Home Technology Science Team Publications Blog

The Raven project, being developed at the University of Victoria Adaptive Optics Lab, will be the first Multi-Object Adaptive Optics (MOAO) technical and science demonstrator installed on an 8 m class telescope, the Subaru Telescope. In partnership with the National Astronomical Observatory of Japan (NAOJ), the Herzberg Institute of Astrophysics and Tohoku University, the Raven team will deliver a carry-in instrument to the near-infrared (NIR) Nasmyth platform of Subaru. The project started in July 2010. RAVEN is proposed to be delivered to the Subaru's Hilo facility early-2014 with on-sky test mid-2014. The block diagram presented below gives an overview of the "Raven concept".

The top level science requirements for Raven are summarized in the table below.

Parameter Requirement
AO System MOAO operation with tomographic reconstructor
Calibration System Capable of testing MOAO during daytime and in lab
Science Instrument Capable of feeding IRCS in imaging, grism and echelle modes
Science spectral range 0.9 - 4 microns
# of science channels 2
# of WFS 3 NGS + 1 on-axis LGS
Field of Regard 3.5 arcminutes diameter
Science Field of View 4 arcseconds diameter per science pick-off
Delivered EE > 30% in 140mas in H-band for r0 = 15 cm
Throughput 80% of AO188 throughput
Image rotation Ability to align each source to the IRCS slit
Zenith angle < 60 degrees
WFS limiting mag R < 14 (goal of R < 15)

Raven is currently being assembled in the UVic AO Lab:

More details on the current status of Raven can be found here.

Raven has three natural guide star (NGS) wavefront sensors (WFSs) and two science pick-off arms that would patrol a ~ 2 arcminute diameter field of regard. The WFSs operate in open loop, sensing light between 500-900 nm, and are used to construct a tomographic representation of the atmosphere. This tomographic model is projected onto deformable mirrors (DMs) that are integrated into the science pick-off arms. The four arcsecond field of view (FOV), MOAO-corrected science images from the two pick-off arms are remapped onto the IRCS slit. Raven's complexity increased somewhat after the addition of a single on-axis laser guide star (LGS) WFS after initial simulations showed that performance, and sky coverage, would be greatly enhanced by employing the existing Subaru LGS.

Compared to previous MOAO demonstrators, sky coverage is an important consideration because Raven is both a technical and science demonstrator. Demonstrating Raven on Subaru, is an opportunity to perform astrophysically compelling research. It is the philosophy of the team that one can learn much more about the strengths and limitations of future facility-class MOAO instruments by also attempting astronomical science. Raven, coupled to IRCS on Subaru, can deliver science at wavelengths between 0.92 and 2.5 μm and at spectral resolutions between ~ 5 (broadband imaging) to ~ 500 (Grism spectroscopy) to ~ 20000 (Echelle spectroscopy). The potential science cases (Raven + IRCS) that are under development are presented under the Science section.

Simulating the performance of Raven at the early stage was critical in establishing whether Raven would be able to meet the scientic requirement that 30 % of the energy of an unresolved PSF be ensquared within a 0.14 arcsecond pixel with existing WFS cameras and DMs. Raven was simulated with independent modeling tools, MAOS and OOMAO, which each gave comparable results. We established that Raven has an order 10 x 10 AO system that was a compromise between performance, sky coverage, and WFS FOV. The importance of minimizing high spatial order wavefront errors was established; tip/tilt and focus dominate the turbulence power spectrum but have little influence on the ensquared energy. This knowledge was employed to establish a high order wavefront error budget for Raven, and this budget was linked to estimates of the ensquared energy. Raven will meet the ensquared energy requirement of 30 % with three NGSs, and will exceed 40 % ensquared energy if the LGS is used as well (assuming median image quality). This is true even if relatively faint (mr < 14) guide stars are used (which corresponds to a ~ 10 % sky coverage). If the LGS is used, the sky coverage jumps by roughly a factor of three (mainly because one can use just 2 NGS plus the LGS to define the Raven asterism).

The optical design meets all the design requirements which were put in place to ensure that Raven could provide some multiplex advantage over AO188. This includes keeping the throughput above 32 % in H-band (the current design has a throughput of ~ 45 % well in excess of the requirement). An initial hope was that Raven could include a common woofer which could be used to provide a closed-loop ground layer correction and be used for calibration purposes, but it proved difficult. It is too difficult to design such a relay, that meets the throughput requirement, and not introduce too many field dependent aberrations. Instead, the common woofer relay was replaced with a calibration unit, which INO designed with our input. The NGS WFS and science pick-offs now receive light directly from the Subaru tertiary. The pick-offs will all track the stars as they rotate during science observations. To ensure that these complicated movements can be achieved smoothly, without introducing additional uncorrectable wavefront errors, the NGS WFS and science pick-off arms have been prototyped during the design phase. A NGS WFS and a science pick-off arm will be working in unison with the calibration unit before the next review. This will also enable early stage investigation of the calibration and alignment plans.

One major risk to MOAO instruments, on extremely large telescopes, is the size their real time controllers (RTCs). The size of the tomographic problem exceeds any commercially available computer. However, Raven presents a much smaller computational problem and, therefore, the Raven RTC will employ a CPU. This baseline architecture gives the design team the necessary freedom to explore several different tomographic reconstruction techniques on Raven. However, the team is also actively exploring alternative, scalable hardware approaches to the MOAO tomography problem. Raven will continue to be designed as a carry-in instrument compatible with the Subaru observatory. Therefore, the designers have been mindful of all the Raven interfaces, particularly those to different Subaru systems. Through on-going discussions with colleagues at Subaru, every effort will be made to minimize the number of interactions needed and to clearly document the remaining important interfaces. Raven is an ambitious project to complete within the four year timeframe and so scheduling risks are being examined with particular attention. Proactive steps to mitigating the scheduling risks have been taken, for example, prototyping complex components early in the preliminary design phase, and purchasing key components (e.g. optical bench, cameras and deformable mirrors).