The Chesapeake Laser Tracker in Industrial: Metrology
The Chesapeake Laser Tracker in Industrial: Metrology
The Chesapeake Laser Tracker in Industrial: Metrology
Robert E. Ruland
Stanford Linear Accelerator Center
Stanford University
0. Abstract
In the summer of 1992, the survey and alignment team at the Stanford Linear Accelerator
Center acquired a CMS3000 laser tracker manufactured by Chesapeake Laser Systems in
Lanham, Maryland. This paper gives a description of the principles of operation and
calibration of the tracker. Several applications are explained and the results shared.
1. Introduction
Until recently, the determination of coordinates in industrial metrology applications has
usually required two instruments or two set-ups. In the most common triangulation
approach, horizontal and vertical angles measured from at least two stations are necessary
to determine object coordinates. The polar method which provides sufficient parameters
from only one station was rarely used because prior to State-of-the-Art the laser tracker,
two different instruments were required to gather the necessary observables, one
instrument for angles, the other for distances. Traditional total stations did not meet the
accuracy requirements of most industrial metrology applications. Promising solutions were
1
not developed through to a commercial product. The laser tracker is now a high accuracy
total station combining horizontal and vertical angle measurements with interferometric
distances, thus providing for real-time coordinate determination.
POWER
IT
The concept of laser trackers goes back to independent studies and efforts undertaken at
2 3
the National Bureau of Standards (now NIST) , at Chesapeake Laser Systems , at the
4
University of Surrey, England and at the Fraunhofer Institut für Informations- und
5
Datenverarbeitung . From these studies, two products have evolved: the Leica SMART310
and the Chesapeake CMS3000.
The material below is based on the Chesapeake CMS3000 tracker, but in a general
technical sense, it is applicable to the Leica tracker as well.
2. System Description
The Chesapeake Laser Systems (CLS) Coordinate Measurement System CMS-3000 (see
Fig. 1) is a servo-controlled tracking laser interferometer measuring tool. The tracker
follows a retroreflective target, providing real-time coordinate information of the target
center location. A laser interferometer measures the linear distance to the target, and two
angular encoders measure the azimuth and elevation angles of the laser beam. This gives
the position of the target in spherical coordinates which are converted to Cartesian
coordinates by the system computer. The tracker system comprises of 2 major components:
the measurement head, and the controller with system software. The remote power unit
and several accessories complete the tool list.
passes straight through the interferometer optics to the servo mirror and hits the
retroreflector. After reflection, the beam returns to the servo mirror and from there through
a beamsplitter to the interferometer where it merges with the interferometer reference
beam. The beamsplitter redirects 15% of the signal to a position sensor. If the laser beam
strikes the retroreflector in its center, the reflected beam in a correctly calibrated system
will focus on the zero position of the position sensor. Moving the retroreflector causes the
reflected beam to move away from the center position. The position sensor measures the
new beam position and translates the offsets into steering signals for the servo motors such
that the laser beam will strike the retroreflector and consequently the position sensor in the
center again. This correction process takes place up to 1000 times per second. The sensor
resolution translates into a 1 arc second angular correction resolution at 5m range. The
measurement head has a field of view of ±60º vertically and ± 135º horizontally. The
mechanical arrangement of the axes permits plunge and reverse measurements without
breaking the beam.
The resolution of the angular encoders is 1 arc second. It should be -pointed out that this
angular resolution applies not only to the encoders and to the mechanical shafts they are
mounted on, but applies directly to the horizontal and vertical angles of the beam itself
This is possible because the unique optical design of the prismatic beam steering
mechanism assures that the beam rotates in a 1: 1 relationship with both the encoders rather
than the 1:2 relationship of a mirror based beam steering.
Option Resolution Velocity The interferometer is of the single beam type. It uses a
helium-neon laser with 632 nm wavelength. The
maximum tracking speed at the highest resolution is
1.06 m/sec. The tracking speed is inversely correlated
to the resolution (Table 1). Additional optical hardware
c a n e x t e n d t h e r e s o l u t i o n t o 0 . 0 7 9 pm. T h e
Table 1: Tracking Speed vs Resolution measurement range is 0.17 m to 40 m.
Mounted to the measurement head is the re-set or home point. The re-set point serves two
purposes. Firstly, it provides a known offset distance from the inaccessible effective
tracking head axis to an easily accessible reference point. The offset distance is used to set
the interferometer’s initial distance. Secondly, it provides for a convenient reference point
on which to close measurement loops or from where to re-start an interrupted
measurement
2.2 Controller with System Software
Data management and tracker control are executed on the same computer. The tracker
system operates on a 286 or higher, MS-DOS based computer. One expansion slot is
required to mount the proprietary communication interface card. 486 Portables with VGA
color screen are preferred choices. TCHost is the standard DOS-based tracker system
operating software. The program contains all of the functions required to operate the
tracker and to collect data. The user views a screen which shows the x, y, z Cartesian
coordinates of the target in real time, and data is saved to a file in response to a function
key or remote push button. TCHost allows for the storage and export of data files in
standard ASCII format which can be imported into most commercial spreadsheet
I/104
programs, and in partial DXF format, which can be imported into commercial CAD
software packages. The user can add labels and comments to the file during the data
collection session for easy future point reference. Any beam interruptions, resets or error
flags are automatically written to the file, giving an active history of the data collection
session. A typical data file will contain the following information for each point.
Data index,
X, Y, Z Cartesian coordinates in user coordinate system and units,
Azimuth, Elevation angles in user selected units,
Interferometer distances, both absolute and relative in user selected units,
Tracker serial number and system number,
File header name, time and date when opened.
TCHost contains a command language which can be used to program the tracker to
perform precise pointing, search patterns, change environmental compensation parameters
or display diagnostic data. TCHost is also available as a function library. Based on this
TM
library, we have built our own front end running under Windows . Fig. 4 shows a typical
screen during the measurement process.
2.4 Accessories
Because the servo loop needs to receive part of the reflected beam, a suitable reflector is
necessary as target. Two kinds of retroreflector are commonly used:
• Magnetic Mount for SMR
In order to provide truly forced center position for the SMR in all three coordinate
axes, a magnetic centering nest with a three point kinematic mount keeps the center of
the SMR in the same accurate position (see Fig. 5). The kinematic mount is realized by
either three conically shaped resting pads or by three pins. The position of the SMR in
a centering nest is independent of the direction in which it is pointing. Centering errors
do not exceed 5 p.
• Spherically Mounted Retro-Reflector (SMR).
Three mirrors are mounted orthogonally inside a 1.5” sphere. The standard SMR has a
finished sphericity of less than 5 p rms. Mechanical centering of the assembly is
within 10 w rms. The advantage of this type of mounting arrangement is significant.
Since the outside surface is a sphere, there is always a known and constant offset
between the actual point being measured (the center of the sphere) and the part surface
in contact with the outside of the ball. The SMR field of view is approx. ±20º. Fig. 6
shows a picture of a SMR sitting in a forced centering mount. The total centering
error budget using different spheres in the same nest is approx. 12 p.
rotated. This error can be determined by the simple procedure of plunge and reverse
measurements of a series of fixed targets.
4. Mechanical runout: If the bearings on which the beam steering assembly rides are
improperly aligned or are of insufficient quality, mechanical runout or bearing wobble will
result. The plunge and reverse measurements will reveal if this is a problem, but it can be
difficult to compensate for.
unit with changes in the ambient temperature affect this distance to an amount measurable
by the interferometer itself. The procedure to determine the reset mount position can be
used to measure the actual condition.
4. Path Length Runout: Slight misalignments of the beam with respect to the axis of the
beam steering assembly will result in a slightly longer or shorter optical path length as the
beam steering assembly is rotated.
5 . Coincidence of axes: The horizontal and vertical axes are constrained to intersect at
the pivot point. Assembly techniques and subsequent precision measurement assures that
this tolerance is held to within a few microns.
6 . Location of the reset mount: The tracker
mounted reset point is the alternative reference
in lieu of the inaccessible true zero point, the
intersection of the horizontal and vertical axes.
The reset mount location (see fig. 8) may be
determined by a simple field procedure.
evenly around the horizon are measured in both positions, plunge and reverse. Additionally,
every target is also measured with a Sphere Mounted Mirror (SMM); this observation type
yields a very accurate collimation measurement. This measurement sequence produces a
total of 80 observations, which are then used to solve for the 14 unknown calibration
condition parameters.
These 14 parameters, some basic constants, and the interferometer offset value are stored
in the tracker configuration file.
Once the calibration parameters are known they can be used to correct the raw
observations. Inverting the mathematical model equation yields the correction equation.
5. Coordinate Measurements
Trackers, in their present designs, cannot be used in a forced centered set-up. The basic
tracker observables are the azimuth [Az] and elevation [El] angles and the radial distance
[d] (see fig. 9). Furthermore, the radial distance in effect is only the radial difference from
the previous point (see fig. 10). At the beginning of a measurement set this is the tracker
reset position or afterwards the last measured point. The tracker control software takes the
raw readings, applies all the required offsets and corrections, and converts the spherical
coordinates into Cartesian coordinates in the local tracker coordinate system. It is possible
to intercept the corrected spherical parameters. The tracker has been interfaced to our
Stanford Industrial Measurement Software (SIMS). This allows us to integrate
measurements from different stations into one least squares bundle solution, and also to
combine additional theodolite observations, distance measurements, leveling data and
hydrostatic height differences into the solution.
In preparation for a measurement session, the tracker should be set up close to the object in
the most appropriate vertical orientation; it does not need to be leveled. For measurements
in the micron domain the floor must be absolutely stable. Before starting the measurement
sequence, it should be verified that all points can be reached without breaking the beam. If
necessary, offset bars should be used. We find that a scale bar should always be integrated
into the set-up.
