Part 1. Optical scheme, main spectral parameters, and test results


Version 1.0, June 28, 2005 


Ilfan Bikmaev, Nail Sakhibullin 
(Kazan State University, Academy of Sciences of Tatarstan, Kazan, Russia),  


Faig Musaev 
(Special Astrophysical Observatory of  Russian Academy of Sciences,  Nizhnij Arkhyz, Russia),  

Zeki Aslan 
(TUBITAK National Observatory, Antalya, Turkey)




This document gives a description of the optical scheme of Coude echelle spectrometer (hereafter CES), its main characteristics, and some results of scientific tests performed during the period of 2004-2005.




Modern spectroscopy uses CCD as the light detector. Geometrical sizes of spectral images have to be in agreement with the geometrical sizes of CCD chips (at least within a factor of 2-3) to prevent excessive light loss during an exposure.

Fig.1 demonstrates the image of low-resolution (~ 10 Å, 2.7 Å/pix) spectrum of a hot star registered by using SAO RAS 1-m telescope + Cassegrain spectrometer and back illuminated CCD. The whole spectrum in ~ 4000-8000 Å spectral range is fully covered by the CCD chip (27 mm in length) at a linear dispersion of 150 Å/mm ( 150 x 27 = 4000 Å), see Fig 2 and Fig.3. (Note the increasing presence of "fringes" (light interference) effect in the near infrared range of 6500-8000 Å when using back illuminated CCD).

It is seen from Fig.2 and 3 that only the main spectral features (like hydrogen lines) are detected in this low resolution spectrum of a hot star (O,B,A spectral classes). All spectral features in the spectra of cool stars (F,G,K,M) are blended at the corresponded resolution of R ~ 500 (R = Lambda / Delta Lambda), 5000 Å/10 Å = 500).

The only way to get detailed information about the spectra of, and physical conditions in, stellar atmospheres is to increase spectral resolution considerably and to use spectrometers with a resolution of R ~ 20000 - 100000 (0.2 - 0.05 Å). At this resolution, a spectral range of 4000-9000 Å has a geometrical length of 1000 - 3000 mm ( 1-3 meters !). In classical spectrometers the registration of long spectra was achieved by using large format photographical plates during the years 1950-1980. Short format CCDs were used to detect these "classical" spectra during 1980-1990, but it was clear that time loss effec was very high in this case (only few percent of the whole spectrum is registered in one exposure, most of the spectrum is situated outside the 10-20 mm size CCD chip "in air").

To solve this "spectrum size - detector size" problem, it was proposed to use echelle gratings (facet's angle of 63 degrees) instead of classical diffraction gratings (facet's angles of 4 - 24 degrees). 

Classical gratings produce spectra comprising few low orders (with numbers N = 1,2,3) with length 1000-3000 Å each. Fig.1 is an example of a spectrum registered within the single (first) order of classical grating.

Echelle gratings work in high orders (with numbers N ~ 50-150) and they form many "short" spectral orders of 50-150 Å each. An Echelle grating disperses the light of all ~ 100 orders into the same geometrical direction. Due to this effect, all spectral orders overlap in the same space direction. To split these 70-100 echelle orders in space, a second dispersing element is needed (called a "cross-disperser"). Depending on the spectrometer design, classical prisms or gratings in low orders are used as cross-disperser. Combination of "echelle grating + prism + CCD" produces "two dimensional" format of echelle-spectrum frame. This combination allows detecting most part of the optical/near infrared spectrum simultaneously in one exposure. Examples of this kind of echelle-frames are shown in Fig. 7, 8. Therefore, echelle-spectrometers are powerful instruments in recording high-resolution stellar spectra, especially in the case of variable stars; most important details are registered at the same time. 

