Adaptive Optics Costs - Update
Jack Baldwin
Ed Loh
Bob Schommer
and
S. M. Simkin
Introduction
Although we do not yet have enough information to make firm decisions about how to proceed with this, we do have enough to get an estimate of the cost for a
PUEO-type system such as they have at CFHT. The email pertaining to this is summarized below and either attached to this discussion or linked as an html file. In addition the web sites for the various systems are included as hyperlinks in the html version of this write-up. The attached discussion is the following:
- 1) Memo from Art Code (WIYN Director) on the WIYN cost estimate of such a system (edited).
- 2) Summary by Jack Baldwin of comments by Francois Rigaut (formerly at CFHT) about his estimates of cost.
- 3) Email from Malcolm Northcott, of Laplacian Optics (which sells components for the PUEO-type systems) giving costs of the hardware and a summary by Bob Schommer of the total hardware costs.
To Summarize:
The cost for the WIYN system (similar to the CFHT Bonnett - ie a low order (19
subapertures, V through I, natural guide stars) was 3.5 \$M with a time for construction of four years.
Bob Schommer's estimate for the 15 actuator system HARDWARE is about \$2M.
(consistent with the WIYN estimate once the software etc. is done). Using the same price base (Laplacian Optics) a 35 actuator system would be \$4.6M plus set up and software costs and a 60 actuator system (which ESO is aiming for) would be \$7.8M plus set up and software costs.
Note the comments about NSF funding in Code's memo.
For a tutorial on AO (by Code) go to codeao.html (on the SAC web site).
Pick up
For a paper on calibration of photometry with AO systems
Links to:
Laplacian Optics:
CFHT-PUEO system:
Adonis (ESO) system:
Arizona Center for Adaptive Optics:
WIYN cost estimate (A.D. Code)
From: code@noao.edu (Art Code - WIYN Visitor)
Sue,
WIYN has indeed been looking at adaptive optics for some time now
although about this time last year we started a serious evaluation of AO. WIYN
is an excellent platform for AO because it is an excellent site, enclosure
environment and optics. WIYN also has active optics and active thermal control.
Active optics sometimes referred to as ao in contrast to adaptive optics AO
differs from AO in bandwidth. That is active optics tunes things up on a time
scale of minutes rather than at a 10 - 100 Hz rate.
Our effort was directed to developing a proposal to be submitted to the
NSF Facilities Instrumentation Activity. That proposal had a deadline of August
29. We did not in fact submit the proposal although there is no doubt that had
we done so it would have been funded. It was a good proposal and NSF received
only two others so they were very anxious for us to propose so that they could
spend their money. It was a good proposal and a realistic one because it had the input from AO experts among other things.
The cost was 3.5 million (we would be asking NSF for 2 million)
Another question you should have asked is how long it would take to
complete and that was four years.
The system was similar to the CFHT Bonnett (ie a low order (19
subapertures), V through I, natural guide stars). Note the CFHT project manager
Derrick Salmon helped us cost our system.
We do not know how much time NSF would want us to give up because that
is a negotiation that occurs after the award, however it is supposed to be
patterned after the recipe given in the McCray Report (A Strategy for
Ground-Based Optical and Infrared Astronomy ,1995 ). Examples would suggest
something like 20\% except if you lived in California. WIYN thought that for a
telescope which already gives away 40% of it's time on the promise that over 15
years NOAO will have put in it's share was excessive. There are other NSF paths
that require cost sharing which is more palatable.
The major reason that we did not submit the proposal, however, was as we came to understand AO better this system proved to be a poor match for the
science objectives that the WIYN astronomers had. We are currently working on a
low cost (less than \$500K and to be completed in less than 2 years) tip-tilt
module.
The decision that was made I believe was a sound one and rather than go
through all the elements that entered in to it I will copy the note that I sent
to Ben Snavely at NSF after our decision.
********************************************************************************
Ben,
We will not be submitting our proposal for AO at WIYN at this time. The
WIYN Science Advisory Committee on reviewing the draft proposal felt that the
cost and execution time were to large to match the scientific gains so long as
the system was limited to small field of view, natural guide stars without IR
capabilities at startup. To achieve these goals would inflate time and budget
even more. It was also true that the SAC felt that the price in giving up
observing time, by the formula that has been used, for an instrument which would already be available for 40 \% of it's time was more than they were prepared to
pay.
The long term goal still remains the same, since WIYN is an outstanding
platform for AO at an excellent site. We are, however, approaching the
development in a phased manner that will yield scientific return and expertise
more rapidly and for these steps we expect to continue funding the program as we have been doing from University contributions and other sources.
*******************************************************************************
I hope this answers the questions you directed to Blair.
-- Art --
Comments by Francois Rigaut (J. Baldwin)
From: SMTP%"jab@ctiow4.ctio.noao.edu" 5-NOV-1997 09:11:10.88
Dear Sue and Ed,
We had a very interesting talk about AO yesterday, given by Francois
Rigaut. He was the head of the CFHT PUEO project, and now is in charge
of AO for the ESO VLT.
He gave a review of AO concepts and showed PUEO results. Some highlights:
* Low-order AO systems like PUEO do not really give a great gain in
limiting sensitivity in the near IR. Although they change the shape of
the image so that it now includes a diffraction-limited core, typically
only about 20\% of the total light from the object is in the core, which
then gets diminished because the AO system has a throughput of 70% (this
is the measured value for PUEO). So instead of getting 50% of the light
in 0.5 arcsec diameter, you have 14% in 0.1 arcsec. This is only about
a 40% gain in s:n for background-limited objects. This is contrary to
what you read in the Roddier et al paper.
Rigaut says the gain is because you can do problems requiring higher
angular resolution, so that is the science case for these instruments.
* Rigaut gave very much lower costs for AO than the ones we have
recently been tossing around. He felt that the best place to buy a
turnkey system is Laplacian Optics in Hawaii. He thought you could get
a full set of the guts of an AO system for \$1.0-1.5M, + 1 FTE at a SOAR
partner site. That would include the deformable mirror, wavefront
sensor, electronics, and computer hardware/software. This was for 35-36
actuators... one step up from the 14(?) used by PUEO. Rigaut advocated
this step up because he thought we would have significantly worse site
seeing than Mauna Kea... he was thinking of the ESO seeing. This is
something we would have to think about very carefully.
He knew about the Canadian estimate of \$2.5M to copy PUEO, but does not
believe it would be that high. He sad that ESO is budgeting \$800K per
system to replicate completely packaged systems with 60 actuators, and
\$800K to develop a prototype system with 35 actuators. But as usual,
there are ambiguities about whether or not all manpower is charged,
overheads, etc.
Bob Schommer will contact Laplacian optics and get some real prices
from them... that is the best way to really get a handle on costs.
* We asked him what he thought we should do on SOAR. He was in the same
camp as Steve Ridgeway about how advisable it is to design for a future
upgrade from natural to laser guide stars... he obviously thought we
should just get a CFHT-like system going quickly without worrying much
about future upgrades. But he also commented that even with laser guide
stars you never need more than 60 actuators... the "cone effect" (the
fact that the light path to laser guide star is not the same as to a
star at infinite distance) prevents the correction from ever getting
better than that. So we ought to compare the costs for 14, 35 and 60
actuator systems.
* He felt that it was important that we have a "science case" for our
AO system, rather than just wanting one because they ae fashionable. I
understood this to mean that we should have the types of observations
we want to make well enough limited and focused so that the system
could be optimized for some things at the expense of not doing others
so well. His example of the science case for PUEO was pre-ms stars and
AGN, but not protoplanetary systems. But I am fuzzy about exactly what
is needed here.
* He talked about doing photometry with PUEO. Their strategy is to use
psf's calculated from what they have recorded about the corrections
which were being applied. Chopping off and measuring the psf from a
bright star does not work so well because the psf changes on 15-30 min
time scales... particularly in the I band. He referred us to a preprint
by Esslinger \& Edmunds who study these issues with simulations. Bob
Schommer will chase down that preprint and distribute it.
He was supposed to leave behind his viewgraphs... if they're here I'll
bring them along to Tucson.
Jack
Hardware Costs from Laplacian Optics ( R. Schommer)
From: schommer@ctiow7.ctio.noao.edu (bob schommer x207)
From: Malcolm Northcott
Dear Bob,
Thanks for your phone call earlier today. Please
do not hesitate to contact me via email or phone if you have
any further questions after reading this email.
As you may know Laplacian optics derives its experience in
adaptive optics largely from the University of Hawaii adaptive
optics project (http://queequeg.ifa.hawaii.edu/).
Key members of Laplacian optics also provided a substantial part
of the design of the CFHT PUAO system (http://www.cfht.hawii.edu).
Laplacian optics manufactures all of the components
needed to produce a curvature adaptive optics system. These
components include the adaptive (curvature/bimorph) mirror,
the wavefront sensor optics and the complete control system
and software. We can supply a complete turnkey AO system to
meet your specifications. For those customers who wish to
perform their own system integration we are happy to quote
for the individual system components.
To date we have built AO systems ranging from 13 to
36 actuators, and are happy to quote for systems in the
range of 7 to 62 actuators at this time.
As a rough guide to costing, we estimate that the basic
AO components will cost approximately US\$130,000 + US\$6500
per AO control channel.
This cost does not include ancillary optics, motor controllers,
optical enclosure design and fabrication, or system integration.
The cost of these latter items, required by a turn-key system,
are sensitive the the system design requirements.
In half arc second seeing on a 4m telescope a 19 actuator
system should provide diffraction limited performance in H and K,
with Strehl ratio's of a few 10's of percent in J. A 36 actuator
system under similar conditions would produce diffraction
limited performance throughout the IR, and give Strehls of a
few 10's of percent in I band. On a 4m telescope performance
of these systems would be level out to about Mv=12, then fall
until tip-tilt equivalent performance is reached at about
Mv=17. We would be happy to give you more exact performance
estimates for specified observing conditions.
Thanks,
Malcolm
So from that pricing a 15 actuator system (similar to
CFHT) costs about \$2M.
Of course as Art Code points out, you need a camera and instrument,
etc. But we will have those anyway, and clearly the idea
would be to make it all part of an integrated system, including
guiding and acquisition.
--bob
Arts descriptions of basic AO sounded good to me. See also references
to Wilson and Jenkins, 1996, MNRAS, 278, p 39 and Tessier in 1995, AO Systems
and Applications, SPIE proceedings Vol 2534, p 178.
Susan Simkin (Simkin@grus.pa.msu.edu) last updated: 12 Nov '97