Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 46 (2016) 525 – 528

7thh HPC 2016

6 – CIRP Coonference on
o High Perrformance C
Cutting

An
A ARM--based Multi-ch
M annel CN
NC Soluution forr Multi-taasking Turning
T and
a
Millling Macchines
Sergeej N. Grigoorieva, Georgi M. Marrtinova*
a
Mosccow State Techno
ological Universitty "STANKIN", Vadkovsky
V per. 3a
a, Moscow 1270555, Russia

* Corresponding autthor. Tel.: +7-4999-972-9440; fax: +7-499-973-3885

+ 5. E-mail addresss: martinov@ncsy
ystems.ru

Abstract

ARRM-based CNC C systems are used u as a low--cost solution ffor controlling 2-3 axis mach hines. The incrrease of compu uting power off ARM
pro
ocessors facilitaated their application to contrrolling multi-taasking and multti-axis machinees with severall independent control c channells. The
papper presents thee results of the exploratory in nvestigation of the transition from f the PC-baased to the AR RM-based CNC C solution. The cross-
plattform architectture of CNC syystems, the porting of CNC kernel softwarre to the singlee-board computters Raspberry Pi 2 with the Linux
opeerating system aand the control of servo drivess and PLC I/O oover EtherCAT fieldbus with cycle c time of 2 mms are investig
gated as well.
Thee dual-channel configuration of o ARM-based CNC designedd to control the multi-tasking turning and millling machine with w inclined layyout is
presented. The pap aper illustrates the
t kinematic schemes and the network archittecture of NAK KLON 535 macchining center for f ARM-basedd CNC
soluution.
© 2016
2016
2 The TheAuthors.
Authoors. Published
Published b Elsevier
by
by Elsevier B.V
B.V This V.is an open access article under the CC BY-NC-ND license
Peeer-review underr responsibility of the Internatiional Scientific Committee of 7th HPC 2016 in the person off the Conferencce Chair Prof.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review
Ma under responsibility of the International Scientific Committee of 7th HPC 2016 in the person of the Conference Chair
atthias Putz.
Prof. Matthias Putz
Keyywords: computeer numerical contrrol (CNC); machine; performancee; ARM processorr; EtherCAT;

1. Introduction
I The trend toowards increaasing the com mputing powwer of
ARRM processorrs facilitated their applicaation to contrrolling
ARM (Advaanced RISC Machine) prrocessors are used muulti-tasking and multi-aaxis machines with seeveral
tod
day to create stand-alone PLC controllers or speciialized ind
dependent control channels (Table 1).
conntrol panels annd remote conntrol panels foor manual hanndling
in the work areea that are coonnected to th he CNC systeem as Tab
ble 1. Single-board computers baseed on ARM proceessors with Linuxx
operating system, wh
hich were used foor running the NC
C kernel.
termminal clientss [1, 2]. A number of recently pubblished
ressearch papers in the area off embedded CNCC devices are of Nam
me CPU
U RAM Ethernet Price
parrticular intereest. Such systtems are used d as low-costt two- Cubbieboard 3 ARRM Cortex-A7 dua
ual-core 2GB 1xRJ45 1G $89
dim
mensional CN NC for laser or
o waterjet cu utting [3] and other (cub
bietruck) 1 GHz
G DDR3 Ethernet
unccomplicated machine toool [4] and ro obotic controol [5] Rasspberry Pi 2 AR
RM v7 quad-core C
Cortex- 1GB 1x RJ45 $35
sysstems. A7 900 MHz SDRAM
M 10/100
Unfortunatelyy, a serious research on the applicability of the Cub
bieboard 4 A800, ARM big.LITT TLE 2GB 1xRJ45 1G $125
em
mbedded CNC C devices forr controlling multi-channeel and (CC
C-A80) octa
a-core (quad-coree DDR3 Ethernet
Corrtex-A15 2.0 GHzz, quad-
muulti axis machine tools is sttill to be develloped. The annalysis
core
e Cortex-A7 1.3 GGHz)
of single board ccomputers bassed on ARM-processors shhows a
steady increasee in computting power by b increasingg the
proocessor speed,, the number of
o cores and memory
m space [6]. An ARM based CNC systeem should imp plement the modern
m
con
ntrol technologies for high performance cutting, e.g. spline

2212-8271 © 2016 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 International Scientific Committee of 7th HPC 2016 in the person of the Conference Chair
Prof. Matthias Putz
doi:10.1016/j.procir.2016.04.036
526 Sergej N. Grigoriev and Georgi M. Martinov / Procedia CIRP 46 (2016) 525 – 528

interp
polation (Akiima spline, cubicc spline, NURBS), loook- Thhe working mode
m of multiiple control channels
c is seet in
aheadd > 150 blocks, claamping corrrections, 5-aaxis the NC
N machine parameters,
p w
where the num mber of availaable
transformation, toool correction, path smoothin ng functions, ssafe contrrol channels is defined aand axes forr each particuular
operaation of the m
machine, etc. chann nel are fixed.
Thhe paper prresents the results of the explorattory Thhe synchronizzation betweeen control channels is carrried
invesstigation of the transition frrom the PC-baased to the AR
RM- out with
w the help of o high-level llanguage constructs (Tablee 2),
basedd CNC solutiion in particuular for the machining
m cennter impleemented by th he NC kernel.
NAK KLON 535. Thhe NAKLON 535 machininng center with h a limited seet of
auxiliary equipmeent, includingg a mechanissm for hydraaulic
2. Teechnical requ
uirements and
d the existing
g CNC solutioon chuck k clamping system, locck/unlock tu urrets, etc. wasw
for machining
m cen
nter NAKLOON 535 with in nclined layou
ut demo onstrated at the exhibition ““Metaloobrabo otka 2015”.
Thhe network architecture
a oof the CNC system (Fig.. 2)
Thhe effectivenness of turnning and milling centerss is includes: the NC kernel baseed on an ind dustrial PC with
w
achieeved by equippping them witth high-speed milling spinddles, operaating system RTR Linux andd NC terminall integrated innto a
which h allows maachine parts with w complex x geometry iin a Soft PLC (consissting of the operator pan nel on the .N NET
single setting. Thhis not only increases accuracy, but aalso platfo
orm, a stand dard and a specialized machine conntrol
reducces the amounnt of required equipment an nd time necesssary panells); buscoupleers and expannsion I/O forr PLC controlled
to perform the auxxiliary transitioons, including g the reductionn of devicces; controllers of seervo drives,, spindle and
the number
n of ppart setups, and a the prep paration timee of countterspindle [8]. Control andd exchange of data in real-ttime
techn
nological equiipment and cutting
c tools, which ultimaately netwo ork devices iss performed vvia the EtherC CAT protocoll. A
increases the utiliization rate of
o equipment and releases the large number of PLC-controlledd devices connected to the NC
produuction area [7]. system required th he use of seveeral bus coupplers that allowwed
Thhe kinematic scheme of thhe machine to ool requires tw
two- he proportional distributionn of the electrrical load am
for th mong
chann nel control (Fig. 1). The firrst control chaannel includess the separrate devices.
turrett 2 (axis Z2, X2), and thee spindle unitt S1. The seccond
contrrol channel is assigned to the t turret 1 (aaxis X1, Z1), the
longiitudinal axis Z and the spinddle unit S2.

Fig.
F 1. The kinem
matic scheme of NAKLON
N 535 maachining center.

Table 2. The high-levell language constrructs for synchron

nization between
ol channels.
contro
Functtion Description
chann
nel_load() Loading a part program
p on a con
ntrol channel
without runningg
chann
nel_run() Loading and runnning a part progrram on a control
channel
chann
nel_start() Starting/continuuing a part program on a control
channel
chann
nel_stop() Stopping a part program
p on a con
ntrol channel
chann
nel_reset() Resetting the coontrol channel
chann
nel_wait() Waiting for the end of the part prrogram on a contrrol
channel
sleep() Suspending the running of a part program for the Fig
g. 2. CNC networrk architecture off NAKLON 535 machining
m centerr.
specified numbeer of millisecondss
 Sergej N. Grigoriev and Georgi M. Martinov / Procedia CIRP 46 (2016) 525 – 528 527

Structurally, the transition from PC-bassed solutions to the of up to 2 contro

ol channels, nno more than 8 interpolatedd axes
ARRM-based soluution is shown in dashed liines and is reeduced (including 2 spinndles), as welll as restrictio
ons on the sizze and
to switching EthherCAT and Ethernet
E cable from the PC to the thee number of elements of thee PLC program m.
Raspberry Pi 2.
4. Conducting
C the
t tests
3. Research
R of tthe cross-plattform architeecture and th
he
porrtability of N
NC kernel on a single boarrd ARM comp puter Bench tests have
h been perrformed using g the configuuration
parrameters of the real machinee tool and its PLC program m.
The portabilitty of the NC kernel software is conceivved in The stand is equipped
e withh 6 servo drivves and buscouuplers
thee model of thhe control systtem created asa a type of vvirtual from different manufactureers connecteed via EtheerCAT
maachine, which has a strict hiierarchy of th
he layers [9,100]. NC inteerface (Fig. 4)).
kerrnel models oof virtual macchines for PC C-based and A ARM-
bassed solutions are built (Figg. 3). The restriction is hanndling
thee EtherCAT prrotocol with a software masster [11].

Fig
g. 3. NC kernel m
models of virtual machines
m for PC-based and ARM-- based
sollutions. Fig. 4. Test
T equipment w
with ARM based NC kernel.

The hardwaree is on the lower level. Th he PC solutionn uses The axis Z is not directtly involved in the proceess of
thee motherboardd GIGABYTE E GA-Z77M-D3H with ann Intel inteerpolation and
d performs auuxiliary functions to interceept the
Coore i3-3250 33.5 GHz processor, 4 Gb b memory annd an parrt from spindle to countersppindle, so forr testing it hass been
add ditional netwwork card PC CI-E D-link DGE-560T/B1
D 1A to connfigured as an
n axis with a vvirtual drive.
con ntrol servo ddrives and auuxiliary devicces via EtheerCAT The standard operator paneel and machin ne control pannel are
prootocol. The iinterpolation and PLC cy ycle is 2 mss. The useed as the term minal part. TThe parameteers such as MTBF M
com mmunication with the opeerator terminaal is performeed via (mean time betw ween failures) > 2000 h, EtherCAT cyclee time
thee built-in mothherboard netw work output. (2 ms), PLC program cycle time, correctt work (2 ms)), etc.
ARM solutionn uses Raspbeerry Pi 2 (Tab ble 1). The neetwork havve been verifieed.
outtput on the bboard is used to control th he servo drivees and The fig. 5 illustrates thhe CNC kerrnel drive system
aux xiliary devicees via EtherCA AT, while thee connection to the perrformance tessts without statarting any PL LC program. CNC
termminal is carrieed out via an adapter
a USB/R RJ45. kerrnel needs ab bout 20% CPPU resourcess for itself before b
At the levell of the opeerating system m, PC and ARM run
nning a part prrogram.
sollutions use thee RT Linux operating
o systeem, which proovides
thee same API (aapplication prrogramming in nterface) funcctions, ARM
M®Cortex™ͲA7
7DualͲCore@
@1GHz
butt has a differeent hardware abstraction laayer for each of the 60
5
51,8
harrdware platforrms. 50 46,2
41,1 6
42,6
Platform inddependent libbraries are identical
i for both 38,9
CPUload[%]

40
sollutions, since the same opeerating system m API interfacces are
27,7
useed. The use of platform m-independentt libraries inn the 30
19,2
arcchitecture enssures the porrtability of the t NC kernnel on 20
diffferent platformms.
The NC kerrnel is identiical for both h solutions, w which 10

enssures the simpple transfer ofo NAKLON 535 NC machhining 0

cennter settings,, including the machinee and kinem matics 1x0 1x1 1x2 1x3 1x4
4 2x4 2x6
parrameters. Thee dashed red linel marks thee identical sofftware Numberrofchanelsandaxxes

moodules that hhave been recompiled

r under
u a partticular
plaatform. Fig
g. 5. Results of CN
NC kernel drive ssystem performan
nce testing withouut PLC
The computational capabbilities of ARM A solutionn are proogram.
infferior to the PPC solutions, but this is compensated bby the
lim
mitations in thee CNC system m machine parrameters on thhe use
528 Sergej N. Grigoriev and Georgi M. Martinov / Procedia CIRP 46 (2016) 525 – 528
The results of testing CNC kernel with 6 interpolated axes
and PLC programs containing up to 20 000 elements in the
FBD language are illustrated on fig. 6.
ARM®Cortex™ͲA7DualͲCore@1GHz
Intel®Core™i74500U@1,8GHzx2
90
80
70,1 72,2
66,9 68,1
70 64
61,3
57,2 59
60
CPUload[%]
50
35
40 34,5
31 32
30
17
20
11
8 based on ARM and FPGA. Procedia Engineering 2011; 16:818-824.
10 3,1
0
5000 10000
NumberofFBDelements
Fig. 6. Results of testing CNC kernel with 6 interpolated axes and PLC
programs containing up to 20 000 elements in the FBD language.
Further tests have been carried out on the machining center
NAKLON 535, a variant without linear encoders. The
EtherCAT cycle is 2 ms, which corresponds to the accuracy
requirements of this machine modification.
The axes have been calibrated with the help of a laser
interferometer Renishaw LX-80 - class 2 as defined in (IEC)
EN60825-1. The position errors for both directions
movements have been automatically inserted in the table of
axes compensation. After calibrating all axes, the positioning
error is reduced to the range of 2 microns [12].
Conclusion
The computing resources of the single board ARM
computers present on the market are sufficient for performing
the key functions of multi-channel and multi-axis machine
tools. The announced release of new ARM processors next
year creates new prospects for the NC system, such as
EtherCAT cycle time reduction up to 1 ms, increasing the
number of control channels up to 4 and simultaneously
interpolated axes up to 12 [13,14].
Properly constructed architectural decision of NC software
kernel allows porting it on a PC platform as well as on an
ARM platform [15,16]. The transition from one solution to
another one is reduced to carrying machine tool settings from
one NC system to another one.
The prototype of ARM based CNC has been demonstrated
in a relevant environment correspondent to TLS 6
(Technology Readiness Level). Further development of
research involves testing the dynamics of the machine,
additional research in the area of precision machining,
adaptive control, diagnostic and forecasting of cutting tool
wear [17,18], etc.
Acknowledgements
This research was supported by the Ministry of Education
and Science of the Russian Federation as a public program in
the sphere of scientific activity and Russian Federal Program
of Supporting Leading Scientific Schools (grant NSh-
3890.2014.9).
75 77,1
74,8
References
[1] Grigoriev SN, Martinov GM. Research and Development of a Cross-
platform CNC Kernel for Multi-axis Machine Tool. Procedia CIRP
2014;14:517-522.
[2] Martinov GM., Kozak NV. Numerical control of large precision
36,3 36,8 37 machining centers by the AxiOMA contol system. Russian Engineering
Research 2015;35(7):534-538.
[3] Geetha A. Modelling a Computer Numerical Control Machine-2 Axis.
Middle-East Journal of Scientific Research 2014;20(1):62-64.
[4] Fei J, Deng R, Zhang Z, Zhou M. Research on embedded CNC device
[5] Vaishak NL, Chandra CR. Embedded Robot Control System Based On an
Embedded Operating System, the Combination of Advanced RISC
15000 20000 Microprocessor (ARM), DSP and ARM Linux. International Journal of
Engineering and Innovative Technology (IJEIT) 2012; 2(6):143-147.
[6] Cloutier MF, Paradis C, Weaver VM. Design and analysis of a 32-bit
embedded high-performance cluster optimized for energy and
performance. In Hardware-Software Co-Design for High Performance
Computing (Co-HPC) IEEE 2014;1-8.
[7] Martinov GM, Nezhmetdinov RA. Modular design of specialized
numerical control systems for inclined machining centers. Russian
Engineering Research 2015;35(5):389-393.
[8] Martinov GM, Martinova LI. Trends in the numerical control of machine-
tool systems. Russian Engineering Research 2010;30(10):1041-1045.
[9] Grigoriev SN, Martinov GM. Scalable Open Cross-Platform Kernel of
PCNC System for Multi-Axis Machine Tool. Procedia CIRP 2012;1:238-
243.
[10] Grigoriev SN, Martinov GM. The Control Platform for Decomposition
and Synthesis of Specialized CNC Systems. Procedia CIRP 2016;41:858-
863.
[11] Martinov GM, Lyubimov AB, Bondarenko AI, Sorokoumov AE,
Kovalev IA. An Approach to Building a Multiprotocol CNC System.
Automation and remote control 2015; 72(10): 345-351.
[12] ISO 230-2: Test code for machine tools - Part 2: Determination of
accuracy and repeatability of positioning numerically controlled axes,
2006.
[13] Martinov GM, Kozak NV. Numerical control of large precision
machining centers by the AxiOMA contol system. Russian Engineering
Research 2015;35(7): 534-538.
[14] Martinova LI, Pushkov RL, Kozak NV, Trofimov ES. Solution to the
problems of axle synchronization and exact positioning in a numerical
control system. Automation and Remote Control 2014; 75(1): 129-138.
[15] Martinov GM, Obuhov AI, Martinova LI, Grigoriev AS. An Approach to
Building Specialized CNC Systems for Non-traditional Processes.
Procedia CIRP 2014;14: 511-516.
[16] Martinova LI, Sokolov SS, Nikishechkin PA. Tools for Monitoring and
Parameter Visualization in Computer Control Systems of Industrial
Robots. Advances in Swarm and Computational Intelligence. Springer
International Publishing 2015; 9142:200-207.
[17] Martinov GM, Grigoriev AS. Diagnostics of cutting tools and prediction
of their life in numerically controlled systems. Russian Engineering
Research 2013;33(7):433-437.
[18] Martinov GM, Grigoryev SA, Nikishechkin PA. Real-Time Diagnosis
and Forecasting Algorithms of the Tool Wear in the CNC Systems.
Advances in Swarm and Computational Intelligence. Springer
International Publishing 2015;9142:115-126.