An important point

Relative light intensity distribution I(β) within a grating's spectral order is proportional to the combinations of sine functions (I(β) ~ [(sin(Nπ))/(Nπ), sin(β)]), see for more details (Gray, 1976), where "N" is the physical number of the grating's spectral order and β is the angle of light diffraction within individual spectral order N. Due to this proportionality to the N, echelle-spectrum is the superposition of many "narrow" orders (Fig. 9,10) as compared with classical single and broad orders with more or less regular (analytical) view of the I(β) function (Fig. 2).

So, the "back side of the coin" in using echelle-gratings ("weakness") is the non-linear and even "non-analytical" intensity distribution within a single and short echelle-order (see, Fig. 11). This demands existence of preliminary knowledge about the main characteristics of the target's spectral lines and very careful echelle-spectra processing procedures. Some details of the processing of RTT150 CES spectra will be given in a separate description (by September 2005).

For example, CES is very optimal for the study of narrow lines spectra, but accurate registration (at the level of 1-2 percents) of broad hydrogen line profiles may be questionable.




RTT150 CES was specially designed and manufactured for the 1.5-m optical telescope in accordance with the Agreements signed between KSU, IKI and TUBITAK and AST and TUBITAK. All the optical-mechanical units were manufactured in Russia during the interval 1995-2002 by Kazan State University and Academy of Sciences of Tatarstan. TUBITAK constructed the RTT150 telescope building which has the special isolated room for the CES, called "coude room" (at the second floor of the building, see Fig. 4), and constructed the metal supports designed by KSU for proper installation of the CES's large scale optical-mechanical parts in the coude room. 

During the design step of RTT150 CES, we have used experiences and high-quality scientific data obtained with three coude-echelle spectrometers. They had been realized earlier for 2-m Zeiss telescope of Shemakha Astronomical Observatory (Musaev, 1993), 1-m Zeiss telescope of SAO RAS (Musaev, 1996, Musaev and Bikmaev, 1995), and 2-m Zeiss telescope of Peak Terskol Observatory (Musaev et al, 1999). Note that all the three telescopes mentioned had initially all the optical parts of a classical coude-spectrometer with photographic detectors (all produced in Germany) and they have been transformed to the coude-echelle spectrometers because of the aforementioned problem of "spectrum size detector size".

Optical scheme of RTT150 CES is shown in Fig. 5 (top view) - all CES units are situated in the horizontal plane in the coude room at the second floor of telescope building. 

The starlight collected by 1.5-m main mirror is directed to the coude-room by the system of secondary mirrors (Fig. 4). Small diagonal mirror (1) reflects the convergent F/48 beam from the telescope ("equivalent" focal length is 72 meters) to the direction of the spectrometer's entrance slit (2), where the starlight is focused. After the slit, the divergent F/48 light beam is directed to the collimator mirror (3) with a focal length of 7.2 meters (1/10 of telescope's 72 meter focal length). Parallel (collimated) beam of 150 mm in diameter (1/10 of main mirror diameter) is reflected from collimator mirror to the direction of flat mirror (4). 

Note that all spectrometer's dispersing elements (gratings, prisms) work properly with parallel beam (to prevent optical aberrations in the spectral line images). So, the stability of 7200 mm distance in between entrance slit and collimator mirror is important.

Flat mirror (4) directs the parallel "white" beam of 150 mm in diameter into echelle grating (5) of 200 x 300 mm in size. Echelle grating has a classical facet's angle of 63.5 degrees (tan 63.5 = 2.0) and 37.5 grooves/mm.

Echelle grating can turn by 6 degrees to the left or right and has two positions depending on the spectral camera used (R = 40000 or R = 100000). Echelle grating unit has special measurers for fine adjustments of proper positions of echelle-spectra on the CCD frame with an accuracy of 0.1 mm (few CCD pixels).

This Echelle grating is the main element of the spectrometer, it disperses "white" parallel beam into approximately 85 echelle "colored" orders with physical numbers N ~ 50 - 135 (LL = 3800-10000 Å). The set of these 85 orders is reflected to the 45 degree large format prism (11) in the case of R = 100000 or twin (45 + 45 degrees) prisms (6) in the case of R = 40000. These prisms split 85 echelle orders in the direction perpendicular to the dispersion of echelle grating.

