater than 0.05
(see Table II), there is no statistically significant difference
between the medians at a 95% confidence level. This confirms
the validity of the test results obtained following the developed
method for measurement of a master microroller artifact.

V. S UMMARY AND C ONCLUSION


The analysis of the results of tests performed to measure the
dynamic accuracy of CMM scanning probes using the new testing method described above allows the following conclusions to
be drawn.
The developed device can be used for the dynamic testing
of scanning probe accuracy. The results of probe tests
obtained using the test stand have been confirmed by measurements on a new master microroller artifact. Statistical
analysis of the measured data has confirmed that at a 95%
confidence level, the difference between the accuracy of
the two methods is insignificant. The total uncertainty
of the developed method and test stand does not exceed
0.35 m at a confidence level of 95%.
Initial tests performed using the Zeiss VAST XT probe
have shown that the dynamic operation errors range from
approximately 2 m up to 20 m. An increase of probe
errors has been noticed as the readings increase. In addition, hysteresis has been observed in all probe axes.
The aim of this paper was not to present the performance
of all types of scanning probes but rather a new method
and setup for probe testing during the scanning mode. This
type of tool can be useful for operators of CMM machines,
for service engineers, or for manufacturers of CMMs. To
gain more knowledge of the metrological properties of
the probes used in CMMs and to optimize the proposed
method for probe testing, further research on the effect of
various factors on probe operating accuracy is necessary.

R EFERENCES
[1] J. A. Bosch, Coordinate Measuring Machines and Systems. New York:
Marcel Dekker, 1995.
[2] C. Butler, An investigation into the performance of probes on coordinate
measuring machines, Ind. Metrol., vol. 2, no. 1, pp. 5970, 1991.
[3] E. Ratajczyk, Coordinate Measuring Technique. Warsaw, Poland:
WPW, 1994 (in Polish).
[4] ISO 10360-4, Geometrical Product Specications (GPS)Acceptance
and Reverication Tests for Coordinate Measuring Machines
(CMM)Part 4: CMMs Used in Scanning Measuring Mode, 2000,
Geneva, Switzerland.
[5] VDI/VDE 2617, Accuracy of Coordinate Measuring Machines Part 3,
1989, Dusseldorf, Germany.
[6] ANSI/ASME B89.1.12M, Methods for Performance Evaluation of Coordinate Measuring Machines, 1990, New York: Amer. Soc. Mech. Eng.
[7] W. G. Weekers and P. H. J. Schellekens, Compensation for dynamic
errors of coordinate measuring machines, Measurement, vol. 20, no. 3,
pp. 197209, Mar. 1997.
[8] R. Liang, O. Jusko, F. Ldicke, and M. Neugebauer, A novel piezo
vibration platform for probe dynamic performance calibration, Meas.
Sci. Technol., vol. 12, no. 9, pp. 15091514, 2001.

2774

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 6, DECEMBER 2007

[9] A. Bendeli, J. Duruz, and E. G. Thwaite, A surface simulator for the precise calibration of surface roughness measuring equipment, Metrologia,
vol. 10, no. 4, pp. 137143, Oct. 1974.
[10] C. Zeiss, ZLM 500 Technical Information, 1998. Carl Zeiss GmbH
Esslingen.
[11] O. Iwasinska and M. Dobosz, Badanie wplywu czynnikw zewnetrznych
na stabilnosci pracy interferometru ZLM 500, in Proc. V SzkolaKonferencja Metrologia wspomagana komputerowo, Warszawa, Poland,
2001, vol. 2, pp. 195200 (in Polish).
[12] A. Wozniak, M. Dobosz, and E. Ratajczyk, Testing the inaccuracy of
CMM scanning probe, in Proc. ISMQCSurface Metrology for Quality
Assurance, Cairo, Egypt, 2001, pp. 5763.
[13] A. Wozniak and M. Dobosz, Methods of testing of static inaccuracy of
the CMM scanning probe, Metrol. Meas. Syst., vol. 10, no. 2, pp. 191
203, 2003.
[14] M. Dobosz, Wspomagana komputerowo statystyczna analiza wynikw
badan. Warszawa, Poland: Akademicka Oficyna Wydawnicza EXIT,
2004 (in Polish).
[15] N. R. Draper and H. Smith, Applied Regression Analysis. New York:
Wiley, 1998.

[16] S. B. Vardeman, Statistics for Engineering Problem Solving. Boston,


MA: PWS, 1994.
[17] C. Zeiss, Prismo Sets the Standard Technical Information, 2002. Carl
Zeiss GmbH Esslingen.

Adam Wozniak received the M.Sc. and Ph.D. degrees in mechanical engineering from the Warsaw
University of Technology (WUT), Warsaw, Poland,
in 1998 and 2002, respectively.
From 2005 to 2006, he was a Visiting Professor
with the Department of Mechanical Engineering,
cole Polytechnique de Montral, Montral, QC,
Canada. He is currently an Assistant Professor with
the Institute of Metrology and Measuring Systems,
WUT, conducting teaching and research. He has
published more than 50 papers. His current research
interests include dimensional metrology, especially coordinate measuring
techniques.

Das könnte Ihnen auch gefallen