Li Fu, Guoxi Liu, Chao Jin, Feng Yan

WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

Surface Accuracy Analysis of Single Panels for the Shanghai 65-M

Radio Telescope
LI FU
Shanghai Astronomical Observatory, Chinese Academy of Sciences, Shanghai
CHINA
Key Laboratory of Radio Astronomy, Chinese Academy of Sciences, Nanjing
CHINA

GUOXI LIU, CHAO JIN, FENG YAN

54th Research Institute of China Electronics Technology Group Corporation, Shijiazhuang
CHINA

TAO AN, ZHIQIANG SHEN

Shanghai Astronomical Observatory, Chinese Academy of Sciences, Shanghai
CHINA
Key Laboratory of Radio Astronomy, Chinese Academy of Sciences, Nanjing
CHINA

fuli@shao.ac.cn

Abstract: - We presented the surface accuracy measurements of 5 single panels of the Shanghai 65-meter radio
telescope by employing the coordinate measuring machine and laser tracker. The measurement data obtained
from the two instruments were analyzed with the common point transformation and CAD surface fitting
techniques, respectively. The derived rms uncertainties of panel accuracy from two methods are consistent with
each other, and both match the design specification. The simulations of the effects of manufacturing error,
gravity, temperature and wind on the panel surface accuracy with the finite element analysis method suggest
that the first two factors account for primary sources of the accuracy uncertainty. The panel deformation under
concentrated load was analyzed through finite element analysis and experiment, and the comparison error is
5.6%. There is not plastic deformation when people of weight below 70kg installs and remedies the panel.

Key-Words: - antennas, finite element methods, measurement, accuracy, panel

1. Introduction small trapezoidal shaped panels (on average, 3.9m2

The surface of a large millimeter-wavelength each) and the average accuracy of individual panels
radio telescope has to be measured and calibrated to is 68m [1].
a very high accuracy in order to guarantee a high The Effelsberg 100-meter radio telescope works
efficiency. The largest solid-panel radio telescopes in the frequency range from 0.3 to 95.5GHz. The
in the world, the Robert C. Byrd Green Bank primary reflector consists of 2353 panels and the
Telescope (GBT) of the US, the Effelsberg radio overall mirror surface accuracy is slightly less than
telescope of the Germany, the Sardinia Radio 0.5mm. The inner cycles (diameter < 60m) are made
Telescope (SRT) of the Italy, and the Shanghai 65- of aluminum honeycomb panels, and the rms (root
meter Radio Telescope (in brief, Sh65RT hereafter) mean square) deviations of this honeycomb-type
of the China, are all composed of high-accuracy sector give a mean value of 0.22mm. The
panels. intermediate diameter range of 60-85m is composed
The GBT is a 100-meter diameter offset- of aluminum cassette panel and the rms is
paraboloid Gregorian reflector radio telescope 0.27mm. The outer rings (diameter > 85m) are made
operating in the frequency range from 290MHz to up of stainless steel mesh with thickness of 6 mm2,
100GHz. The primary reflector consists of 2004

E-ISSN: 2224-3429 54 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

and the surface deviations are up to 0.55mm (rms) placed in a room with constant temperature (20C)
[2]. for longer than eight hours.
The SRT is a shaped Gregorian 64-meter radio
telescope recently completed in Sardinian Island in
Italy. The frequency band continuously covers from
300MHz to 100GHz. The primary mirror composed
of 1008 panels distributed on 14 rings [3]. The
manufacturing error of individual panel is about
65m, the thermal-induced error 11m, the gravity-
induced error 29m and the wind-induce error 4m
[4].
GBT and Effelsberg radio telescopes have
achieved a lot of profound-impact scientific results
[5,6] and SRT is conducting commissioning
observations [7]. All these advanced sciences
depend on high-sensitivity observations which in
turn rely on the large high-accuracy reflectors. Fig. 1 The 3D model of Sh65RT
Although the generic antenna structure design [8],
overall reflector accuracy and alignment of panels First we used coordinate measuring machine
[9] are important for maintaining the perfect (CMM) for the panel surface accuracy measurement.
parabolic surface, the surface accuracy of single According to the dimension of the single panel and
panels is also of essential importance. In this paper, accuracy required, ALPHA IMAGE 25.50.18 CMM
we report the surface accuracy measurements of was adopted. The work environment temperature of
single panels of the Sh65RT. The two different this instrument is 20C and the precision is
measurement methods and the data processing (8.0+8.0L/1000)m.
methods are described in section 2. In section 3, the
measurement data combined to the finite element
analysis (FEA) data are used to calculate the rms
value of the manufacturing error using the common
point coordinate transformation and CAD surface
fitting techniques. The effects of manufacturing
error, gravity, temperature and wind on panel
precision are evaluated with the FEA method. We
also discuss the effect of concentrated load on panel
deflection in section 4.

2. Measurements and Data Analysis of

Fig. 2 The measurement of panel under CMM.
Panel Surface Accuracy
2.1 Measurement methods Ideally, we set four panel corners to the
designed paraboloid by adjusting the supporting
The panel surface accuracy of Sh65RT was
screws near each corner so as to separate the panel
measured independently with two different
error from the backup structure (BUS) error. But, a
instruments and techniques, the coordinate
different way to accept a panel is to measure the
measuring machine and laser tracker.
panel in the factory. We need to duplicate what we
2.1.1 Coordinate measuring machine define above. First, we established the panel
The primary reflector of the Sh65RT showed in coordinate system. Reference surface is defined by
Figure 1 consists of 1008 panels distributed in 14 four corner points close to the installing holes. The y
rings. The targets to measure are from different axis is perpendicular to the reference surface. One
sectors of the shaped reflector surface and the side of the panel was projected on the reference
largest area is about 5m2. As shown in Figure 2, surface to form the z axis. The coordinate origin is
the panel is formed by steel frame supporting the set to a point on the line connecting z axis and y axis.
aluminum sheet, and is placed on four supports. Next, we switched the reflector coordinate system to
Before the measurements, the panels have been the panel coordinate system. The four corner points
were repeatedly measured and adjusted to ensure

E-ISSN: 2224-3429 55 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

that all four corners fell at y=0. Finally, according to model of the panel produces the deviations of x,
the preassigned points on the ideal model, operators y and z and the error was calculated with (1).
manually controlled the CMM to determine the x 2 y 2 z 2 (1)
relative positions of other measuring points.
The accuracy of the panel surface is expressed
2.1.2 Laser tracker with the rms value of the measuring errors at each
Laser tracker is a portable precision metrology point. That is
tool enabling to achieve accuracy as high as 5m/m.
The T3-60 system was adopted in our measurements. (2)
It composes of a laser tracker and a spherically
mounted retroreflector which emits the laser beam. where i is the measuring error of each sampling
As shown in Figure 3, the same panels used in point, the average value, n the number of the
above experiment were measured by the laser points.
tracker under the same work environment. The
positions of spherically mounted retroreflector are Panel
the measuring points which obey the principle of
distribution of equal areas and the distance of two Measuring instrumentation
consecutive points is between 50mm and 100mm. CMM Laser tracker
Firstly, we set reasonable position between the Data processing method
laser tracker and panel. After the instrumentation
turned on for half an hour, the panel can be Common point transformation CAD surface fitting
measured. The environment temperature and Advantage and disadvantage
pressure during preheating were recorded. The four and are larger and are smaller
corner points were also repeatedly measured and
adjusted to ensure the corner points lie on a same ideal status repeatedly adjust
plane. More than 400 points have been measured.
Fig. 4 Illustration of the measurement and data
processing method for the individual panel.

The data process of the laser tracker made use of

the CAD surface fitting method [11]. The CAD
surface model and measuring data were compared
and adjusted in order to determine the approximate
coordinate transformation parameters. Once these
transformation parameters were found, the software
calculated automatically and stopped until the
square sum of the distance between measuring point
and its projection reaches the minimum. This
method does not solve the errors of four corner
points but is sensitive to the minimization of the
square sum of the errors.
Fig. 3 The measurement of panel under Laser Tracker.
Normally the CAD surface fitting method places
the origin of the measuring coordinate system in the
mirror or panel center. If one wants to obtain the
2.2 Data processing
results associated with four corner points, the
Figure 4 displays the flow chart of the data
following steps are adopted:
processing. The data measured from the CMM and
(i) establishing the panel coordinate system
Laser track were analyzed separately. In the
according to three corner points;
common point coordinate transformation [10], the
measurement data obtained from the CMM were (ii) transforming the measuring coordinate system to
converted from the panel coordinate system to the panel coordinate system;
theoretical coordinate system. The reference points (iii) converting the panel coordinate to theoretical
are the four corner points and their positional errors coordinate system;
are approximately zero. Comparison between the (iv) calculating the error of each measuring point
measurement values obtained by a CMM and a solid and rms value.
As shown in Figure 5, oxyz and o'x'y'z' represent

