DESpec spectrographs

DESpec spectrographs

Jennifer MarshallDarren DePoy

Texas A&M University

Prototype design: VIRUS clone

• 10 fiber-fed unit spectrographs, 400 fibers each

• Wavelength range 550-950 nm in one arm

• Resolution at 950 nm = 3167• Uses 2 DECam CCDs in each arm• Based on VIRUS design

VIRUS

• The first highly-replicated instrument in optical astronomy

• 150+ channel fiber-fed Integral Field Spectrograph placing >33,000 1.5” dia fibers on sky

• 350-550 nm coverage and R~700

VIRUS spectrographs

• Simple design– Single reflection spherical collimator– Schmidt camera

• Two lenses + one spherical mirror– VPH grating

• High throughput

Unit spectrographs packaged in pairs

Texas A&M’s role in HETDEX

• Participate in optical and mechanical design of VIRUS

• Fabrication and procurement of VIRUS components

• Assemble VIRUS unit spectrographs• Optically align instruments in lab• Ship to McDonald

HETDEX+VIRUS specs

• Wavelength: 350 – 550 nm• Resolution: R~700• Integration time: t=20 minute• Fiber diameter: 1.5” on sky• Sensitivity

– Line flux limit 3.5e-17 – Continuum detection gAB~22 mag

Flexibility of VIRUS design

• VIRUS design is readily adaptable to other fiber-fed spectrograph systems– Easy to change resolution, wavelength

range, etc. with simple redesigns• Has already been used as basis of new

spectrograph design– LRS2, a moderate resolution red-optimized

spectrograph for HET

DESpec as VIRUS clone

• Relatively straightforward redesign of VIRUS can produce DESpec– Change grating– Reoptimize coatings– Refractive camera?

Prototype design: VIRUS clone

• 10 fiber-fed unit spectrographs, 400 fibers each

• Wavelength range 550-950 nm in one arm

• Resolution at 950 nm = 3167• Uses 2 DECam CCDs in each arm• Based on VIRUS design

Alternate design: two arms

• 10 fiber-fed unit spectrographs, 400 fibers each• Increased wavelength range• Two arms, blue (500-760) and red (760-1050)• Different resolution in each arm

– 625 nm, R~1923– 950 nm, R~3276

• Uses 2 DECam CCDs in each arm• Significant design modification from VIRUS

– Similar optical layout to GMACS

GMACS

• Wide-field, multi-object optical spectrograph for GMT

• Four quadrants with two arms (red and blue) each– One quadrant could be

modified to become DESpec unit spectrographs

How to decide

• Need science input to provide instrument requirements:– Wavelength range– Resolution– Density of targets/number of fibers– Fiber size on sky

Work required to design DESpec as VIRUS clone

• Science input for instrument requirements

• New optical design for camera• Mechanical redesign of camera• Mechanical design of instrument

mounting scheme on telescope• Cooling system redesign

Work required to design DESpec as VIRUS clone

• We would need about 2 years of engineering effort for redesign

• A&M could assemble and test spectrographs in ~2 years– Lots of experience from VIRUS!

• These are estimates; will require more careful schedule/planning

Work required to design DESpec two-arm design

• More optical and mechanical design work required– Increases cost

• May need non-DECam CCDs for blue channel– Increases cost

Summary

• VIRUS design could be easily and relatively cheaply adapted to DESpec spectrographs– Two-arm re-design is more involved but

possible• Would need ~10 spectrographs• 3-4 years of effort in redesign and

assembly

Optimal Spectral Resolution

Jennifer MarshallDarren DePoy

Steven VillanuevaTexas A&M University

What is the “best” spectral resolution (λ/Δλ)?

• Science objectives set broad constraints• Various considerations suggest low resolution

– Easier optics– Smaller CCD format– Cheaper spectrographs

• Low means R=1000-1500– 200-300 km/sec

• Night sky emission lines are bright in the red– Suggest resolution should be higher– Isolates lines and allows for more “clean” pixels– What does “higher” mean?

Low resolution red spectra compromised by night sky emission lines

Fewer compromised pixels at higher resolution

Much less of a problem at bluer wavelengths

Lower resolution in “blue” not substantially compromised

Fraction of “uncontaminated” pixels (SNR > 0.9 relative to no night sky emission lines)

SNR per pixel versus resolution

SNR per pixel versus resolution

Conclusions

• Red spectra require relatively high resolution– R > 2500– Optimization is soft

• Blue spectra can be lower resolution– R > 500