HEIDENHAINAccuracy of Feed Axes 349 843-20

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Technical Information

Accuracy of Feed Axes

Every year’s new machine tools have been showing improvements in efficiency and power. Increasing feed rates and acceleration values
have reduced machining times. At the same time, increasing accuracy has permitted ever-closer tolerances. On the one hand, these
developments enable the machining of increasingly critical parts; on the other hand they simplify the manufacture of complex
assemblies. Selective manual assembly and bench work can often be reduced. Also, higher accuracy of parts normally results in
improved function of component assemblies. Dimensional accuracy in motor transmissions, for example, has increased service life and
reduced noise emission.

In the total error budget of a machine tool the positioning error values of the feed axes play a critical role. The following text discusses
these errors and compares them with other types of error. The primary problem involved with position measurement using rotary
encoder and ball screw is the thermal expansion of the ball screw. The resulting positioning error often outweighs the thermally induced
structural deformation and geometric error of machining centers. Several tests show the influence of the heating of the ball screw on the
results of machining – also with respect to the ball screw bearing. When a linear encoder is used for position measurement, the thermal
expansion of the ball screw has no influence and the position drift is negligible. As requirements for machine tool accuracy and velocity
increase, the role of linear encoders for position measurement grows increasingly important.

The accuracy of modern machine tools is Considering the increasing frequency of


measured with an increasing number of changing jobs and the concomitant
new and revised inspection and acceptance reduction in batch sizes, reducing the
tests. Where years ago purely geometric thermal or systematic error of a machine
acceptance tests predominated, today’s tool through tedious optimization of
routine methods include dynamic tests individual production steps is seldom
such as circular interpolation and free-form feasible. The ´accuracy of the first part´ is
tests, thermal tests such as described in gaining in importance. In particular the
ISO/DIS 230-3, and for production thermal error of machine tools is drawing
machines, capability testing during ever more interest.
acceptance or regular inspection. The
influences of the cutting processes, the The following text shows that thermal error
geometric accuracy of the machine, its can be quite significant, especially for the
static and dynamic rigidity and the feed axes. Unlike structural deformation,
positioning response of the feed axes on errors of the feed axes can be dramatically
the attainable accuracy of the workpiece reduced through a choice of simple and
can be more specifically analyzed. Machine readily available measuring devices.
errors are therefore becoming increasingly
transparent to the user.

Feed drive mechanism of a milling


machining center

August 2006
Feed-Drive System Design

An exact error analysis of position measure- motor and the slide must have a known Trend toward digitally driven axes
ment via rotary encoder and feed screw and, above all, reproducible mechanical As a result of the trend toward digital axes
begins with a consideration of prevalent transfer behavior. A position control loop in drive technology, a large share of new
mechanical feed-drive systems. Although with a linear encoder, on the other hand, servo motors feature rotary encoders,
machine tool designs vary immensely, the includes the entire mechanical feed-drive which in principle can serve together with
mechanical configuration of their feed drive system. Transfer errors from the mechanics the feed screw for position control. With
is largely standardized (Fig. 1). In almost are detected by the linear encoder on the such a drive configuration the decision must
all cases, the recirculating ball screw slide, and are corrected by the controller be made as to whether to add a linear
has established itself as the solution for electronics. encoder or simply to use a ball screw
converting the rotary motion of the servo working in combination with the already
motor into linear slide motion. Its bearing Differing terminology existing motor encoder.
takes up all axial forces of the slide. The Different terms are used to distinguish
servo motor and ball screw drive are usually between these two methods of position One should remember to consider the
directly coupled. Toothed-belt drives are also control. German-speaking and some problems discussed in the following text
widely used to achieve a compact design English-speaking communities generally regarding position measurement using a
and better adapt the speed. refer to them somewhat inaccurately as rotary encoder/ball screw system. They
”direct and indirect measurement.” can quickly increase the cost of an ”eco-
For position measurement of feed axes on However, these terms are rather poorly nomical” machine if the owner finds that
NC machine tools it is possible to use either chosen because, strictly speaking, both the accuracy does not suffice in certain
linear encoders or recirculating ball screws methods are direct. One method uses the applications.
in conjunction with rotary encoders. A line grating on the linear scale as the
position control loop via rotary encoder and measuring standard, the other the pitch of
ball screw includes only the servo motor the ball screw. The rotary encoder simply
(Fig. 1 dashed line). In other words, there serves as an interpolating aid. Here the
is no actual position control of the slide, Japanese concepts of "semi-closed-loop
because only the position of the servo and closed-loop control" seem appropriate,
motor rotor is being controlled. To be able since they more aptly describe the actual
to extrapolate the slide position, the problem.
mechanical system between the servo

