12k x 8k mosaic for the oschin schmidt on palomar caltech optical observatories ernest cromer (tech)...

12K x 8K Mosaicfor the Oschin Schmidt on Palomar

Caltech Optical Observatories

Ernest Cromer (Tech) Richard Dekany (Lead)

Anna Moore (Optics) Hal Petrie (ME)

Gustavo Rahmer (EE) Roger Smith (EE)

Palomar Transient Factory meeting

2007-09-17

2007-09-1712K*8K Mosaic for PTF

2 of 37

Star Light Corrector Plate

QUEST Large Area Camera

Primary Mirror

TelescopeYoke

Figure 1.1 The QUEST Large Area Camera at the Prime Focus of the Samuel Oschin Schmidt Telescope at Palomar

Telescope layout

Clear Aperture Diameter = 1.26 m

Focal Length = 3.06 m

f ratio = 2.5

Plate scale = 15 μm/arcsec

Latitude = 3321’ N

Longitude = 11651’ W

Elevation = 1706 m

http://en.wikipedia.org/wiki/Samuel_Oschin_telescope

Spherical primary is larger than beam to accommodate the wide field (diagram at top right)

Focal surface radius = focal length of primary = 3m.

Field flattener design for QUEST:http://hepwww.physics.yale.edu/quest/quest_docs/hardware_flattener_model.html

4th order polynomial on outer surface, thicker at center and edge. Flat on side facing primary. Placed at radius of curvature of primary.

Spherical

Focus at half radius of curvature of primary


3 of 37

Telescope Field of View

Active area:

QUEST: 112* 600*1200pix * 0.87”/pix = 10.4 deg2

PTF: 12* 2048*4096pixels * 1“/pixel = 7.77 deg2

30% smaller, better quality CCDs, smaller gaps

20% vignetting at 430mm,

QUEST image (includes gaps)

10% vignetting at 380mm

Unvignetted to 307mm diameter2.86o radius

Since beam moves off primary at large field angles

30% vignetting at 480mm

PTF 12K * 8K …only 500um gaps


4 of 37

Requirements Implications

• Maximize survey speed, given fixed budget use 12K*8K mosaic of MITLL CCDs purchased from CHFT re-use as much of this system as possible: eg focal plane assy; front

section of dewar, dewar wiring, ~half the electronics, data acquisition software (ccdcom) and instremental calibration s/w (ELIXIR).

• Minimize telescope modifications (expensive) Replace large LN2 cooled dewar with compact dewar with closed

cycle cooling so it mounts on Schmidt focusing hub (only 6” from hub to focal surface); goal is to load fully assembled instrument through side of telescope without removing Schmidt corrector.

• Filter mechanism is NOT REQUIRED … right ? Manual filter insertion is cheaper and causes no vignetting.

• 30-90s exposure times reduce read time from 60s to <30s; read noise can increased from 5e-

to 10e- since sky noise > ~15e-/min in B. Need greater analog BW, faster CCD waveforms and faster driver for DMA card.


5 of 37

Requirements Implications• Accurate photometry using standard stars in science fields

This and ability to use short (5s) exposures for flats, sets shutter timing accuracy requirements. For super-flats just need stability.

• Minimize obscuration of beam by instrument Mount electronics outside telescope; longer cable probably requires warm video

preamps at hermetic connectors.

• High reliability with minimal maintenance Custom low impact shutter (timing requirements imply dual blade); purchase from

experienced vendor.

• Image quality better than 2” (goal 1.7”) with minimal interruption to observing for focus & tracking adjustment. We will attempt to derive focus (and tracking?) corrections from science images;

using temperature sensors to predict focus is expected to be problematic on this telescope at least until data has been accumulated over a long period.

Can we do better? Tracking is possibly the limit. Best image quality during photographic survey was 1.1” but with frequent mirror support tweaks. Seeing is worse than this about half the time.


6 of 37

While one can write requirements for productivity in terms of down time and throughput, a judgment call is required to decide how much to invest now to possibly “save later”.

Invest now to save later ?

CCD SELF TEST

• Need daytime flats with stable long term intensity ... Temperature regulated LEDs on back of instrument shine back to screen inside corrector cover.