I/110
All tracker measurements should be started from the re-set position. This will set the
absolute distance and lock the laser beam on to the SMR. Now the SMR is moved to the
first object point with the beam tracking. For point measurements the SMR is centered on
to a point by e.g. setting it into a magnetic mount. There, it should be carefully pointed
towards the tracker. To randomize the SMR internal systematic alignment residuals, the
SMR should be rotated around the beam axis without lifting out of the mount in about 4
increments while taking readings in each position. If the laser beam is interrupted, the
interferometric fringe count is lost, and measurements must be re-started from the reset
point. It is possible to define any object point as reset point.
For dynamic continuous measurements, the data rate needs to be selected and start/stop
commands will control the measurement sequence.
6. Measurement Results
The following results have been obtained with a CMS-3000 tracker with was updated to
the latest revision in March/April this year. The modifications yielded a significant
improvement in measurement accuracy.
Time [h:min]
Distance [mm]
redundancy. Fig. 18 shows the coordinate differences between the first and the third data
set. The histogram in Fig. 19 shows more directly that the majority of the resulting
coordinates are less than 20 p. From this we may deduct that the measurement precision
is in the 20 p range.
The bundle adjustment did not show any abnormal results, the normalized chi-square test
yielded a value very close to one. Fig. 21 shows a graph of the coordinate differences
between the survey 1 of test 1 and this new survey 6. Fig. 22 again shows a histogram Fig.
plot of the coordinate differences and the normal distribution at a 95% probability level.
From the histogram we can see that the differences are normally distributed, indicating that
the data is not systematically biased. From these results we can deduct that careful laser
tracker measurements can yield coordinate accuracies in the 30 w domain. However, we
must note that a considerable number of points may carry offsets three times that amount.
If these points cannot be tolerated, the observation plan and geometry has to be changed
accordingly.
7. Conclusion
The above described tests represent only a small subset of all the tracker surveys which
have been performed over the last year at the Stanford Linear Accelerator Center to
familiarize our survey staff with this new tool and to qualify itscapabilities. We feel that we
now can routinely achieve results in the 100 ~U+Z range and with appropriate care and
planning perform surveys at the 50 pm level. In this regard, the laser tracker has opened a
domain which was in most cases not accessible with theodolite systems. The tracker also
brought an increase in productivity. What used to take with a theodolite system two
experienced theodolite operators one shift takes with a tracker 75% of a shift for one
operator and a helper. However, all of the above does not mean that the theodolite systems
have become obsolete. They might not longer be State-of-the-Art, but on the other hand a
tracker requires the equivalent investment of two theodolite systems.
The Chesapeake tracker has seen almost constant improvement over the last years.
Fortunately, we were able to have our model upgraded to the most recent version. But, as
e.g. the plunge and reverse measurements indicate, there is still ample room for further
I/118
improvements. Axis alignment conditions, thermal stability and weight are very high on our
wish list. As a long range goal, I wish to see a substitution of the interferometric distance
measurement with a method which can tolerate an interrupted beam.
ACKNOWLEDGMENTS
It is a pleasure to acknowledge the many helpful discussions with Larry Brown, Scott
Ackerson and their team at Chesapeake Laser Systems, the tremendous effort in tracker
programming by Robert Pushor, Mike Gaydosh’s support in running the data analysis,
Bernard Bell’s helpful suggestions and careful proof-reading, and the careful field
measurements by our field personnel, especially by Hans Imfeld and Levirt Griffin.
REFERENCES
[1] Greve, A. and Harth W. (1984). “Laser-diode distance meter in a KERN DKM 3A
theodolite”, Applied Optics, Vol. 23, No. 17, pp. 2982-2984.
[2] Lau, K., Hocken, R., and Haight, W. (1986). “Automatic Laser Tracking Interferometer
system for Robot Metrology”, Journal of Precision Engineering, Vol. 8 No. 1, pp. 3-8
[ 3 ] Brown, L.B., Merry, J.B., and Wells, D.N. (1986). “Coordinate Measurement With a
Tracking Laser Interferometer”, Laser & Applications, Oct.86, pp. 69-71.
[ 4 ] Gilby, J.H., and Parker, G.A. (1982). “Laser Tracking System to Measure Robot Arm
Performance”, Sensor Review, Oct.82
[ 5 ] Lasertriangulations-Meßsystem zur genauen dynamischen 3D-Bewegungsanalyse”,
Fraunhofer-Institut für Informations- und Datenverarbeitung”, Karlsruhe, 1989
[ 6 ] CMS-3000 Manual: Description of a CMS-3000 system, Chesapeake Laser Systems,
Lanham, 1990
[ 7 ] In this context, horizontal and vertical do not necessarily refer to the direction of the
vertical. Rather, they apply to the orientation of the local coordinate system in which the
tracker is set-up.
[ 8 ] Ruland, R., Fischer, G. (1990). “The proposed Alignment system for the Final Focus
Test Beam at SLAC”, Proceedings of the Second International Workshop on
Accelerator Alignment, Hamburg, pp. 4 19-453