xactl
Position Velocity
controller controller

xnom

Fig. 1: Typical drive system of a numerically controlled machine tool with linear scale on the slide and rotary encoder on the motor.
Unlike in position feedback control with rotary encoder and ball screw, a linear encoder includes the feed drive mechanism in the
control loop.

2
Positioning Errors Caused by Mechanical Influences

Kinematics error 5 μm/div 5 μm/div


Kinematics errors that can be directly
attributed to position measurement using
feed screw and rotary encoder result from
ball screw pitch error, from play in the feed Y Y
elements, and from the so-called pitch
HEIDENHAIN HEIDENHAIN
loss. Ball screw pitch errors directly influe-
nce the result of measurement because X X

the pitch of the ball screw is being used as ACCOM ACCOM


a standard for linear measurement. Play in
the feed transfer elements causes back-
lash. The pitch loss [1] results from a shift
of the balls during the positioning of ball
screw drives with two-point preloading
and can lead to reversal error in the order New After 1 year
of 1 to 10 µm.
Fig. 2: Circular tests of a machining center without linear encoders in new condition and
Error compensation after one year. The reversal error of the X axis has increased significantly.
Most controls are capable of compensating
such pitch error and reversal error. How-
ever, to determine the compensation
values it is necessary to make elaborate
measurements with comparative measur-
ing devices such as interferometers and
grid encoders. In addition, the reversal error
is often unstable over long periods of time
and must be regularly recalibrated (Fig. 2). 5 μm/div f = 2, 4, 6, m/min 5 μm/div
R = 10 mm
Strain in drive mechanisms
Forces leading to the deformation of feed Y Y
drive mechanisms cause a shift in the
actual axis slide position relative to the HEIDENHAIN HEIDENHAIN
position measured with the ball screw and X X
rotary encoder. They are essentially inertia ACCOM ACCOM
forces resulting from acceleration of the
slide, cutting process forces, and friction in
the guideways. The mean axial rigidity of a
feed drive mechanism as shown in Fig. 1
lies in the range of 100 to 200 N/µm (with
a distance between ball nut and fixed Position control with rotary encoder Position control with linear encoder
bearing of 0.5 m and a ball screw diameter
of 40 mm). Fig. 3: Circular tests of a machining center that has been retrofitted with linear encoders.
With position feedback control by rotary encoder and ball screw, the circles deviate
Cutting force significantly from the ideal path at high feed rates. With linear encoders, the
The cutting force can quite possibly lie in contour accuracy is considerably better.
the kN range, but its effect is distributed
not only in the feed drive system, but also
over the entire structure of the machine Force of acceleration 5 N therefore results in feed drive
between the workpiece and the tool. The A typical slide mass of 500 kg and a mod- deformation of only 0.25 to 6 µm.
deformation of the feed drive system erate acceleration of 4 m/s2 result in de-
therefore normally has only a small share formations of 10 to 20 µm that cannot Circular test for inspecting machine
in the total deformation of the machine. be recognized by the rotary encoder/ball tools
A linear encoder can recognize and correct screw system. The present industry trend A typical example for errors dependent on
only this small portion of the total defor- toward accelerations in significantly higher acceleration and velocity can be recorded
mation. Critical component dimensions, ranges will result in increasingly large de- in a circular interpolation test on a vertical
however, are normally finished at low feed formation values. machining center (Fig. 3). With position
rates with correspondingly low deformation feedback control by rotary encoder and ball
of the feed drive system. Force of friction screw, the circles deviate significantly from
The force of friction in the guideways lies the ideal path at high contour speeds. The
between 1% and 2% of weight for roller same machining center shows significantly
guideways and 3% to 12% of weight for better contour accuracy when equipped
sliding guideways [2]. A weight exerting with linear encoders.