• Tie together lab scripts to measure and log: offset stability, noise in image and overscan; dark current; charge injection, e-/ADU, Shutter timing error vs field position, serial and parallel CTE. QE stability in current filter, linearity, well capacity.

• Include images in archive along with calibrations.

DATA LOGGING

• Collect time tagged diagnostic information to diagnose problems with image quality or reliability.

• Data collected may include: pointing, measured image quality, pointing tracking & focus settings for each image, telescope and environmental temperatures, dewar pressure & temperatures, results of CCD self test, system up time and exposure duty cycle.

• There is some motion towards this under way on Palomar already.

Repackaging Concept


8 of 37

The current CFH12K

Move the electronics boxes outside the beam

Only the blue bits fit on the Schmidt instrument support hub.


9 of 37

Manually installed filter slide. Use filters from CFH12K New dual blade servo

controlled shutter from U.Bonn

Use existing CFH12K front plate, but change window to serve as field flattener. Tweak QUEST optical design for different filter position & thickness, and different (?) distance from window to CCD.

Approximate Layout (cartoon, not to scale)

Vacuum housing. Re-use front section of CFH12K dewar.

Connector sub-plates: cables are hidden behind shutter overhang

Access hatch for bolted thermal links?

This surface attaches to telescope’s instrument focus hub. Attachment must allow for tilt adjustment.

25W Cryotiger, =114mm (no moving parts). Thead= 135K, TCCD=153K


10 of 37

Mosaic footprint (looking towards primary from corrector, approximate scale)

Shutter housing

Central obscuration will be dominated by shutter blades. Barn door shutter for quest avoids this but timing error is greater and variable across field.

Connectors and cables hide behind shutter overhang.

60mm

338mm

60mm

Dewar photo with window removed

facing wrong way and only 4 CCDs installed

Support vanes Instrument mounting hub (focus stage) On-axis beam

Hide Cryocooler head in corner?

Simpler installation than QUEST since cooling

system does not span hub and vanes.


11 of 37

Solid modeling of space allocation

Focus mechanism

Spider

Shutter

Filter holder

Cryocooler (not in final

position)


12 of 37

Other views

Cryocooler might fit behind CCDs, coming in from side.

Tricky to fit cryocooler here

Rear view


13 of 37

Manual filter holder

Existing CFH filter frame: no problem with size in faster beam

Filter slide, for manual insertion

Retention device.


14 of 37

CCD Performance


15 of 37

From http://www.cfht.hawaii.edu/Instruments/Imaging/CFH12K ….

CFH12K characteristics

• MITLL CCDs from foundry run

• 15um pixel 1”/pixel

• One amplifier per CCD

• 9 standard, 3 deep depletion (better red response, less fringing)

MASK IS N

OW

BLACK TO R

EDUCE

REFELECTIONS

http://www.cfht.hawaii.edu/Instruments/Imaging/CFH12K


16 of 37

QE patterns

• Blue flats have quilting due to laser annealing of boron implant which overcomes surface potential which would otherwise trap photogenerated charge. This effect is strongest in B band and still presnt in V.

• Red flats show fringing, except in 3 thick CCDs. AR coating differentiates others. Fringing in Z band is twice R band.

Blue extreme (10%) Red extreme (5%)


17 of 37

Cosmetics, etc.

• Only 0.4% pixels are bad, including 200 bad columns, mostly in lower right CCD.

• Compare to gaps between CCDs which are ~10% of area.

• Bad pixel/column clusters are all narrower than CCD gaps so dither that fills the gaps also fixes bad columns.

CFH have produced many beautiful images like this one by dithering to fill in gaps.

Full well ~150,000e-; Saturation leaves ~4e-/min image persistence


18 of 37

Initial tests at Caltech

• Cabled together, pumped and cooled, initially in shipping container with window cover used for shipping. The light leak is minor.

• A sealed light box has been built and internal light source is under construction.

Photo here

Small vacuum leaks were found at the hermetic connectors (6mT/hr when warm). This was absorbed by the getter at 77K, and hold time was still 24 hours after several days of operation. Since we will be more sensitive to leaks when the activated carbon getter is only cooled to 133K, these leaks (which have probably been present for a long time) will have to be fixed.


