Application possibility of 3.8 m telescope for radio …diono/meetings/NRO...Grinding Machine...
Transcript of Application possibility of 3.8 m telescope for radio …diono/meetings/NRO...Grinding Machine...
Application possibility of 3.8 m telescope for radio telescopes
20150730
Mikio Kurita
Kyoto Univ.
Optical and Radio Telescope
Optical Radio
Similarity • Cassegrain system Primary and Secondary Reflector • Alt-Az mount Difference • Wavelength Accuracy of Refs. • Size of primary Ref. • F-ratio Length between the Refs. →location of Altitude axis and Nasmyth focus • Enclosure
Introduction
Cost scaling law Single is more economy.
Minel 1981
∝D2-2.7
Belle & Minel 2004
∝D2.45
• Important thing is innovation which shift the line to downward.
Cost scaling law
∝D2.45
National foundation is weakly limited Technology to break size limitation is more meaningful than technology to cost reduction
In optical telescope, innovations happened in • mount type(Equatorial to Alt-Az mount) • F-ratio (3 to ~1) • Mirror configuration (monolithic to
segment or multiple mirror)
Quiz Which direction is stiff?
𝑓 𝑓
Quiz Stiffness of cantilever
F
When you load constant magnitude force on the end of the rod, and change the direction of the force. What shape of trajectory does the end draw?
Quiz Stiffness of cantilever
Quiz Stiffness of cantilever
Every end describes a circle
Quiz more general shape
A: Ellipse Ellipse is more general
Quiz more general shape
Displacement ellipsoid in case of 3-D
Under condition of
• constant boundary condition
• elastic body
(follow the Hooke’s law)
• concentrated load and volume force
国宝東大寺南大門金剛力士像
An arbitrary point on any object follows the law.
Displacement ellipsoid Application
• Deformation can be grasped by analysis in three elevation angle.
• Sag of the dish is axis-symmetry.
• Notice: The boundary condition must be constant or approximately constant to the whole structure.
Natural Frequency
• Natural frequency indicates the mass-stiffness ratio of the structure
• Natural frequency of truss structure depends its geometry only
=
Stiffness of truss structure against gravity
> =
Cross section scaled
>
Cross section scaled
Natural Frequency with mass
Now you design the supporting structure (purple members) for the box structure. Which structure have higher natural frequency? How size of the cross section of the member is more moderate?
3-bipod 4-bipod 4-tripod
Natural Frequency with mass
mM
ff
c /1
max
Nat
ura
l fre
qu
en
cy
Mass of purple member: 𝑚 (proportional to the cross section)
Natural frequency can be described by only two parameters 𝑓𝑚𝑎𝑥 and 𝑀𝑐 , which are governed by geometry of truss structure Once you decide the geometry, the highest natural frequency can be estimated by 𝑓𝑚𝑎𝑥 We can estimate the character of the structure from this curve
Highly optimized but fragile
Safety
Higher frequency available but confusing
3.8 mTelescope
Aperture: 3.8 m
Focus: Nasmyth × 2 F/6
Field of view: 10’ , 1°
Observational
Wavelength: 0.4 to 4.2 um
Adaptive Optics: J, H bands
Pointing speed < 1 min (whole sky)
Elevation speed 2°/s
Azimuth speed 3°/s
8 m
8 m
20 ton
Mirror Factory
• A venture company was newly build for research and development of large optical elements.
• We aim to shorten processing a large optical element with high precise grinding system instead of polishing.
Facility Astroaero Space Inc.
Grinding Machine (N2C-1300D)
Machining and Test System
10 m
Interferometer &
Anti vibration system
Grinding Machine
Chamber T:23±0.1 deg
Mirror Factory
Motion accuracy of the grinder
1um
Typical error in P-V = 0.1 um/m ※The value includes the error of the gauge
Mirror Factory
Grinding accuracy
Grinding P-V = 0.4um
• Size: Φ610
• Form: Flat
• Material: Clearceram
(zero expansion ceramics, OHARA)
• Processing time: Few hours
Mirror Factory
Problem of Direct Support
The error induced by a direct support comes up to several microns, in case of large but thin optics.
Before setting
After setting (Stress loaded)
Grinding
After grinding (Stress released)
Mirror Factory
blank
machinery table
Kinematic Support Grinding
• Three fixed points for kinematic support (without over constraint). • Multi springs to assist of the support and to decrease friction
between the segment and fixed points • The deformation induced by grinding pressure was simulated
beforehand, and then a stone is controlled to correct the deformation.