3
Positioning Error due to Rising Ball-Screw Temperature

Positioning error resulting from thermal gutter. The result is a continual pressing and The moment of friction of a ground
expansion of the ball screw presents the pushing with occasional slipping of the balls. precision recirculating ball screw with 40
greatest problem for position measurement The friction among the balls is aggravated mm diameter and 10 mm pitch was
via rotary encoder and ball screw. This is by high surface pressure due to the measured by Golz [4] for various preload
because the ball screw drive must serve a absence of a retaining device to separate forces and rotational speeds (Fig. 4). The
double function: On the one hand it must them. As in every angular-contact ball Stribeck characteristic of frictional moment
be as rigid as possible to convert the rotary bearing a spinning friction results from a is clearly recognizable. It confirms the high
motion of the servo motor to linear feed contact diameter that is not orthogonal to share of solid-body friction and mixed
motion. On the other hand it must serve as the axis of ball rotation. Each ball therefore friction in ball screw drives at low speeds.
a precision measuring standard. The two- rotates about its contact diameter. Recent Viscous friction dominates at high speeds.
fold function therefore forces a compro– studies have also shown that the balls can It is interesting to note that for this typical
mise because both the rigidity and the move in the thread only because of an ball screw the normal machining feed rates
thermal expansion depend on the pre– additional slip component brought on by the lie far below the speeds at which the
loading of the ball nut and the fixed bear– thread pitch [3]. moment of friction is at its minimum. The
ing. Both the axial rigidity and the moment rapid traverse feed rates, however, lie far
of friction are roughly proportional to the The recirculation system is a special above it. The feed rates at which this ball
preloading. problem zone for ball screws. With every screw is at optimum efficiency therefore
entrance into the recirculation channel, just seldom occur. The moment of friction is
Friction in the ball nut as with every exit, the movement of the only slightly dependent on the axial load of
The largest portion of the friction in a feed ball changes entirely. The rotational energy the ball nut [4].
drive system is generated in the ball nut. of the balls, which in rapid traverse typically
This is because of the complex kinematics rotate with 8 000 rpm, must be respec- Frictional heat generated in the ball nut
of a recirculating ball nut. Although at first tively started and stopped. In contrast to With a typical preload of 3 kN and allowing
glance the balls may seem only to be rolling, the preloaded thread zone, in the recircula- for the missing wiper, this results in a no-
they are in fact subjected to a great deal of tion zone the balls are not under stress. load or frictional moment of 0.5 to 1 Nm.
friction. Besides the microslip resulting from The play of energy causes the balls to This means that in rapid traverse at a ball
relative motion in the compressed contact collect in the recirculation channel. Without screw speed of 2000 rpm approximately
areas, the greatest effect is from the macro- elaborate measures to reintroduce the balls 100 to 200 W of frictional heat is generated
slip due to kinematics exigencies. The balls into the thread at the end of the channel it in the ball nut.
are not completely held in the races and tends to congest, causing the familiar
wobble much like tennis balls rolling down a jamming of the ball screw drive.

Nominal diameter 40.00 mm


Pitch 10.00 mm
Axial force 0.00 kN
Angle of contact 45.00 degrees
Ball diameter 6.35 mm
Internal ball recirculation, no wiper
Moment of friction 

Moment of friction

Mixed Viscous friction


friction
e
rc
fo
g
in
oad
el
Pr
rpm
Spindle speed  Rotational speed n

Fig. 4: Measured moment of friction of a two-point preloaded ball screw [4].


The Stribeck characteristics is plainly visible.

4
More frictional heat to be expected Deformation in the machine structure
To increase rapid traverse velocity, either resulting from changes in ambient
the pitch or the rotational speed of the conditions or through heat generation in
recirculating ball screw must increase. In the spindle drive are recorded with the aid
the last 5 years the maximum permissible of 5 probes that measure against a cylinder
speed of recirculating ball screws has mounted in the tool holder. This makes it
doubled. Due to the continually increasing possible to measure all 5 relevant degrees
requirement for acceleration, the pre- of freedom. To test the feed axes, it pro-
loading and therefore the friction of the poses a repeated positioning to 2 points
ball nut could not be reduced. Recirculating that lie as near as possible to the traverse
ball screw drives therefore generate sig- range at an agreed percentage of rapid
nificantly more heat than before and will traverse velocity. The change of the posi-
generate even more in the future. tions with respect to the initial value is
recorded. The test is to be conducted until
Measurement of positioning accuracy a satiation effect is clearly observable.
according to ISO 230-3 Simpler test equipment than the laser
The influence of frictional heat on the interferometer, such as dial gauges, can
positioning response of the feed axis also be used for the axis test. These tests
becomes apparent when positioning tests enable any workshop to conduct such
are conducted in accordance with the new inspections at a reasonable cost.
international standard ISO/DIS 230-3. This
standard contains proposals for making
uniform measurements of thermal shifts of
lathes and milling machines as a result of
external and internal heat sources (Fig. 5).