E-ISSN: 2224-3429 56 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

the measuring and panel coordinate system, We specially investigated the 2-dimensional
respectively. The three corner points are P1(x1,y1,z1), distribution of the measuring errors in the panel 11-
P2(x2,y2,z2) and P3(x3,y3,z3). The x' axis is defined by 21 (Figure 7). Figure 7(a) shows the CMM
the unit vector of P1P2. The z' axis is defined by the measurement errors. It is obviously that in most
cross product of vectors P1P2 and P1P3. In a similar parts i is rather smooth in the range of0.05mm.
way, the y' axis (P1P4) is also established The large errors (denoted with red color) only
x x y y z z appear at a few isolated positions, and the largest
x 2 1 x 2 1 y 2 1 z P x P y P z (3) xx xy xz
l0 l0 l0 one in the left edge. Figure 7(b) shows the
As a result, the transformation matrix from distribution of laser track errors. It agrees with the
measuring coordinate system to panel coordinate CMM errors in Figure 7(a) in a gross manner. The
system is given by most significant errors are concentrated close to the
x Pxx Pxy Pxz x x1 x x1 left edge.

Pyz y y1 A y y1
(4)
y Pxy Pyy
z Pzx Pzy Pzz z z1 z z1 0.10
CMM
0.08
0.06 LaserTracker
o
x 0.04

mm 0.02
P2 CMM
0.00
o x -0.02
y z P1
-0.04
-0.06

z y P3 -0.08
P4
-0.10
01-09 05-33 06-28 11-21 14-18
Panel number
Fig. 5 Schematic of the measuring coordinate system
oxyz and panel coordinate system o'x'y'z'. Fig. 6 The RMS value and error average value of
five panels. CMM are shown as Figure 7 and LaserTracker are
zero.
2.3 Results and conclusions
All together five panels from different rings Comparison between Figure 7(a) and 7(b) gives
have been measured using the CMM and Laser a clue that the maximum of the absolute value of
Tracker. The rms uncertainty and mean derived from the CMM is in general larger than
measurement error are shown in Figure 6. We those from the laser tracker. A possible reason is
should note that includes both the manufacturing that the CMM contains a systematic error of the
and gravity-induced errors. The absolute values of
installation of the reference points. The whole
CMM of five panels are in the range of 0.01~0.08mm.
reflector surface errors include single panel error,
The average value of LT is approximately zero since BUS error and installation error. Although the
the CAD surface fitting ensures the square sum of errors CMM and CMM are relatively larger, the
error to be minimized, therefore LT is not shown in condition in the factory is consistent with that on the
Figure 6. The rms uncertainties LT and CMM of BUS. Only the installation error needs to be
the five panels were calculated from (2), ranging adjusted if the single panel and BUS have been well
between 0.07 and 0.09mm. The deviations between calibrated. We caution that the overall errors LT is
two measurements are rather small, indicating that smaller, but the condition of LT measurement in the
both measurements satisfy the design specification factory is not same with that on the BUS. In
(s<0.10mm: 1st-10th ring; s<0.13mm: 11th-14th addition to the installing error, the adjustment has to
ring), and either method is applicable for measuring take into account of panel error. Consequently, the
panel surface smoothness with high accuracy. The CAD surface fitting method makes the panel
largest deviation ((LT-CMM)/LT~12.5%,) is installation/adjustment more difficult to achieve the
associated with panel 11-21, probably resulted from same setting accuracy of the telescope.
large local mechanical errors in certain positions.

E-ISSN: 2224-3429 57 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

seam in the sheet and stiffeners were neglected

when constructing the finite element model.
Figure 8 shows the contour of gravity-induced
deformation under 90 elevation angle. The
deformation accords with the deformation law of
structure under uniform load that the profile is a
parabola. The deformations in the middle are largest
and the ones at the two edges are minimal. The
surface accuracy of the panel is about 0.056mm.
The surface accuracies of the panel under
elevation angle of 0, 15, 30, 45, 60, 75 and
90are obtained. The results are shown as Figure 9
The rms value increases with the changing of
(a)
elevation angle. Combining the measuring data with
simulated data, we achieve that the manufacturing
error is 0.065mm.

(b)
Fig. 7 (a) The panel deformation calculated through the
common point coordinate transformation and (b) the
CAD surface fitting method. The units of the horizontal
and vertical coordinate axis are mm. Fig. 8 The contour of gravity deformation under 90
elevation angle.