19 of 37

Recent Bias Frame @ Caltech


20 of 37

Recent 1200s Dark Frame @ Caltech5-10e- in 2 min at -85C (warm)


21 of 37Recent Flat @ CaltechRed LED; gradient is illumination

Hot columns on this CCD are still sensitive


22 of 37

650nm flat

Lower left CCD in previous image


23 of 37

Noise

• These are very low noise CCDs

• No sign of electrical interference in spite of low random noise floor (see slide background)

• Only moderate noise degradation is expected at higher pixel rate.

Design issues


25 of 37

Vacuum envelope & mounting

• There is 6.4” from mounting hub to focal surface. The existing dewar front end is 7” high with CCDs recessed 1” behind the window exterior and thus fits without modification.

• A new back plate will replace the LN2 dewar and provide attachment to focus hub. It may hold the Cryotiger . Flexure TBD.

• The attachment to the focus hub must support static adjustment of tilt to <20um across 236mm diagonal of image area. See focus budget allocation. Screws or shims?

• A thermally isolated (passively cooled) radiation shield with minimal gaps is highly desirable to reduce load on the cryocooler.

• New/modified heat spreaders are required within dewar to bring CCDs closer to cold head temperature.


26 of 37

Mounting to focus hub and tilt

• A stiff instrument support plate (orange) is attached to the focus hub.

• The back plate of the dewar (green) overhangs so that it can be bolted to the instrument support plate at the edges.

• This back plate can be “thin” since bow in center doesn’t matter. It is stiffened by the dewar wall at the edges where flexure counts.

• The contact surfacecontact surface between the two plates will be adjusted (machined or shimmed) to achieve correct tilt.

Mosaic dewar base:

just fits !

Telescope spider

Focus hub


27 of 37

New cooling system• Polycold Closed Cycle Cooler (Cryotiger)

with no moving parts in cold head.

• Rough thermal budget– 11 W radiation to CCDs assuming 90%

emissivity for CCD and window

– 5 W radiation to shield (3% emissivity product)

– 3 W conduction by wiring

– 2W conduction in supports

– 5 W CCD heater (servo); TCCD=-163K

50% safety margin. CCD heater power goes to zero if load increases from 21W to 33W.

Expect ~135 C at cold head and getter.

It looks like one cold head will work and can be made to fit (with some effort).

Note: IMACS 8Kx8K mosaic reaches -85C with one Cryotiger and no radiation shield. We could run this

warm if necessary but we aim for more margin.


28 of 37

U.Bonn Precision 2 Blade Shutter

No new technology: our requirements are well within U.Bonn experience range

~ $US 25K for size required. Estimated by Klaus Reif, U.Bonn.

Image area = 200 * 126 mm

Opening = ~215 * 141 mm

Footprint = ~335 * 493 mm

1 ms

Uniform, low exposure error

Precision servo matches velocity profile for open and close. Low acceleration for low telescope vibration and long life: tested to 107 exposures !

14% vignetting (>11% just for dewar)


30 of 37

Optical analysis tasks

• Re-optimize QUEST field flattener prescription for new filter position and distance to CCD.

• Allocate image quality budget:– Aberrations (various sources)– Seeing– Residual tilt (static)– Flexure– Residual CCD height variations– Focus motion error (resolution, hysteresis)– Focus measurement error– Tracking error & wind shake.– Differential track rate due to atmospheric refraction


31 of 37

Current focal plane flatness is not good enough

• Degradation in image size from 0.53” to 0.7” implies 0.45” defocus or 144 um CCD height variation. On P48 this would produce 3.8” images !

• We have been offered use of the Carnegie Observatory’s optical profilometer to scan the full focal surface to 0.5um accuracy.

• Once this height variation has been verified, the focal plane will be disassembled to adjust CCDs to be closer to coplanar.

CFHT P48

Blur circle 0.45” 3.8”

Arcsec / 15um pixel

0.2 1.0

f-ratio 4.2 2.5

Height change 144um 144um

Image size vs position on CFHT


32 of 37

Automate focus

Find focus:• Read one band per CCD, each

~20% of area, so read time per 10s exposure at each focus position is only 6s.

• Automatic 7 point focus sequence takes just 2 minutes.

