HEIDENHAINAccuracy of Feed Axes 349 843-20
HEIDENHAINAccuracy of Feed Axes 349 843-20
HEIDENHAINAccuracy of Feed Axes 349 843-20
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.
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
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.
Moment of friction
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).
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.
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).
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
8
Influence of the Temperature Distribution along the Ball Screw
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.
-30
-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
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
349 843-20 · 10 · 8/2006 · F&W · Printed in Germany · Subject to change without notice
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: