RADIO SCIENCE, VOL. 43, RS1010, doi:10.

1029/2007RS003692, 2008

Design and applications of a versatile HF radar calibration target

in low Earth orbit
Paul A. Bernhardt,1 Carl L. Siefring,1 Joe F. Thomason,2 Serafin P. Rodriquez,2
Andrew C. Nicholas,3 Steven M. Koss,4 Mike Nurnberger,4 Chuck Hoberman,5
Matthew Davis,5 David L. Hysell,6 and Michael C. Kelley6
Received 11 May 2007; revised 27 September 2007; accepted 25 October 2007; published 7 February 2008.
[1] High frequency (HF) radars are used to detect ionospheric irregularities, meteor trails,
and moving targets. The Precision Expandable Radar Calibration Sphere (PERCS) is a
simple radar target in space to help determine the operational parameters of ground HF
radars. PERCS will have a known radar cross section that is independent of observation
direction within 0.5 dB. The PERCS satellite can be launched in a stowed configuration
that has about 1 m in diameter. After launch, the PERCS will expand to a diameter of
almost 10 m. Upon expansion, a stable wire frame is formed to act as a radar scatter target
in the form of a polyhedral sphere. The simplest version of the sphere has 60 vertices
(V60) that are joined to 90 rigid segments. Each segment is hinged so that the PERCS can
be folded into a compact package for launch. Analysis of the V60 wire frame with a
10 m diameter shows that the radar cross section (RCS) is nearly independent of viewing
angle up to 30 MHz. Another design with 240 vertices produces even better
performance. Radar systems will be calibrated using the radar echo data and the precise
knowledge of the target RCS, position, and velocity. The PERCS can reflect radar signals
from natural targets such as field aligned and current driven irregularities not
presently accessible from ground-based radars. The wire frame structure has several
advantages over a metalized spheroid ‘‘balloon’’ with (1) much less drag, (2) larger radar
cross section, and (3) lower fabrication cost.
Citation: Bernhardt, P. A., et al. (2008), Design and applications of a versatile HF radar calibration target in low Earth orbit,
Radio Sci., 43, RS1010, doi:10.1029/2007RS003692.

1. Introduction disturbances to measure convection electric fields used in

[2] In the radio frequency range 2 to 40 MHz, Medium
Frequency (MF), High Frequency (HF), and Very High
Frequency (VHF) radars are used by atmospheric
researchers to study natural and artificial ionospheric
irregularities, and backscatter from meteoric ionization.
The network of SuperDARN radars tracks high latitude
1
Plasma Physics Division, Naval Research Laboratory, Washington,
D. C., USA.
2
Radar Division, Naval Research Laboratory, Washington, D. C.,
USA.
3
Space Science Division, Naval Research Laboratory, Washington,
D. C., USA.
4
Space Engineering Department, Naval Research Laboratory,
Washington, D. C., USA.
5
Hoberman Associates, Inc., New York, New York, USA.
6
Department of Earth and Atmospheric Sciences and School of
Electrical Engineering, Cornell University, Ithaca, New York, USA.
Copyright 2008 by the American Geophysical Union.
0048-6604/08/2007RS003692
space weather models. HF radars in Alaska support the
Alaskan High Frequency Active Auroral Research Pro-
gram (HAARP) transmitter by detecting backscatter off
field aligned irregularities generated by high power radio
waves. All of these HF radars lack a precise calibration
target with both a known position and well defined radar
cross section.
[ 3 ] The Precision Expandable Radar Calibration
Sphere (PERCS) will provide an HF radar target with
an isotropic radar cross section that can be used by all
ground based HF radars. The PERCS target will make
use of Hoberman sphere technology for deployment of
large objects in space. The absolute accuracy for scatter
from field aligned irregularities or meteor trails will be
greatly improved after ground MF and HF radars are
absolutely calibrated with the PERCS target.
[4] The PERCS is a deployable sphere that has <1 m
diameter in stowed configuration and a 10 m diameter in
deployed state. The PERCS can be designed to yield a

RS1010 1 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 1. Geometric structures of potential radar calibration targets. (a) The icosahedron has each
vertex truncated to yield a hexagonal face. The rest of the spherical surface is filled with hexagons
to yield spherical polyhedra with (b) 60 vertices and (c) 240 vertices.

radar cross section that is constant at any viewing angle
within 0.5 dB. The fully deployed sphere has a precise
radar cross section for frequencies less than 23 MHz and
some frequencies up to 40 MHz. The sphere can be placed
in a high inclination orbit to service (1) the SuperDARN
radars located around the north and south polar regions as
well as the new SuperDARN radars being setup in the
United States and (2) the HF radar diagnostics that are part
of the HAARP program. PERCS is a geodesic sphere with
60 or 240 vertices and 12 pentagonal faces. In chemistry,
these structures are denoted as C60 and C240 fullerenes
with the C60 specifically named Buckminsterfullerene or
‘‘Buckyball’’. The next section describes spherical poly-
hedra that form a wire frame suitable for radar scatter. The
calculated radar cross sections of these polyhedral struc-
tures are shown in section 3 to be suitable for radar
calibration target in space over a restricted frequency
range. The monostatic radar cross section is computed as
a function of radar frequency and viewing orientation. In
section 4, deployable polyhedra are presented using a
technique developed by Chuck Hoberman. The PERCS
target is launched into space with a relatively small stowed
size and, with spring loaded hinges, which allows it to
expand into a large wire frame. The final sections outline
the construction of the radar target, effects of atmospheric
drag, optical tracking of the sphere and potential uses of the
PERCS for calibration of backscatter radar systems. A
large, well calibrated radar target in orbit around the Earth
will be available for many new research applications.
2. Spherical Polyhedra With Pentagon and
Hexagon Faces
[5] All of the spherical radar calibration targets will be
based on wire frame approximations to a smooth
spheroid. These wire frame representations of a nearly
spherical object use conducting wires at the edges of
spherical polyhedra. All of the geometric structures are
derived from the icosahedron (Figure 1a) which has
12 vertices, 20 faces and 30 edges.
[6] Truncation at each vertex of the icosahedron yields
a well known Archimedean solid or ‘‘soccer ball’’ with
60 vertices (V60) as illustrated in Figure 1b. During the
truncation process, each edge of the triangular face of an
icosahedron is divided by 1/3rd and each corner is
removed leaving a hexagon. All the vertices of this
structure lie at a common distance from the center and
all the 90 edges have the same length. The faces are thus
composed of 12 pentagons and 20 hexagons. The radar
cross section from the V60 truncated icosahedron will
vary to some degree if the vertex, edge, hexagon, or
pentagon face is rotated toward the radar beam.
[7] The polyhedron becomes more like a perfect
sphere as the number of faces increases. Consider if
the edges of the icosahedron are truncated at a distance
1/6th from the vertex. The remaining long edge is
divided into two edges on either side of a hexagon.
Using the procedure outlined in Appendix A and de-
scribed by Wang and Chiu [1993], the resulting spherical
V240 polyhedron is generated with 240 vertices, 360
edges, 12 pentagons, and 110 hexagonal faces (Figure
1c). As the number of edges increases, the target becomes
more spherical and the variation of the radar cross section
with viewing direction is reduced.
[8] The physical dimensions of the polyhedra deter-
mine their mechanical and radar scattering properties.
The V60 truncated icosahedron has 90 edges of identical
length meeting at 60 identical vertices. The higher order
spherical polyhedra are comprised of hexagon and pen-
tagon faces with several length edges. Length of a radius