• [ We don’t need multiple exposure with charge shift since read time is < exposure time. We don’t want it because source confusion makes automatic analysis complicated. ]

Then follow focus:• Automatically analyze science frames

to measure image size and astigmatism versus CCD height which will be known very accurately from profilometry. (CCDs cannot be made perfectly coplanar)

• Tracking and focus error will be distinguishable, since tracking error is invariant with CCD height.

• By using all the stars in the frame (computing the major and minor axes of the 2D autocorrelation function) one should be able to resolve small focus and tracking errors relative to seeing and static aberrations.

• It may be possible to trim track rate as well as focus !


33 of 37

Electronics:

• Move electronics box outside telescope tube to reduce obscuration and heat in beam:– New cables– Install a preamp at hermetic connectors (in air) to drive longer cable

length.

• Reduce pixel time from 6.5us to 3us (readout from 58s to 27s).– Same video card: video BW limits pixel rate

1MHz Datel ADS937 ADC converter is not limiting.– Shorter dwell times (definitely more noise)– Faster charge dump (less settling time may degrade linearity) – Overlap serial clocks with entire pixel (maybe more noise)– AD conversion must dodge transients to avoid code drop outs.– Existing two fiber links run at 50Mb/s limit pixel time to >2.4us

We don’t need to upgrade fiber links and DSP code as first thought.


34 of 37

Software by camera team

Photons to FITS:

• Move CFHT’s CCD control program, ccdcom, to a modern computer and Linux version.

– A command line interface will be supplied for for manual operation: no GUI’s. ccdcom provides an interface for camera control by scripts or the Observatory Control Software package

• Install new driver to support higher data rate (10 MB/s)

• Setup VNC for remote access and IRAF for image analysis in lab and when commissioning.

• Augment existing lab scripts to support automatic self test.

[ .. ] = if we decide to do this


35 of 37

Tasks for science team(s)

• Scheduling and Observation Control software.– command telescope and dome– initiate exposures– Work form existing solutions: QUEST or P60?

• Near real time image analysis for focus and tracking adjustments

• Collect and append information to image headers (eg pointing, UT)What time accuracy is required?

• Data logging and analysis for ongoing performance improvement.This has much in common with image header creation. Note that “data logging” allows for collection of information in one place which, if only stored in headers has to be mined from the image archive. It also allows for recording of information which is collected asynchronously to exposures or collected when no exposures are being taken.

• Manage local data storage and transfer– Tell camera where to store data; check when disks fill.– Temporary backup to second spindle, when space available– Image transfer to IPAC– Image deletion after transfer.

• Data archiving, reduction, and community access.


36 of 37

Palomar Roles ….TBC

• Removal of QUEST [ not part of this project ]

• Contract with Vertex-RSI to complete the telescope control system installation by calibrating pointing and tuning tracking. ($20-30K?)

• Advise observation control software authors regarding TCS interfacing.

• Assist with instrument installation, alignment and acceptance test.

As budget permits (to support image quality improvement):

• Add telescope temperature sensors to existing system for data logging.

• Add primary mirror positions sensors and logging.


37 of 37

Decisions needed (summary)

• Filters: When changed? Manual change ok? (Can we get by without a filter change mechanism?)

• Image quality– How hard do we work ? [Science team]– Note that primary mirror support adjustments by Bob Thicksten, during the

photographic surveys, based purely on mirror position measurements were successful in making image quality better than the 1.8” limit seen for QUEST.

• ACTION [Palomar]: (Later, as budget permits?) install electronic sensors to measure primary mirror position, log it and flag when adjustment is needed.

• ACTION [COO] : Decide who will have overall responsibility for image quality.• ACTION: Allocate image quality budget and assign responsibilities for each element.

• ACTION: assign responsibilities for software tasks. Slides 33-35.

• ACTION: Allocate photometric accuracy budget (3%, goal=2%). E.g: Shutter timing accuracy wont be an issue. Scattered light probably will be.

~ E N D ~

12k x 8k mosaic for the oschin schmidt on palomar caltech optical observatories ernest cromer (tech)...

Documents

focal length of primary

mosaic of mitll ccds

focal surface goal

better quality ccds

compact dewar

section of dewar

dewar wiring

oschin schmidt