Structural deformation Drift of positioning axis

– Ambient conditions – Heat generated in the ball screw


– Heat generated by the spindle

Fig. 5: Measurement of thermal displacement of a machining center in accordance with


ISO/DIS 230-3.

5
Influence of the Ball Screw Bearing on Positioning Accuracy

Differing types of behavior can be expected Fixed bearing at one end (fixed/floating), see Fig. 7
depending on whether the ball screw can
expand freely. The various types of bearings
for recirculating ball screws are illustrated in
Fig. 6.
Fixed bearing at both ends (fixed/fixed), see Fig. 10
Fixed bearing at one end
In the case of fixed/floating bearings, the
ball screw will expand freely away from the
fixed bearing in accordance with its tem-
perature profile. The thermal zero point of Fixed bearing at one end (fixed/preloaded), see Fig. 11
such a feed axis is at the fixed bearing.
This means that theoretically no thermal
shift would be found if the ball nut is 50 µm/m
located at the fixed bearing. All other
positions will be affected by the thermal Fig. 6: The various types of bearings for recirculating ball screws.
expansion of the ball screw.

Fig. 7 shows the result of a positioning


test as per ISO/DIS 230-3 on a vertical
machining center (built in 1998) retrofitted 40 min
with linear encoders. An X axis was 20
Drift of position [µm] 

positioned to three points a total of 100


times at 10 m/min. Taking the standstill
0 Linear encoder
periods for measured value acquisition into 0 10 20 30 40 50 60 70 80 90
account, the mean traversing speed during
the test was approx. 4 m/min. In addition -20
0 mm
to the two positions at the ends of traverse
as recommended in the standard, a third -40
position at the midpoint of traverse was
measured. Fig. 7 shows the position -60
values with respect to their initial values. 360 mm Ball screw/
At first the ball screw / rotary encoder rotary encoder
-80
system was used for position feedback
control. In a second test under otherwise
-100
identical conditions, linear encoders were
used. The comparator system was a 720 mm
-120
VM 101 from HEIDENHAIN.
Time [min] 

In spite of the moderate feet rate of 10 m/


min (rapid traverse 24 m/min), the position
farthest from the fixed bearing of the ball Fig. 7: Drift of three positions during positioning accuracy measurement in accordance
screw shifted by more than 110 µm within with ISO/DIS 230-3 on a machining center with ball screw in fixed/floating
40 minutes. It is interesting to note that the bearings. Position measurement via rotary encoder and ball screw shows a distinct
drift increases very quickly immediately drift of positions due to the thermal growth of the screw.
after switch-on. Any change in the mean
feed rate in a production process therefore
immediately affects positioning accuracy.
Similar results were published by Schmitt
[5]. No drift in position values measured Batch production
with linear encoders To demonstrate the applicability of this
The measured positioning accuracy experiment to actual production conditions,
therefore depends directly on the number a small batch of aluminum workpieces was
of repetitions, particularly after the first few machined on the same machine. Eight
repetitions. The measurements taken by 70 mm x 70 mm workpieces were fixed on
the retrofitted linear encoders show no a vertical machining center. Four pockets
drift. and two radii were machined using 4 tools
with an infeed of 1 mm in the Z axis (Fig. 8).

6
After the 6-minute machining operation the
8 parts were not exchanged. Rather, the
infeed in Z was increased by 1 mm and the
operation repeated. As a result of the
thermal expansion of the ball screw, all 8 workpieces
workpieces show a step pattern on the left Tool 1
side. This pattern is particularly pronounced
on the workpiece farthest away from the
fixed bearing. The right sides of the work-
pieces are smooth because with each shift
Tool 2
in the positive X direction the previous step
was also removed. In principle the same
effect could be observed in the Y direction
as in the X direction, but because of the
lesser amount of movement in the Y axis Tool 3
the step pattern is significantly less pro-
nounced. In the X direction the compara-
tive measurement of the step pattern
shows a drift of approx. 90 µm with a time
constant of thermal expansion slightly less Tool 4
Time / 8 workpieces = 6 min
than an hour (Fig. 9).