2 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Table 1. Distribution of Edges and Vertices on 10-m Spherical Polyhedra

Edges Vertices
Polyhedron
Vertex Vertex Vertex Angle With
Designation Radius Length Number Radius, deg Planar Angles, deg Number

V60 5.0 2.1077 90 78.389 111.4, 124.3, 124.3 60

V240 5.0 0.9524 240 84.535 108.7, 126.6, 126.6 60
1.0814 60 83.793 117.9, 117.9, 124.2 60
1.1319 60 83.500 119.37, 119.40, 121.23 120

vector is defined as the constant distance to each vertex
from the center of the sphere for each polyhedron. The
edges at the vertices for the spherical polyhedra have
come together with a finite number of angles in the plane
perpendicular to the radius vector at each angle. These
angles add up to 360°. The angles between the radius
angle and edge at a vertex also have a small number of
values which increase with the number of vertices. The
properties of the V60 and V240 spherical polyhedra with
a 5 m radius are given in Table 1.
[9] For simplicity of construction and reduction of
atmospheric drag, the polyhedral surfaces shown in
Figure 1 will be replaced by the wire frame targets of
Figure 2. The targets are 10 m in diameter with 60 and
240 vertices joined with conducting wires. If the indi-
vidual wire segments or edges are much smaller than an
HF radar wavelength, they act together to form a good
approximation to a perfectly spherical radar target. The
polygons on the geodesic structure are open so the wire
frame has much lower drag than a similar design using a
spheroid such as metalized balloon due to the dramatic
decrease in presented frontal area. Other polyhedral
configurations using triangle faces could be used for
the radar calibration target their consideration will be
postponed to the future.
3. Radar Cross Section of Spherical
Polyhedra
[10] Monostatic radar cross section determines the
amount of power reflected from an object back to the
transmitter. The radar equation incorporates the radar
cross section with transmitting system and receiving
system parameters and range to the target
Pr G2 1 l20 C0 ðf ; q; fÞ
¼ 2
 2
s¼ s ðf Þ ð1Þ
Pt L 4p R2 4p R4
where P is received (r) and transmitted (t) power, G is
gain, L is loss, R is range, l0 is wavelength, s is radar
cross section and C0 is the radar system parameter. The
directional antenna gain G(f, q, f) for a ground radar
system is dependent on the radar frequency (f), on the
zenith angle to the target (q) and the azimuth angle to the
target (f). The system losses L(f) are only frequency

Figure 2. Wire frame configurations for HF radar calibration targets with (a) 60 vertices and
(b) 240 vertices. The dimensions are shown for 10-m diameter spherical polyhedral structures.

3 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

dependent. The monostatic cross section s(f) of a A perfect sphere does not change the polarization of the
spherical radar target is dependent on frequency but is reflected wave so for a RHCP incident wave, sRH = sTotal
independent of q and f. In terms of incident (Ei) and and sLH = 0. For the PERCS to be a useful radar
scattered (Es) electric fields, the total bistatic radar cross calibration target, most of the energy radiated should come
section is defined as back with the same polarization. A measure of the amount
of electromagnetic energy coupled into the other polariza-
jEs ð R; fs ; qs Þj2 tion is the ratio of the polarization RCS given by
sTotal ðfi ; qi ; fs ; qs Þ ¼ 4p R2 ð2Þ
jEi ðfi ; qi Þj2  2
jELC ð R; fs ; qs Þj2 Eq  jEf 
where the incident field Ei(fi, qi) is a plane wave rLHRH ðfi ; qi ; fs ; qs Þ ¼ ¼ 
jERC ð R; fs ; qs Þj2 Eq þ jEf 2
propagating along a direction given in spherical polar
coordinates by the angles fi and qi, and the scattered ð7Þ
field Es(R, fs, qs) is a spherical wave in the far field with
the form Es(R, fs, qs) = es(fs, qs) Exp(jk0R)/R. For where rLHRH = 0 for a perfect sphere and rLHRH = 1 if
backscatter the incident and scattered directions lie along have the initial RHCP wave is scattered into the LHCP
the same path with opposite directions. wave.
[11] Measurements of the ratio Pr/Pt from (1) with a [12] A 10-m spheroid or metal balloon was selected as
known s over a range of look directions and frequencies the baseline for radar cross-section (RCS) comparisons.
permits estimation of the system parameter C0(f, q, f). A sphere with another dimension will have radar cross
The radar cross section for any scattering medium is section that scales in frequency with ratio of sphere
found from diameter to radio wavelength (2r/l) and scales with cross
 section magnitude as the projected area of the sphere
R4 Pr  (pr2). This scaling applies to the Rayleigh, Mie, and
sMeasured ¼  ð3Þ geometric optics regions of scatter. The 10-m sphere
C0 ðf ; q; fÞ Pt 
Measured should be large enough to provide usable echoes for
Another parameter that is affected by scatter from a ground HF radars. The Mie scattering for the radar cross
target is the polarization. The complex electric field section of a perfectly conducting sphere is given in
vector are completely determined by it components Ef Appendix B based on the theory of Ruck et al. [1970].
and Eq with the formula [13] The radar cross section (RCS) is computed for
frequencies up to 50 MHz (Figure 3). Below 4 MHz, the
Eq cross section monotonically drops off as l4 where l, the
R¼ : ð4Þ
Ef radio wavelength, is greater less than the sphere radius r.
This is called Rayleigh scattering. For high frequencies
With R = ±j, the waves are right-handed (RHCP) or left- with l  r, the radar cross section is approximately pr2
handed (LHCP) circular polarized, respectively. Each in the asymptotic geometric optics limit. Strong localized
electromagnetic wave E can be decomposed into these minima in RCS for the 10-m sphere are found near
two circularly polarized waves according to 17 MHz and 29 MHz.
[14] Next, the RCS is computed for a conducting
E ¼ Eq q þ Ef f
polyhedron as a radar target. All of the 10-m diameter
Eq þ jEf wire frames in Figure 2 will reflect HF radar signals.
ERC ¼ ERC ðq  jfÞ where ERC ¼ ð5Þ Their radar cross section (RCS), however, will fluctuate
2
as the wire frame is rotated. The objective of the
Eq  jEf calibration target design is to produce a minimal varia-
ELC ¼ ELC ðq þ jfÞ where ELC ¼
2 tion in radar cross-section, 0.5 dB, as the target orienta-
where q and f are unit vectors. An initial electric field that tion is changed. This will first be studied using the V60
is right hand circular polarized (RHCP) can scatter into wire frame shown in Figure 2a. The computational
both RHCP and LHCP modes. The radar cross sections for frequency range will be from 0 to 50 MHz to match
scattering into the same and different modes are the operational frequencies for most HF and lower-band
VHF radars on the ground.
jERC ð R; fs ; qs Þj2 [15] The estimations of RCS were obtained by apply-
sRH ðfi ; qi ; fs ; qs Þ ¼ 8p R2 ing the method of moments solutions to the electric field
jEi ðfi ; qi Þj2
ð6Þ integral equations (EFIEs) using the WIPL-D 3-D Elec-
2
tromagnetic Solver [Koludzija et al., 2004]. The polyhe-
jELC ð R; fs ; qs Þj dral wire frame is excited by a right-hand circular
sLH ðfi ; qi ; fs ; qs Þ ¼ 8p R2
jEi ðfi ; qi Þj2 polarized (RHCP) electromagnetic wave. The incident