3. FEA of Panel Accuracy 0.06

The surface unevenness of panels under gravity,
wind and temperature was conventionally analyzed 0.05
by the finite element software. The analysis of
gravity-induced deformation can deduce the 0.04
manufacturing error from the measuring data.
Gravity(mm)

The materials of the panel which are built using 0.03

sheet glued in two longitudinal and several
transversal z-shaped stiffeners are aluminum. The 0.02

elastic module is typically 7.0107Pa, the Poissons

0.01
ratio is 0.33, and the density is 0.0027kg/m3. The y
direction displacements of the four supporting 0.00
points were constrained. According to the structure 0 10 20 30 40 50 60 70 80 90
and deformation characteristics, shell63 and Elevation angle (degree)
beam188 elements were adopted for the finite
element analysis. When we meshed we should
Fig. 9 The change of panel surface accuracy derived
ensure the sheet and stiffener have common nodes
from gravity deformation at different elevation angles.
on the intersected line. The sizes of the glue
between the sheet and z-shaped stiffeners and the Temperature shift of 5C was applied on the

E-ISSN: 2224-3429 58 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

panel at the elevation angle 90. The thermal

deformation contour is shown as Figure 10. The
deformation is uniform within the reflector surface
and only becomes larger in the vicinity of the edge.
The uncertainty of the panel surface accuracy
induced by temperature change is 0.011mm. There
are some arguments that the thermal-induced
deformation is the case of the above FEA. In fact,
the panel and the BUS are made of different
materials (aluminum and steel). When the panels are
constrained by the BUS, a uniform temperature
change of the BUS and the panels causes panel
buckling. The buckling amplitude zmax is derived
by Christiansen & Hgbom[12]
zmax 0.6l p BUS T
12
(5) Fig. 11 The contour of wind-induced deformation under
where l is the panel length; p and BUS are the 90elevation angle.
thermal expansion coefficients of the panel and Table 1 lists the accuracy budgets resulted from
BUS, respectively. According to the parameters of different factors. Obviously, the manufacturing and
the panel and BUS we used, zmax is about 0.112mm gravity-induced errors are the primary error sources.
and the rms value is =zmax/3= 0.037mm. As a These two kinds of errors are repeatable, which can
result, the change of surface accuracy is a little be modeled and compensated by the active surface
larger after the panel is constrained by the BUS. system. In principle, the manufacturing error could
be further improved by enhancing the
manufacturing process and by modifying molds.
Wind- and thermal-induced errors have
instantaneous and uncertain characteristics; they are
not able to be fit with a solid model, and need
intense monitoring and on-line calibration.

TABLE 1
ACCURACY BUDGET

Error source RMS(mm) Remark

Manufacturing 0.065 None

Gravity 0.056 Normal

Fig. 10 The contour of thermal-induced deformation
under 90elevation angle. Wind 0.046 10m/s normal

We considered the effect of the vertical wind of Thermal 0.037 5 temperature

10m/s on the panel accuracy. Assuming a uniform difference
wind load, it is calculated following the formula: [13]
F CF qA (6) 4. Experiment and FEA of
where CF is the wind power coefficient; q=1/16v2 is Concentrated Load on the Panel
the dynamic pressure and v is the wind speed; A is It is inevitable to step on the panel during
the characteristic area. From (6), we determined F installing and servicing. Accordingly, it is necessary
of 36N/m2. The deformation contour is shown as to analyze if it is elastic deformation when people
Figure 11. Accordingly, the wind-induced step on the panel. In this paper, we obtain the
uncertainty is 0.046mm. deformation by experiment and simulation.
The experiment is shown as Figure 12. One
people steps on the middle of the panel and dial