Segment
Fixed point
Fixed point
Spring Spring
Without correction
With correction
Figure Error
Mirror Factory
Kinematic Support Grinding
Deformation map with grinding force of 5N
P-V = 5 um The black points show the position of fixed points
Figure error without the grinding pressure correction Each color of the lines corresponds to that of the right figure (1div = 1 um)
Mirror Factory
6 μm
Kinematic Support Grinding
Figure error with the grinding pressure correction Each line is described in a same manner of the preceding page
(1div = 1 um)
Error map -1 um to 1 um
Mirror Factory
Test the Mirror CGH Interferometer
Test beam Goes through the CGH as 1th order and back through as 0st order Reference beam Goes through the CGH as 0th order and back through as 1st order • Semi common pass → Robust against turbulence and vibration
Reference
Imaging lens
Test the Mirror
CGH (Computer Generated Hologram)
Grating
1st order diffraction
Incident
Designed grating pattern produces intentionally distorted wave front
CGH (Computer Generated Hologram)
CGH for the inner (left) and outer (right) segments Each line represents 30 lines in the real pattern
Figure error of the mirror (Structure Function)
Requirement
Result
10
100
RM
S e
rro
r (n
m)
Separation (m)
1 0.1 0.01
Test the Mirror
Mechanical Probe System (under developing)
• Applicable to freeform – Interferometer needs each reference surface meeting test surface
• Wide range and area
• Robust to the environment
• Small Linear stage
Rotary table
Probe
Test the Mirror
Mechanical Probe System Result of φ800 Spherical mirror
Figure map integrated 60 lines scanning The mirror supported by three points (indicated red points)
p-v 900nm Deformation by three point supporting (FEM simulation)
Test the Mirror
200
-200
[nm]
Mechanical Probe System Comparing with interferometer
• The small scale structure (centrosymmetric pattern) with amplitude of 50 nm is successfully detected
Interferometer
Test the Mirror
This work
New data reduction idea
• Multi continuous data generally conflict at overlap region.
• This problem can be removed efficiently and reliably by an idea that recognize the data as an elastic body. from JAXA
Example Scanning data of flat surface
Scanning path Actual data analyzed by current algorithm
by this algorithm
Predecessor well know nature…
Salvage the embedded information
Current techniques
Spatial filter or data binning techniques lost some information but this technique keep them
This techniques
Supporting mechanics
・SHWFS ・Phase Camera
Component of Segment Control System
Edge Sensor
Processor
Supporting mechanics
Segment
Segment
Actuator
Actuator
Kinematic Support for Segment
Actuator connected
Supporting point ×9
Segment
Segment Mirror
Flexure Pivot
Axial rod ×9 (Flexure)
• Segment mirror is supported by multi points in such a way that the segment floats without over constraint.
• Tip-tilt and piston of the segment are actuated via
the three pivots.
Actuator ×3
Segment Technics
Segment Techniques Supporting Structure
Deformation @ 0 deg RMS = 30 nm, P-V = 150 nm
1
10
100
1000
1 10 100 1000
Fig
ure
erro
r rm
s [n
m]
Separation[mm]
Deformation @ 90 deg RMS = 12 nm, P-V = 81 nm
Red:Requirement Blue:Support
[email protected] [email protected]
[email protected] [email protected]
Simulated SR considering the actual figure error of segments and supporting system
Segment Technics
Segment Techniques Edge Sensor
Modified DS2001
Requirement Resolution (RMS nm) < 10 Stability (P-V nm/10hr) 50 Linearity > 90% Sample rate Hz > 10 Range mm 1
13 hr
30 nm
Result under temperature changes of 5 degree
Segment Technics
Warping Harness
Simulation
Result obtained by CGH interferometer
Segment Mirror
Warping harness provides torque on the flexure pivot in order to • reduce the figure error of low order • correct the positioning error of the segment on the
mount(power and astigmatism) • correct the seasonal distortion
Segment Technics
Actuator
Lever (Flexure pivot)
output 1/30
Actuator
Decelerator
Input at actuator (μm)
Ou
tpu
t (
μm
)
Time (second)
Ou
tpu
t (μ
m)
Reverse motion test
Segment Technics
11/13
Feedback OFF
Actuator
Feedback ON
Put a load on and off the segment
Time (second)