4 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 3. Radar cross section for a 10-m diameter
sphere with a perfectly conducting surface. The solid red
line is the analytic theory described in Appendix B, and
the black dots are the numerical solutions using the 3-D
electromagnetic computations with the WIPL-D model.
and reflected waves are compared to yield the monostatic
RCS. In the model, currents in both wire segments and
plates are expressed as finite polynomial sums with
orders between 1 and 9 automatically determined by
the electrical length of the conducting component. If the
wire becomes longer than two wavelengths, it is divided
into small segments each shorter than l/2. The diameter
of each wire in the model was set to 4 mm to resemble
the mechanical structure. The WIPL electromagnetic
Figure 4. Comparison of the radar cross section (RCS)
for a V60 wire frame (blue and green) with a
conducting spherical surface with the same diameter
(red). The RCS for the wire frame only significantly
depends on observation direction for frequencies from
23 to 30 MHz and above 32 MHz.
5 of 23
RS1010
model was validated by comparing the analytic MIE
theory for the conducting spheroid with the numerical
RCS computations (Figure 3).
[16] Comparison of the V60 wire frame and conduct-
ing spheroid models show many differences in RCS
(Figure 4). In the 9 to 24 MHz frequency range, the
wire frame has a larger RCS than the metalized spheroid.
Below 9 MHz, the wire frame RCS is less. Also, the dip
in RCS near 17 MHz is less for the V60 wire frame. The
RCS is illustrated for two different spherical coordinate
directions (blue and green curves) in Figure 4. The
incident wave makes an angle q with the x-y plane and
an angle f with the x-axis in the x-y plane. If (q, f) = (0, 0),
the wave is incident in the negative x-direction. If (q, f) =
(p/2, p/2), the wave is incident in the negative z-direction.
Only for frequencies below 23 MHz and between 30 and
34 MHz there minimal change in RCS for the two
viewing angles. The variation of RCS near 20 MHz for
the two view directions is less than 0.5 dB. Rotation of
the wire frame produces large variations in RCS in the 23
to 30 MHz frequency range and for frequencies above
34 MHz. To better illustrate the dynamic range for
backscatter variations with frequency and orientation,
the RCS will be displayed as dB m2 rather than m2.
[17] The full range of RCS changes variations with
radar observation direction is obtained by stepping the
WIPL simulations through all target angles. The numer-
ical values for RCS were computed over at total of 648
viewing directions using 10 degree steps in both q and f.
The average value of RCS for all viewing angles is given
in Figure 5. Between 9 and 33 MHz, the wire frame has a
larger radar cross section than the spheroid with a
continuous conducting surface.
[18] The maximum RCS deviation based on the WIPL
computations shows that the 10-m diameter V60 wire
Figure 5. Computed monostatic RCS averaged over all
viewing directions from a V60 polyhedron with equal
edge lengths.
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 6. Maximum change with orientation of the
monostatic RCS for the V60 wire frame radar reflector
with equal length edges.
frame is well suited for a calibration target for frequen-
cies less than 23 MHz where this deviation is less than
0.5 dB (Figure 6). Above 23 MHz, a local maximum in
RCS variation with orientation is found to be 9.6 dB at
26 MHz. This makes the V60 of less use in the 26 to
29 MHz band unless the orientation of the sphere is
known. Figure 6, however, shows that that in the 30 to
34 MHz the variation in RCS with is at a local minimum
so that at this frequency the V60 conducting frame may
be a useful radar calibration target.
[19] The RCS behavior at different frequencies is
explained in terms of resonances with the edges and
faces of the pentagons and hexagons in the wire frame.
Figure 7. Radar cross section patterns for the V60 wire frame at selected frequencies.
6 of 23
RS1010
Below 16 MHz, a spherical backscatter pattern is pro-
duced by the v60 wire frame. Between 16 and 20 MHz,
the small lumps in the backscatter pattern (Figures 7a)
are aligned with the faces of the 12 pentagons in the V60
sphere. Between 23 and 24 MHz, the backscatter bulges
are associated with the hexagon faces. Near 30 MHz
when the sphere is 1=2 wavelength in diameter, the
backscatter has even contributions from the pentagons
and hexagons on the surface of the V60 (Figure 7d).
From 31 to 34 MHz the peaks in the RCS pattern
corresponds to the locations of the edges between the
hexagons (Figure 7e). From 35 to 40 MHz, backscatter
from the faces of the hexagons dominates the pattern
(Figure 7f). At higher frequencies, the edges and faces
contribute to the radiation with increasingly complex
patterns.
[20] Figure 7 shows that the change in RCS with look
direction is associated with the transitions between
hexagon and polygon faces of the wire frame. To reduce
this difference, the wire frame was redesigned with an
increase pentagon face area and a decrease in the
hexagon face area to make all the polygons have ap-
proximately the same area. For the 10-m diameter
sphere, this redesign yielded a spherical polyhedron with
a radius of 5 m to each vertex, 30 Edges with length
1.53 m, 60 edges with length 2.29 m, pentagons with
area 9.03 m2 and hexagons with area 9.34 m2 (Figure 8).
[21] The revised V60 sphere has an aspect independent
RCS out to 30 MHz. Figure 9 shows the average RCS
and the maximum deviation in RCS with viewing
direction. A comparison with Figures 5 and 6 shows
that the revised design has much lower deviations in
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