If additional work is to be done on


previously machined workpieces with
critical dimensions, the machine datum
Fig. 8: Experimental setup for batch production with multiple workpieces. 4 pockets
must be continually inspected and cor-
and 2 radii were machined using 4 tools with an infeed of 1 mm in the Z axis. To
rected. The machine achieves thermal
illustrate drift resulting from thermal expansion of the ball screw, the workpieces
equilibrium after one hour, but after an
were not exchanged after machining. Instead, the part program was run repeatedly
interruption in machining it begins to drift
at successively increasing depth.
in the reverse direction. If the part pro-
gram and with it the mean feed rate are
changed, it again takes approx. 1 hour
for the ball screw to regain thermal
equilibrium.

Fixed bearing at both ends


The situation is more complex in the case of
fixed/fixed bearings. Ideally rigid bearings 2h
would prevent expansion of the ball screw
Drift
at its end points. However, this would re- 100
Drift of machine datum [µm] 

quire considerable force. To prevent expan-


sion of a ball screw with 40-mm diameter, 80
2.6 KN must be applied per degree Celsius
of temperature increase. A typical angular- 60
contact ball bearing would quickly fail under
any large increase in temperature. Under
real conditions, the rigidity of the pur- 40
portedly fixed bearings with their seats lies
in the area of 800 N/µm. This means that as 20 Ball screw / rotary encoder
the temperature of the ball screw increases,
the bearings deform significantly. The end 0 Linear encoder
points of the ball screw do not remain at 0 5 10 15 20
their original position. The same experiment -20
as in Fig. 7 was conducted on a vertical Cycle and depth [mm] 
machining center (built in 1998) with fixed
bearings at both ends. The tested 1-m long
feed axis was mechanically designed to be
very rigid. At each end of the ball screw the
same bearing was built into seats that were Fig. 9: Result of the experiment in Fig. 8. The left pocket of the workpiece plainly shows a
machined directly into the machine’s cast step pattern resulting from thermal expansion of the ball screw.
frame.

7
The results of measurement in Fig. 10
show curves as theoretically expected.
The end points of the ball screw cannot
be kept in their original positions. They 30

Drift in position [µm] 


each move by 20 to 30 µm in the direc-
tion of the force generated by heat. The 20 0 mm
total expansion of the ball screw is about
50 % less than that shown in Fig. 7. This
10
means that by designing fixed bearings
at both ends, the expansion could be
halved. The thermal zero point of the feed 0
axis seems to lie at the midpoint of the 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 450 mm
traverse range. This is also expected
-10
because the bearings have approximately
equal rigidity and the ball screw was
heated evenly over its entire length. -20

Fixed/preloaded bearing -30 900 mm


This type of bearing causes problems
for traversing programs with high mean
velocities because the bearing load is -40
detrimental to service life and the forces Time [min] 
to be withstood result in deformation of
the machine structure. A fixed/preloaded
bearing design is therefore often used as
Fig. 10: Drift of three positions during positioning accuracy measurement in accordance
a sort of pressure valve (Fig. 6). With a
with ISO/DIS 230-3 on a feed axis with a ball screw in fixed bearings at both ends.
typical preload of 50 µm/m, one would
Position measurement via rotary encoder and ball screw causes a distinct drift of
expect that such a bearing configuration
positions due to the thermal growth of the ball screw.
would behave like a fixed/fixed combination
up to a temperature increase of approx.
5 K, and beyond that, like a fixed/floating
combination.