Dis
pla
cem
ent
(nm
) Segment Technics
Processor
Optimized Configuration of Edge Sensors
A part of the result of control stiffness of the segment system
Cross section of the Segment and Edge Sensor Arrangement
Segment Technics
Lightweight Structure
• Truss structure
• Large arc rails for elevation bearing
• Homologous deformation optimized by genetic algorithm
Moving mass around elevation :8 ton (4 ton optics) 1/5 of conventional telescope
Truss Structure
• The telescope tube consists of truss structure only
• Advantages
• Light but stiff
• Small cross section and low air drag
• Low heat capacity
• Large surface area and low heat inertia
Back view of primary mirror
Lightweight Structure
Large Arc rails
• The primary mirror is directly supported by large elevation rings without center section and mirror cell
center section
cell
This mount Conventional mount
In conventional mount ,because the center section and cell are supported by its periphery, bending moments load on them
Conventional radio telescope mount
Lightweight Structure
Homologous Deformation
• Serrurier Truss for keeping the optical alignment between the primary and secondary mirrors
– similar but more primitive idea to homologous deformation utilized in radio antennae
primary Secondary
Optical axis
Lightweight Structure
Kunda Masashi @ 夏ゼミ
Parent
88.51 kg
1st Generation
84.14 kg
3rd Generation
82.71 kg
10th Generation
77.52 kg
18th Generation
76.22 kg
Genetic Algorithm (GA) for lightweight and homologous deformation
Lightweight Structure
GA is a way to optimize a subject which has multi conflict object. In GA, a model devolves like a evolution of life by using Selection, Mutation, & Crossover
Cantilever example
1 tf
4 tf
Genetic Algorithm for lightweight and homologous deformation
Lightweight Structure
Requirement in homologous performance (El. 88~20 deg) • nodes for primary mirror in normal <0.1 mm • nodes for secondary mirror in lateral <2mm→0.4 mm • nodes for tertiary mirror in normal <0.05 mm Variable • Position of node (deleting is possible) • Cross section of straight member (deleting is possible) Material is selected from JIS member only Rate of crossover 0.8 Rate of Mutation 0.01
進化後のモデル
ホモ
ロガ
ス変
形(m
m)
重量(ton)
仕様範囲
Genetic Algorithm for lightweight and homologous deformation
First model
Final model
Lightweight Structure
初期解 最適解の一例
Acceptable deformation
0% 200%
Lightweight Structure
Genetic Algorithm for lightweight and homologous deformation
Homologous Deformation
• Dish must keep its shape under dead loads to satisfy optical performance.
Fixed elevation Active Support Rigid Structure Homologous Deformation
Homologous Deformation
Classical method for Homologous Deformation
• Element for Optimization – Objects: Homology of selected points 𝒏 on the
reflector • Sufficient 𝑛 must be selected, 𝑛 ∝ 𝐷2/𝜆2
• Weight and stiffness are not primary
– Design parameters: Degree of freedom of the model 𝑵
– Formulation: Stiffness matrix 𝐾 of the model and Newton method on the derivative of 𝐾−1
– Constraints: Geometry of the first model is fixed
𝒏
𝑁 > 𝑛
Homologous Deformation
Discussion Panel shape: Hexagon or Annular
• Hexagon – more efficient in process – mechanically stiffer (wider angle
of 120 deg at the corner) – more robust for segment control
system – identical supporting system for
all segments – easy to design the back structure
• Annular – less degradation of image due to
the diffraction – smaller number of the spare
segments
Dark area: waste region under rotary motion processing
Annular segment Hexagonal segment
Discussion Panel
• Panel – Material: Aluminum
– Support point: 3
– Thickness: 80 mm
– 30% Weight: 90 kgf • Solid 300 kg
– 30m dish: ~500 panels
• Our facility can deliver this size of panel in a few days per panel.
P-V=5 um, RMS=1.2 um (Solid)
1500
Discussion Configuration of Panel
All panels independently supported
Group panels supported by sub cells (Each sub cell supported by 3 points)
All panels connected to a single panel
Independent Group Single
Control type positioning positioning force
Sensor and Actuator
high precision long range
high precision short range/ Low precision Long range
Low precision
Homologous structure
yes no yes