One strong candidate for an optical design is
a Gregorian optics in which a secondary mirror can be placed in a position
conjugate with the lower atmospheric turbulent layer. However, in this
configuration, a telescope tube would become long and a dome would
become large. As an alternative, a new optical system involving three
aspherical surfaces is being considered.
1. Three aspherical surface optical system
Figure 1 shows a layout of an optical system
in which three aspherical surfaces are used to
produce a flat focal plane without aberrations
across a field of view spanning 10 arcmin.
Element mirrors comprising this optical system
are listed in Table 2. In contrast to a three
aspherical surface optical system based on the
Korsch system used in Japan Astrometry Satellite Mission
for Infrared Exploration (JASMINE), a magnification
of the third aspherical mirror in the present
design is close to 1, ensuring that a wide field
of view is maintained.
The primary mirror is ellipsoidal with a diameter of 30m and focal ratio
of F/1.5, and is composed of 1080 hexagonal mirror segments. A 4m hyperboloidal
secondary mirror, a 4m flat folding tertiary mirror, and a fourth mirror
produce an ordinary Nasmyth focus and a virtual image plane in the vertical
direction. This virtual plane becomes the final image plane after the
light is refocused by a fifth 4m ellipsoidal mirror.
As the first virtual image plane and the final image plane produced by
a two-mirror system are effectively in the same
position, bent light rays are superimposed.
However, since the final image is an inverted
mirror image of the virtual image, superposition
of light rays can be avoided by using only half
of the circular field of view. If a fourth
mirror is added and the opposite side of the
Nasmyth platform is used, two semicircular fields
of view can be produced at one Nasmyth focus, making it possible to sample
a full 20 arcmin field of view.
2. Imaging performance
Figure. 1: Three aspherical surfaceoptical system
A field of view occupies an effective area
of approximately 1m in radius, allowing several instruments to be
placed at each focus. A new focus can be established on the opposite
Nasmyth platform, providing suitable locations for additional instruments.
As the beam is bifurcated at the center of the optical axis, a maximum
50% of the total light is cut off by vignetting. However, as this vignetting
occurs only 1mm or less from the center of the field of view, which
is beyond the diffraction limit, this optical system in practice can
be regarded to be free of vignetting.
A spot diagram is shown in Fig. 2. Within a radius of 8 arcmin, the size
of a spot is within the diffraction limit of visible light.
3. Peripheral optical system
Figure. 2: Spot diagram of JELT three
aspherical mirror optical system.
Note that spots out to a radius of 8 arcmin fall within a diffraction limit of 30m telescope.
By introducing a lightweight truss structure,
the telescope structure can be constructed as a natural extension of
present 8m-class telescopes. However, as a diffraction size of a 30m
telescope is 1/3.75 that of an 8m telescope, and with advanced adaptive
optics providing ultra-high compensation, ultra-high resolution observations
are expected to be possible. It should be noted, however, that requirements
for atmospheric dispersion compensation in a 30m-class telescope will
be an order of magnitude more demanding than those for the present 8m-class
telescopes. Such requirements are being examined in parallel with other
1. Basic investigation of overall structure
A change in the direction of gravity for any part of the mechanical structure
will result in structural deformation. To eliminate
this deformation, it is necessary to increase
stiffness of structural elements. However,
increasing the stiffness often results in increased
weight, which exacerbates deformation. Furthermore,
as a telescope requires a precise drive system
for tracking of celestial objects, limiting
an increase in weight is critical to the design
of the drive system.
Such deformation is generally
dealt with by employing a Serrurier truss structure,
which ensures that deflection of the primary
mirror section and the secondary mirror section
are matched under any orientation and degree
of deformation. Serrurier truss is a structural
arrangement of triangles that support the top
ring and bottom mirror cell with respect to the
center section of the telescope. A correction
optics for primary focus or the secondary mirror
are supported by tension applied by a spider (plate
or rod-shaped structure) suspended from the top
ring. The plane of the center of gravity of the
top ring, including the secondary mirror, is
connected to the center section by the Serrurier
truss, which is coupled by hinged joints to ensure that external
force does not exert a bending moment on incident
beams. Thus, when the telescope is tilted, the
top ring undergoes parallel shift in a lateral
direction while retaining its perpendicular orientation
with respect to the axis. The primary mirror is also
connected to the center section by the Serrurier
truss structure and undergoes parallel shift when tilted. The system
is designed such that this shift matches that of the top ring.
The Serrurier truss structure effectively compensates
for deformation that occurs when the telescope
is tilted. However, to ensure that the weight
of the top ring balances the weight of the entire
structure, including the section that
supports the primary mirror, the top ring must
be made very heavy. Large telescopes built using
the Serrurier truss system will therefore be
very heavy, limiting an aperture size to around 10m.