Fig. 11 shows the results of a positioning


test on a machining center with a ball screw
with fixed/preloaded bearings conducted
along the pattern of the previously described
experiments. Surprisingly, in spite of the 50
Drift in position [µm] 

fixed/preloaded bearing configuration, a


position drift similar to that in Fig. 7
becomes apparent. This means that the 40 500 mm
feed axis behaves roughly like one with
fixed/floating bearings. The thermal zero 30
point seems to lie near the fixed bearing.
Unlike the axes in the two previous experi- 250 mm
ments, this axis had a travel of only 500 mm 20
instead of one meter. The magnitude of the
drift is therefore not comparable. 10
This experiment shows that the simple 0 mm
model of the ball screw with fixed/pre-
loaded bearings does not stand up to 0
reality. As a rule, the end with the movable 0 6 12 18 24 30 36 42 48 54
bearing is much less rigid than the end
-10
with the fixed bearing. The cause lies in
Time [min] 
the difference in the bearing designs. While
the first end with a genuine, inherently
preloaded fixed bearing must continue
to remain rigid when the second end has
begun to move, with increasing tempera- Fig. 11: Drift of three positions during positioning accuracy measurement in accordance
ture the second end loses its preload and with ISO/DIS 230-3 on a feed axis with a ball screw in fixed/preloaded bearings.
therefore also its rigidity. The results show a distinct drift due to position measurement by ball screw and
rotary encoder.

8
Influence of the Temperature Distribution along the Ball Screw

Apart from the ratio of bearing stiffness,


the position of the thermal zero point
depends particularly on the distribution of
temperature along the ball screw. Fig. 12
shows a thermographic snapshot of a ball
screw drive after several hours of recipro-
cating traverse between two points at a
distance of 150 mm. As the thermograph
shows, even after several hours the
temperature increase remains almost
exclusively in the area of ball nut traverse.
The temperature of the ball screw and
therefore the thermal expansion is very
local.

Because the bearings of the ball screw can


provide at best only an evenly distributed
mechanical tension and ensure constant
expansion along the ball screw, they cannot
compensate the expansion resulting from
local temperature changes.

A simple calculation shows this clearly Fig. 12: Local heating of a ball screw in the traverse range of the ball nut after 6 hours of
(Fig. 13). On a 1-m long ball screw with a reciprocating traverse at 24 m/min between two positions separated by 150 mm
fixed bearing at one end, a local tempera- [6]. For this thermographic snapshot, the machine table was moved aside at the
ture increase of 10 K as indicated by the end of the traverse program.
red curve in Fig. 13 would result in a The illustration shows the higher temperatures of the belt drive, locating bearing,
positioning error as indicated by the green and ball screw.
curve in Fig. 13 right. A fixed/fixed bearing
configuration with a rigidity of 700 N/µm
results in an error curve as indicated by the
blue curve. As a result of the forces
exercised by the bearing, the ball screw is
compressed at its ends where the temper-
+10 K
ature is not increased. The area of the ball
screw near its midpoint expands due to the
temperature increase at almost the same
rate as with the fixed/floating configuration.
¬ 40 (35)mm
At 22 µm, the maximum positioning error
from the fixed/fixed configuration is 1m
roughly 2/3 of the error that occurs from
the fixed/floating configuration. 700 N/μm 700 N/μm
0 N/μm

40
Positioning error [µm] 

20

0
0.5 1

20
Position [m] 

Fig. 13: Positioning error in a semi-closed loop as a result of local temperature rise in the
recirculating ball screw.
9
Countermeasures