only 2.2 dB at 28 MHz. As expected, the addition of

small hexagon facets to the spherical surface reduces the
dependence of RCS on target orientation. The new V240
target meets the 0.5 dB design requirements between 30
and 36 MHz.
[25] Samples of the RCS patterns are shown in
Figure 13 for selected frequencies. There is a one-to-
one correspondence with the features of the RCS pattern
and the locations of the polygons on the surface of the
sphere. For frequencies less that 30 MHz, the peak faces
in the RCS pattern are collocated with the pentagon
faces. The smooth RCS surface between 30 and 35 MHz
shows small valleys in the RCS at the locations of the
pentagons. For instance, near 42 MHz, the peak RCS
faces are located directly over the hexagons located in
the center of a triangle formed by three pentagon faces on
Figure 8. Redistribution of polyhedral edges to make the surface of the sphere.
all faces have approximately equal areas. [26] An attempt is made to reduce the RCS variations
with look direction by redesigning the wire frame to
make the area pentagon faces equal to most of the area of
RCS with viewing direction and it meets the 0.5 dB the hexagon faces. Figure 14 shows the new design.
specification out to 30 MHz whereas the old design met [27] The average RCS for the two designs of the V240
the specification only to 24 MHz. Figure 9 also shows are nearly identical (Figures 11 and 15a). The RCS for
that the total RCS for the wire frame is larger than the the revised V240 design does provides better RCS
RCS for the reference spherical ball at all frequencies behavior at frequencies up to 35 MHz (Figure 15) but
between 10 and 32 MHz. the deviations with viewing direction are larger above
[22] The nonspherical imperfections in the V60 wire
frame may change the polarization by scattering of
individual elements of the sphere. The definition of
polarization coupling from RHCL to LHCP given by
(7) was computed with the WIPL code for scattering
equal-edge V60 from all directions. Figure 10 shows that
the frequencies of low average (green) and maximum
(blue) polarization coupling coincide with the frequen-
cies of low variations of total RCS. The frequencies
below 23 MHz which have low variation with radar view
direction also have negligible change in polarization by
scatter. Similar results are obtained for scattering of left
hand circular polarization as well as with both the f and
q directions of linear polarization from all of the wire
frame structures.
[23] To reduce the viewing angle variations in RCS
and the polarization changes above 23 MHz, the RCS
properties of the V240 wire frame were examined. First
consider the V240 sphere with the three lengths given in
Table 1. The direction-averaged RCS for the V240 is
closer to that of a conducting spherical surface (Figure 11)
than for the V60. The deep RCS minimum near 18 MHz in
the V240 RCS is close to that of the reference spheroid.
There is another deep reduction in RCS near 38 MHz. For
most of the frequencies above 20 MHz, the wire frame has
about 5 dB larger cross section than the solid sphere.
[24] The V240 has better uniformity in RCS pattern. Figure 9. HF RCS for the spherical polyhedron with
The RCS deviation is less than 0.5 dB for frequencies up equal area polygon faces. The useful frequency range
to 26 MHz (Figure 12). The maximum RCS deviation is extends to 30 MHz.

7 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 10. Polarization conversion from right hand
circular polar (RHCP) to left hand circular polar (LHCP)
by scatter from the V60 radar target.
28 MHz that shown in Figure 12. As discussed in the
next section, the difficulties for deployment of the
mechanical structure with a few extremely short sides
may affect the decision of which wire frame design is
used as a radar target.
[28] For both versions of the V240, in the band for
constant radar cross section below 26 MHz, the polari-
zation conversion is less than 1%. Figure 16 illustrates
the polarization change for the first version of the V240
sphere. In the frequencies above 30 MHz, the V240 has
much less polarization change than produced by the V60
wire frame.
[29] As a practical consideration, the signal to noise
ratio for a V60 wire frame flying over the SuperDARN
Figure 11. Comparison of the RCS a for 10-m diameter
V240 wire frame with a reference 10-m metalized solid
sphere.
8 of 23
RS1010
Figure 12. Maximum variation in monostatic RCS for
the V240 wire frame radar reflector.
radars is considered. The signal to noise ration deter-
mines the usable size for the radar target that can be
detected at a large range with the ground based HF
radar. The radar equation (1) divided by the sky noise
power can be manipulated to give the signal-to-noise
s 0 P T G 2 l2
ratio as SNR ¼ where R is the range, k
ð4pÞ3 R4 kTSky Bn
is the Boltzmann Constant, TSky is the radio sky
temperature in Kelvin, Bn is the radar receiver band-
width, PT is the transmitter power, G is the antenna
gain, and l is the radio wavelength. The radio sky
noise that comes from atmospheric, galactic, and man-
made sources is estimated by the standard International
Radio Consultative Committee (CCIR) model. The
RCS (s0) is given in Figures 5 or 9a as a function of
operating frequency. Using parameters representative of
the SuperDARN radar (Bn = 3.3 kHz, PT = 10 kW).
The antenna gain varies from G = 100 at 11 MHz to G =
250 at 17 MHz [Walker et al., 1987; Lester et al., 2004].
With target ranges of R = 300 to 1500 km, the required
radar signal-to-noise for detection of the target is illus-
trated in Figure 17. The signal to noise ratio is larger than
0 dB with frequencies near 20 MHz for all ranges.
Consequently, a 10-m sphere at an orbit of 450 km
altitude is big enough to provide useful calibration for
SuperDARN radar systems looking with low elevation
angles at the higher frequencies where the sky noise
drops and the antenna gain increases.
[30] In summary, wire frame polyhedra with 60 and
240 vertices provide excellent calibration targets for
radar wavelengths up to the target diameter. The higher
order spheres have regions near where, with radar wave-
lengths near 1.26 the target diameter, RCS is nearly
independent of viewing direction. For use in space, the
spheres must be deployed from a compact configuration.
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 13. Large variations with frequency in radar cross section for the V240 wire frame with
relatively small variations with direction.

The mechanical scheme for stowing and deploying these

spheres is discussed next.

4. Deployment of a Spherical Polyhedron

in Space
[31] Launching a fully deployed 10-m diameter wire
frame in low Earth orbit (LEO) is not feasible. The
polyhedral structure must be collapsed into a small

Figure 14. Tuned V240 wire frame with 5-m radius to

each vertex, 60 edges with length 0.45 m, 240 ledges Figure 15. HF RCS for the spherical V240 polyhedron
with length 1.08 m, 60 edges with length 1.28 m, with larger pentagon faces. The useful frequency range
pentagon areas of 2.01 m2 and three types of hexagons extends to 35 MHz.
with areas 1.81, 1.85, and 3.22 m2.

9 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 16. Polarization conversion from right hand
circular polar (RHCP) to left hand circular polar (LHCP)
by scatter from the V240 radar target.
package and then deployed after injection into orbit. For
this purpose, hinges are introduced each vertex and along
the edges allow the structure to be stowed for launch.
Using the technique patented by Hoberman [1990,
1991], two concentric geodesic spheres are interleaved
Figure 17. Signal-to-noise ratio (bottom) for Super-
DARN radar detection of the PERCS 10-m sphere based
on atmospheric, galactic, and manmade sky noise (top)
given by the CCIR model.
10 of 23
RS1010
to form a stable structure with alternating vertices of each
sphere folding either inward or outward. Hoberman
Designs Inc. has marketed this structure as a Trans-
forming Sphere#.
[32 ] The mechanically stable structures for both
stowed and deployed configurations are produced by
introducing ‘‘scissors’’ along each edge. As illustrated in
Figure 18 adapted from Hoberman [1991], each scissors
edge is an angled blue and red arm (10 and 20) that is
connected by a common vertex point (12, 22). As the
bracket pivots around the vertex, the end points (16, 26)
and (14, 24) begin to separate. During this process, the
left end points, the center pivot point, and the right end
points are restrained to follow straight lines (40, 30, 50)
that converge to one origin. When one of the end points
of the red arm reaches this origin, the scissors are stowed
and the polyhedron is fully collapsed. The minimum
stowed size is determined by the distance between the
ends of the scissor arms and the compacted spacing of
the vertex hinges around the stowed center. The geo-
Figure 18. Structure of the Hoberman Scissors
(adapted from Hoberman [1991]). The ends of the two
scissor arms move along lines 40 and 50 while their
pivot point stays on line 30.
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 19. Motion of the two scissor arms for a 5 m radius sphere with a distance between edges
of 2.018 as given by Table 1 for the V60 wire frame.

metrical design of the scissor arms is discussed in
Appendix C.
[33] The motion of a single set of scissor arms with
equal length struts is shown in Figure 19. The dimen-
sions of the arms are taken from Table 1 for the V60 wire
frame. The computed length of the struts is 1.03 and the
angle between the struts is 156.7 degrees. As the scissors
ends and pivots move along radial lines, they form a
collapsible frame for the wire frame.
[34] Full collapse and full extension is never reached
because of hinges located at each vertex. Table 1 gives
the design parameters for all the wire frame hinges.
Figure 20 illustrates on hinge configuration for a V60.
Each of the vertex hinges and the edges for the V60 wire
frame is identical. The angle at the pentagon face is
smaller than the angle for the hexagon faces. The angles
for the V60 hinges given by Figure 20 are in the plane
normal to the radius vector from the center of the sphere.
Each fully deployed edge makes a 78.4 degree angle
with this radius vector for the V60. The V240 and higher
order Hoberman sphere, besides having more elements,
have a more complex design. As shown in Table 1 for the
V240 wire frame, more than one type of hinge needed at
different hexagon and pentagon vertices. In all designs,
small torsion springs are added to each segment
(Figure 21) so that the precision expandable radar
calibration sphere (PERCS) will be self expanding
after release into space.
[35] PERCS can be considered for launch as a second-
ary small-satellite payload from a rocket if the diameter
of the stowed configuration is about 1 m. To achieve this
stowed diameter, the wire frame is converted into a
Hoberman sphere with spring loaded edges. The limita-
tion in stowed size is determined by the length of the
edges and the gathering of the hinges at the center.
Consider a V60 wire frame with extremely large vertex
(33 cm) hinges. Figure 22 shows a comparison of the

Figure 20. (a) Top view and (b) side view of vertex
hinges used to fold the V60 wire frames into a small Figure 21. Detail of torsion spring inserted around
package. scissors pivot to provide opening force.

11 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 22. Comparison of stowed and expand config-
urations of a V60 Hoberman sphere with 10 m diameter
and intentionally large 33 cm vertex hinges. For a 10 m
diameter sphere, the internal vertex sphere has a diameter
of 0.86 m and the outside diameter of the collapsed
sphere is 4.88 m.
expanded and collapsed spheres with straight line approx-
imations to the scissor arms. The central hinges come
together into a small sphere which is structurally related
to the dual of the original polyhedral structure. A poly-
hedron dual is constructed with edges between the center
of each polygon face that forms the original polyhedron.
The dual of each structure given in Figure 2 has triangles
as faces. To obtain the six-sided vertex hinge shown in
Figure 20, these triangles are converted into hexagon
hinges by truncating each triangular vertex. The total
diameter of the stowed configuration is the sum of the
Figure 23. Comparison of stowed and expand config-
urations of a V60 Hoberman sphere with 10 m diameter
and 3 cm vertex hinges. For a 10 m diameter sphere, the
internal vertex sphere has a diameter of 0.08 m and the
outside diameter of the collapsed sphere is 4.10 m.
12 of 23
RS1010
diameter of the dual sphere at the center and twice the
length of the scissor arms between each vertex hinge. In
Figure 21, where the hinge size of 33 cm has been chosen
for the purpose of illustrating the internal sphere, the
diameter of the stowed radially extended arms is 4.88 m.
[36] Greater compression for the stowed configuration
is achieved with smaller vertex hinges and higher order
polyhedra. Figures 23 and 24 illustrate the V60 and
V240 wire frames collapsed using 3 cm hinges. As the
order of the polyhedra increases, the stowed structure has
a reduced diameter. With a 3 cm hinge, all of the spheres
have stowed diameters greater than 1 m so further size
reduction may be required for placement into orbit.
[37] Additional reduction in size can be achieved by
using multiple scissors on each edge. The V60S2, which
designates a 60 vertex structure with two scissors per
edge, is illustrated in several stages of deployment in
Figure 25. The vertex hinges are deleted and the scissor
arms are colored red and blue to indicate the two
polyhedral spheres that are interlaced. The 10 m dual-
scissors V60 would have a minimum stowed diameter of
2.1 m using 3 cm hinges. Table 2 lists the stowed
diameters of all the wire frame polyhedra for both single
and dual scissors implementations. The configurations
near or under 1 m diameter can be accommodated as a
small satellite.
5. Performance of the Hoberman Sphere
Wire Frame as a Radar Target
[38] When the wire frame is deployed as a Hoberman
sphere, two factors must be considered. First the struts
Figure 24. Comparison of stowed and expand config-
urations of a V240 Hoberman sphere with 10 m diameter
and 3 cm vertex hinges. For a 10 m diameter sphere, the
internal vertex sphere has a diameter of 0.16 m and the
outside diameter of the collapsed sphere is 2.08 m.
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 25. V60 radar reflector with two scissors per edge.

along the edges are joined together at mechanical hinges about 0.5 dB for frequencies above 10 MHz (Figure 26)
where electrical resistance may be a factor. Second, the and the variation in RCS with viewing angle was
Hoberman sphere does not fully deploy into a polyhedra changes by a maximum of 0.12 dB (Figure 27). From
because the internal and external hinges cannot merge this study, it is concluded that if care is taken to keep the
together. To study these effects, the WIPL electromag- resistance at each contact point less that 10 Ohms, the
netic solver was used to calculate the RCS of practical Hoberman sphere radar target will meet a requirement to
mechanical configurations. know the absolute RCS within 0.5 dB at each frequency.
[39] The resistance between arms of the Hoberman
sphere will be minimized by design with high conduc-
tivity (i.e., gold) surfaces with maximum pressure over
maximum area at the hinge interfaces. The torsion Table 2. Minimum Stowed Diameters for the Hoberman
springs on each scissors will be chosen to apply pressure Spheres Using 3 cm Hinges
at the hinges when the sphere is in the deployed state. To
Parameter Edge Type V60 V240
obtain an understanding of the effects of hinge resis-
tance, the RCS for the wire frame was computed with the External Spokes Single Scissors 4.10 2.08
addition of fixed resistance inserted at the ends of 30 Sphere Diameter Dual Scissors 2.10 1.12
wires of the 90 edges in the V60 wire frame. A resistance Internal Hinge Single and 0.08 0.16
Sphere Diameter Dual Scissors
of 1 Ohm in 30 places had insignificant effects on the
RCS. With 10 Ohms, the total RCS was changed by
13 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 26. Effect of contact resistance at 30 vertices on
computed radar cross sections for the V60 radar target.
For this reason, all of the metal components of the wire
frame should be gold plated to prevent corrosion at the
conductive interfaces.
[40] The RCS calculations used wires with infinite
conductivity. It may be profitable to introduce a small
amount of resistance into a wire to reduce the ‘‘Q’’ of the
sphere so that some of the frequency dependencies
become broader.
[41] To identify the effects of ‘‘crinkle’’ in the surface
of the Hoberman sphere, a partially deployed configura-
tion was considered. Figure 28 shows the affects on RCS
of a V60 sphere with 95% deployment (0.25 m out of
5 m radius). This is a substantial exaggeration of the
actual surface imperfections. The actual deployed Hober-
man sphere is expected to be deployed within 0.01 m of
a full 5 m radius. The average RCS and the variations
in RCS with viewing direction are affected by less than
0.2 dB with this large fluctuation in surface structure. In
fact, this calculation shows that partial deployment of the
sphere yields a structure with reduced absolute cross
section with increased range to 25 MHz for less than
0.5 dB variations with viewing direction. In summary,
for applications as a radar target, imperfections in the
Hoberman sphere such as vertex hinge resistance and
slight crinkle of the surface can be minimized so as to
have negligible impact on the radar RCS.
6. Atmospheric Drag and Optical Tracking
of the Spherical Radar Target
[42] Precise orbit determination is needed for accurate
validation of target location algorithms for ground HF
radars. The primary perturbation on the orbit of the
spherical radar target will be atmospheric drag. Atmo-
spheric drag will ultimately limit the lifetime of the target
in orbit. The satellite acceleration (or deceleration) due to
14 of 23
RS1010
drag is given by the equation [Montenbruck and Gill,
2000]
@2r CD A 2
¼ rv er ð8Þ
@t 2 2 m r
where r is geocentric radius, CD is the drag coefficient
assumed to have a value of 2, A is the projected 2-D
cross section area of the metal object, m is the mass, r is
the neutral mass density of the atmosphere, vr is the
speed of the sphere with respect to the medium and er is
the unit vector in the direction of the orbit. The mass
density model used for this study was obtained from the
NRL MSIS Model described by Hedin [1987] and is
shown in Figure 29. The actual neutral density profiles
will vary with season and solar cycle but the one selected
for the drag calculations is representative of minimum
solar activity at equinox.
[43] Numerically integrating (3) for circular orbits
yield a time history of the orbiting sphere. The mass
and drag area for a fully metalized target with a 30 micron
thick spherical shell is estimated to be 43 kg and 78 m2,
respectively resulting in an area to mass ratio of 1.84 m2/kg.
When this solid-surface spheroid is injected to orbit at
450 km, it will have a lifetime of only 5 days (see red line
in Figure 30). The V60 Hoberman sphere made with
8 mm diameter aluminum arms has an estimated mass of
49 kg and a frontal area of 2.3 m2. This corresponds to an
area to mass ratio of 0.047 m2/kg, which is 40 times
smaller than the value for the fully metalized target. Due
to the dramatic reduction area to mass ratio, the spherical
wire frame will remain in orbit for over 1/2 year (see blue
line in Figure 30). For this model, the wire frame consists
of gold plated aluminum struts with 8 cm thickness. If
the wire frame were made from a more dense material
like gold plated steel, the mass would increase to 143 kg
Figure 27. Effect of contact resistance (30 vertices) on
maximum variation in radar cross section.
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 28. Radar cross section for a V60 wire frame with a 5% variation radius at the scissor
hinges.

and the sphere would stay in orbit almost 2 years (see
green curve in Figure 30). The area to mass ratio in (3)
will be similar to the V60 for the higher order wire
frames. Consequently, the orbit lifetime should be greater
than 600 days for orbits above 450 km altitude. Even
though the details of the orbit lifetime will vary with
number of vertices and background neutral density, the
wire frame has an obvious advantage a metal spherical
balloon.
[44] The addition of small, optical corner-cube-
reflectors to each vertex of the Hoberman sphere permits
satellite laser ranging of the radar target. Corner reflec-
tors provide a reflection cross section for visible light
that depends on the incident angle (Figure 31). With one
corner cube on each vertex, those that are facing toward a

Figure 29. Neutral density used for atmospheric drag Figure 30. Limitations on the lifetime of a 10-m
calculations for the spherical radar target in space. spherical radar target because of atmospheric drag.

15 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 31. Laser light reflection cross section for a small corner cube reflector with less than 1 cm
diameter.

ground laser system will reflect visible light back to the
source (Figure 32). The total intensity of the reflected
light will vary as the spherical target rotates. Using the
V60 Hoberman sphere and the tilt angle dependence on
optical cross section, the fluctuations in total optical
cross section were calculated for rotation around several
axes on the sphere (Figure 32). The rotation rate for the
target can be determined for the temporal fluctuations in
the reflected light (Figures 33a and 33c). The lowest

Figure 32. Laser illumination of a V60 sphere with corner cube reflectors at each vertex.

16 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure 33. Illustration of fluctuations in optical reflection cross section (a and c) and frequency
spectra (b and d) as the spherical polyhedra rotates about an axis with tilts of 60 degrees (a and b)
and 18 degrees (c and d).

harmonic component of the frequency spectrum gives the
rotation period (Figures 33b and 33d).
[45] Satellite laser ranging sites can therefore provide
both position and rotation information on the orbiting
polyhedral sphere in space. This information can provide
precise location of the target for HF radar calibration.
Any small fluctuations in the HF radar return signal can
be correlated with the independently measured rotation
of the spherical target.
target location will be tested and validated. The space
weather community is developing both empirical and
physics based models for the Global Assimilation of
Ionospheric Measurements (GAIM). The High Latitude
Data Assimilation Model developed by Utah State Uni-
versity uses high latitude convection from the National
Science Foundation (NSF) SuperDARN radars as inputs
to this physics based calculation that is under develop-
ment as part of the GAIM program. A PERCS calibration

7. Applications Summary
[46] The Precise Expandable Radar Calibration Sphere
(PERCS) has many applications that would be of use to
several scientific organizations. The National Science
Foundation funds the operation of many ground radars
for upper atmospheric studies. The Air Force and Navy
jointly manage the High Frequency Active Auroral
Research Program (HAARP). Jointly the Air Force,
Navy and NSF are funding new ionospheric modification
facilities in Arecibo, Puerto Rico and Jicamarca, Peru.
PERCS will provide the first long range calibration target
for HF radars and transmitter antennas associated with
these research systems. All HF radars are affected by
ionospheric refraction. The refraction effects are estimated
using plasma density data from backscatter ionograms
and other independent measurements. With radar scatter Figure 34. PERCS calibration sphere passing over a
from the PERCS at a known location, the algorithms for ground HF radar for calibration.

17 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
Figure 35. Radar oblique scatter to detect irregularities
from two-stream and gradient drift instabilities.
sphere can improve the accuracy of convection radar data
used in these and other space weather models.
[47] Several options for a orbiting calibration targets in
space have been considered. A deployable Mylar sphere
is a coating of gold was found to cost about $1M from
commercial vendors. Atmospheric drag on a 10 m
metalized ‘‘balloon’’ requires that the orbit altitude be
700 km to give a lifetime of 1 year. The alternative
sphere using the Hoberman configuration should have a
substantial lower cost. A 10 m PERCS will have a
lifetime of over 5 years for altitudes above 600 km.
The PERCS solution is cheaper, lighter and has a much
longer lifetime than the Mylar balloon solution. Another
alternative is to scatter radar off large orbiting satellites
like the International Space Station. This provides pre-
cise target location but cannot give an monostatic radar
cross section with better than 6 dB accuracy. PERCS is
the best target solution to meet the HF radar community
requirements.
[48] Once PERCS is in orbit, most of the applications
will involve calibration of ground HF transmitter anten-
nas and radars. Figure 34 illustrates the concept of
operations where the sphere would fly over the ground
radar for measurements of the radar system sensitivity.
Both vertical and oblique antenna patters can be cali-
brated with PERCS. Most publications of HF radar
results use ‘‘Signal to Noise Ratio – SNR’’ or ‘‘Arbritary
Units’’ to describe the radar backscatter data. With
calibration using PERCS, the units can be expressed a
absolute cross section in dB m2. This is especially
important in study of meteor head echoes where the
strength of the radar backscatter is used to determine
the number of electrons produced by meteor entering
the atmosphere [Close et al., 2002]. High latitude
applications of the HF SuperDARN radars to observe
backscatter from auroral irregularities [Bristow and
Lummerzheim, 2001] and from heater induced irregu-
larities [Hughes et al., 2004].
18 of 23
RS1010
[49] For other applications, nearly vertical radar signals
can be deflected by the PERCS to permit radar measure-
ments at directions not accessible from the ground.
Figure 35 illustrates side scatter from the PERCS to
view horizontally propagating electrostatic waves such
as may be produced with the electrojet currents found
near the Earth’s equator. Both modified two-stream (i.e.,
Farley-Buneman) and gradient drift instabilities are used
to explain irregularities driven by the equatorial electro-
ject [Kelley, 1989]. Observing these irregularities with
ground radar is difficult because the vertical or oblique
radar beam is at an angle with the primary direction of
the horizontal irregularity structure. By using the geom-
etry illustrated in Figure 35, ground radar waves can be
scattered into the horizontal direction with the PERCS.
After backscatter off the irregularities, the radar signal is
scattered downward into the aperture of the ground radar.
Since the RCS for PERCS is accurately specified for
bistatic scatter and the PERCS sphere has been used to
calibration the ground radar system, the absolute inten-
sity of the electroject instabilities can be determined.
Similar applications can be used to study high latitude,
field-aligned irregularities where the magnetic field is
nearly vertical. High power transmitters such as available
with the HAARP facility in Alaska, the SPEAR very
high latitude heating facility in Norway, and the planned
future HF facilities for Arecibo, Puerto Rico and Jica-
marca, Peru will be required to give strong enough
signals for the PERCS oblique-scatter mode to work.
[50] In summary, the design of the wire-frame radar
calibration target uses both mechanical and electromag-
netic modeling. The PERCS mechanical design employs
polygons to tile the surface of a sphere. Deployment of a
large PERCS as a satellite with scissored hinges is based
on the Hoberman transformable sphere design. Expan-
sion after release into orbit is accomplished with torsion
springs at the pivot points of the scissors. Once the
PERCS is expanded into a 10-m sphere, it will provide a
well calibrated radar target for high frequency radars.
The WIPL EM code has shown that the radar cross
section will be nearly independent of orientation for
frequencies less than 23 MHz. At 600 km altitude, drag
computations with the NRL MSIS atmosphere demon-
strate that the sphere will remain in orbit for at least
5 years. Once in space, the PERCS target will provide
ground HF radars the ability to routinely check system
sensitivities and target location algorithms.
Appendix A: Geometric Construction of
Spherical Polyhedra With Hexagonal Faces
From Truncated Icosahedra
[51] To illustrate the construction of a high order
polyhedral approximation to a sphere, the V240 design
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure A1. Truncation of an icosahedron to yield pentagons.

will be outlined. Each edge of an icosahedron is trun- new hexagon. The equations for the angles in Figure A2
cated at a distance of 1/6th. With an icosahedron’s edge are
length = 2, the truncation yields line segments of length
1/3. For illustration, three of the pentagons resulting from  
1 b
the truncation are projected on to a flat surface q1A ¼ 2 Sin ;
(Figure A1). 2 R1
 
[52] Next the remaining long edges are divided into a L12
series of lines and hexagons. The lines between the q12 ¼ 2 Sin1 ; ðA1Þ
2 R1
pentagon vertices lie on a great circle between points 1  0 
b
and 2 in Figure A2. The side b is the length of one q12  2q1A ¼ 2 Sin1
pentagon edge. The side 2b0 is the diagonal across each 2 R1
qA2 ¼ q12  q1A

Figure A2. Angles and spherical edges for construction of the hexagon surfaces on a sphere.

19 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure A3. Initial line segments between pentagons and yet undefined hexagons.

where the quantities b, R1, L12 are defined by the initial
truncated icosahedron. Because of the curved surface, the
length b0 is greater than b.
[53] Equations (A1) are used to define the points A and
B along the great circle. The red segments (1-A) and (B-2)
are drawn both in the planar projections and the spherical
surface of Figure A3. Additional line segments between
the free ends of the red lines are added with the orange
lines attached to single hexagons, yielding the segments
shown in Figure A4. Finally all the undefined edges are
obtained by joining the uncommitted end points of the
orange and green lines to give the wire frame in
Figures 2b. All of the red and orange edges are of
the same length while the other edges will be longer
because of the curved spherical surface.
[54] As a result of this procedure, a spherical polyhe-
dral surface with 240 vertices is formed with 240 edges
of equal length and two sets of 60 edges with slightly
longer dimensions. Figure A5 shows the sphere edges
color coded by length and the distribution of these
lengths. Using a similar procedure, the edges for the

Figure A4. Line segments obtained along great circle paths of lines and hexagons between
existing red line-segment end points.

20 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

infinite series. In the Rayleigh region with k0r < 0.4, the
backscatter cross section is given by
sðl0 rÞ ¼ pr2 9ðk0 rÞ4 ðB2Þ
In the high frequency (or short wavelength limit k0r >
20) the spherical radar cross section is simply the
geometric projection area
sðl0  rÞ ¼ pr2 ðB3Þ

Appendix C: Design of the Hoberman

Figure A5. Distribution of edge lengths for the V240 Scissor Arms
spherical polyhedron based on an icosahedron with an
edge length of 2 units. [56] The angles and length for each strut in the Hober-
man scissor arms are determined by geometric consid-
erations. Consider the diagram shown in Figure C1. Two
higher order wire frames can be constructed [Wang and identical scissor arms have lengths L1
L2 joined at a
Chiu, 1993]. pivot with angle q12. In addition, L3 is defined the
distance between the end points of each arm. The pivot
point moves along a vertical line with a radial distance
Appendix B: Far Field Backscatter Cross Rb from the origin. The end points follow converging
Section for a Perfectly Conducting Sphere lines that have radial distances Ra, Rc, Rd, and Re from
the origin. Using the sine law, a constant L0 of the
[55] The general Mie scattering for radar echoes from scissors is given by
spheres is given by Ruck et al. [1970]. Far field solutions
exist when the distance to the sphere (R) is much larger L1 L2 L3
L0 ¼ ¼ ¼ ðC1Þ
than the radius of the sphere (r). For a perfectly con- Sinq23 Sinq13 Sinq12
ducting sphere with R r, the monostatic cross section
looking in the direction of the incident polarization is where the angles are defined in Figure C1. When the
given by scissors are fully, Rb = RH where
 2  
q13  q23
4p X1
n1 nðn þ 1Þ