Figure. 3: Telescope structure of JELT
Therefore, an alternative to the Serrurier truss
structure is being investigated. The 45m radio
telescope at Nobeyama Radio
Observatory is a good example, in which the
primary mirror is supported by a structure that
resembles a truss structure, but the telescope
itself is supported not by the center section
but by an R-shaped rail structure
attached to the bottom of the primary mirror.
Therefore, the lateral displacement of the primary
mirror that occurs when the telescope is tilted
is practically negligible in this configuration.
However, whereas a top ring of radio telescopes
can be made very light, allowing lateral displacement
to be neglected, the top ring including the secondary
mirror in an optical/infrared telescope
is very heavy, and lateral displacement must
be considered. For JELT, it is proposed that
the entire top ring be made heavier and that
the secondary mirror itself be moved in the lateral
direction. Using this system it is expected that
the telescope will perform well without relying
on the Serrurier truss structure. A conceptual
design of a telescope with this support structure
is shown in Fig. 3.
As a telescope becomes larger, its weight increases,
shifting its eigenfrequency to lower values.
A relationship between tube length and eigenfrequency
is shown in Fig. 4.
3. Elevation bearing section and horizontal rotation
Selection of Bearing:
Figure. 4: Relationship between telescope
tube length and eigenfrequency
Use of a roller bearing
makes it possible to drive rotational motion
of a large load smoothly with high accuracy.
Roller bearings are therefore commonly employed
for altitude and azimuthal axes of small
telescopes. However, the largest
roller bearings in use today are 7m in diameter,
for use in tunnel excavation. It is practically
infeasible to fabricate larger roller bearings,
and bearings of such size will be very heavy.
Hydrostatic bearings are therefore used for larger
telescopes, such as Subaru telescope. For
JELT project, a drive method based on hydrostatic bearings,
similar to that used for Subaru telescope,
is being considered.
Figure 6 shows the concept of the hydrostatic
bearing . An amount of gap (h) of support
provided by a hydrostatic bearing is given by
where Q is an oil flow rate, eta is dynamic viscosity of oil, sigma
p is a loss of pressure in the gap, and b is
a gap width ( =2(a1+a2) ). In the OWL project,
a 100m telescope proposed by European Southern
Observatory (ESO), a bogie car has been proposed
instead of a hydrostatic bearing. Advantages
of a bogie car are twofold: (1) the length of
the bogie car can be changed dynamically to compensate
for any lack of flatness of the base rail, and
(2) the bogie car itself provides the drive power,
whereas a separate motor is required to drive
conventional hydrostatic bearings. However, as
the bogie car system requires very complex
technology, and has never been implemented in
practice, it is not being considered for the
4. Overall structure of JELT
Figure. 5: Conceptual diagram of a hydrostatic bearing
Required specifications and design parameters
for JELT structure are listed in Table 3.
Examples of the structure design for JELT
are detailed below.
Elevation bearing section:
Table 3: Required specifications and design parameters for JELT structure
||20 m or more (nominal 30 m)
||diagonal 1 m (hexagon) or fan-shaped
||Gregorian (or Cassegrain)
||Three aspherical mirrors, 2 Nasmyth foci
(1 each optical and infrared)
|two Nasmyth foci
|Field of view
(non-block field of view)
|5 arcmin or more
|Field of view
(adaptive optics field of view)
|1 arcmin or more at 2 μm
||1 arcmin at 2 μm
||0.39 - 5 μm
||0.39 - 5 μm
||Altitude: 20 - 80 deg
||Altitude: 20 - 80 deg
(range within which performance is guaranteed)
rigid horseshoes are arranged on both sides,
suppressing deformation of the mirror cells by gravity while maintaining
high rigidity around the elevation axis. Two
points on left and right sides of the horseshoe
cylindrical surface (four points total) are supported
in the radial direction by hydrostatic
bearing pads. The thrust direction is constrained
by clamping one of the horseshoes using two hydrostatic
bearing pads. This kind of support system constrains
5 degrees of freedom, providing support
without excessive constraint. A weight is to
be reduced by changing a weight distribution
of the horseshoe and mirror cell.
Primary mirror cell:
most important issue to be resolved in designing
the structure of the primary mirror cell is a number of mirror
support points. Three support points are required
for each 1m-diameter hexagonal mirror, giving
a total of 3000 support points for 1000 mirrors.
One conceivable way to reduce the number of support
points is to combine a number of mirrors into
a cluster, as has been proposed for TMT project.
If a cluster structure is employed, assuming
19 mirrors per cluster, a total of 50 clusters
will be still necessary; corresponding to 150 support
points given 3 points per cluster. A question
then becomes how to support these 150 points.
