The Orthogonal
Transfer Array
Astronomy & Astrophysics Decadal Survey
Large Synoptic Survey Telescope (LSST)
 6-8m equivalent aperture
 3 degree Field of View (FOV)
 1-3 Gpixel detector ($30-100M at present costs)
 Data pipeline, data mining, “National Virtual Observatory” (NVO)
The Panoramic Optical Imager
http://poi.ifa.hawaii.edu
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Science Goals – NEOs
 Near Earth (“killer”) asteroids
1907
1908
 Requires whole sky survey, good PSF, <10 sec readout
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Science Goals – SNIa
 Supernovae – The Future of the Universe – is it dominated by dark energy?
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Science Goals – SNIa
 Supernovae – What is the dark energy, and the future of physics
 What are the relative roles of
 Dark matter
 Dark energy (“Quintessence”)
 What is the dark energy equation of state?

Need ~300 SNIa at z~0.7 (Z band ~23 mag)
 red response and PSF essential!
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Science Goals – PN
 From the Main Sequence to the White Dwarfs
 will the Sun make a planetary nebula?
 It will, but only if white dwarf remnant has >0.55 solar masses
 We don’t know what sort of white dwarf the sun will make
 clusters, like M67, fill blanks in our knowledge
 need to survey lots of sky (~1 degree)
for very faint white dwarfs (V ~ 26)
 Need good PSF to separate galaxies from stars
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Many Other Science Goals
 Near Earth Objects
 Space junk / fast objects
 Trans-Neptunian objects / Asteroid belt
 Variability / microlensing / transients
 Proper motions / binaries / parallax
 High-z supernovae (z~0.4, z~0.8, z~1.2)
 Strong lensing
 Weak lensing
 Star properties / galactic structure
 Galaxy evolution / formation
 Large-scale structure / clusters
 z > 6 (first light in the universe)
 Rare unknown objects
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Traditional Mosaic Imagers
 Too expensive
 Too slow
 Poor red response
 Figure of merit:
M=
AWe
dq 2
 Collecting area A may be fixed, but
we can improve all three other factors
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Targeted Detector Enhancements
 Cost
Current mosaics ~0.5 ¢ / pixel CCD only
2-3 ¢ / pixel, with electronics, shutter, filter
 Speed
 Current mosaics have 30 –100 sec readout
 NEO searches call for <20 sec exposures
 Red response
 Current mosaics 25% QE at Z band, poor fringing
 Image motion compensation
 Only slow (~1 Hz) telescope guiding, but atmospheric
image motion and telescope shake at 10-20 Hz

National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Our Proposed Solution
Bring state of the art and system approach to bear
 Detector
 Bring yield to >50%; currently ~25%
 Extend red sensitivity by 3x
 Add image motion compensation  improve PSF by 20%
 CCD Packaging -- make it efficient for 100's of detectors
 Cryostat -- make it efficient for 100's of detectors
 Electronics -- design a cheap, scalable module
 Computers -- 1GHz Linux PC per 8K x 8K imager is enough
 Software -- pay attention to scalability and efficiency
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
The Orthogonal Transfer Array (OTA) –
A New Technology CCD Imager
 A new paradigm in large imagers
OTCCD pixel
structure
Basic OTCCD cell
OTA:
8x8 array of OTCCDs
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Components of an OTA
o Bond to carrying wafer
o Flip over
o Thin backside
o AR coat
One 512x512 CCD
One 4K x 4K
monolithic OTA
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Detector Details – Overview
Each CCD cell of a 4Kx4K OTA
 Independent 512x512 CCD
Individual or collective
addressing

1 arcmin field of view
 Dead cells excised, yield >50%

Bad columns confined to cells
 Cells with bright stars for guiding
 8 output channels per OTA

Fast readout (8 amps, 2 sec)
5cm

12 um pixels
 Disadvantage -- 0.1 mm gaps, but
gaps and dead cells are dithered
out anyway
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Detector Details – CCD Output
 Wrap output amplifier
and JFET follower
around under serial
register.
 Run clock, bias, and
address lines between
cells.
 Limit to ~100 micron
gap between cells
 Good metrology over
entire OTA
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Detector Details – Clock and
Analog Signals and Cell Addressing
Clocks and biases to OTA cells
8 channels of output from OTA
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Detector Details – Enhanced Red
 45um high-resistivity Si


extended red
response
very low fringing
National Science Foundation, July 23, 2001
Transfer Array
The Orthogonal
Detector Details –
Orthogonal Transfer
 Orthogonal Transfer


remove image motion
high speed (few usec)
Normal guiding (0.73”)
OT tracking (0.50”)
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Detector Details –
Image Motion Compensation
 “Rubber” focal plane permits low
order “AO” over degrees

Every 30 msec collect guide
star info (Guide stars are
plentiful for WIYN)

Compute a “displacement
surface” for all cells

Perform OT shifts of all cells
 Removes image motion from
atmosphere and telescope shake
on arbitrarily large angular scale
 Improvement in PSF (20-30%) is
very significant

Exposure time to a given S/N
decreases by 30-40%
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Package and
Demonstration Camera (QUOTA)
 4-side buttable package with multilayer ceramic substrate
 Flexprint to hermetic or
through wall
 Cryocooled bars
 Four OTAs = QUOTA
(8K x 8K = 15 x 15 arcmin)
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Electronics – Signal Chain
 SDSU dual channel
video board
 2 channels
 150 kpixel/sec
 CDS, 16 bit ADC
 15 W power
 Analog Devices 9826




