Optical Engineering 46共5兲, 051016 共May 2007兲

High-accuracy method for measuring

two-dimensional angles of a linear guideway
Cuifang Kuang Abstract. A single method for simultaneously measuring 2-D rotational
Beijing Jiaotong University angles is proposed by using a collimation laser. Compared with the tra-
School of Science ditional autocollimator measurement method, the proposed method can
Beijing 100044, China also be applied to longer distance and higher precision angular measure-
ment. The 2-D angular drift of the reference datum beam is suppressed
by a factor of 10 via expansion. A traditional single lens in front of the
En Hong detector is replaced by combination lenses to increase the sensitivity and
University of Michigan compactibility. The measurement principle of the 2-D angles is also ana-
Mechanical Engineering Department lyzed. The feasibility and stability of the presented system are verified by
Ann Arbor, Michigan 48109-2125 comparison with the Renishaw ML10 laser interference measuring sys-
tem. The experimental results show that the linear correlativity is
0.99999 and the point-to-point standard deviation error is less than
Qibo Feng ±0.2 arcsec in a ±60-arcsec range and 2000-mm distance. This proposal
Beijing Jiaotong University is most advantageous for simultaneous long distance and high accuracy
School of Science of 2-D angular errors measurement of a linear guide manner. © 2007
Beijing 100044, China Society of Photo-Optical Instrumentation Engineers. 关DOI: 10.1117/1.2740137兴
E-mail: [email protected]
Subject terms: two-dimensional rotation angles; pitch; yaw; linear guideway; laser
measurement.
Paper 060477SSR received Jun. 16, 2006; revised manuscript received Sep. 26,
2006; accepted for publication Oct. 5, 2006; published online May 16, 2007.

1 Introduction 2 Measurement Principle and System

Configuration
The slide axis of machine tools, coordinate measuring ma-
2.1 Measurement Principle
chines 共CMMs兲, and the like has three rotational errors
共pitch, yaw, and roll兲 pertaining to the linear movement. The fundamental measurement principle of the 2-D angles
Rotational errors would manifest in a linear form via am- is shown in Fig. 1. When a collimated beam is projected
plification by an Abbe arm, and thus remarkably influence onto a plane reflector that is perpendicular to the beam, the
beam will be reflected back along the same path. When the
the machining or measurement accuracy. Therefore, it is
reflector is tilted by a small angle ␪ with respect to the
significantly important to identify these rotational errors.
incident beam, the deflection of the beam is twice the an-
For measurement of small angles, an autocollimator or in- gular displacement of the reflector. The relationship be-
terferometer is commonly used. Based on interferometry, a tween the deviation ⌬ and the angle ␪ is given by
larger space is required when higher accuracy is necessary,
because this method requires two optical paths separately.
In recent decades, some straightforward and noninterfero- ⌬ = F tan 共2␪兲, 共1兲
metric methods1–11 have been proposed. As for the mea-
surement of yaw and pitch errors, a 2-D measurement where F is the focal length of the lens, and ⌬ is the dis-
method was used based on a semiconductor laser and opti- placement of the beam on the detector. Considering that ␪
cal units.1,2 Because of ambient effects and electrical noise is a small tilt angle, Eq. 共1兲 can be expressed approximately
errors, however, it is difficult to further improve the accu- as follows:
racy for the measurement of slides with long travel ranges.
In this paper, two sets of combination lenses are used to 共1兲
suppress the angular drift of the reference datum beam and
alleviate the ambient effects due to the larger diameter laser
and 共2兲 increase the measurement resolution and reduce the
installation space. This system makes use of a charge-
coupled device 共CCD兲 instead of a position-sensitive detec-
tor 共PSD兲 for further suppression of background noise.

0091-3286/2007/$25.00 © 2007 SPIE Fig. 1 Principle of angle measurement.

Optical Engineering 051016-1 May 2007/Vol. 46共5兲

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/17/2015 Terms of Use: http://spiedl.org/terms

 Kuang, Hong, and Feng: High-accuracy method for measuring…

Fig. 2 Schematic diagram of the system for measuring 2-D rotation

angles.

