OptoMechanicalDesign_BuildToCostMeeting_Feb5,2009.ppt

Download Report

Transcript OptoMechanicalDesign_BuildToCostMeeting_Feb5,2009.ppt

AO opto-mechanical design architectures
Cost Impact
Don Gavel
NGAO Team Meeting #5
February 5, 2009
Impact of Build to Cost on the Optomechanical design
• One tier (140 mm beam) vs two-tier (100 mm beam)
designs
– Woofer DM cost – impacted by the likely inability to reuse the
current system’s DM
• Cilas:
– increasing DM from 100mm to 140mm could increase cost 15%
– increasing diamter 100mm to 140mm decreases TT bandwidth
by 30%
– cost of TT stage 50-100% of DM cost.
• Xinetics
– Not much difference between 100mm and 140mm DM
regarding cost, probably also not resonant frequency.
– No ROM cost response yet
2
One-tier vs Two-tier
• Support structure for the K-mirror derotator - $90K
engineering cost makes it a wash with eliminating the
second tier.
• Switchyard choice impact (essentially none – both
switchyard options doable for both first relay design
options
3
Tip/Tilt Stages
• DM on Tip/tilt stage
– Tip/tilt stage cost (baseline is to put woofer on a t/t stage,
possibly a “woofer” tt stage)
– Cilas: cost of TT stage 50-100% of DM cost - I believe this is
what we costed at SDR: $200K on woofer, $20K on tweeter.
– Xinetics: Have done 2 designs of TT with heavy mirror insert –
100Hz bandwidth at 0.5 mrad mechanical displacement, or
0.62mrad of TT to 1.4kHz
– PI: to get back to us
– Lick shops: gimbaled TT stages for both woofer and tweeter look
reasonably doable at low cost
– Bottom line: there is room for reducing the $220K cost
significantly, perhaps by as much as $100K, but at some risk
4
Switchyard
• Dichroic set (6 keyhole designs) and changer mechanism vs single
focal plane and common pickoff design
– Dichroics
• Custom Scientific
– T>80% in the transmission region and R>90% in the reflection region
– Tavg>90% and Ravg>98% but that would require more layers in the coating and
that could significantly degrade the lambda/10 surface
– Cost $10,000-$25,000 per piece
• Barr Associates
– J trans/HK reflect will be difficult to make
– Cost ROM pending
– Cost of changer mechanisms
(NGAO_SD_Cost_Estimate_Don_Gavel_Penult6.xls 4.2.4FSD) $30K
– Cost of dichroics $10K per, 6 post relay 1 dichroics (KAON 550) = $60K
5
Switchyard
• Pickoffs
– Cost of pickoff mechanisms – note: cost of mechanisms is same as in
split to IFS/LOWFS design, so zero delta cost
– Single focal plane pickoff approach requires care in packing LOWFS
and IFS packages to allow narrow field second relay room to be
mounted. Preliminary layout design shows this is doable. There is still
plenty of room for post second relay switchouts (to NGSWFS, IR
Imager/Spectrograph, VisImager)
– Science impact:
• Pickoffs may need to enter 30 arcsec narrow field if a tt star is there
No provision to use the science object itself as the tt star in LGS mode
observing. This effects only some science scenarios (asteroids,
companions to low mass stars, AGNs, between v=12-20)
• Should be able to use 3 TT stars in the 30-120 arcsec annulus and still get
good sky coverage
• And we have a proposed work-around to this problem
6
Refrigerator-free low surface count
single relay option?
• Memo of Jan 28 shows high leverage of cooling leads to no
improvement or even worse emissivity for this option
• Gain in LGS throughput with no windows and no K-mirror (or Kmirror after LGS splitter). But new data concerning window AR
coatings mitigates LGS throughput (see Drew Phillip’s chart). (1/2%
per surface vs 1% loss per surface assumed earler). K-mirror has 3reflections loose 1% each, so total double pane window plus Kmirror loss is 5% instead of pessimistic 7% in memo. +$273K cost of
laser light compared to +$383K.
7
Cost Impact of not refrigerating the
LGS WFS assembly
• Requires another double-pane window that LGS light must pass
through. (2% loss on 75 W = 1.5 W at $73K/W = $109K)
8
Cost Impact of Reduced RTC
• 4-plus-3, with 4 on 1 arcmin field, vs 9 on 2 arcmin field
– Reduces size of tomography engine (which scales with field)
– Reduces number of processors needed for wfs (7 vs 9) but this may be
negligible because these processors are reused in the tomography
engine step of the algorithm, and the tomography engine processor
need is larger
– Lower cost of the telemetry disk array (fewer raw camera outputs)
• We are still investigating this. Possible cost gains:
– Less tomography hardware (but capped by $450K total cost of all
computer and interface boards)
– Fewer WFS cameras, and camera interface boards ($2632K savings on
cameras, $5K on interface boards)
– Less complexity and hence lower I&T costs
– We do not anticipate any decrease in programming costs, perhaps even
a modest increase due to need to add point-and-shoot algorithm
support.
9