October 2020, Vol. 20, No.

4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

DOI: 10.21062/mft.2020.074 © 2020 Manufacturing Technology. All rights reserved. http://www.journalmt.com

Influence of CMM Scanning Speed and Inspected Feature Size on an Accuracy of

Size and Form Measurement

Jan Urban1, Libor Beranek1, Michal Koptiš1, Jan Šimota1, Ondřej Košťák2
1Faculty of Mechanical Engineering, Czech Technical University in Prague. Technicka 4, 160 00 Prague 6. Czech

Republic. E-mail: Jan.Urban@fs.cvut.cz, Libor.Beranek@fs.cvut.cz, Michal.Koptis@fs.cvut.cz,

Jan.Simota@fs.cvut.cz
2Skoda Auto a.s., Trida Vaclava Klementa 869, 293 01 Mlada Boleslav, Czech Republic. E-mail: On-

drej.Kostak@skoda-auto.cz

Modern production systems requires high effectivity and flexibility with always increasing demands for
precision as an imperative for more efficient components. The same apply for quality inspection provid-
ing data for feedback regulation of production processes. CMMs (coordinate measuring machines) which
are flexible and universal in use yet very accurate and easy to automate are a standard mean for quality
inspection. With many sensors available on the market, central fixed scanning heads with tactile scanning
probes are a reference equipment for inspection of precise production of engine and transmission com-
ponents. Tactile probes are right choice where very high accuracies and stability of results is required.
Effectivity was allways a target in production processes and today the same pressure for effectivity and
productivity is required from measuring machines, yet measurement strategies are often taken from
measurement plans even 10 years old. This means that in old programs low scan speeds are used based
on capability of older technologies and the approach of don´t change it when it works is common. This
limits productivity of the whole quality control process. Motivation for this paper and whole research is
to increase productivity and thus capacity of quliaty inspection without compromising process capability.
Lack of measuring capacity is usually solved by purchase of a new machine which may not be allways
necessary. Primary motivation of companies supplying these technologies is not maximum efficiency of
quality inspection, which in context of spare capacities ultimately means lower sales. Aim of this article
is to describe influence of scanning speed and size of inspected feature on CMMs accuracy. High-preci-
sion CMM control itself is not easy because with decreasing path radius dynamic effects of machine
construction itself increase on measurement results. Accuracy of CMM measurement is then function of
feature size being checked. This knowledge can be used for optimization of measurement plans in terms
of productivity while maintaining sufficient measurement accuracy depending on required tolerance.

Keywords: CMM, Productivity of measurement, Scanning speed, capability of measurement

correct implementation is important. This is investiga-

Introduction ted via coefficients cg and cgk (gauge capability) which
describes level of accuracy and precision of measuring
As a result of fundamental development from re-
system. Standardly actual measurement programs are
cent years it is no problem to produce very complica-
designed to fulfil coefficients values, however time
ted parts with high precision where we need tolerance
required for measurement is for some companies mar-
in hundredths of millimetre. Highly specialized ma-
ginally. From previously published experiments is
chines are then able to produce even below this limit.
known that scanning speed affects productivity and
However, we must be able to control/measure this ac-
measurement capability. In practice, values of these
curate production productively and correctly. CMMs
coefficients exceeding even more than ten times ne-
are appropriate choice for this purpose and in recent
cessary limit of gage capability (1.33). This situation
years they have been used a lot. Main reason for their
then provides potential to increase speed and thus re-
deployment is to exclude human factor from dimensi-
duce time required for measurement. Modern times
onal quality control process which is significant source
require high demands on productivity in production
of measurement variability. Other reasons for using
and productivity of CMMs must correspond accor-
CMM are versatility, accuracy and productivity. Ma-
dingly. In some publications is describe influence of
chine will rapidly increase productivity compared to
stylus system configuration on variability of measure-
conventional devices (like callipers, etc.) but input co-
ment result [1], description of dynamic effects CMM
sts of this machine are in thousands euros. Of course,
machine on sensor system [2], or first indications