␪ = ⌬/共2F兲. 共2兲 Fig. 3 Drift after collimation expanded: 共a兲 angular and 共b兲 lateral
drift after the beam is expanded.
Using a detector to measure the deviation ⌬, it is evident
from Eq. 共2兲 that a high resolution is achieved by a long
focal length of the lens and a high spatial resolution of the 3 Theoretical Analysis of Two Sets of
detector.6 Combination Lenses
3.1 Combination Lenses for Beam Expansion
The beam drift of the system will be affected by the quality
2.2 System Configuration of the reference laser beam. As we know, the pointing drift
Figure 2 shows a schematic diagram of experiment system. of a laser beam can be divided into two parts, namely, the
This system is composed of a fixed part and a moving part. angular drift and the lateral drift 共Ref. 12, pp. 17–21兲.
A beam emitted by a He-Ne collimating laser source be- When a laser beam goes through the focus lens 共5兲 and the
comes linearly polarized light after reflecting from the mir- collimator lens 共6兲, the divergence angle of the laser beam
ror 共2兲 and passing through a polarization beamsplitter 共3兲. and angular drift will be highly suppressed. According to
If we let the angle between the fast axis of the quarter-wave Fig. 3共a兲, when a laser has a drift or divergence angle of ␪1,
plate 共4兲 and the direction of laser polarization be 45 deg, after beam expansion through combination lenses, the drift
when linearly polarized light passes through the quarter- or divergence angle ␪2 will become smaller, and
wave plate, it becomes circularly polarized light. After
passing through the focus lens 共5兲 and the collimator lens
␪2 = 共f 1/f 2兲␪1 . 共3兲
共6兲, the divergence angle of the laser beam and angular According to Fig. 3共b兲, when a laser has lateral drift, the
drifts will be suppressed, hence providing a highly stable lateral drift will expand by f 2 / f 1, that is,
measurement datum. The reflected light from the angle-
detecting mirror 共11兲 of the moving part can be improved d2 = 共f 2/f 1兲d1 , 共4兲
by a factor of 2 compared with the rotational angle of the
mirror itself; after inversely passing through lenses 共5兲 and where f 1 and f 2 are the foci of lenses 共5兲 and 共6兲, respec-
共6兲 again, the measured angle is further amplified 10 times. tively, and ␪1 and d1 and ␪2 and d2 are the angular and
Circularly polarized light becomes linearly polarized light lateral drifts of the ingoing and outgoing beams. Therefore,
after it is transmitted again through the quarter-wave plate if we select a relatively large focal length f 2 and a small
共4兲, but the direction of polarization rotates by 90 deg. focal length f 1, the divergence angle and angular drift will
Therefore, the laser can be reflected only via the polariza- be highly suppressed. Moreover, for angular error measure-
tion beamsplitter 共3兲. The high-resolution information con- ment, the error contributed by the angular drift obviously
cerning the pitch and yaw of the reflected beam is detected becomes the dominant part, and the lateral drift can be
by the CCD detector after passing through the mirror 共7兲, neglected. With the use of a beam expander, this angular
the convex lens 共8兲, and the concave lens 共9兲. The output drift can be reduced by a ratio of f 2 / f 1.
signals of the CCD detector through analog-to-digital
共A/D兲 converters are processed by a computer. From the 3.2 Combination Lenses for Angular Measurement
output signals, the angular displacement of the object is The combination lenses in front of the CCD detector can
calculated. not only improve the measurement resolution, but can also

Optical Engineering 051016-2 May 2007/Vol. 46共5兲

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/17/2015 Terms of Use: http://spiedl.org/terms

 Kuang, Hong, and Feng: High-accuracy method for measuring…

Fig. 4 Measuring angle principle with the combinational lens.

