Focal plane alignment strategy: metrology

1
Focal plane alignment strategy metrology
assembly
A. Rasmussen, SLAC

• Tolerances (detailed PSF investigations)
• Required geometry (tolerance) of the focal plane
  FP specification placeholder is FLAT to 5µm,
  based on beam convergence etc.
• Stability of FP Actuation capability necessary?
  (piston,tip,tilt)
• (Assy of rafts entirely glossed over here)
• Buildup of 25 rafts (3x3 each) onto Integrating
  Structure
• Lab Support for Assy (Stitching vs. Direct
  Metrology)
• In situ alignment feedback/monitor (for finite
  stability over ?T ?g
• possibility for active feedback or band dependent
  focal plane model geometry..
• Thurston Laser Array (TLA) Thurston Segmented
  Diode Array (TSDA) built into FP structure..

LSST Camera F2F meeting (SLAC) 050323-24
2
LSST camera focal depth surface of spot size
minima and surface of zero ellipticity raytrace
calculations
Andy Rasmussen 050304 SLAC

• A brief description of the ray trace calculation
  and the derived surfaces of minimum spot size and
  of zero ellipticity. A large part of the
  motivation to perform this study was to confirm
  the tight focal plane positioning specifications
  quoted previously (5µm) based on scaling
  arguments. Curves are derived for several camera
  configurations (designations following each
  filter name B,V,R,I and Z.) The Z band
  configuration was used for performing the Y band
  filter calculation.

3
A Flexible and Articulate Ray-trace was
configured to perform simulations of the nominal
3.5-Short LSST baseline. All distances,
curvatures and coefficients were taken from
Lynn's presentation materials. The specific
configuration for using the V filter is
represented in the script LSST_V below
arasmus_at_foot.phys.columbia.edu cat
LSST_V asphere -R -19200 -k -1.254809 -z 0 -ro
4180 -ri 2573 -A6 5.8773e-9 \ asphere -R
-6032.68 -z -6034.41 -ro 1600 -ri 820 -k
-0.284514 -A6 -1.2799e-5 -A8 -9.1574e-7
\ asphere -R -8577.43 -z 224.09 -ro 2573 -ri 660
-k 0.129492 -A6 -2.2717e-7 -A8 -3.6963e-9
\ asphere -L -R1 -2739.40 -R2 -3803.20 -z1
-3814.619 -z2 -3882.931 -ri 0 -ro 800 -p1 1 -p2 1
\ asphere -L -R1 -5198.60 -R2 -2058.50 -z1
-4391.598 -z2 -4421.598 -ri 0 -ro 550 -p1 1 -p2 1
\ asphere -L -R1 -5630.32 -R2 -5630.32 -z1
-4788.965 -z2 -4807.465 -ri 0 -ro 390 -p1 1 -p2 1
\ asphere -L -R1 -3625.50 -R2 -17192.0 -z1
-4849.965 -z2 -4909.965 -ri 0 -ro 365 -p1 1 -p2 0
\ tran_ray -t 0 0 -4934.965 arasmus_at_foot.phys.c
olumbia.edu
where the parts not shown are the photon
source(s) and the focal plane module, each of
which either pipes photons to LSST_V or from
LSST_V to evaluate various performance parameters
of the configuration. This approach was taken to
determine details of the focal plane geometry,
image structure and so forth. NB the CCD
component bi_ccd, which simulates diffraction
into the Si bulk and lateral charge drift during
collection time, broadens the simulation
parameter space significantly. For the time
being, we assumed three separate detector models
No CCD (rays simply collapse onto a plane
parallel to and nearby the design focal plane),
CCD1 (100µm thick, 5E9 e-/cm2 bound charge
density providing the finite electric field
there) and CCD2 (100µm thick, fully depleted,
bound charge density provided by the epitaxial
bulk NA 1E12/cm3).
4
A sample focal spot size curve given for the
three focal plane models. Optics are described by
the baseline 3.5-short LSST V band
configuration (5370Å illumination).
Plotted are the radial and linear spot sizes as
the position of the focal plane surface is
varied. The primary and tertiary mirror are in
the Z direction. Recall that a single pixel size
is 10µm, or 0.01 mm, which corresponds to roughly
0.2 arcseconds..
5
A comparison of the radial spot size function and
the difference between the sagittal and
tangential RMS measures.
This is the data for the no CCD model, 5370Å, V
band filter configuration. NB the positions for
the smallest radial spot size and the symmetric
spot are different by about 13µm, more than twice
the focal plane tolerance placeholder of 5µm. The
difference curve changes qualitatively with
off-axis angle, while the smallest spot position
also depends on wavelength.
6
(No Transcript)
7
(No Transcript)
8
Parametric plot of focal spot behavior with focal
plane position.
A single monochromatic wavelength within the R
filter passband was used to generate these
measures of focal spot size and asymmetry. A
curve is given for each off-axis angle
considered, out to the FOV edge. An astigmatic
focus is represented by a curve with a nearly
flat bottom (e.g., 1.43) where sx trades for sy
and vice versa. Data points obtained from PSF
measurements are given for 1µm steps. Beyond 1.5
off-axis or so, the PSF loses its astigmatic
quality and no solution for sxsy exists.
(sagittal extent dominates)
(tangential extent dominates)
9
Effects to measured ellipticity values
In a flat focal plane geometry, sx!sy in general
over the FOV. However, typical values for the
asymmetry measure sx-sy are 0.6µm. According
to the ellipticity expressions introduced by weak
lensing cosmology, such asymmetries in PSF can
induce systematics in ellipticity values that are
readily estimated. For a 2-D Gaussian, the HPD is
equal to the FWHM, which is in turn Under
0.6'' seeing conditions, linear s values will be
0.25'' or 12.9µm. E1 and E2 are given by
and the ellipticity The degree to
which ellipticity distorts (due to systematic
effects) depends precisely on the light
distribution second moment (or simply the PSF) so
we must in general discuss effects to ellipticity
in terms of ?s2sx2-sy2 and the other effects to
the PSF extent, which we'll just call ss for now
(s for seeing). In the cases raytraced here,
E20 explicitly, and eE1. With ?s2 ! 0, e
should be e ?s2/(2ss22sy?s2) with position
angle (phase) value solutions of 0 and 90.
10
Using symmetry arguments, we normally expect even
numbers of solutions where sxsy.
Two solutions exist within 50µm of the nominal
focus position for off-axis positions between
0.6 and 1.4. Beyond 1.5 off axis, no solutions
appear in this range, for the V filter
configuration, at least.
sx gt sy
sx sy
sx lt sy
sx gt sy
sx gt sy
11
(No Transcript)
12
Surface functions for a CCD model including
refraction and finite penetration are
qualitatively similar to the no ccd model
We use the Y band filter passband (and LSST in
its Z filter configuration) to estimate effects
on the sx sy solution conditions. This choice
is based on the more extreme transparency of the
Si bulk in the CCD. The focal plane position
range was extended out to /- 100 microns to
perform these searches. The dominant effect is a
moderate piston value toward the primary mirror
(9µm shift), when using a CCD. Slight effects
in the dependence of ellipticity parameter E1 on
off-axis angle could arise from non-zero
incidence angle of light to the CCD surface.
Below are plots showing the Y band curves (in
Z-band configuration), for the model CCD1 (right)
compared to no ccd (left). Averaged over the Z
band, the 100µm CCDs had about 31 QE (for 173K).
E1gt0
E1gt0
Y band no ccd
Y band ccd1
E1lt0
E1lt0
E1gt0
E1gt0
13
Conclusions

• This work provides input that should fuel
  discussions relevant to focal plane flatness and
  sensor positioning tolerances, weak lensing
  science requirements and the trade between
  producing symmetric point spread functions at the
  cost of operating with a sub-optimal image size.

14
Focal Plane Assembly

• Tentatively there are 2 approaches for assembly
  lab support
• stitching metrology support
• direct metrology support

15
stitching metrology support
up-looking and down-looking confocal optical
sensors. m1z1(x,y)s(x,y) m2z2(x,y,f)-s(x,y) m2
'z2(x,y,fp)-s(x,y)-(axbyc) z1(x,y) is the
local FP surface function that would be solved
for and stitched together with adjacent z1
measures. entire FP is not mapped in 1 shot, and
propagation of errors may be estimated readily.
16
direct metrology support (1)
optical table carriages
17
direct metrology support (2)
optical table carriages laser turret system
18
direct metrology support (3)
metrology turret schematic
19
direct metrology support (4)
optical table, carriages, laser turrets in
assembly configuration
20
in situ FP alignment monitor (1)

• laser alignment plugs can be built into the
  individual CCD rafts and used as alignment
  monitors TLA TSDA.
• SLAC Group E members (R. Schindler, E. Lee, P.
  Kim, M. Perl) have expressed motivation to define
  and perform some of the necessary R/D work to
  flesh out this idea.

21
in situ FP alignment monitor (2)
single TSDA component (100 plugs total)
22
in situ FP alignment monitor (3)
Reference straightedge beam can be provided by
laser diodes making up the TLA..
23
in situ FP alignment monitor (4)

• 100 readouts
• 3 dof per raft
• 1 dof per laser diode
• 85 dof (7510)
• 75 raft actuators