The test results show that the thermal of a positioning test on a vertical machining feed axis shows the variations on input
expansion of the ball screw as a result of center with liquid-cooled ball screws in parameters to be considered.
friction in the bearings and particularly in fixed/floating bearings. During the test, the
the ball nut results in significant positioning axis was moved slowly at 2.5 m/min be- To compensate the expansion of the ball
errors if the axis is controlled in a semi- tween two points at a distance of 500 mm. screw, its temperature must be known with
closed loop. Besides the use of linear The maximum traverse range was 800 mm. respect to its position, since the local
encoders, countermeasures aimed at The position drift of the position farther temperature depends on the traversing
avoiding this error include coolant-conduct- away from the fixed bearing was recorded. program. Direct temperature measurement
ing hollow ball screws and purely electronic The switching of the chiller is plainly visible. of the rotating ball screw, however, is very
compensation in the control software. Its hysteresis was 1 K. Compared with the difficult. Machine tool builders therefore
noncooled semi-closed loop design the often attempt to calculate the temperature
Cooled ball screws position drift was significantly reduced. distribution. This is theoretically possible if a
The circulation of the coolant requires a However, the switch operations produce heat analysis can be prepared for individual
hole in the ball screw and rotating bushings relatively quick changes in position, which sections of the ball screw. The heat in such
near the screw bearings. Apart from the have a stronger effect during the machining a section is generated by friction in the ball
sealing problems, the hole reduces the ball of workpieces with short machining times nut through thermal conductance along the
screw’s mechanical rigidity in its already than the slow position drift evident in the ball screw, and through heat exchange with
weak axial direction. The greatest problem, noncooled semi-closed loop design. the environment. The friction of the ball nut
however, is a sufficiently accurately temper- depends almost proportionately on the
ature control of the coolant. A 1° C change Software compensation preload of the ball nut and, in a complex
in temperature changes the length of a 1-m Research is underway on compensation of manner, from the type, quantity and
long ball screw by 11 µm. In the light of the thermal deformation with the aid of analytic temperature of the lubricant. The preload
considerable amount of heat to be removed models, neural networks and empirical of the ball nut normally changes by ±10 to
it is not an easy task to maintain a temper- equations. However, the main focus of 20% over its traverse range in an manner
ature stability of < 1 K. This is particularly these studies is in the deformation of the depending on the individual ball screw. In
so when the spindle or its bearings are machine tool structure as a result of internal the course of the first 6 months, the mean
cooled with the same system. In such a and external sources of heat. There is little preload typically decreases to 50% of its
case, the required cooling capacity can interest in investigation into compensation individual value. Due to the complex inter-
easily lie in the kilowatt range. The of axis drift. action of static forces at play on the ball
temperature constancy of existing ball screw, certain jamming effects and an
screw chillers is usually significantly worse As a whole, the possibilities of such soft- associated increase in friction are unavoid-
than 1 K. It is therefore often not possible ware compensation are frequently overes- able. Even these few examples show that
to use them to control the temperature of timated in today’s general atmosphere of the calculation of the actual frictional heat
the ball screw. Switching controllers are enthusiasm for software capabilities. presents formidable problems. Calculating
often used in the chillers to reduce cost. Successful compensation in the laboratory the heat dissipation is similarly difficult
Since each switching operation is triggered is usually achieved only after elaborate because is depends strongly on largely
by a violation of temperature limits, the special adjustments on the test machine. unknown ambient conditions. Even the
individual switching operation can be It is usually not possible to apply such temperature of the air surrounding the ball
considered to be an expansion of the methods to machines from series produc- screw is normally unknown, although it
cooled ball screw and therefore an axis tion without time-consuming adjustment of plays a decisive role in any calculation of
positioning error. Fig. 14 shows the result the individual machines. The example of the heat dissipation.

On the whole it seems certain that, even


10 in the relatively simple case of a ball screw
Position error [µm] 

Linear encoder with fixed/floating bearings, software


compensation of ball screw expansion
0
0 10 20 30 40 without additional temperature sensors
has little chance of success. In the case
-10 of fixed/fixed and fixed/preloaded bearing
Cooled ball screw / rotary encoder one must additionally take into account the
bearing rigidity and the preload-dependent
-20 friction in the bearings. These factors make
compensation even more difficult.

-30

Noncooled ball screw / rotary encoder


-40

-50
Time [min] 

Fig. 14: X axis of a vertical machining center with liquid-cooled ball screw in fixed/floating
bearings. The diagram shows the drift of the position farthest away from the fixed
bearing during reciprocating traverse over 500 mm (800-mm traverse range) at 2.5
m/min. The axis was also equipped with a linear encoder for test purposes.
10
Comparison of Positioning Error with Other Types of Error

After this discussion of temperature-


20

Linear deformation [µm] 


dependent positioning error of feed
drives, it remains to classify these types
10
of error with the other types of static and X
quasistatic error in the total error budget 0
of the tested machining centers. The
frame deformation resulting from the -10
heat generated by the spindle drive
was examined on all three machines in -20
accordance with ISO/DIS 230-3. After Y
several hours of operation at a maximum -30
spindle speed of 6000 rpm, the first
machining center showed a linear defor- -40
mation of {x: 5 µm, y: 60 µm, z: 15 µm}. The
rotational deformation reached a maximum -50
Z
of {a: 40 µm/m, b: 70 µm/m}. Spindle on Spindle off
-60
The deformation of the second machining -70
center is shown in Fig. 15. Under the same 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
conditions, also with 6 000 rpm, it shows a Time [h] 
maximum linear deformation of {x: 5 µm,
y: 45 µm, z: 55 µm}. The rotational deforma- 40
Rotational deformation [µm] 