decrease the installation room required as well. The follow-

ing discussion shows how to determine the optimal focal
lengths of the convex lens 共8兲 and concave lens 共9兲 and the
distance between them. As shown in Fig. 4, a conventional
single lens is replaced by a convex lens 共8兲 and a concave
lens 共9兲. The parameters involved in the discussion are de-
fined. Based on the geometric optics, we obtain the follow-
ing equations:

f = f 3 f 4/共D1 − f 3 + f 4兲, 共5兲

D2 = f 4共f 3 − D1兲/共D1 − f 3 + f 4兲, 共6兲

D = D1 + D2 , 共7兲
Fig. 6 Stability of the 共a兲 pitch and 共b兲 yaw before expansion.
where f 3 and f 4 are the foci of lenses 共8兲 and 共9兲, respec-
tively; f is the equivalent focal length of the combination
lenses; D1, D2, and D are the distances between lenses 共8兲
and 共9兲, lens 共9兲 and the CCD detector, and lens 共8兲 and
CCD detector, respectively. From Eqs. 共5兲 to 共7兲, with ap-
propriate f 3, f 4, and D, one can obtain a larger equivalent
focal length f and a small physical size D.

3.3 Analysis of Angular Measurement Resolution

According to Fig. 2, the focal length of lens 共5兲 and lens 共6兲
are 10 and 100 mm. The autocollimator, which is com-
posed of two lenses with focal length f 3 = 50 mm and
f 4 = −6 mm, respectively, was designed to improve the an-
gular resolution in the pitch and yaw measurements. The
pixels of CCD are 6.4⫻ 7.4 ␮m. From the preceding analy-
sis, when the moving part of this system has a pitch of ␣,
the normal line of mirror 共11兲 will rotate an angle of ␣
correspondingly, therefore the reflected laser will rotate an
angle of 2␣. After inversely passing the combination lenses
of lens 共6兲 and lens 共5兲, the measured angle will be ampli-
fied 10 times, according to Eq. 共1兲. When the distance D1
between lens 共8兲 and lens 共9兲 is set to be 45 mm, from Eqs.
共5兲 and 共6兲, the equivalent focal length of combination
lenses 共8兲 and 共9兲 equals 300 mm, the distance between
lens 共9兲 and the CCD detector was 30 mm. Hence, pitch ␣
and D will be

Fig. 5 共a兲 Original beam on the CCD and 共b兲 the beam after being
filtered on the CCD. ␣ = x/共2 ⫻ 10 ⫻ f兲, 共8兲

Optical Engineering 051016-3 May 2007/Vol. 46共5兲

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/17/2015 Terms of Use: http://spiedl.org/terms

 Kuang, Hong, and Feng: High-accuracy method for measuring…

Fig. 7 Stability of the 共a兲 pitch and 共b兲 yaw after expansion.
Fig. 8 共a兲 Results and 共b兲 residuals of the pitch comparison.

D = D1 + D2 = 45 + 30 = 75 mm, 共9兲
gray distribution is nearly a Gaussian distribution, more-
where x is the lateral displacement of the light on the CCD over the light spot noise is eliminated. A special data-
detector. Differentiating Eq. 共8兲, one obtains the resolution processing algorithm was used to reduce the noise. Data
of pitch ␣ to be processing was done automatically by software. By apply-
ing the energy gravity method, the center of gravity of the
d␣ = dx/共20f兲. 共10兲 light spot could be identified; the position stability of the
Similarly, the resolution of yaw ␤ is light spot is further stabilized via mean-value filtering.

4.1 Stability Experiments

d␤ = dy/共20f兲. 共11兲
To determine the measurement accuracy, we performed a
Applying the image processing method, one can determine stability test. When the distance to the target is 2000 mm,
the resolution of the CCD detector with respect to the light the results of the stability experiment are as shown in Figs.
to be in the subpixel range, dx = dy = 0.1 ␮m, and substitut- 6 and 7. From Figs. 6共a兲 and 6共b兲, before beam expansion,
ing into Eqs. 共10兲 and 共11兲, one obtains the 2-D angle mea- the pitch angular drift is from 0.9 to 3 . 2⬙ and the yaw
surement resolution to be 0 . 003⬙ in theory. It is evident angular drift is 0.4 to 2 . 5⬙ in 2 min. Moreover, the standard
from the preceding analysis that a higher angular resolution deviations of the pitch and yaw are 0.4 and 0 . 3⬙, respec-
is achieved by a longer equivalent focal length of the com- tively. With the use of a beam expander, which consists of
bination lenses f, larger magnitude of f 2 / f 1, and higher lens 共5兲 and lens 共6兲, this angular drifts can be reduced by a
spatial resolution of the detector. ratio of 10. The experimental results are illustrated in Figs.
7共a兲 and 7共b兲, which show that the drifts of pitch and yaw
4 Experiments and Discussion are ±0.15 and ±0 . 14⬙, respectively, and the standard devia-
To validate the feasibility, stability, and accuracy of this tions are about 0 . 06⬙ during a period of 50 min.
system, a series of experiments were performed under the
laboratory conditions. The light spot received by the CCD 4.2 Comparison Experiments
detector reflected from the target mirror is shown in Fig. Figure 8 and 9 show the comparison results with Renishaw
5共a兲; the distribution is not very smooth due to the light ML10 laser interferometer system when the measure-
diffraction, stochastic vibration, and electronic noise. After ment distance is 2000 mm. The sensitivity of ML10
passing a low-pass filter and setting the gray threshold be- system is 0 . 01⬙ and measurement accuracy is
tween 127 and 225, one obtains Fig. 5共b兲. In this figure, the 共±0.6% R ± 0.1± 0.02D兲 arcsec, where here R is the reading

Optical Engineering 051016-4 May 2007/Vol. 46共5兲

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/17/2015 Terms of Use: http://spiedl.org/terms

 Kuang, Hong, and Feng: High-accuracy method for measuring…

Fig. 9 共a兲 Results and 共b兲 residuals of the yaw comparison.

and D is the measurement distance. Figure 8共a兲 shows the

comparison results when pitch is within ±60⬙; the linear Fig. 10 Comparison 共a兲 results and 共b兲 average residuals of the
correlativity is 0.99999. And the point-to-point error range pitch between the proposed system and Renishaw ML10 system for
is from −0.20 to 0 . 33⬙ and standard deviation is about three runs.
0 . 19⬙, which is shown in Fig. 8共b兲. Figure 9共a兲 shows the
comparison results when yaw is within ±60⬙; the linear
correlativity is also 0.99999. In Fig. 9共b兲, the point-to-point 1500-mm distance, respectively. The standard deviation er-
error range is from −0.22 to 0 . 19⬙, standard deviation is rors of the pitch and yaw direction are 0.43 and 0.25 arcsec,
0 . 18⬙. From the experimental results, the standard devia- respectively.
tion of the 2-D angular measurement method proposed in
5 Conclusion
this paper is better than 0 . 2⬙. Good repeatability was also
found in comparison experiments. We applied a combination lens to collimate a laser beam.
This proposed novel method is able to improve the laser
temporal and spatial stability, suppress the angular drift due
4.3 Validation Experiments to thermal drift and vibrations of the reference datum beam,
In a laboratory environment, measurements were per- and alleviate the ambient effects on the measurement. By
formed by moving the stage along a linear axis for a dis- using CCD as the detector and an image-processing
tance of 1500 mm, which was divided into 15 steps for method, the energy center of the laser light can be identi-
three runs. For comparison, the Renishaw ML10 laser in- fied, thus reducing the measurement noise and fulfilling the
terferometer was also used for the pitch and yaw measure- long-distance, high-precision and real-time 2-D angle mea-
ments. In Figs. 10共a兲 and 11共a兲, the experimental results surement. The experimental results show that the linear cor-
show that the pitch and yaw measured by the proposed relativity is 0.99999 and the standard deviation the point-
system are consistent with those measured by the Renishaw to-point error is better than 0.2 arcsec within a ±60 arcsec
system. The average residuals errors of three runs measure- range compared to the Renishaw ML10 laser interferometer
ment of the two measuring system are shown in the Figs. system. If a single-model fiber collimating laser is chosen
10共b兲 and 11共b兲. The average residuals of the pitch and yaw as the light source, the measurement stability and accuracy
are within the range of ±0.62 and ±0.45 arcsec over the could be further improved. The system proposed in this

Optical Engineering 051016-5 May 2007/Vol. 46共5兲

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/17/2015 Terms of Use: http://spiedl.org/terms