This system of dispersed parallel beams enters one of the spectral cameras. The spectral cameras are Schmidt telescopes with spherical , mirrors (9,13) of ~ 40 and 50 cm in diameter and Schmidt's spherical aberration correcting plates (7,12).

The task of spectral cameras is to transform dispersed 150 mm parallel beams into the "two-dimensional" spectrum images a "continuous" set of "colored" optical images of the spectrometer's entrance slit on the focal surface of each camera.

R = 40000 (short focus, F = 400 mm) camera (7,8,9) is a "folded Schmidt" camera. An additional flat mirror (8) is used in such systems to get focal surface outside the "fast" (F/2.5) camera. This flat mirror has the special round hole in its center (See Fig.6).

The CCD detector (10) is attached in the backside of this flat mirror and has the possibility of fine movements for proper spectrum focusing on the chip surface.

R = 100000 (long focus, F = 1000 mm) camera (12,13,14) is the Schmidt telescope with Newtonian focus. A small diagonal mirror (14) directs the convergent spectrum light into the CCD chip (15).

Note that collimated beams have cross-sections like on the main mirror (shadowed by the secondary mirror), so collimated beams have a shadow of ~ 50-60 mm in the center. Diagonal mirror (14) is situated within this shadowed zone. Above mentioned round hole in the flat mirror of R = 40000 camera is situated within this shadow too. These specific positions of the indicated elements in the shadowed zone prevents light loses in the cameras.

CES has a calibrating unit mounted in the same box as the entrance slit. The calibration sources - Thorium + Argon (17) and Halogen Flat field (18) lamps - direct their lights to the spectrometer's entrance slit by additional lens F/48 optical system and movable flat mirror (16). All calibrating images are exposed before or after the observational night, so the mirror (16) is in "off" position during the scientific exposures.

CES has also a remote guiding system. The slit unit has an aluminum coated surface and tilted by ~ 10 degrees relative to the telescope's collimator direction. Due to the long equivalent focal length of 72 meters in the coude focus, the angular scale is 3 arcsec / mm and field of view is only a few arcminutes. Therefore, as a rule, there will not be any bright star (brighter than 9th mag, say,) near the target star. This prevents off-set autoguiding in the coude focus as normally used in Cassegrain.

Because of this limitation, the wings of the target's image itself (part of the light not entering the spectrometer) are used for guiding purposes during the observations in Coude. A special optical system (19) directs the slit zone image to the special low-light video CCD (20) and this slit image zone can be viewed in the observers' room on the first floor of the building (near the remote control PC), see Fig.18.

The Echelle-spectra images detected by the CCD are stored in the PC in the computer room on the 1st floor (see Fig.17). TUG's Night Assistants have participated in most of the test observations with the CES and they have been provided already with a separate manual to operate the CES CCD during the observations.




a) Spectral image coverage by CCDs.

Test observations with CES have been performed within a few periods of KSU time during the interval 2004-2005 using two CCDs ANDOR 2Kx2K TE CCD (2004) and SAO RAS 1Kx1K nitrogen cooled CCD (2004-2005) and both cameras (R = 40000 and R = 100000).

ANDOR CCD was used first to check the main CES parameters. 

It was found that R = 40000 camera + ANDOR CCD combination (high-resolution mode) fully registers the CES's whole working optical/near infrared spectral range of 3800 - 10000 Å (Fig.7). Echelle spectral orders overlap in wavelength in the 3800- 8000 Å region with small gaps in 8000-10000 Å region. 

But back illuminated ANDOR CCD chip produces systematically increasing "fringe" effect in the near infrared region of 7000-10000 Å (as in Fig. 3). This fringe effect is partially eliminated in the 7000-9000 Å region by the use of flat field calibrating lamp.