538 indexed on: http://www.scopus.com

October 2020, Vol. 20, No. 4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

about scanning speed effect for straightness measure- cg and cgk have been introduced. It is mathematical-
ments [3]. However, most of measured elements are statistical tool (type 1 gage study from MSA) assesses
circular on engineering products. In overall context, effects of bias and repeatability on measurements
there are currently no systematic efforts to make from one operator and one reference part. Coefficient
CMM measurement process more productive. Moti- cg describes ratio between precision and tolerance (po-
vation for this article was also long-term cooperation tential capability) while cgk describes ratio between ac-
with Škoda Auto a.s., where pressure on maximum curacy and tolerance (actual capability). Precision (re-
possible useable of available measurement capacities peatability) is expressed as a standard deviation found
is high. At beginning of this cooperation was study by simply taking standard deviation of all repeated
which describe theoretical potential to increase capa- measurements. Calculation is then performed accord-
city of dimensional quality control about 10% by sim- ing to relations (formula 1 and formula 2). Process is
ply reprogram measurement strategy on CMMs while capable if both coefficients are greater than 1.33.
maintaining capability. !."∗$
• = (1)
%∗ •
Measurement strategy
$.%∗'()*̅ (*̅ , )
Actual most common measuring systems mounted !"# = (2)
-∗•
on CMMs are tactile systems. Across industrial appli- Where:
cations, they can be found in versions with fixed or T...process tolerance [mm]
rotating heads (mean mounted on stationary CMMs). Sg…standard deviation of measurements (precision)
Thanks to their flexibility a rotating heads are used for [mm]
measuring components with complex shapes (free- /̅" …mean of all measurements [mm]
forms) where a lot tactile styluses would be required. /̅0 …reference value [mm]
Fixed systems are used in cases where high accuracy is
required and this publication is focused on parts with
Feasibility study
high requirements. The industry habit for controlling
precision manufacturing applications is uses fixed sys- In cooperation with Škoda Auto a.s. and based on
tems. Measurement strategy using scanning systems above findings a feasibility study of intent to optimize
represents scan setting parameters such as path type CMM measurement program was conducted and
(which points are extracted from part surface), scan- main focus was only in scanning speed changes. Opti-
ning speed, distance between each point (number of mization was carried out on CMM Zeiss UMC 850
points), parameters for filtering measured points, Carat which has quality control department for pro-
eliminating outliers and method of numerical calculat- duction of aggregate components. It´s machine with
ing geometrical elements. Strategy as a complex is de- active Vast Gold scanning system and specifications
fined by machine operator and is absolutely important according to VDI / VDE 2617 u3 = 1.8 + L / 450
from point of ensuring measurement accuracy. Some µm. Measurement program was optimized for dimen-
parameters affect measurement productivity. Mainly sional inspection of cylinder heads 1.0 MPI engine af-
length of scanned path and scanning speed. Scan path ter manufacturing operations 10 and 20. This part
length adjustment leads to reduce number of points shows Figure 1. Optimization process was divided
on individual elements and thus also to reduce infor- into several steps. First step was analyse strategy set-
mation about elements as a whole. For this reason ar- tings of both programs. Subsequently ten repetitions
ticle focuses attention to scanning speed because it has measurement of same part was measured for provide
major impact on productivity of measurement while information about accuracy and precision of all con-
whole information remains unchanged. Other ways to trolled characteristics. These properties represent the
increase measurement productivity can be optimize coefficients cg and cgk as described above. In order to
measurement configuration, for example optimize determine reference values required for cgk coefficient
safety positions, non-measurement paths around part. calculation, measurement of same part was performed
Accelerate stylus system configuration replacement on ZEISS UPMC 850 calibration machine which has
and reduce computing times with more powerful specification 0.7 + L / 600 µm. Measurement on this
computer equipment. However, these methods do not machine was five times repeated and arithmetic mean
affect the value of the measurement process capability of results was used as a reference value. Measurement
and therefore they will not be in our interest. plan optimization was performed only for geometrical
elements whose value of coefficients Cg and Cgk ex-
Gage capability Cg and Cgk ceeded value 5. Adjustment affected only scan speed
parameters. Values of optimized speeds was chosen
Whole dimensional inspection process needs to be only on operator's experience because this is untested
evaluated in some way to see if it is suitable for specific methodology of program optimization. Results of re-
application. For this reason gage capability coefficients