 RH ¼ L0 Cos ðC2Þ
sðl0 Þ ¼ 2  ðiÞ ðAn þ iBn Þ ðB1Þ 2
k0  n¼1 2 
When Rb = L1, the scissors are fully collapsed. Notice
where the at the internal angles (q12, q13, q23) of the triangle
with sides (L1, L2, L3), also determine the angles lines
2n þ 1 jn ðk0 rÞ
An ¼ ðiÞn ; emerging form the origin. Using the Sine and Cosine
nðn þ 1Þ hðn1Þ ðk0 rÞ laws yields the variation of the radial distances to the
2n þ 1 ½k0 rjn ðk0 rÞ0 scissors end points as the pivot radius, Rb is varied
Bn ¼ ðiÞnþ1 h i ;
nðn þ 1Þ k rhð1Þ ðk rÞ 0 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 n 0  ffi
Ra ¼ L0 1  Cos q13 þ Sin1 ðRb =L0 Þ
2p 2pf @ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
k0 ¼ ¼ ; ½ 0 ¼ ;
l0 c @ ð k0 r Þ Rc ¼ L0 1  Cos q23  Sin1 ðRb =L0 Þ
rffiffiffiffiffi rffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi ðC3Þ
p ð1Þ p ð1Þ Rd ¼ L0 1  Cos q23 þ Sin1 ðRb =L0 Þ
jn ð zÞ ¼ J 1 ð zÞ; hn ð zÞ ¼ H 1 ð zÞ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2z nþ2 2z nþ2
Jn(z) is the cylindrical Bessel Function of the first kind,
Hn(z) is the Spherical Bessel function of the third kind
and the wavelength is related by the speed of light c to
the radar frequency by l0 = c/f. The RCS is computed
numerically from (A1) with a finite truncation of the
21 of 23
 ffi
Re ¼ L0 1  Cos q13  Sin1 ðRb =L0 Þ
The choice of the dimensions and angles for the arms can
be determined by the selection of three parameters R1 the
fully deployed radius to an end point with the two scissor
arms merge together, L3, the distance between neighbor-
ing vertices at deployment and L4 the minimum value of
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT RS1010

Figure C1. Geometry for the design of Hoberman scissors.

Rd when the sphere is stowed. With these given and Technology Organisation. The authors thank one of the
parameters all the other mechanical specifications are referees for suggesting the discussion on the SuperDARN radar
given by signal-to-noise for scatter from PERCS.

  References
q12 L3 L4
Cos ¼ ; Sinð2 q23 Þ ¼ Sinð2 q12 Þ; Bristow, W. A., and D. Lummerzheim (2001), Determination of
2 2 R1 L3 field-aligned currents using the Super Dual Auroral Radar
q13 ¼ p  q12  q23 Network and the UVI ultraviolet imager, J. Geophys. Res.,
L3 ðC4Þ
L0 ¼ ; L1 ¼ L0 Sinðq23 Þ; L2 ¼ L0 Sinðq13 Þ 106, 18,577 – 18,588.
Sinq12   Close, S., M. Oppenhein, S. Hunt, and L. Dyrud (2002), Scatter-
q13  q23 ing characteristics of high-resolution meteor head echoes
and RH ¼ L0 Cos
2 detected at multiple frequencies, J. Geophys. Res.,
107(A10), 1295, doi:10.1029/2002JA009253.
Hedin, A. E. (1987), MSIS-86 thermospheric model, J. Geophys.
[57] Acknowledgments. The research was funded by the Res., 92, 4649.
Office of Naval Research at the Naval Research Laboratory. Hoberman, C. (1990), Reversibly expandable doubly-curved
The authors would like to thank Benjamin T Root of the Naval truss structure, United States Patent 4942700, 24 July.
Research Laboratory and Robert M. Robinson of the National Hoberman, C. (1991), Radial expansion/retraction truss struc-
Science Foundation for helpful discussions on the PERCS tures, United States Patent 5024013, 18 June.
concept. The original idea of using a large sphere in low Earth Hughes, J. M., W. A. Bristow, and R. T. Parris (2004), Super-
orbit for global calibration of ground HF radars was first DARN observations of spectral enhancements excited during
suggested by Gordon Frazer of the Australian Defence Science

22 of 23
RS1010 BERNHARDT ET AL.: HF RADAR CALIBRATION TARGET IN ORBIT
an ionospheric heating experiment, Geophys. Res. Lett., 31,
L08808, doi:10.1029/2004GL019613.
Kelley, M. C. (1989), The Earth’s Ionosphere, Academic, San
Diego, Calif.
Koludzija, B., J. Oganjanovic, T. Sarkar, M. Tasic, D. Olcan,
B. Janic, and D. Sumic (2004), WIPLE-D Software Users
Manual, WIPL-D Ltd., Belgrade.
Lester, M., et al. (2004), Stereo CUTLASS – A new capability
for the SuperDARN HF radars, Ann. Geophys., 22, 459 –
473.
Montenbruck, O., and E. Gill (2000), Satellite Orbits: Models,
Methods and Applications, Springer, Berlin.
Ruck, G. T., D. E. Barrick, W. D. Stuart, and C. K. Krichbaum
(1970), Radar Cross Section Handbook, Plenum, New York.
Walker, A. D. M., R. A. Greenwald, and K. D. Baker (1987),
Determination of the fluctuation level of ionospheric irregu-
larities from radar backscatter measurements, Radio Sci., 22,
689 – 705.
23 of 23
RS1010
Wang, B.-C., and Y.-N. Chiu (1993), Geometry and orbital
symmetry relationships of giant and hyperfullerenes: C240,
C540, C960, and C1500, Synthetic Metals, 55 – 57, 2949 – 2954.

P. A. Bernhardt and C. L. Siefring, Plasma Physics Division,
Naval Research Laboratory, Washington, DC 20375, USA.
(bern@ppd.nrl.navy.mil)
M. Davis and C. Hoberman, Hoberman Associates, Inc.,
40 Worth Street, Suite 1680, New York, NY 10013, USA.
D. L. Hysell and M. C. Kelley, Department of Earth and
Atmospheric Sciences, Cornell University, Ithaca, NY 14853,
USA.
S. M. Koss and M. Nurnberger, Space Engineering
Department, Naval Research Laboratory, Washington, DC
20375, USA.
A. C. Nicholas, Space Science Division, Naval Research
Laboratory, Washington, DC 20375, USA.
S. P. Rodriquez and J. F. Thomason, Radar Division, Naval
Research Laboratory, Washington, DC 20375, USA.