It was found that, to cover the whole spectrum in 3800 - 9000 Å region with the R = 100000 camera + ANDOR 2Kx2K CCD combination (very high resolution mode), 3-4 overlapping images have to be obtained by carefully changing the echelle grating working angles.

Starting from August 2004, ANDOR CCD is permanently assigned to the Cassegrain focus of RTT150 for photometric and astrometric observations with high angular resolution. 

Test observations with 1K x 1K CCD and R = 100000 camera have shown that ~ 8-9 overlapped exposures are needed to cover the spectral range of 4000-8500 Å implemented by changing the inclination angles of the echelle grating in between the exposures. This is a very inconvenient case from the point of view of both the stability of spectra between the nights and/or targets and accuracy of wavelength calibration.

For these reasons, to start with regular observations by CES during 2005-2006, it was agreed that 1K x 1K N2 CCD should be attached permanently on the R = 40000 camera. Special test observations were performed in April 2005 to optimize the inclination of the echelle-grating. The results show that spectral coverage is ~ 3900 - 8700 Å (with some gaps between echelle-orders in the range 5000-8700 Å), so approximately 70 percent of useful CES spectral range is detected now by 1K x 1K CCD (Fig.8).

SAO RAS CCD has "thick" (front illuminated) chip, so that it has no "fringe" effect in 6500-8700 Å as compared to back illuminated CCDs.

68 echelle orders (with physical numbers from 53 to 120) are registered simultaneously in the one "two-dimensional" echelle-frame (Fig. 8).

Table 1 gives the spectral coverage of all these 68 orders. During the data processing, extraction of the spectral orders is started from the red part of the spectrum (top of the frame), so the first (red) extracted orders have numbers of 1,2,3, etc (~ 8700 Å), and blue orders end with numbers 66, 67,68 (~ 3900 Å). Test observations have shown that highest Signal-to-Noise (S/N) ratio is realized within the 4500-6500 Å region due to the combination of CCD sensitivity, CES optics transparency and stellar flux distribution, see Fig.9. This effect has to be kept in mind when planning observations additional exposures are needed to reach high S/N ratio in the blue and infrared region, especially for very hot (O,B) or cool (K,M) stars.

Table 2 gives approximate exposure times and the recommended number of single exposures for the indicated range of visual stellar magnitudes to detect high S/N ~ 100 - 200 in the whole spectral range.

Tables 3 gives the results of radial velocity measurements for Vr-standard stars.


Table 1. Spectral coverage of 68 echelle orders for the R = 40000 camera with 1Kx1K CCD. (spectral resolution is 0.13- 0.25 Å depending on wavelength), see Fig. 8.



Order number Wavelength range (Å) Order number Wavelength range (Å) Order number Wavelength range (Å)


8688 - 8772


6127 - 6186


4733 - 4778


8533 - 8615


6050 - 6108


4686 - 4731


8383 - 8464


5974 - 6031


4641 - 4685


8239 - 8318


5900 - 5957


4596 - 4640


8099 - 8177


5829 - 5885


4553 - 4596


7964 - 8041


5758 - 5814


4510 - 4553


7834 - 7909


5690 - 5745


4468 - 4510


7707 - 7782


5623 - 5677


4426 - 4469


7585 - 7658


5558 - 5611


4386 - 4428


7467 - 7539


5494 - 5547


4346 - 4388


7352 - 7423


5431 - 5484


4307 - 4348


7241 - 7310


5370 - 5422


4269 - 4309


7133 - 7201


5311 - 5362


4231 - 4271


7028 - 7095


5252 - 5303


4194 - 4234


6926 - 6992


5195 - 5245


4157 - 4197


6827 - 6893


5140 - 5189


4121 - 4161


6731 - 6796


5085 - 5134


4086 - 4125


6638 - 6701


5031 - 5080


4052 - 4090


6547 - 6610


4979 - 5027


4018 - 4056


6458 - 6520


4928 - 4975


3984 - 4022


6372 - 6434


4878 - 4924


3951 - 3989


6288 - 6349


4828 - 4875


3919 - 3956


6207 - 6267


4780 - 4826




Table 2. Approximate exposure times for the SAO RAS CCD. 



Visual magnitude range Single exposure duration (sec) Recommended number of exp. Total exp. time (min.) Remarks

0 - 1

5 - 10


10 - 15


1 - 2

20 - 30




2 - 3

60 - 120




3 - 4

180 - 300


25 - 40


4 - 5

300 - 600


20 - 40


5 - 6





6 - 7

900 (15 min)



7 - 8

1800 (30 min)

3 - 4 

90 - 120

Seeing < 2 

8 - 9

1800 (30 min)

4 - 6

120 - 180 (2-3 h) 

S/N ~ 30-100


Table 3. Results of radial velocity measurements of Vr standard stars by ELODIE spectrometer and RTT150 CES.


Date/Star, HD Vr, ELODIE Vr, RTT150 RTT - ELODIE (RTT-ELO)-0.5
May 06, 2004 km/s km/s km/s km/s
















May 07, 2004















May 08, 2004






+ 9.55














June 06, 2004

































+0.50 mean

+/-0.10 rms







b) Tests of positional accuracy.


To obtain data for Table 3, spectra of radial velocity standard stars have been obtained in 2004 from a new list of very accurate (~ 50 meter/sec) radial velocity determinations. This list is based on observations obtained with ELODIE Coude echelle-spectrometer (Baranne et al, 1996) attached to the 1.93 m telescope of Haute Provence Observatory (France).

Processing of these data have shown the following results:

1) Zero point of RTT150 CES is -0.5 km/s +/-0.1 km/s. This value has to be added with proper sign to the results of Vr measurements based on CES data.

2) Initial errors of Vr measurements are +/- (1-2) km/sec due to the existence of weak (but measurable) flexibility effects in the optical-mechanical system "RTT150 + CES"

This flexibility effect is controlled by measurements of positions of telluric lines of O2 and H2O molecules in the orders N 9, 16, 22 (Fig.8,13,14). Radial velocity value measured from the telluric lines has to be equal zero. Deviation of this Vtell from zero gives the effect of flexibility. Data given in Table 3 are corrected for these flexibility effects.

Adjustment of echelle-grating angle was made to expose the orders with telluric lines.


c) Tests of photometric accuracy.


Unfortunately, there are no lists of published "standard" equivalent widths of spectral lines in EW "standard" stars. In this situation we have compared our EW data measured from RTT150 CES spectra and those measured from spectra of high quality spectrometers like ELODIE (for the common stars). A comparison of equivalent widths measurements is shown in Fig.15, which indicate a very high systematic agreement (at the 1 percent level) between these two coude-echelle spectrometers.

In Fig.16 part of spectral range is shown for the same star obtained with ELODIE (Bruntt et al, 2004) and RTT150 CES.

These results of test observations with the RTT150 CES are very promising: in the case of detecting spectra with S/N ~ 100-200 and with proper echelle data reductions, it is possible to measure radial velocities of cool stars with narrow sharp lines with an accuracy of ~ 0.1-0.2 km/sec, to detect weak spectral lines and measure EWs with systematic errors within 1 percent and rms errors within 2-4 mÅin the "working" range of 10 - 300 mÅ.

All data reductions have been made by using DECH software realized by Dr. G. Galazutdinov (1992) for echelle spectra reductions. We highly recommend this PC based software for RTT150 CES data reductions. This DECH software is freely distributed now; 



Acknowledgements. Authors are grateful to Kazan State University, Academy of Sciences of Tatarstan (Russia), and TUBITAK National Observatory (Turkey) for financial supports of RTT150 Coude-echelle spectrometer construction.





Baranne A. et al, 1996, Astron.Astrophys.Suppl.Ser., v.119, p.373 (ELODIE).

Bruntt H., Bikmaev I.F., Catala C, et al., 2004, Astron.Astrophys., v.425, p.683.

Galazutdinov G.A., 1992 // DECH software, Preprint SAO RAS N 92.

Gray D., 1976, The Observations and Analysis of Stellar Atmospheres, Univ. of Western Ontario, Canada.

Musaev F.A., 1993, Astronomy Letters, v.19, p.776.

Musaev F.A., Bikmaev I.F., 1995,  ASP Conference Series, v.81, p.146.

Musaev F.A., 1996, Astronomy Letters, v. 22, p.795.

Musaev F.A. , Galazutdinov G.A., Sergeev A.V., Karpov N.V., Pod'yachev Yu.V., 1999, Kinematics and Physics of Celestial Bodies, v.13, p.282.

Schroeder D.J., Hillard R.L., 1980, Applied. Optics, v.19, p.2833.




Fig. 1 CCD image of the single order low-resolution spectrum. (vertical line is the Earth atmosphere emission line) 



Fig. 2 The profile of first-order spectrum of classical grating. 



Fig. 3 Main spectral details of Hot star spectrum, telluric bands and fringes.



Fig. 4 Cross-section of RTT150 building with coude-room and the positions of CES units.



Fig. 5 The optical scheme of RTT150 CES. Top view.



Fig. 6 R = 40000 spectrum view from the backside of camera's flat mirror (echelle orders are not resolved due to autofocusing regime of digital camera.



Fig. 7 Full echelle frame registered by 2K x 2K Andor CCD in the whole 3700-10000 Å range (85 orders). Fringes (interference) effect is seen at the top part of the frame (L > 7600 Å)



Fig. 8 Full echelle-frame of V ~ 7 mag star registered with 1K x 1K nitrogen cooled CCD in 3900-8700 Å spectral range (68 orders).

Single exposure time is 15 min. White dots are cosmic spike events. Because of the influence of these events, exposure times over 30 min are not recommended. To eliminate them, at least two single exposures have to be obtained for each star during the night. Vertical thin lines are cosmetics effects of the CCD chip. They are eliminated during processing procedures within the DECH software. On top of the frame echelle order N 9 with deep telluric lines at ~ 7600 Å is seen, and the last two orders in the bottom of the frame contain H and K CaII lines.


Fig. 9 Relative Intensity distribution over the whole set of 68 orders extracted from spectra of hot and cool stars. Each individual echelle orders have ~ 40-90 Å in length. Compare with Fig.2.
Note the strong dependence on spectral class of the star for cool star blue part of spectrum has lower intensity and for hot star near infrared part (6000-8500 Å) has lower intensity. To get high S/N ~ 100 in the whole spectral range, a set of multiple (4-10, but not 1-2) exposures are recommended. 


Fig. 10 Enlarged part of Fig.9 with only 7 echelle orders. 



Fig. 11 Enlarged part of Fig. 10 - the single order N 41 before local continuum rectification.



Fig. 12 Order N 41 after continuum rectification, S/N ~ 200 at continuum level.



Fig. 13 The cross-section of normalized order N 9 with deep telluric lines.



Fig. 14 An example of detection of telluric line shifts in spectra of two stars (Fig.9) due to instrumental effects (between two nights).



Fig. 15 Example of EW comparison for the common star (Bruntt et al., 2004) as measured from RTT150 CES and ELODIE echelle spectra.



Fig. 16 Example of direct comparison of RTT150 and ELODIE spectra of the same F-star.



Fig. 17 The view of echelle frame on the PC screen after the end of exposure and reading data from CCD.



Fig. 18 The CES entrance slit view during the remote control guiding procedure using wings of image of targeting star (here 10 arcsec diaphragm illuminated by Thorium Argon lamp is shown). Slit width is 1.5 arcsec for R = 40000 resolution.

-A +A