 Kuang, Hong, and Feng: High-accuracy method for measuring…

2. J. S. Yeung, “Portable autocollimators using the laser diode and the

position sensitive detector,” Rev. Sci. Instrum. 69共2兲, 402–405
共1998兲.
3. T. R. Armstrong and M. P. Fitzgerald, “An autocollimator based on
the laser head of a compact disc player,” Meas. Sci. Technol. 3, 1072–
1076 共1992兲.
4. S. Kiyono and Y. Asakawa, “A differential laser autocollimation
probe for on-machine measurement,” Precis. Eng. 15共2兲, 68–76
共1993兲.
5. J. Yuan and X. Long, “Temperature-autocollimator with ultrahigh an-
gular measuring precision,” Rev. Sci. Instrum. 76共12兲, 125106 共2005兲.
6. J. Yuan and X. Long, “CCD-area-based autocollimator for precision
small-angle measurement,” Rev. Sci. Instrum. 74共3兲, 1362–1365
共2005兲.
7. A. Just and M. Krause, “Calibration of high-resolution electronic
autocollimators against an angle comparator,” Metrologia 40, 288–
294 共2003兲.
8. C. Zhang and X. Wang, “Sinusoidal phase-modulating laser diode
interferometer for measuring angular displacement,” Opt. Eng.
43共12兲, 3008–3013 共2004兲.
9. D. B. Leviton, “Ultra-high resolution, absolute, Cartesian electronic
autocollimator,” Proc. SPIE 5190, 468–475 共2003兲.
10. Y.-Y. Zhang, M.-H., Zhang, and Y.-F. Qiao, “Research of CCD laser
collimation system with high precision,” J. Optoelectron., Laser
14共2兲, 168–170 共2003兲 共in Chinese兲.
11. T. Suzuki, “Two-dimensional small-rotation-angle measurement us-
ing an imaging method,” Opt. Eng. 45共4兲, 043604 共2006兲.
12. P. Huang, “Laser optical measurement systems and their application
to the on-line error compensation of coordinate measurement ma-
chines,” PhD Dissertation, University of Michigan, Ann Arbor
共1993兲.

Cuifang Kuang received his BS degree

from the Department of Physics, Jiangxi
Normal University, China, in 1999 and he is
currently working toward his PhD degree at
Beijing Jiaotong University, China. He has
published more than 20 journal and confer-
ence papers in the area of laser and opto-
electronic measurement and his current re-
search interests are optoelectronic sensors
and optical measurement technology.

En Hong received his PhD degree in optical

Fig. 11 Comparison 共a兲 results and 共b兲 average residuals of the yaw science and engineering from the University
between the proposed system and Renishaw ML10 system for three of Alabama at Huntsville in 2005 and is a
runs. research fellow with the University of Michi-
gan, in the Engineering Research Center. In
1998 he joined the Optical Pattern Recogni-
tion Applications under the National Scien-
paper could be used to measure the 2-D angles of slides tific Research Programs of China and later
within a long travel range and to identify the on-line 2-D received the National Science and Technol-
ogy Award for Real Time Image and Signal
rotational angles of moving targets. Processing Technology. His research inter-
ests are optical alignment systems used in high-accuracy measure-
ment, electro-optical slip rings, and optical pattern recognition.
Acknowledgments
The author wishes to express deep appreciation to Dr. Jie
Zhu for his support during the measurement. The research Qibo Feng received his BS and MS de-
grees from Hefei University of Technology,
was supported by the doctoral innovation foundation of China, in 1983 and 1986, respectively, and
Beijing Jiaotong University with Grant No. 48003 and the his PhD degree in optical engineering from
National Natural Science Foundation of China under Grant Tsinghua University, China, in 1993. Since
No. 50675017. 1993 he has been an associate professor
and a professor with School of Science,
Beijing Jiaotong University, where he is cur-
rently, dean of the School of Science. He is
References a member of SPIE, the American Society of
1. C.-F. Kuang and Q.-B. Feng, “Development of equipment for simul- Precision Engineering, the Chinese Optical
taneously measuring two-dimensional rotation angles of the linear Society, Chinese Mechanical Engineering, and the Chinese Railway
guide track,” J. Optoelectron., Laser 15共8兲, 965–968 共2005兲 共in Chi- Society. His research is focused on optical measurement technology
nese兲. and nondestructive testing.

Optical Engineering 051016-6 May 2007/Vol. 46共5兲

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 05/17/2015 Terms of Use: http://spiedl.org/terms