This presentation is dealing with some improvements of the Gecko coude spectrograph to increase its versatility and expand its performance. The goal is to have access to Gecko from any telescope focus with a quick connection time. This is possible through the use of optical fibres coupling. We present the different solutions whose feasibity relate to the wavelength domain. In a second part, we present a Gecko upgrade to a cross-dispersed spectrograph, by feeding it from the AOB.
This contribution will be divided in two parts. The first one is an
answer to the CFHT corporation request and deals with an optical
fibre coupling between the Cassegrain focus and the Gecko coude
spectrograph. The second one is a suggestion to upgrade the coude
spectrograph to a cross-dispersed instrument. The fibre link between
the Cassegrain and Gecko was requested by the CFHT because it had been
foreseen that this set-up will lower the pressure on the corporation
staff, in the focus exchange, which will not be needed any more.
Our point of view is that this is a partial solution, it kills the
UV which is a tremendous tool at CFHT and does not allow to use Gecko
in conjunction with the PF and the AOB foci, which will be among
the most used ones in the near future. So, we suggest to add to the
CFHT request the link of PF and AOB to Gecko and to use a single
calibration box implemented at the Cassegrain focus. This is also
Recommendation # 4 of the 53rd SAC meeting. The second proposal
is a more ambitious one and based on the new opportunity open
by the Adaptive Optics Bonnette (AOB). The AOB is able to deliver 
images in the red. We suggest to pick-up the light at this focus
with an optical fibre and to bring it to the Gecko slit. Doing
that, we can drop the image slicer and use a 
slit
that takes all the light coming out from the fibre,
still delivering the original spectral resolution above 
.
Now, a 2k x 4k CCD is ready to collect a cross-dispersed spectrum.
The present Gecko is using a small grism to select the
different orders. We suggest to replace this grism by a
different one and to change the grating to an echelle one.
By doing that, we will be able to record at least 250 nm in
a single shot and booster the CFHT at a front position in the
extra-solar planet search.
To build an efficient optical fibre link, the choice of the fibres and the beam aperture have to be carefully analysed. Fibres can be used on long distances in the red, but in the UV, a very short connection should be used. We have also to try to feed the fibre entrance with as fast a beam as possible. At the fibre output, a microlense is used to adapt the focal ratio to the spectrograph collimator one.
This link needs a 42 m fibre and a fibre connector to allow easy PF disconnection. This link will have a transmission of 85% in the V and R bands to compare to the 81% with the present mirror train. This link is totally inefficient in the UV and should not be used for this purpose. The PF focal ratio is F/4 and this beam can enter directly the fibre without much Focal Ratio Degradation (FRD).
This link is shorter than the previous one and does not exceed 30 m. It has a total transmission of 95% in the V and R bands as compared to the present 81% with the mirror train, but still poor in the UV, with a throughput as low as 18% that precludes its use in this spectral range. To avoid FRD, a small focal reducer is used at the fibre input.
This link is of the same length as the previous one and can not be used in the UV. The adaptive optics is not working at this wavelength. A 30 m fibre is needed for the link. To avoid FRD, a small focal reducer is used at the fibre input.
The only way to have a valuable UV link is to decrease as much as possible the fibre length. This is possible by mixing a fibre link and a path using some mirrors of the present UV mirror train. It needs a fibre not longer than 12 m. This fibre collects the light at the Cassegrain focus and sends it, with a two lenses optical relay, to mirror M7. Doing that, we can achieve an optical throughput of 58%, slightly better than the 55% of the present link. In the solution suggested, the mirror tunnel is only used for the UV and no optical components have to be moved to go to the UV. To avoid FRD, a small focal reducer is used at the fibre input.
The only complete study made in this field was the one requested by CFHT. This is the one when the light is collected at Cassegrain. Acquisition and guiding are made with Cassegrain bonnette and offset guiding is performed. An auxiliary camera is used to conjugate the fibre centre with the offset guiding.
The calibration lamps are located in the Cassegrain environment. They are sending the flat-field and wavelength beams through optical fibres to the different foci and feed the object fibre at its entrance, with the help of a movable mirror.
At the fibres output, a new image slicer, of the Bowen-Walraven type is implemented. This image slicer, made of silica, is coupled to each different fibre and moved in front of the slit environment.
Increasing the spectral range of Gecko without losing its spectral resolution capability would be a goal of great importance for all radial velocity studies. Indeed, any precision work improves with the number of spectral lines that are correlated. This is particularly true for the extra-solar planets search, and this is our main driver. In the interstellar domain, when people are looking only at a few lines spread far apart, increasing the spectral domain would allow simultaneous observations to serve many different programmes and to save telescope time. Using optical fibres to link with Gecko would increase the temporal versatility and optimise telescope use with relation to weather and sky brightness.
In its present configuration of single order observations, the
Gecko coude spectrograph is able to reach and even, in some
conditions, to surpass a spectral resolution of 
over a spectral range covering 4096 pixels. It was checked
with the UBC CCD. This performance is only possible when the
equivalent slit is of .
To be able to use such a narrow slit and still to get a
reasonable throughput, an image slicer is implemented at
the spectrograph slit plane. This procedure is supplying an
adequate narrow equivalent slit, but
is spreading the light in the direction perpendicular to the
dispersion and consumes space on the detector. Our suggestion
is to remove the image slicer and to pick-up the light,
with an optical fibre, at the adaptive optic focus.
To be able to retain the spectral resolution capability
with full light collection, we suggest to feed an optical
fibre at the adaptive optic focus and to throw it on the
Gecko collimator. The AOB is able to deliver 
images in the red, and if we assume a Gaussian profile
of the stellar image, the total light is collected over a 
disc. This is far enough to save the original spectral resolution
with a 100% efficiency. At the AOB output, we collect
of the F/20 beam on a 140 micron microlense and launch it at
F/6 to a 50 micron fibre core. At the output, we collect the
beam at F/4, with a 96% efficiency, and we expand it to F/20,
the collimator F ratio. We need a 30 m fibre and this
gives 84% transmission at 600 nm. Using AOB injection to the
fibre and the scrambling properties of it, we increase greatly
the spectral stability, which is the main noise in the planet
searches.
To introduce a spectral cross dispersion to Gecko, we have to replace the present grating by an echelle one. But before doing that, we have to check that the present optical configuration is able to maintain the spectral performances in the perpendicular direction with respect to the spectral dispersion. Going to the archives, and using the 3 slices delivered by the present image slicer, we checked that on 4096 x 200 pixel window, this was true. It was also cross checked with some long slit spectra over the full 200 pixel window. This means that without any computation, we know that a spectral range of 140 nm is at least reachable. Going to optical ray tracing software, we have seen that a range going from 470 - 710 nm is possible if we tilt the detector 15 deg. to its present position. This is inside of the mechanical mount. More sophisticated software optimisations will certainly expand this spectral range.
To be able to work in a cross-dispersed mode, we need a grating working in high orders. Gecko has a pupil of 300 mm, and due to the fact that we need and R2 grating, the projected pupil is an ellipse that falls into to a 300 x 600 mm area. If we accept a 4% vignetting, a grating of a ruled area of 300 x 500 mm is acceptable. This is the larger monolithic grating that the Vavilov Institute in Russia is able to made. It is a grating blazed at 63 degrees with 50 gr/mm. Using a monolithic grating, is a large saving on the staff manpower that is consumed to align the present gratings mosaic. The cross-disperser, is made of two LLF1 prisms used in a parallel beam. The prisms angle is 29 degrees and the total transmission is 97%.
All this study is made on the basis of 4096 x 200 pixels CCD.
This is completely covered by the new 4096 x 2048 EEV CCD that
CFHT will soon integrate in its dewars![[*]](foot_motif.gif)
.
Even if after
optimisation we would like to expand the spectral range,
we will be still far from the CCD size limitation.
The preliminary budget evaluation with no manpower costs, would be as follows:
Item Costs [0.5ex]Grating 300 x 500 mm 75 K$ Prisms, lenses, components 15 K$ Total 90 K$
It looks to us that for a relatively
low expense, we would be able to equip the CFHT of the most
powerful cross-dispersed spectrograph in the world. This is due
to the excellence of the AOB and of its first use
for high spectral resolution. Doing that, we are making advantage
of the AOB investment to open a new front-rank field, the
extrasolar planet search, and to push ourselves to the Keck
telescope level.
Furthermore, we could also reopen the debate about coupling
Gemini with Gecko through an optical fibre link. This is
feasible much better than in the previous project due to the
adaptive optic approach. A 
image at the Gemini
adaptive optic focus, between 600 nm and 1 micron, could reach
Gecko through a 300m fibre with a transmission of 85%.
in size.
