CFHT, Instruments, Spectroscopy, Gecko: The Coudé f/4 Spectrograph

Web Manual for Gecko - The f/4 Spectrograph

This section presents the different components of the spectrograph, in the order seen by the beam of light:

  1. Slicer
  2. Order Sorting Filters
  3. Order Sorting Grisms
  4. Collimator
  5. Hartmann Mask
  6. The 316 l/mm Echellette Mosaic Grating
  7. The f/4 Camera and Detector Environment
  8. Detectors

Next are presented the Exposure Meter used to center the object in the spectrograph and the Comparison and Flat Field Lamps used for calibration. Finally Performance and Throughput Estimates are given.

NOTE: you may click on some of the images to get a better quality EPS file.

Unlike most echelle spectrographs, Gecko has been optimized for use with a single spectral order (between 5 and 18) from a 316 groove/mm echellette mosaic. Since spectra for these orders normally overlap, order sorting is achieved with interference filters or by one of three variable grisms. The average spectral resolving power (R) is of 120,000.

For observations between 3700┼ and 10,000┼, the mirror train has been replaced by optical fibers running from the Cassegrain focus to the CoudÚ room. This CAssegrain Fiber Environment is called CAFE. A four-slice Bowen-Wallraven slicer is used to optimize the instrument throughput.

The entire spectrograph and CAFE unit are remotely controlled.

As seen in the accompanying figure, the f/20 beam coming out of the slicer passes through an order sorting filter or a grism, and then goes to the collimator. The parallel beam is sent to the mosaic grating, the camera, a field curvature lens, and the detector.

1. Slicer

The Coudé focus is about f/19.6, giving a scale of 342 µm/arcsec (2.92 arcsec/mm). The minification of the spectrograph is wavelength dependent, but it is about 0.17 in the dispersion direction at the center of an order. In the direction perpendicular to the dispersion, the minification is 0.209. In other words, 1 arcsec on the sky corresponds to 71 µm. The spatial resolution on the sky is limited by the natural astigmatism of the optical system to no better than 0.5 arcsec.


CAFE is used with a Bowen-Wallraven image slicer, all made of silica. The fiber-microlens system forms at the microlens focus a gaussian-shaped image that includes more than 95% of the signal inside a 800 µm diameter. This image is then reshaped by the slicer into an elongated image 200 µm wide and 3.2 mm long. Placed at the Gecko collimator focus, this four-slice image become the entrance slit for the spectrograph and produces the required f/20 beam.

2. Order Sorting Filters

Since this spectrograph has been designed to observe only a single order at a time, there are two methods to isolate a single order from the echelle mosaic. The first involves the use of order sorting [interference] filters, particularly for the red orders. Prospective observers can inspect a subset of the CFHT Interference Filter Database to decide on filters needed for their run. The following table gives a list of filters that have been used successfully as order sorters for the free spectral ranges of the orders defined below.

Order CFHT
Filter #
Efficiency (%)
Coverage (Å)

3. Order Sorting Grisms

Because the free spectral ranges of the echelle's orders become relatively narrow in the UV-blue, grisms provide the most efficient means of isolating individual orders for those blue wavelengths. The following table gives the general properties of the three grisms available.

Efficiency (%)
Coverage (Å)

4. Collimator

A spherical collimator mirror sends a parallel beam 318mm wide to the grating. Two collimators are available: the first has an enhanced silver coating for use from about 3800Å through the near-IR; the second has an aluminum coating for use from 3000 to 4000Å. Changes of the collimator (and camera) mirrors are only performed by CFHT staff during the daytime when a new configuration is needed.

The following figure shows the reflective properties of the Ag + MgF2 coating (Gecko RED). The reflectivity is higher than 93% between 4000 and 7000Å, and above 98% above 6000Å.

The following figure shows the reflective properties of the UV enhanced aluminum coating (Gecko UV) shown along with a regular Al coating, between 3000 and 5000Å. The UV coating gives a few percent more reflectivity below 4500Å than a regular Al coating. The reflectivity is still good up to and beyond 5000Å (above 85%).

5. Hartmann Mask

A Hartmann mask is used for precise focusing of the spectrograph. The Hartmann Masks are automatically opened and closed when the Focus Tool is used. A good focus sequence (with 1+3 open, and 2+4 open) shows a difference of centroids smaller than 0.1 px, and a FWHM of about 2.8 px at the center of the detector (lines are thinner at the blue end of the spectrum, and fatter at the red end - note also that lines may be very large if a very wide slit is used). To speed up things, use a smaller raster. If you want to do the focus with the lamp, turn the lamp ON (not in AUTO mode), otherwise, it will not go on when the focus tool is started.

6. The 316 l/mm Echellette Mosaic Grating

For high spectral resolving power a mosaic of four 316g/mm echelles is used, each having a ruled area of 154 X 320 square mm. The approximate relationship between wavelength (in Å) and the grating incidence angle (in degrees) for the 316g/mm grating in order k is:

The grating rotation and, hence, the central wavelength, is set using a rotation table that supports the mosaic cell.

The procedure followed to align the four grating in the mosaic can be performed only by CFHT staff. Guest observers must ensure that their support astronomer is aware of every wavelength they plan to observe in order that the spectrograph setup can be optimized for these settings.

The following table gives the approximate central wavelength, linear dispersion and spectral coverage for each order between 3000Å and 10,000Å.

Wavelength (Å)
4600 13.5 µm
The data given in the preceding stable are approximate because of the changing dispersion within each order. Thus, relatively small changes in the grating rotation within the same order may cause noticeable changes in the linear dispersion scale. This is illustrated in the accompanying figure for the whole wavelength range of the Gecko spectrograph.
The corresponding changes in resolution (assumed equal to 2.5 pixels) are shown on the accompanying figure. These changes are relatively large within each order, ranging between R=90,000 and R=150,000. Therefore, the popular number for Gecko's resolution of 120,000 must be taken only as a rough estimate and not the actual number. [A Better Figure Will be Added]

The theoretical responses of the echelle are shown in the next figure, below.

7. The f/4 Camera and Detector Environment

The 600mm diameter camera mirror gives an unvignetted field of about 60mm in the camera focal plane. There are two camera mirrors, each with a radius of curvature of 2500mm. One has a silver coating with a protective overcoating of MgF2 (Gecko RED), for use from about 3800Å through the near-IR; the other has a UV enhanced aluminum coating (Gecko UV) from Denton Vacuum for use from 3000 to 4000Å. Changes of the camera (and collimator) mirrors are only performed by CFHT staff during the daytime when a new configuration is needed.

The following figure shows the reflective properties of the Ag + MgF2 coating (Gecko RED). The reflectivity is higher than 96% between 4000 and 7000Å.

The following figure shows the reflective properties of the UV enhanced aluminum coating (Gecko UV) shown along with the regular Al coating, between 3000 and 5000Å. The UV coating gives a few percent more reflectivity below 4500Å than a regular Al coating. The reflectivity is still good up to and beyond 5000Å (above 85%).

The beam from the camera mirror is reflected vertically downward by a prism-lens towards the detector environment. The detector position and orientation can be controlled about all three axes during alignment of the spectrograph and to fine-tune wavelength settings during observations. There are two corrector lenses, one for UV use (3000-4500Å) and the other for blue-red use (4000-10,000Å). The camera has an effective focal length of 1250mm which results in a slit-to-detector demagnification of 0.208 perpendicular to the direction of dispersion, and 0.170 parallel to the dispersion direction at the center of any order. One arcsecond on the sky corresponds to 342 µm on the slit and 71 µm perpendicular to the dispersion at the detector focal plane.

Adjustments are available for the observer to focus the spectrograph (movement along the Z-axis) and to position the spectrum on the detector by moving it along either the Y-axis (perpendicular to the direction of dispersion) or the X-axis (parallel to dispersion) under computer control. Rotational adjustments may also be made about any of these three axes to align the spectra along rows or columns of the detector, or, during instrument setup, to make the image and detector planes parallel.

8. Detectors

Currently, 2 possibles CCDs may be installed on Gecko. EEV1 is used for all wavelengths, although it shows fringing starting at about 6000Å (see the available information on fringing). If fringing is a problem, MIT2 is offered on a shared-risk basis, but its QE is not as good as EEV1's in the blue.

Detector Format Pixel
Noise (e) QE @
QE @
QE @
QE @
EEV12048 x 450013.5 Thin4.63 42%82%78%27%
MIT22048 x 409615 Thin7.5 <5%68%92%62%

9. Exposure Meter

The spectrograph is equipped with a pulse-counting exposure meter. The Exposure Meter is fed by reflection from one of up to three pellicles/filters located behind the slicer. The "clear" pellicle is used in most of the cases; the other options offered are filters used to feed the E.M. and also to cut some extra-orders that overlap the order of interest when using a grism in the red. It is important to select the "clear" pellicle and not a filter if no extra orders are present.