indexed on: http://www.scopus.com 539

October 2020, Vol. 20, No. 4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

peated measurements confirmed original assump- at Škoda Auto a.s. on Zeiss Prismo machine. This ma-
tions. On average optimized programs saved 10% chine is equipped with active Vast Gold scanning sys-
time in measurement process. This was impetus for tem and is located in climate controlled metrology
further experiments that would extent which measure- centre with passive damping of ambient vibrations.
ment results were affected by this intervention. [4] Laboratory temperature was 21.2 °C at the time of ex-
periment and maximum deviation was ± 0.15 °C, tem-
perature compensation was used on measurement re-
sults with CTE 11.5*10-6K-1 for a steel calibration
rings. Sensor with 1.5 mm sphere diameter and 20 mm
shaft length was used for scanning. Entire configura-
tion was assembled from clamping plate for VAST
head and lightweight steel extension (80x11 mm).
Configuration of a sensor was designed for real meas-
urement conditions. Results from individual rings cal-
ibration were used for cgk coefficient calculation as a
reference value. Tolerances used to evaluate capability
of process were on the level of ±0.01mm for size and
0.02mm for roundness in both cases the width of to-
lerance was 0.02mm. Calibration was performed on
Zeiss UPMC 850 reference machine and Hommel
Fig. 1 Cylinder head EA 211 1.0 MPi calibration rings gauge. [5]
Experiment description and realization Experiment evaluation
The study showed that it is necessary to describe Evaluation of experiment revealed that scanning
effect of scanning system on overall measurement re- speed value used in combination with different radius
sults using higher scanning speeds. On this basis, full of measured circular path quite fundamentally affects
factorial experiment was designed to define maximum accuracy and precision of measurement. We will
usable scanning speed ensuring capability of measure- demonstrate this effect on 4 diameter values which oc-
ments depending on circular path diameter. Two pa- cur at engine components production and at same
rameters were chosen as variables. First parameter is time represent whole dimensions range of experiment.
diameter (factor A), this parameter was ensure by us- Selected values were: 5 mm diameter corresponding to
ing adjusting rings with range from 5 mm to 100 mm. mounting pins diameter, 20 mm diameter correspond-
Step between individual diameters was 5 mm. Second ing to valve seat on cylinder head dimension, 50 mm
parameter was scanning speed (factor B) which corre- diameter corresponding to crankshaft dimension and
sponded to range 2 - 40 mm/s with 2 mm/s step. 75 mm diameter corresponding to internal combus-
Each of these two factors contained 20 levels and each tion engine block dimension. Graph in Figure 2 shows
combination of these factors had 10 replications. Re- deviations obtained by subtracting reference value
sponse quantities were average deviatin from refer- from arithmetic value of ten measurements for each
ence size and roundness. Experiment was performed scan speed level.

Fig. 2 Influence of scanning speed on diameter deviation from reference value

540 indexed on: http://www.scopus.com

October 2020, Vol. 20, No. 4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

Here we can see increasing deviation of individual path radius. With larger measured path radius the ma-
rings in direct connection with increasing scanning chine control reacts more smoothly due to normal
speed. This increase is more pronounced for smaller vector direction changes. Normal vector defines scan
diameters. As diameter increases, this effect gradually points direction and thus also pressing force in indi-
changes. The roundness deviations are calculated in vidual axes. For larger diameters is smoother change
same process. The deviations are shown in Figure 3. of pressing force direction than smaller diameters.
Same trend is confirmed here as well. We can also no- With increasing scan speed is more difficult to control
tice that up to 6 mm/min speed the deviations are ap- movement of machine. Another influence that will be
proximately same in both cases. Above this speed, in- reflected in process is movement inertia of CMM con-
creasing difference in deviations from reference value struction, which increases with increasing speed.
can then be observed. This is due to different scanned

Fig. 3 Influence of scanning speed on form deviation from reference value

D – 20 mm D – 50 mm D – 75 mm
v – 10 mm/s
v – 20 mm/s
v – 30 mm/s

Fig. 4 Measured data graphical report for specified speeds and diameters of adjusting rings

Evaluation of roundness is displayed in Figure 4. mm). Comparison clearly presents how important is
On this figure we can see scanning path deformation correct measurement strategy setting for accurate re-
with increasing scanning speed (10, 20 and 30 mm/s) sults and how dynamics of machine influence meas-
and also dependence on path diameter (20, 50 and 75 urement result. This is illustrated by step changes

indexed on: http://www.scopus.com 541

October 2020, Vol. 20, No. 4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

around circumference. Then we can see in Figure 4 is still good. Conversely, situation is different for co-
changes of measured points in X and Y axes. These efficient Cgk which evaluates accuracy of the measure-
are axes where (in case of CMM with moving bridge) ment. Here it can be observed that individual diame-
inertia of CMM construction is most pronounced. ters have different limit of scanning speeds when
Results of measured dimensions also confirm cal- measurement process is still capable. This indicates
culated values of gage capability Cg and Cgk for diam- maximum scan speeds for particular measured diame-
eter evaluation. It can be seen in Figure 5. For better ter. An interesting difference is between maximum
illustration, the values of the coefficients Cg and Cgk speed for 20 mm diameter which can be measured
were approximated by a polynomial of the fourth de- correctly and accurately at speeds up to 20 mm/s and
gree. Cg coefficient is satisfactory for all selected path 75 mm diameter where scanning speed can be used up
scanning speeds. Reason is that coefficient evaluates to 32 mm/s. So it is confirmation about effect of com-
precision of measurement. Cg refers to ratio between bination scanning speed and diameter of measured el-
precision and tolerance. Due to fact that this is meas- ement.
urement without human factor influence, repeatability

Fig. 5 Comparison of calculated capability of coefficients Cg and Cgk

Fig. 6 Calculated standard deviation Sg for each evaluated calibration ring: on left - measurement of diameter on right measurement
of roundness

Figure 5 also shows large differences in coeffi- possible to involve standart deviations from 10 repli-
cients Cg and Cgk calculated for different scan speed cations for each measurement in figures 2 and 3 be-
levels. This is due to dependence on standard devia- cause of their size, compared to measured values error
tion Sg which is used for calculation. In Figure 6 we bars were not visible. Standart Differences between
can see standard deviation values for each scan speed them are very small but they still have significant effect
level when measuring 20 mm diameter ring. It was not on value of calculated coefficients. This proves how

542 indexed on: http://www.scopus.com

October 2020, Vol. 20, No. 4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

strict a capability coefficient evaluate measurement viation from calibration still on acceptable level for se-
process. rial measurement. Since all calibration rings were me-
It should also be noted that surface of adjusting asured on the same machine and negative values at low
ring is lapped and is almost ideal for scanning on speeds occurred in all measured rings compared to
CMM with active scanning system. Therefore, it is ad- their calibrated values, this is probably some systema-
visable to take scan speed setting limits as maximum. tic cause, which can be caused by a worse condition of
In real application, surface quality will be worse and the aerostatic bearing surface in one of the machine
therefore it is necessary to determine safety coefficient drives. The trend was subsequently confirmed by the
which will guarantee applicability of this methodology calculations of the coefficients Cg and Cgk, where at
even on qualitatively worse surface. Here we would lower scanning speeds higher standard deviations of
lean towards correction factor 0.66 which corresponds repeated measurements were calculated and subseque-
to 2/3 of maximum speed and should safely ensure ntly also lower values of the coefficients of the capa-
applicability of this knowledge to optimize measure- bility of gauge. However, it should be emphasized
ment plans. Although introduction of this factor will that, despite the lower values of capability, the coeffi-
reduce potential benefit of optimized programs but it cients were still above 1.33 and the measurement was
still increases productivity. therefore eligible for selected tolerance width of
0.02mm.
Conclusion The Figure 5 show the values of the coefficients
Cg evaluating the precision and Cgk evaluating the
The experiment clearly demonstrated a combina- precision and accuracy of the measurement together,
tion of the effect of the diameter of the measured cir- which we generally refer to as the measurement ac-
cular path and the scanning speed on the measure- curacy. In particular, it can be seen from the values of
ment result accuracy. The graphs in Figures 2 and 3 the coefficient Cgk that they are quite high for certain
show the calculated deviations from the average of the values of scanning speed (normally exceeding 20
ten measurements for each scanning speed level. The mm/s). This is due to a very small standard deviation,
increase in these deviations when using increasing spe- which is in the order of hundreds of µm and for selec-
eds varies according to the measured diameter. The ted diameters it’s displayed in Figure 6. After excee-
effect is noticeable both when evaluating the diameter ding a certain speed limit, however, there is a fairly
and roundness and is caused by the controller and dy- steep drop below the limit capability value of 1.33. The
namical influences of the machine. When measuring speed at which the capability value falls below the limit
smaller diameters, it is increasingly difficult to control of 1.33 can then be referred to as the limit scanning
the whole measuring machine with increasing speed. speed of a specific diameter, where sufficient accuracy
The reason is the rapidly changing values of the axial of the measurement represented by the coefficient
components of the normal vector, which is crucial for Cgk is ensured. Any knowledge of these speed limits
the control of the sensor pressure. For small diame- is important from the point of view of productive use
ters, the change is more pronounced with increasing of the CMM machine. This is because it provides the
speed than for larger diameters. Therefore, the control potential to increase the scanning speed of commonly
of the machine at higher speeds does not manage to compiled measurement plans in order to increase the
regulate the components of the measuring force in the productivity of dimensional inspection. As the scan-
individual axes. In addition, the dynamic effects of ning speed increases or measured diameter decreases,
CMM machine design are affecting measured data, the trueness of measurement, represented by the
which is clearly shown in Figure 4. The problem of the average deviation from roundness reference value and
influence of sensor head control at higher scanning average deviation from reference size, decreases as
speeds of smaller diameters in combination with dy- well. Unfortunately, precision described by standard
namic effects of machine design is clearly demonstra- deviation is increasing as displayed in Figure 6, this is
ted. The direct influence of the diameter of the mea- common for CMM machines yet very unintuitive and
sured path on the recording of the measured points is thus for accuracy evaluation precision is the least rele-
evident. vant and trueness has to be considered.
Furthermore, it can be seen in graphs 2 and 3 that If we keep the capability values above the limit of
at lower speeds the calculated deviation of the average 1.33 with some slight safe margin for example use as
value from the reference dimension is negative and a limit value of 3, we have a room for productivity op-
only from a certain scanning speed does the deviation timization and still have some safe space for potential
reach positive values. This may be due to the machine errors. However, it should be noted that the level of
used, on which serial measurement of the manufactu- the coefficients Cg and Cgk depends on the width of
red parts is normally carried out in continuous opera- the tolerance field. If the width of the tolerance field
tion, and the machine may already show a certain de- is larger than 0.02mm used in this study, the potential
for optimization will be even more significant.