It is difficult to obtain high rigidity using
a hexagonal support pattern proposed for
TMT. It is therefore proposed in the present
case that the primary mirror be placed in square
layout, which is expected to provide high rigidity
with a smaller number of cells. One of issues that remains
to be investigated is an optimum structure
to connect the square structural members of the grid to the horseshoe.
than using wires to apply tension to the upper
telescope tube as is generally the case, a self-supporting structure
will be employed. In the current proposal, the
spider will have an elliptical cross-section,
allowing vignetting to be minimized while maximizing rigidity.
Horizontal rotation section:
rotation section is supported by six thrust hydrostatic
bearing pads, with a radial hydrostatic
bearing pad in their center constraining the horizontal rotation
section within the plane. The thrust hydrostatic bearing
pads are designed to slide on a top surface
of a rail attached to its base. Magnets
to read encoders will be mounted on the outside
or inside of the rail, with coil assemblies mounted
on the telescope side. A tape encoder to detect rotation
speed will be installed on the rail side. To
compensate for changes in gap distance between
an encoder head and the tape due to effects such as
differences in thermal expansion, a gap-maintaining
mechanism is being considered. Furthermore, arrangement
of drive points near detection points (to
achieve co-location) to achieve precise control is being considered.
same yoke structure as used for Subaru telescope
is also to be applied in JELT. The yoke structure, which supports
the Nasmyth table and an elevation shaft, consists
of two groups of tripod trusses set on hexatruss
structure on the horizontal rotation rail. This
structure limits effects on the elevation
shaft thrust pressure by movement of the six
nodal points following any swell in the horizontal
rotation rail, which can result in undulation.
It is also possible to suppress fluctuations
in pointing of the telescope. The features
of this type of structure are listed below.
5. Investigation of eigenfrequencies and oscillation
modes using a simplified model
The load is transmitted
directly from the point
at which force is applied to the point
where it acts.
Forces that support the telescope structure
are transmitted directly and
linearly to the hydrostatic bearing pads.
Rigidity of an oil film in a hydrostatic
bearing is very high, and the
structure makes effective use of this high
||The mechanism consists of a minimum number of parts,
facilitating assembly and improving cost performance.
||The effect of swell in
the azimuth rotation rail (which may be present due
to error in manufacturing the system and/or due to change
in the system over time) is effectively compensated by this structure.
Using an analysis model shown in Fig.6, eigenvalue
and static solution analyses (for the case in
which a static acceleration of 1G acts on) were
performed to determine eigenfrequencies,
oscillation modes and deformation due to the
telescope's own weight as a check of appropriateness
of the basic structure.
Figure. 6: Model for eigenfrequency analysis
(1). Deformation due to gravity
||Deformation due to gravity,
under a static acceleration of 1G,
occurs at the Nasmyth table.
The maximum load is applied to the yoke legs,
under the weight of the Nasmyth table.
(2) Eigenfrequencies and oscillation modes
6. Secondary mirror support mechanism
||The oscillation modes of the principal members are as follows.
Oscillation in the direction parallel to the table.
Oscillation in the direction parallel to the tube.
Rotational oscillation around the azimuth(Az) axis.
Rotational oscillation around the elevation(EL) axis.
The primary oscillation mode is forward translation of the table, which occurs
with an eigenfrequency of 2.3Hz. The lowest eigenfrequency of tube oscillation
is 3.2Hz. The eigenfrequency of rotational oscillation around the azimuth axis
is 4.4Hz, and that of rotational oscillation around the elevation axis 4.5Hz
Figure. 7: Trade-offs in secondary mirror support mechanisms
Figure 7 illustrates trade-offs involved
in possible secondary mirror support mechanisms. A
support system shown on the left side of the
figure is Schwesinger support system used
in the primary mirror of VLT, consisting
of a rear surface axial support and an outer circumference
lateral support. This system requires a lateral
support mechanism in the outer circumference
of a secondary mirror, which introduces a source
of heat radiation in infrared wavelength. A
mechanism shown in the center of the figure
is a support mechanism used for the primary
mirror of Subaru telescope. In this system,
holes in the mirror are used for lateral support
at a center in thickness direction. This
system requires holes to be made in the mirror,
which affects a fabrication
process and a cost of the system.
In a system shown on the right of the figure,
two support mechanisms are adopted to provide
support in a lateral direction, positioned
such that an intersection of the support force
lines lies at the center of the mirror in
thickness direction. This system does not introduce unnecessary
infrared heat radiation, and also does not require
holes to be made in the mirror. This system is therefore the most desirable
Figure. 8: Secondary mirror support mechanism
Figure 8 shows specific proposals for secondary
mirror support mechanisms. The proposal involves
45 passive support points, 9 of which are for
the tip-tilt drive, with 3 fitted with sensors
to detect rigid positions of the secondary
mirror. A mirror cell is positioned with
6 degrees of freedom by 6 jacks. Reaction force
compensators are provided on top of the mirror
cell, on the opposite side from the secondary
mirror, to cancel the reaction force from the secondary mirror.
The features of the secondary mirror support
mechanism are listed below.
By arranging an actuator
in anti-nodes of oscillation mode, an effect
of the lower-order oscillation modes can be eliminated.
This means that the eigenfrequency of the secondary
mirror is effectively increased, allowing
control bandwidth to be increased substantially.
System without a support structure around the secondary mirror:
No thermal noise.
Large thrust with compact actuator:
type as the tip-tilt drive actuator used in Subaru telescope)
Maximum thrust: 500N
Dimension: 90mm diameter x 60mm height
Table. 4: Comparison of requirements for tip-tilt drive actuators
A comparison of requirements for the tip-tilt
drive actuators in JELT and Subaru telescope,
and numbers of actuators required,
are given in Table 4. If chopping frequency
is set at around 1Hz, a total of 6 - 9 actuators
are expected to be sufficient.
7. Drive mechanism
Features of the proposed drive system are as follows.
Direct-drive (DD) system.
- Lower cost than a gear-drive system.
- The drive system of Subaru telescope, which
has already performed successfully, can be applied.
- High resolution: Encoder resolution is 0.0005".
Features of the DD system are as follows.
General features of the DD system
- Zero friction
- High rigidity
The DD features of Subaru telescope are
inherited without alteration.
- Low heat production, high torque
A praseodymium magnet is used to obtain high
Low torque ripple
- Optimized skew of the magnet.
- Coil layout for minimum torque ripple.
For reference, comparison of a DD system,
a gear-drive system and a friction-drive
system is given in Fig. 37. In the gear-drive
system, it is necessary to continuously apply
torque via two gears in order to prevent backlash.
Thus, large fluctuations in friction occur when
the drive system is operating at low speed, particularly
upon start-up. Error in gear shape becomes
a source of external disturbance. Similarly,
in the friction-drive system, friction fluctuation
is also generated due to pressure applied
between rollers. In the case of JELT,
the roller would be too large to be made in one piece, and therefore must
be divided into several pieces. However, the
resulting seams may become a source of external torque disturbances. The
DD system is completely
free from such external torque disturbances.
For gear- and friction-drive systems,
rigidity of the drive system decreases by an
amount depending on rigidity of the mechanism
itself, whereas the DD system has no mechanism
other than motors, resulting in absolutely no loss of rigidity.
Considering the features of these systems, the
DD system allows high tracking accuracy to be
achieved relatively easily. Furthermore, a
cost of fabricating gears, as required for the
gear-drive system, is much higher per unit length
than a cost of magnets in the DD system, a cost of
which is comparable to the friction-drive system.
Accordingly, based on an overall evaluation, the DD system has been adopted
for JELT. The specifications of Subaru telescope DD system
are listed in Table 6.
Performance required for the DD system is
given below, in reference to Technical Report
No. 4 of TMT project.
||Slewing requirement 360deg Azimuth, 65deg Elevation / 5min
||1" on sky 1s
||10" on sky 3s
||100" on sky 10s
||1000" on sky 30s
||Maximum speed and maximum acceleration :
speed (using a rough precision encoder) : 1.2deg/s
speed (using a high precision encoder) : 0.2deg/s
Maximum acceleration : 0.05deg/s2
||Maximum torque :
Azimuth axis : 1x105 Nm ( ˜10 ton m)
Elevation axis : 2x105 Nm ( ˜20 ton m)
Azimuth axis : 1x108 kg m2
Elevation axis : 2x105 kg m2
Specifications of DD system proposed
for JELT, based on the DD system in Subaru
telescope, are compared with these required specifications in Table 5.
8. Tracking control system
Table. 5: Specifications of DD system required for JELT
The tracking control system developed for
Subaru telescope will be used for JELT. The features
of the control system are listed below.
||High-order filtering of control devices to obtain
a large control bandwidth
||Compensation for structural resonance and anti-resonance
||Speed estimation unit inside a control unit :
speeds, pulses from encoders are counted to estimate the speed.
At very low speed
(very low frequency), the speed is estimated by an observer's speed estimation unit.
||Total system analysis to degenerate structural modal order
||Structural design based on frequency response function
||Incorporation of predictive control
including converging acceleration during slewing, converging acceleration
during guiding, slewing, ADC driving,
ImR/InR driving and guide probe driving are made possible.
In addition, following additional functions must be incorporated :
||A high-order compensator to suppress external disturbances
such as those due to wind (H1 control)
||Automatic compensation for directional error due to
swell in the horizontal rotation rail
Adaptive optics system
Control software and computer system
Figure. 9: Block diagram of tracking control system for JELT