A limited number of filters are available for use with the exposure meter. Since guiding is best done by referring to the exposure meter ratemeter, it is advisable to use a filter combination giving a system response peak close to the observed wavelength. This is particularly important in the blue and ultraviolet where atmospheric refraction is large.

The pellicles and filters are selected according to the wavelengths requested by the observers and installed during the instrument's setup. The choice of E.M. pellicle and E.M. filter for a particular wavelength is automatically made when a wavelength is setup using the "gs" script.

The Exposure Meter is switched ON at the beginning of the night and OFF at the end of the night by the Observing Assistant. Once again, to prevent damage to the photomultiplier tube, be sure to close the E.M. shutter before using any lights around the slit area or in the Coudé room. There are software interlocks in the control computer to close the exposure meter shutter prior to turning on either the flat field lamp or the comparison lamps, but there is no interlock with the room lights or protection against use of flashlights around the exposure meter.

In the Control Room, there is an ORTEC crate with a Ratemeter, Counter and Power unit. First, power ON the crate. Then, on the Counter, set the counter function to count and hit "Reset" to initialize. On the Ratemeter, adjust the range and audio level to suit your requirements. The time constant should be set to 0.3 for bright stars and 1 to 3s for faint objects. The "Gain" on the Amp/Disc module should be 20.

10. Comparison and Flat Field Lamps


The CAFE head mounted on the Cassegrain Bonnette contains a flat field Halogen lamp with 7 intensities and a Th/Ar hollow cathode source. For each of these lamps, a small, precisely aligned mirror (in the Central Opto-Mechanical Sub-Assembly) is positioned so that the light gets into the fiber optic, and then to the Bowen-Wallraven slicer.

Lamps are selected via the "Gecko Configuration" form of the observing session. A lamp may be left in an automatic mode (AUTO), in which the proper mirror is rotated in place and the lamp is turned on and off automatically when the "comparison" type or "flat" is selected in the "Expose" window. However, for certain tests it is possible to select ON, which leaves the lamp always ON.

The CFHT Coudé Comparison Arc Spectral Atlases is available on-line.

11. Performance and Throughput Estimates

The following table gives some Gecko sensitivity estimates, based on old data taken with the LORAL3 thick CCD and ORBIT1 thin CCDs, as well as with the Loral5 thick and EEV2 thin CCDs. Counts are given in terms of electrons/sec/column (i.e. the spectral data have been integrated perpendicular to the dispersion) for a V = B-V = U-B = 0 star. Note that throughputs at a given wavelength are sometimes given for two different orders, and with various filter or grism combinations. A more recent set of measurements is also available.

Wavelength Order Loral3 Orbit1 Loral5 EEV2
(Å)  e/s/colOrder sorting e/s/colOrder sorting e/s/colOrder sorting e/s/colOrder sorting
313018 360UV Grism2250UV Grism     
398314   33200UV Grism     
403214   32800UV Grism     
404714 6000Blue Grism       
405514   39000UV Grism     
451212   43400Blue Grism     
517011 42000#1508       
578610      41800#1519  
58859   122100Blue Grism     
58909 34400#1519       
58949      83160 #1519  
59579      91600 #1519  
61509 53000#1508       
61509 52000Red Grism       
61509 43000Blue Grism       
61509 40800#1519       
63779        28900#1604
64289 22400#1617       
66138      74200#1621  
67078 69000#1626       
69858        35885Red Grism
77407        42848Red Grism
77737 42000#1704       
81007 23160#1803       
85427 5100#1803       
85426 19800#1803       
90866 5100#1901       

Limited tests have shown that the spectrograph appears to have very good scattering properties. Analysis of the spectra of several heavily reddened early-type stars with saturated Na I D absorption lines indicates that the percentage of scattered light in the spectrograph at this wavelength, and with the blue grism as the order-sorting device, is only about 1.4% if the interorder light is carefully subtracted from the spectrum. The accompanying figure shows the residual intensity in the core of one of the Na I D lines for one of these objects.


Here is the theoretical comparison of CAFE's and the mirror train's efficiency:

  4500Å 5700Å 8000Å
Coudé train throughput (from M3 to M7) 73%83%82%
CAFE throughput (from CAFE to fiber output) 79%84%84%
Overall throughputs (slicer and Gecko central obscuration included)    
Coudé train overall throughput 45% 51% 50%
CAFE overall throughput 51% 55% 53%

Back to Gecko's Web Manual
Send comments to: manset -=AT=-
Copyright © 2004. CFHT Corporation. All rights reserved.