tion reached a maximum of {a: 25 µm/m,


b: 10 µm/m}. The third machine was 30
equipped with a high-speed spindle and
20
jacket cooling. At 12 000 rpm it showed Rotation about X
linear deformations of {x: 5 µm, y: 5 µm,
10
z: 40 µm} and rotational deformations of
max. {a: 20 µm/m, b: 30 µm/m}. The 0 Rotation about Y
measured axis drift values attain at least
the same magnitude as the structural -10
deformation. Particularly on spindles with
fixed/floating bearings, or machines with -20
effective cooling of the spindle, the posi-
tioning error of the feed axes driven in a -30
semi-closed loop is significantly greater Spindle on Spindle off
than the measured structural deformation. -40

A comparison with the usual geometric -50


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
error leads to similar results. If one observes
Time [h] 
the pitch, roll and yaw error of the feed axes
of 16 different NC machines one sees that
Fig. 15: Structural deformation of a vertical machining center as a result of heat generated
these types of error usually lie in the range
in the spindle drive at 6000 rpm without load.
of 10 to 50 µm/m (Fig. 16). The positioning
error is found by multiplying these values
by the respective Abbe distance. The error
does not attain the values of the feed axis
160
Deviation [µm/m] 

until over 1 meter traverse.


140  Roll
 Pitch
120  Yaw
100

80

60

40

20

0
500 550 600 650 700 750 900 950 1000 1200 1900 2000 2600 3200
Travel range of X axis [mm] 

Fig. 16: Pitch, roll and yaw angles of feed axes of 16 different NC machines

11
Conclusion

The primary problem involved with position


measurement using rotary encoder and ball Literature
screw is the thermal expansion of the ball
screw. With typical time constants of 1 to 1) Schröder, Wilhelm, Feinpositionierung
2 hours, thermal expansion causes posi- mit Kugelgewindetrieben, Fortschritts-
tioning error in the magnitude of 0.1 mm, bericht VDI Reihe 1 Nr. 277, Düsseldorf;
depending on the nature of the part pro- VDI Verlag 1997
gram. This positioning error therefore out- 2) VDW-Bericht 0153, „Untersuchung
weighs the thermally induced structural von Wälzführungen zur Verbesserung des
deformation and geometric error of statischen und dynamischen Verhaltens
machining centers. von Werkzeugmaschinen”
3) Weule, Hartmut, Rosum, Jens,
After every new part program the ball screw Optimization of the friction behaviour of
requires approx. 1 hour to attain a thermally ball screw drives through WC/C coated
stable condition. This also applies for inter- roller bodies, Production Engineering Vol.
ruptions in machining. A rule of thumb for 1/1 (1993)
thermal expansion is that, over the entire 4) Golz, Hans Ulrich, Analyse, Modell-
length of a cold ball screw 1 meter in length, bildung und Optimierung des
the ball screw grows by approx. 0.5 to 1 µm Betriebsverhaltens von
after every double stroke. This expansion Kugelgewindetrieben, Dissertation Uni
accumulates within the time constant. Karlsruhe, 1990
5) Schmitt, Thomas, Modell der Wärme-
As requirements for machine tool accuracy übertragungsvorgänge in der
and velocity increase, the role of linear mechanischen Struktur von CNC-
encoders for position measurement grows gesteuerten Vorschubsystemen, Verlag
increasingly important. This should be taken Shaker, 1996
into consideration when deciding on the 6) A. Frank, F. Ruech, Position
proper feedback system design. measurement in CNC Machines ...
Lamdamap conference, Newcastle 1999

For more information:


• Brochure: Linear Encoders for
DR. JOHANNES HEIDENHAIN GmbH Numerically Controlled Machine Tools
Dr.-Johannes-Heidenhain-Straße 5 • Brochure: Rotary Encoders
83301 Traunreut, Germany • Brochure: Position Encoders for Servo
{ +49 (86 69) 31-0 Drives
| +49 (86 69) 50 61
E-Mail: info@heidenhain.de
www.heidenhain.de

349 843-20 · 10 · 8/2006 · F&W · Printed in Germany · Subject to change without notice

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy