See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281321980

Benchmarking of FDM Machines through Part Quality Using IT Grades

Conference Paper · June 2015

DOI: 10.1016/j.procir.2015.12.075

CITATIONS READS

14 298

3 authors, including:

Paolo Minetola Giovanni Marchiandi

Politecnico di Torino Politecnico di Torino
53 PUBLICATIONS   562 CITATIONS    5 PUBLICATIONS   31 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Production of polymeric components by additive manufacturing: study of integrated kinematic couplings, assessment of wear and life, value analysis and sustainability
View project

CIRAM Italian Interuniversity Center for Additive Manufacturing Research View project

All content following this page was uploaded by Paolo Minetola on 19 February 2016.

The user has requested enhancement of the downloaded file.

 Available online at www.sciencedirect.com

ScienceDirect
Procedia CIRP 41 (2016) 1027 – 1032

48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015

Benchmarking of FDM machines through part quality using IT grades

Paolo Minetolaa,*, Luca Iulianoa, Giovanni Marchiandia
a
Politecnico di Torino, Department of Management and Production Engineering (DIGEP), Torino, Italy

* Corresponding author. Tel.: +39 (011) 090.7210; fax: +39 (011) 090.7299. E-mail address: [email protected]

Abstract

The diffusion of Fused Deposition Modeling (FDM) was recently boosted by the expiration of the FDM patent and the subsequent worldwide
development of low cost FDM machines by a huge number of small companies. In most of the cases, FDM machines are worth what they cost.
Thus the performance of expensive industrial FDM systems is better than that of low cost machines, also known as 3D printers.
In this paper a benchmarking is carried out between a Dimension EliteTM by Stratasys and a 3D TouchTM by Bits from Bytes (BFB). The study
and comparison is based on a reference part that was designed to fit into the building volume of most of low cost FDM machines. The part
includes several classic geometries (planes, cylinders, spheres and cones) of different sizes to cover several ranges of basic sizes as defined by
the ISO 286 standard. Geometric features appear both in the concave and convex shapes to account for all design possibilities. The proposed
reference part allows to consider a higher number of features for each range of basic sizes with respect to other benchmarking models presented
in the literature. Moreover the part does not require support structures for its production, allowing for manufacturing on 3D printers that come
with a unique extruder. Replicas of the reference part are printed out of ABS (acrylonitrile butadiene styrene) material with different layer
thicknesses using the compared machines. After inspecting the replicas by means of a Coordinate Measuring Machine (CMM), the dimensional
accuracy of the compared FDM systems is reported through part quality using IT grades associated with the ISO basic sizes. GD&T values are
also evaluated for some of the geometric features appearing on the reference part.
©©2015
2015The
TheAuthors.
Authors.Published
Published byby Elsevier
Elsevier B.V.
B.V.This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the Scientific Committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
2015.
Peer-review under responsibility of the scientific committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015
Keywords: Benchmarking; 3D printing; Fused Deposition Modelling; IT grades; GD&T.

1. Introduction ranges from 0.10 mm to 0.70 mm. The cost raises when the
3D printer is equipped with multiple extruders, a hot table or a
Since 2008 after the expiration of fused deposition hot working chamber. The machine set-up is often manual in
modelling patents, a great number of low-cost FDM machines case of calibration operations or material change.
have been developed worldwide. These machines are usually On the contrary, industrial FDM systems, that have been
referred as 3D printers, most of which are open source developed since the early 90s by Stratasys, come with a hot
systems whose development was sometimes driven by working chamber and advanced mechanic solutions for
crowdfunding campaigns. improved positioning accuracy and speed. They have at least
3D printers are sold at prices starting from some hundreds two extruder heads, one for depositing the build material and
euros for kits that the user should self-assembly. Most of 3D the other for the support material, while the price starts from
printers come with a three axis architecture, a unique above 15,000 euros. The machine set-up is automatic and the
extrusion head and the building table and working volume are material change is quite easy and fast since the filament is
at room temperature. Two standard sizes are used for the supplied in chipped cartridges.
plastic filament that measures around 1.75 mm or 3 mm in With such a variety of FDM systems and commercial
diameter. Usually the hot nozzle of the extruder can heat the offers, there is the need to quantitatively compare the
plastic filament up to about 280 °C, while the nozzle diameter performances of different machines. The first impression that

2212-8271 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015
doi:10.1016/j.procir.2015.12.075
1028 Paolo Minetola et al. / Procedia CIRP 41 (2016) 1027 – 1032

users get when visiting 3D printing fairs is that FDM reference part are printed out of ABS material using the
machines are worth what they cost from the qualitative point compared machines. After inspecting the replicas by means of
of view of part surface accuracy and finishing. a Coordinate Measuring Machine (CMM), the dimensional
Since the diffusion of first rapid prototyping (RP) systems accuracy of the compared FDM machines is reported in terms
in the late 80s, the lack of an international standard guideline of IT grades associated with the ISO basic sizes of the
for quantitatively assessing the performance and accuracy of reference part’s geometric features and tolerances.
RP machines has driven researchers to adopt benchmarking
procedures. Benchmarking is carried out for comparing the 2. Benchmarking
performance of similar diverse systems (products, processes,
machines, services and organizations). In the case of products 2.1. The compared FDM machines
and machines, the comparison is often based on real
measurements because although technical characteristics are Two FDM machines are available at the Rapid
reported on datasheets, information is not always comparable Manufacturing Laboratory (RMLab) of the Department of
or trustworthy of real performances of the compared systems. Management and Production Engineering of the Politecnico di
As regards the comparison of RP machines, their Torino. The former machine is the Dimension EliteTM (Fig.
performance is evaluated upon the dimensional and geometric 1a) by Stratasys. This industrial FDM system has two
accuracy of manufactured parts. Thus several reference extruders that are fed with the 1.75 mm ABS build filament
artifacts with different geometries were defined and proposed and the soluble support material respectively. The working
in the literature to be used for benchmarking [1-12]. volume is 200 x 200 x 250 mm, the layer thickness can be set
When dealing with the part’s dimensional accuracy and to 0.178 mm or 0.254 mm and the system costs about 20,000
tolerances, it is particular convenient to refer to the ISO euros.
standard IT grades [13, 14]. The IT grades allow for The latter is an entry-level machine 3D TouchTM (Fig. 1b)
comparison of the geometric accuracy of different whose production was stopped last year by 3D Systems to
manufacturing processes as also reported by other studies [15- convert it to the new product line CubeXTM. Disregarding
17]. Nevertheless only a few works [2, 8, 9, 11, 12] use the IT small details, the 3D TouchTM is now available on the market
grades to summarize the results of a benchmarking study on at about 4,500 euros with the name CubeX TrioTM since its
the geometric accuracy of additive manufacturing (AM) or RP extrusion head is composed by three extruders. The extruders
processes or machines. In addition to this, most of the can be fed with 3 mm filaments, for deposition through a 0.5
proposed artifacts presents several different geometries that mm nozzle into a working volume of 185 x 265 x 240 mm,
have similar sizes, thus it is impossible to cover different with a layer thickness of 0.125 mm, 0.25 mm or 0.50 mm.
ranges of ISO basic sizes or the number of features for each
range is low. 2.2. The reference part
With the aim of overcoming those limitations, an
innovative reference part is proposed in this paper for The proposed reference part is designed to fit into the
benchmarking purposes. The proposed reference part includes building volume of most of low cost FDM machines and its
a higher number of features or dimensions for each range of overall dimensions are 110 x 110 x 33 mm. Moreover the part
basic sizes with respect to other benchmarking artifacts that does not require support structures for its production, allowing
have been proposed in the literature. for manufacturing on 3D printers that come with a unique
The following indications, that were defined by Moylan et extruder without soluble support material.
al. [18, 19], are kept into account in the definition of the As concerns dimensional inspection, the presence of
reference part’s geometry: simple classic geometries is imperative, since form errors and
geometrical tolerances are defined on them [20]. Simple
x have a considerable number of small, medium and large classic geometries (planes, cylinders, spheres and cones) are
features; represented in the reference part in both concave and convex
x not consume a large quantity of material;
x have many features of a ‘real’ part;
x have simple geometrical shapes, allowing perfect
definition and easy control of the geometry;
x allow repeatability measurements;
x require no post-treatment or manual intervention (no
support structures);
x not take long to build.

Nevertheless it should be noted that the last aspect

concerning the building time depends on the layer thickness
and on the machine speed, so it is not exclusively affected by
part geometry.
For benchmarking purposes between an industrial FDM
system and an entry-level 3D printer, two replicas of the Fig. 1. The compared FDM machines: (a) Dimension Elite; (b) 3D Touch.
 Paolo Minetola et al. / Procedia CIRP 41 (2016) 1027 – 1032 1029

16 mm in diameter plus two blind holes that have a

diameter of 8 mm and 24 mm. All these features have a
height of 8 mm and the set is inscribed into a 16 mm high
hexagonal prism whose hexagonal base has an edge length
of 16 mm. The second set CC2 has the reverse shape but
same size of the cylinders and holes of set CC1;
x two sets of hemicylinders (HC1 and HC2): unlike the
cylinders of sets CC1 and CC2, the hemicylinders have an
horizontal axis. The first set HC1 consists of four convex
hemicylinders that have a length of 10 mm, while their
diameter decreases in the sequence 24 mm, 16 mm, 8 mm,
and 4 mm. The four concave hemicylinders that compose
the second set HC2 measure 10 mm in length and their
diameter increases from 4 mm to 24 mm in the reverse
sequence but same sizes of the features of set HC1;
x four sets of quarters of spheres (SP1, SP2, SP3 and SP4):
each set is composed of a concave quarter and a convex
quarter of sphere that have the same size. The diameter of
the spheres decreases from set SP1 to set SP4 in the
sequence 4 mm, 8 mm, 16 mm and 24 mm;
Fig. 2. CAD model and geometries of the reference part. x three sets of tilted planes (TP1, TP2 and TP3): the first set
TP1 is fan shaped and consists of 10 couples of planes. The
shapes accounting for different design opportunities (Fig. 2). inclinations of the two tilted planes within each couple are
The change in the size of similar geometries is introduced complementary. The ten couples fan out with an inclination
for the measures to fit into different ISO ranges for the basic to the part base plane increasing from 0 degrees to 45
sizes 1 to 3 mm, 3 to 6 mm, 6 to 10 mm, 10 to 18 mm, 18 to degrees through steps of 5 degrees. The second set TP2 is
30 mm, 30 to 50 mm, 50 to 80 mm, 80 to 120 mm. located on the vertical face of the rectangular block of the
The following features appear on a 5 mm thick square base slots SL. TP2 is used to appreciate changes in the
plate whose thickness is used to prevent deformation of the inclination smaller than 5 degrees, as it is composed of five
part during fabrication and after separation from the building planes whose inclination increases in the sequence 0
platform of the FDM machine: degrees, 2 degrees, 4.5 degrees, 7 degrees, 9 degrees to the
vertical plane. Last set TP3 consists of five planes that are
x a set of seven rectangular blocks (BL): the blocks that have located on the vertical face of the second set of
a 4 mm x 15 mm rectangular base are separated by each hemicylinders HC2. The inclination of these five planes
other and their height increases in the sequence 1 mm, 3 ranges from 70 degrees to 90 degrees to the part base plane
mm, 6 mm, 10 mm, 15 mm, 21 mm and 28 mm; with steps of 5 degrees. TP3 set is needed because in the
x a set of seven rectangular slots (SL): the slots, that have a 5 fan-shaped set TP1 the planes with those inclinations do
mm x 15 mm rectangular base, are included into a 28 mm not have a sufficient extension as to allow for a correct
height rectangular block and their depth changes from 1 measurement of the features.
mm to 21 mm in the same sequence of the previous blocks x several other vertical or horizontal planes, that are parallel
(BL) with a repetition of the 3 mm dimension; or orthogonal to the square base of the reference part.
x a set of seven steps (ST): the steps are connected to form a
stair and their height decreases from 28 mm to 1 mm in the The geometric features have been organised, located and
reverse sequence but same sizes of the blocks (BL); oriented rationally to be representative for the evaluation of
x two couples of coaxial truncated cones (TC1 and TC2): geometrical tolerances during the inspection phase. The
each set consists of an outer convex truncated cone and an staircase effect, that is typical of layer-by-layer manufacturing
inner concave truncated cone. The outer truncated cone of processes, can also be evaluated on different surfaces having
set TC1 has a major diameter of 20 mm, a minor diameter diverse curvature.
of 13.6 mm and it is 12 mm high. The coaxial inner In summary, more than eighty geometric features appear
truncated cone has an height of 10 mm, a major diameter on the reference part. They allow for the measurement of over
of 10 mm and a minor diameter of 6.5 mm. The second set one thousand sizes and distances over the considered ISO
TC2 has a smaller size than the first set. The inner ranges of basic sizes from 1 mm up to 120 mm, if one
truncated cone of the second set has a major diameter of 5 considers all measurable entities and surfaces. It should be
mm, a minor diameter of 4 mm and it is 6 mm high. The also remarked that a unique configuration of the CMM probe
coaxial outer truncated cone of set TC2 has an height of 9 with the axis orthogonal to the part base plane is sufficient to
mm, a major diameter of 10 mm and a minor diameter of easily carry out the part inspection because enough space
6.8 mm. between adjacent features is provided for the approach and
x two sets of coaxial cylinders (CC1 and CC2): the first set retraction of the probe using a tip diameter of 2 mm.
CC1 is composed of two cylinders that measure 4 mm and
1030 Paolo Minetola et al. / Procedia CIRP 41 (2016) 1027 – 1032

3. Results of dimensional inspection

The results of the measurements of the geometric features

of the replicas are used to evaluate the dimensional accuracy
of the compared machines in terms of ISO IT grades [13]. The
values of standard tolerances corresponding to IT 5 – IT 18
grades for nominal sizes up to 500 mm are evaluated through
the standard tolerance factor i that is expressed in micrometres
by the following formula:

i 0.45˜ 3 D  0.001˜ D (1)

Fig. 3. Replicas fabricated by Dimension EliteTM (a) and 3D TouchTM (b).

where D is the geometric mean of the range of nominal

2.3. Fabrication and inspection sizes in millimetres (Table 1):

A white filament was used in the Dimension EliteTM D D1 ˜ D2 (2)

machine and the part was fabricated in approximately 7 hours
adopting a layer thickness of 0.254 mm (Fig. 3a). About 120 The IT grades are classified according to Table 2 by the
cm3 of ABS material were required to build the part, number n of times that the tolerance factor i fits into the
corresponding to around 1/10 of a brand new material dimensional deviation. For example the grade IT12
cartridge. corresponds to a minimum of 160i with n = 160.
In order to distinguish the two replicas, a black filament For a generic nominal dimension Djn, the number nj of
was instead used in the 3D TouchTM printer (Fig. 3b). The tolerance units is computed as follows:
fabrication of the reference part took almost 18 hours with a
layer thickness of 0.125 mm. About 80 gr of ABS material out 1000 D jn  D jm
of a brand new spool of 1 kg were used for part fabrication. nj (3)
i
Therefore for both machines, the amount of material required
is acceptable. In both cases, the part was manufactured in the
centre of the building platform, with the sets of blocks (BL), where Djm is the corresponding measured dimension.
slots (SL) and steps (ST) aligned to the machine Y axis.
Replicas were not finished or polished, not to alter their For each feature’s dimension or distance between features,
surfaces or dimensions, since the as-manufactured condition the nj value is computed and attributed to the range of ISO
should be considered for a correct benchmarking of the FDM basic sizes to which the dimension or distance belongs. A
machines. certain distribution of numbers of units is so obtained for each
After manufacturing, the replicas are inspected by means ISO range for the two replicas. Within each range the n value
of a DEA CMM model GLOBAL Image 07.07.07 that has a corresponding to the 95th percentile of the distribution is
declared volumetric length measuring uncertainty MPEE assumed as the maximum dimensional error of the FDM
according to ISO-10360/2 [21] of 1.5 + L/333 μm, where machine to assess the accuracy through the IT grades
MPE is the acronym for Maximum Permissible Error and L is consistently with previous studies in the literature [2, 15-17].
the measured length. Results are summarized by the bar chart of Fig. 4.
Three replications were made for the inspection of each Despite the smaller layer thickness used for part fabrication
replica and then average values are considered in the analysis in the low cost 3D TouchTM machine, the dimensional
of the results. At least ten inspection points were measured on accuracy of the industrial system Dimension EliteTM is better
each geometric feature and each replication took almost one for most ranges of the ISO basic sizes. The difference is
hour. significative for smaller basic sizes up to 30 mm, whereas for
bigger sizes the accuracy of the compared machines is similar.

Table 1. Ranges of ISO basic sizes and corresponding tolerance factor i.

Range Basic sizes
Above D1 (mm) 1 3 6 10 18 30 50 80
Up to and including D2 (mm) 3 6 10 18 30 50 80 120
Standard tolerance factor i (Pm) 0.542 0.733 0.898 1.083 1.307 1.561 1.856 2.173

Table 2. Classification of IT grades according to ISO 286-1:1988.

Basic size Standard tolerance grades
Above Up to IT 5 IT 6 IT 7 IT 8 IT 9 IT 10 IT 11 IT 12 IT 13 IT 14 IT 15 IT 16 IT 17 IT 18
1 mm 500 mm 7i 10i 16i 25i 40i 64i 100i 160i 250i 400i 640i 1000i 1600i 2500i
 Paolo Minetola et al. / Procedia CIRP 41 (2016) 1027 – 1032 1031

Fig. 4. Dimensional accuracy (95th percentile) of the compared FDM machines in terms of IT grades for different ranges of ISO basic sizes.

Fig. 5. Comparison of GD&T values for different geometric tolerances of reference part’s features.
 1032 Paolo Minetola et al. / Procedia CIRP 41 (2016) 1027 – 1032

The IT grade for Dimension EliteTM ranges from IT 13 for References

smaller sizes to IT 11 for bigger sizes and it is always better
or at least equal to the one of the 3D TouchTM printer for the [1] Childs THC, Juster NP. Linear and Geometric Accuracies from Layer
same range of ISO basic sizes. The IT grade of 3D TouchTM Manufacturing. CIRP Ann Manuf Technol 1994;43(1):163-166.
[2] Ippolito R, Iuliano L, Gatto A. Benchmarking of Rapid Prototyping
worsens from IT 11 to IT 15 as the basic size decreases. Techniques in Terms of Dimensional Accuracy and Surface Finish.
As concerns GD&T values (Fig. 5), the flatness is CIRP Ann Manuf Technol 1995;44(1):157-160.
evaluated separately for horizontal planes, vertical planes and [3] Xu F, Wong YS, Loh HT. Toward generic models for comparative
tilted planes to investigate the dependence on the staircase evaluation and process selection in rapid prototyping and
effect. Parallelism between couples of parallel planes and manufacturing. J Manuf Syst 2000;19(5):283-296.
[4] Mahesh M, Wong Y, Fuh JYH, Loh HT. Benchmarking for
coaxiality of coaxial cones and cylinders are also considered. comparative evaluation of RP systems and processes. Rapid
The GD&T values confirm the results of the IT grades in Prototyping J 2004;10(2):123-135.
terms of better accuracy of the industrial FDM system.The [5] Dimitrov D, van Wijck W, Schreve K, de Beer N. Investigating the
variation for the considered geometric tolerances is also much achievable accuracy of three dimensional printing. Rapid Prototyping J
wider for the 3D TouchTM printer when compared to the one 2006;12(1):42-52.
[6] Scaravetti D, Dubois P, Duchamp R. Qualification of rapid prototyping
of the Dimension EliteTM. As it could be expected, the flatness tools: proposition of a procedure and a test part. Int J Adv Manuf Tech
of horizontal planes is better than that of other planes because 2008;38(7-8):683-690.
of the absence of the staircase effect. In fact, horizontal planes [7] Brajlih T, Valentan B, Balic J, Drstvensek I. Speed and accuracy
are fabricated within the same layer of material, while the evaluation of additive manufacturing machines. Rapid Prototyping J
presence of different layers affects the worse quality of 2011;17(1):64-75.
[8] Garg HK, Singh R. Pattern development for manufacturing
vertical and tilted planes. For the Dimension EliteTM machine, applications with fused deposition modelling-a case study. Int. J.
parallelism tolerance is also good and has a small variation Automot. Mech. Eng. 2013;7(1):981-992.
like the flatness of horizontal planes. Coaxiality of coaxial [9] Singh R, Singh JP. Comparison of rapid casting solutions for lead and
features is the worst tolerance for both machines. brass alloys using three-dimensional printing. P I Mech Eng C-J Mec
2009;223(9):2117-2123.
[10] Johnson WM, Rowell M, Deason B, Eubanks M. Comparative
4. Conclusions evaluation of an open-source FDM system. Rapid Prototyping J
2014;20(3):205-214.
The benchmarking activity carried out through the [11] Singh R, Singh G. Investigations for statistically controlled investment
proposed innovative reference part highlights the greater casting solution of FDM-based ABS replicas. Rapid Prototyping J
geometric accuracy of the Dimension EliteTM machine when 2014;20(3):215-220.
[12] Cruz Sanchez FA, Boudaoud H, Muller L, Camargo M. Towards a
compared to the 3D TouchTM printer. The difference in the standard experimental protocol for open source additive
performance of Dimension EliteTM can be ascribed to manufacturing: This paper proposes a benchmarking model for
advanced mechanic solutions, but also to the smaller diameter evaluating accuracy performance of 3D printers. Virtual and Physical
of ABS filament and extrusion nozzle which ensure a higher Prototyping 2014;9(3):151-167.
accuracy and detail for features of minor size. The hot [13] ISO 286-1:1988 - ISO system of limits and fits - Part 1: Basis of
tolerances, deviations and fit. 1988: International Organization for
chamber of that industrial FDM system also plays its role in Standardization (ISO).
reducing the shrinkage of deposited material layers that [14] ISO 286-1:2010 - Geometrical product specifications (GPS) - ISO code
cannot cool down to room temperature, thus preventing part’s system for tolerances on linear sizes - Part 1: Basis of tolerances,
deformation during fabrication. deviations and fits. 2010: International Organization for
As regards GD&T values, the median and interquartile Standardization (ISO).
[15] Bassoli E, Atzeni E. Direct metal rapid casting: mechanical
range is very similar for both machines in the case of flatness optimization and tolerance calculation. Rapid Prototyping J
tolerance of vertical planes. On these features, the smaller 2009;15(4):238-243.
layer thickness adopted for the 3D TouchTM printer helps [16] Bassoli E, Gatto A, Iuliano L, Violante MG. 3D printing technique
compensating the poorer geometric accuracy. applied to rapid casting. Rapid Prototyping J 2007;13(3):148-155.
Over the first benchmarking results presented in this paper, [17] Violante MG, Iuliano L, Minetola P. Design and production of fixtures
for free-form components using selective laser sintering. Rapid
the reference part will be used for comparing other FDM Prototyping J 2007;13(1):30-37.
machines by fabrication of new replicas. The machine speed [18] Moylan S, Slotwinski J, Cooke A, Jurrens K, Donmez MA. Proposal
or the position and the orientation of the part into the machine for a standardized test artifact for additive manufacturing machines and
working volume could be changed, not only for benchmarking processes. in 23rd Annual International Solid Freeform Fabrication
purposes but also for optimizing the machine performances. Symposium - An Additive Manufacturing Conference, SFF 2012.
2012. Austin, TX, 902-920.
The proposed methodology has a general applicability and [19] Moylan S, Slotwinski J, Cooke A, Jurrens K, Donmez MA. An
could also be extended to other AM systems that are used for additive manufacturing test artifact. J Res Natl Inst Stan 2014;119:429-
production of metal parts. 459.
[20] ISO 1101:2012 - Geometrical product specifications (GPS) -
Acknowledgements Geometrical tolerancing - Tolerances of form, orientation, location and
run-out. 2012: International Organization for Standardization (ISO).
[21] ISO 10360-2:2009 - Geometrical product specifications (GPS) -
The authors gratefully acknowledge Mr. Kai Yue for Acceptance and reverification tests for coordinate measuring machines
having carried out the CMM measurements of the replicas. (CMM) - Part 2: CMMs used for measuring linear dimensions. 2009:
International Organization for Standardization (ISO).

View publication stats