indexed on: http://www.scopus.com 543

October 2020, Vol. 20, No. 4 MANUFACTURING TECHNOLOGY ISSN 1213–2489

Acknowledgement DAAAM International Vienna, 2017, 2017,

0400-0409. DAAAM Proceedings. DOI:
At this point, we would like to thank Škoda Auto 10.2507/28th.daaam.proceedings.056. ISBN
a.s. which provided technological equipment for 9783902734112. Dostupné také z:
experiments and helped to create this contribu- http://www.daaam.info/Downlo-
tion. ads/Pdfs/proceedings/procee-
dings_2017/056.pdf
References
PEREIRA, P.H. a R.J. HOCKEN. (2007). Cha-
MIKEŠ, P. (2016). Influence of stylus system racterization and compensation of dynamic er-
configuration on the variability of measurement rors of a scanning coordinate measuring ma-
result on CMM. Manufacturing Technology Journal. chine. Precision Engineering. 2007, 31(1), 22-32.
2016, (Volume 16), 184 - 188. ISSN 12132489. DOI: 10.1016/j.precisioneng.2006.01.006.
KUMAR, A. S., C. CHANDRASEKARAN. ISSN 01416359. Dostupné také z: https://lin-
(2019). Analysis of Dynamic Probing Errors in kinghub.elsevier.com/re-
Measuring Machines. Innovative Design, Analysis trieve/pii/S014163590600047X
and Development Practices in Aerospace and Automo- GUNIŠ, Z., VAGOVSKÝ, J., GÖRÖG, A.
tive Engineering (I-DAD 2018). Singapore: Sprin- (2013). Effect of Scanning Speed on the Ac-
ger Singapore, 2019, 2019-12-15, 473-482. curacy of Measu-Red Values using Coordinate
Lecture Notes in Mechanical Engineering. Measuring Machine. Technological Engineering.
DOI: 10.1007/978-981-13-2697-4_51. ISBN 2013, 10(2), 2-5. DOI: 10.2478/teen-2013-
978-981-13-2696-7. Dostupné také z: 0011. ISSN 1336-5967. Dostupné také z:
http://link.springer.com/10.1007/978-981- https://con-
13-2697-4_51 tent.sciendo.com/doi/10.2478/teen-2013-
STĘPIEŃ, K. (2015). An analysis of influence 0011
of sampling strategy and scanning speed on es- FACTORS AFFECTING MEASUREMENT
timation of straightness and flatness deviations UNCERTAINTY IN INDUSTRIAL CMM
with CMMs. Advanced Technologies in Mechanics. WORK. Statistica Applicata - Italian Journal of Ap-
2015, 2(2(3), 2-8. DOI: plied Statistics. 2008, 3(22), 341-352. Dostupné
10.17814/atim.2015.2(3).17. ISSN 2392-0327. také z: https://pdfs.semanticscho-
Dostupné také z: http://atim.media.pl/in- lar.org/7318/2b950ba0df37075f11bcc39e1760
dex.php/ATiM/article/view/17 640fa2b5.pdf
Statistické metody řízení kvality. 1. Praha: Česká DIAN, M., HODINÁŘ, L. (2018). The GRR a
společnost pro jakost, 2007. Fundamental Tool for Dealing with Measure-
URBAN, J., DVOŘÁK, R., BERÁNEK, L., ment System Variability. Manufacturing Techno-
ŠIMOTA, J., MIKEŠ, P., TILLINGER, K. logy. 2018, 18(1), 29-34. ISSN 12132489. Do-
(2017). Optimierung der Qualitätskontrolle bei Koor- stupné z: doi:10.21062/ujep/48.2018/a/1213-
dinatenmessgeräten hinsichtlich der Messproduktivität: 2489/MT/18/1/29
Studentská tvůrčí činnost 2017. V Praze: Fakulta VIT, J., NOVAK, M. (2018).. A Roundness
strojní ČVUT, 2017, 2017. ISBN 978-80-01- Machine Measuring Probe Calibration. Manu-
06143-5. facturing Technology. 2018, 18(6), 1053-1059.
MELICHAR, M., KUBATOVA, D., ISSN 12132489. Dostupné z:
KUTLWASER, J. (2017). Influence of CMM doi:10.21062/ujep/223.2018/a/1213-
Velocity on Scanning Precision. Proceedings of the 2489/MT/18/6/1053
28th International DAAAM Symposium 2017.

544 indexed on: http://www.scopus.com