E-ISSN: 2224-3429 59 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

indicators are placed at the four corners and under tracker. The rms uncertainties obtained from the two
the position of the people. The weight of the people approaches show excellent consistency. The ideal
is 70kg. The dial indicators show that the whole reflector surface accuracy will be achieved by
deformation in the middle of the panel is 0.71mm, repeated measurement and adjustment. The work
the horizontal deformations of the supporting bolts condition of the CMM method in factory is more
are zero and the vertical deformation is 0.027mm. coincident with that on the BUS. Only four
The above experiment is simulated using finite reference points of each panel need to be measured
element software. The establishment of the finite to compensate the installation errors. In contrast, the
element model is similar in section 3 and only the four adjusting points are not zero when the laser
load is different. The deformation of the panel in the tracker method is used. Not only the four adjusting
position of concentrated load is 0.67mm. points but also the points on the panel need to be
The comparison error between the experiment measured. The evaluation of the effects of wind,
and simulation is 5.6%. The stress of the maximum temperature, gravity and manufacturing error on the
deformation point is = E = 49.7MPa < []. panel surface accuracy with the FEA software
Therefore, the panel is elastic deformation under suggests that the latter two factors are of the primary
70kg concentrated load. The experimental result care. The accuracy budget in this study offers a
show that the dial indicator is zero after the people useful reference for other large antennas. The result
left the panel. of panel deformation under concentrated load shows
that there is not plastic deformation when people of
weight below 70kg installs and remedies the panel.
But, people should prefer to pull on big shoes to
distribute the pressure. The accuracies of single
panels under different cases totally satisfy the
design requirement, which assures the accuracy of
primary reflector surface (less than 0.6mm).
We hope the experts and correlated organization
can define the criterion of antenna structure design,
i.e. the definition of panel surface accuracy and
determination of the grade of wind speed and
temperature shift according to different accuracy
requirement.
Fig. 12 The experiment of concentrated load applied
on the panel.

Acknowledgment
This work was supported in part by the National
Natural Science Foundation of China (Grant NO.
Y347201001), Science and Technology
Commission of Shanghai Municipality
(08DZ1160100), and Knowledge Innovation
Program of the Chinese Academy of Sciences
(KJCX-YW-18). We appreciate the collaboration
with Shanghai Shen Mo Die & Mold Manufacturing
Co. Ltd. We thank Profs. Zhihan Qian (SHAO) and
Dr. King Lee (NRAO) for the guidance.
Fig. 13 The contour of deformation induced by
concentrated load under 90elevation angle. References
[1] R. M. Prestage, et al., The Green Bank
telescope, IEEE Proceeding, 2009, pp. 1382-
5. Conclusion 1390.
The panel surface accuracy of Sh65RT was
[2] R. Wielebinshi, The Effelsberg 100-m radio
measured independently with two different telescope, Naturwissenschaften, vol.58, 1971,
instruments and techniques, the CMM and laser pp. 109-116.

E-ISSN: 2224-3429 60 Volume 10, 2015

 Li Fu, Guoxi Liu, Chao Jin, Feng Yan
WSEAS TRANSACTIONS on APPLIED and THEORETICAL MECHANICS Tao An, Zhiqiang Shen

[3] E. Cenacchi. SRT Project Book. Available:

http://www.ca.astro.it/srt/project-
book/index.htm
[4] A. Orfei, Updates on SRT, the Sardinia radio
telescope, in 8th Radionet Eng. Forum, Yebes,
2008, pp.1-40.
[5] J. M. Hollis, P. R. Jewell, F. J. Lovas, and A.
Remijan. Green Bank telescope observations of
interstellar Glycolaldehyde: Low-temperature
sugar, The Astrophysical Journal, 2004, pp. 45-
48.
[6] S. Cichowolski, E. M. Arnal. An Effelsberg HI
study of the ISM around WR 126, WR 154 and
WR 155, A&A, vol 414, 2004, pp. 203-209.
[7] G. Zacchiroli, et al., The panels for primary and
secondary mirror reectors and the Active
Surface System for the new Sardinia Radio
Telescope, Mem. S.A.It. Suppl., vol. 10, 2006,
pp. 126-130.
[8] Jr. Hoagg, et al. Method for fabricating antenna
reflector panels, United States Patent,1991, pp.
1-7.
[9] D. H. Parker, J. M. Payne, J. W. Shelton, and T.
L. Weadon. Instrument for setting radio
telescope surfaces, GBT MEMO 206, pp. 1-4.
[10] H. Kutoglu, C. Mekik, H. Akcin. Effects of
errors in coordinates on transformation
parameters, J. Surv. Eng., vol. 129, 2003, pp.
9194.
[11] Z. C. Li, G. Y. Li, C. Jin. On the data
processing methods of surface antennas
inspection, FIG XXII International Congress,
2002, pp.1-9.
[12] W. N. Christiansen, J. A. Hgbom.
Radiotelescopes, Cambridge University Press,
1985.
[13] Z. G. Zhu, S. H. Ye. Antenna structure design,
The National Defence Industry Press, 1980.

E-ISSN: 2224-3429 61 Volume 10, 2015