3 channels (RGB)
15 Mpixel/sec
CDS, 16 bit ADC
250 mW power
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Electronics – Computer
Communications
 Four OTA served by
QUAD OTA Interface Unit (QOIU)
8 CH
ACQ UNIT
CE LL
8
OTA
CLK
DISTRIBUTION
NETWORK
C LKS/C TL
32 DAT A
40
CL K&BIAS
UNIT
CE LL
EMBEDDED
MICRO
32 DAT A
8 CH
ACQ UNIT
CE LL
8
OTA
C LKS/C TL
32 DAT A
40
CL K&BIAS
UNIT
CE LL
C LKS/C TL
32 DAT A
8 CH
ACQ UNIT
CE LL
8
OTA
USED T O SYNC H
OTHER QOIUS
LVDS
C LKS
C LKS/C TL
C LKS/C TL
HIGH-SPEE D BAC KPLANE
an Interface Unit
and Gbit fiber
 Decodes
computer
commands
 Synchronizes
readout
 Formats data for
computer
transmission
 64 Mpixel = 128 Mb
10 Mb/ S
ET HE RNET
USED FOR PIXEL
DATA AND
C OM M AND DAT A
TO DATA AC Q PC
FIBER
INTERFACE
FPGA
1Gb/ S
FIBE ROPTIC
I/O
32 DAT A
40
CL K&BIAS
UNIT
CE LL
C LKS/C TL
32 DAT A
8 CH
ACQ UNIT
CE LL
8
OTA
C LKS/C TL
32 DAT A
40
CL K&BIAS
UNIT
CE LL
COMMAND
DATA
MUX
LOGIC
C LKS/C TL
PIXEL
DATA
MUX
LOGIC
FIBER
INTERFACE
LOGIC
USED FOR C ONFIG
AND DIAGNOSTICS
OF QOIUS
32 DAT A
32 DAT A
SYSTRAN
FIBER
XVCR
(DAUGHTER
CARD)
FIBE R
C TL
32 DAT A
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Electronics – Block Diagram

Scalable system
architecture
QUOTA
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Acquisition Software Tasks
 Observation shift and guide loop
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Reduction Software Tasks
 Read out science arrays; organize data
 Flat fielding: OT shifting integrates over many pixels
 Adding dithered images
Determine offsets (Continuous? Piecewise constant?)

Remove bad data (dead cells, gaps, etc)

Remove cosmic rays
 Combine (Remap? Piecewise constant? Integer pixel?)
 Ideally want many dithers but only one final image

National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
The WIYN One Degree Imager (ODI)
 PI-driven science for WIY (U. Wisconsin, Indiana U., Yale) and the
US community (via NOAO)
 Pathfinder for LSST detector and data pipeline
 Why WIYN?

Excellent “seeing”; median ~ 0.7”

1 degree FOV, inherent to telescope

Hydra spectrograph – 1 degree FOV

US access
NGC 7620 in R, I
Seeing = 0.38”
Sky Survey
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
WIYN One Degree Imager
Instrumentation
goal for WIYN
 64 OTAs = ODI (32K
x 32K = 1 x 1 deg)
 QUOTA does the
R&D, different funding
for large cryostat,
additional devices,
filters, shutter, etc.
 Deployment in 2005
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
The OTA Team
 CCD Design and Fabrication: Barry Burke (Lincoln Labs),
John Tonry (IFA) – CCDs for 20+ years, inventors of OTCCD
 Project management: George Jacoby (WIYN) – 20+ years of
instrumentation at NOAO, Director of WIYN Observatory
 Package and Cryostat design: Gerry Luppino (IFA) – 8K,
12K CCD mosaics for CFHT, UH
 Electronics, device testing: Barry Starr (NOAO) – 12K at
CFHT
 Optics, filter, shutter, commissioning: Chuck Claver
(NOAO) – WIYN optics, AO, George Jacoby (WIYN)
 Software: John Tonry (IFA)
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Experience
 OTCCD camera used at
MDM (Tonry, Burke,
Schechter 1997)
 Extensive modelling of
image motion
compensation (Kaiser,
Tonry, Luppino 2000)
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Experience
 OTCCD camera used at
MDM (Tonry, Burke,
Schechter 1997)
 Extensive modelling of
image motion
compensation (Kaiser,
Tonry, Luppino 2000)
 OPTIC camera nearing
completion (two 2K x 4K
OTCCDs)
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Timeline to Deployment
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Budget
 QUOTA (Quad OTA Camera)  ~$700K + Detectors ($750K)

2 ¢ / pixel (but large R&D costs)
 ODI (WIYN One Degree Imager)  additional $3.7M

0.4 ¢ / pixel (includes corrector, shutter, and filters)
 Additional funding now would allow




Multiple foundry runs to reduce risk, produce more devices
Additional engineering for improved hardware and software
design, earlier and more reliable implementation
On-chip shuttering
Integrated CMOS ASICs for focal plane controllers
(read out NxN CCDs of OTA instead of N)
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Budget Details: ODI
 Total -- $3.7M







Detectors (2 add’l foundry runs)
Hardware Design
Dewar plus cooling system
Shutter plus BVRIZ filters
Software Design
Acquisition computers/software
Optical corrector for WIYN
$1000K
$ 300K
$ 300K
$ 400K
$ 200K
$1000K
$ 500K
National Science Foundation, July 23, 2001
The Orthogonal
Transfer Array
Summary
 QUOTA: 8K
 ODI: 32K (~$4M)
 LSST: 48K
National Science Foundation, July 23, 2001