Miscellaneous information

• Airglow emission lines in the red arm
• FITS headers with telescope parked
• Scattered light and ghost images
• Known scratches on the gratings
• Dichroic focus offsets and ripple profiles
• Available colour filters and their focus offsets
• Image rotation caused by the dichroics
• Slit view and CCD orientation
• ISIS red arm astigmatism problem
• Focus offset caused by gratings

The sky background contributes numerous emission lines in the red arm from airglow, predominantly from the OH Meinel bands. Street lighting contributes to the strength of the sodium doublet at ~5890/6Å. The detailed appearance of the airglow spectrum depends on resolution, and this figure depicts it at a resolution of ~1000 (R158R grating, courtesy of Nik Humphreys and Chris Benn).

A detailed critique of all contributions to the La Palma Night-Sky Brightness is given here.

At the end of each night the telescope is parked very close to zenith and left in engineering mode. In this mode TCS commands to move the telescope are ineffective, and FITS header entries ordinarily provided by the TCS such as RA, Dec, parallactic angle and sky position angle, are wrong in images taken in this configuration. However, raw FITS header entries such as telescope azimuth and zenith distance, and mount position angle, are reliable.

On some occasions it's necessary to know the sky position angle of the slit in calibration images taken with the telescope parked at zenith, for example, in spectropolarimetric twilight flat fields. The calculation of the sky position angle from raw FITS header entries is described below.

The Cassegrain mount position angle, m, slit sky position angle, θ, ISIS slit offset, θ₀, and parallactic angle, q, are related by

m = (θ - θ₀) - q (1)

and so the slit sky position angle is

θ = m + θ₀ + q (2)

The parallactic angle is

sin (q) = sgn (h) cos (φ) sin (a) / cos (δ) (3)

where h is hour angle, φ is latitude, a is azimuth, δ is declination and sgn (h) = -1 for h<0 and sgn (h) = +1 for h>0. The telescope's latitude is φ=28.760470^o, and the fixed ISIS slit offset is θ₀=-92.1^o, as written in FITS header keywords LATITUDE and PAOFFSET, respectively.

When the telescope tracks a target during normal ISIS observing, the mount moves to maintain a fixed slit sky position angle, θ, and the values of mount position angle, m, and parallactic angle, q, at the start and end of each exposure are written in FITS header keywords MNTPASTA, MNTPAEND and PARANSTA, PARANEND repectively.

When the telescope is parked at zenith, it's parked just outside the formal zenith blind spot at a zenith distance z_p=0.28^o, as listed in FITS header keywords ZDSTART, ZDEND, and in azimuth a_p=298.64^o, as listed in keywords AZSTART, AZEND. The parked mount position angle is usually, but not always, m_p=45^o, as listed in keywords MNTPASTA, MNTPAEND. The telescope's declination therefore must satisfy (φ-0.28^o) ≤ δ ≤ (φ+0.28^o), and so the ratio cos(φ) / cos(δ) must be in the range ~0.997 to ~1.003. Therefore from (3), with the telescope parked 0.28^o from zenith, its azimuth must be within ~4.4^o of the parallactic angle over all values of azimuth, and so

θ ~ m + θ₀ + a (4)

Therefore, using the approximation q_p~a_p~298.64^o, the slit sky position angle from (4) is ~251.5^o, with the mount parked at m_p=45^o. For a more precise calculation, the parked telescope's horizontal coordinates can be transformed to equatorial coordinates, h_p and δ_p. These are

h_p=00^h01^m07.3^s and δ=28.8944^o

The parallactic angle from (3) is then 298.50^o, which differs from the telescope's azimuth by only 0.14^o. In the standard zenith park position azimuth is therefore a very accurate proxy for parallactic angle. With the mount parked at m_p=45^o, the precise slit sky position angle from (2) is 251.4^o.

The slit sky position angle can be calculated from (2) when the telescope is parked in an arbitrary position, e.g., at zenith distance 45^o to take dome flat field images, but of course telescope azimuth is no longer a suitable proxy for parallactic angle, and the latter must be calculated explicitly.

Scattered light in ISIS is minimised by the use of anti-reflection coatings on the optics, and of course by deploying a dekker above-slit. Diffuse scattered light has been demonstrated to be less than 2% from observations of totally black absorption lines in quasars.

ISIS does, however, exhibit a number of generally low-level ghost images, caused by stray reflections within the spectrograph. The main ghost images which can manifest themselves, listed more-or-less in order of relative significance, are:

• In the blue arm when a dichroic filter is deployed and when observing bright targets such as standard stars, a ghost spectum which runs parallel to the primary spectrum. The ghost is strongest at wavelengths in the crossover region. It is caused by light reflected off the back surface, as opposed to the front reflective surface, of the dichroic. The displacement from the primary spectrum is defined by the thickness of the dichroic, and is ~10.8-arcsec (54 unbinned EEV pixels). Under certain circumstances, e.g., an especially bright target, further, equally-spaced ghost spectra may be apparent; these are caused by multiple reflections within the dichroic. This image shows an example of two ghosts, with a hint of a third, in the blue arm (the primary spectrum is the rightmost). The observation was taken with the D5300 dischroic and R300B grating, and the CCD was binned 2x2. The intensity of the first ghost is ~1/14 that of the primary spectrum, and the intensity of the second, fainter ghost is ~1/30 that of the first ghost. The separations of primary, first and second ghost are 26.8 binned pixels.
• In the blue arm with a dichroic deployed, the light passing through the dekker slots is reflected into the blue arm by the mirror which lies adjacent to the dichroic. This effect can be seen as in-focus zeroth-order image of the slit-dekker area at the blue end of the R158B grating in the form of bright rectangular patterns, plus low-level spectra of these images slightly spatially displaced at the red end of the R158B grating. The same effect is seen also with other gratings in the blue arm when the central wavelength is set very red. An example of this ghost with the H2400B grating and the central wavelength 6000Å can be found here. This ghost vanishes when the observing dekker is deployed.
• In the blue arm without a dichroic deployed, when the central wavelength of the R1200B grating is set to a wavelength redder than 5300Å a slightly-diffuse, ghost image is seen at the red end of the spectrum. Of course, the ghost appears brighter when the detector is binned; this is an example with 2x2 binning.
• In the red arm, a reflection of stray skylight between the R158R grating and the camera can cause an undispersed ghost centred at ~9300Å, which extends over ~340 pixels in the spectral direction. It occurs when the mount position angle is in the range -8 to -64 degrees, and has maximum intensity of a few tens of counts when the mount position angle is in the range -25 to -50 degrees. The ghost does not occur at any mount position angle outside the range -8 to -64 degrees. Observers working redward of ~9000Å should therefore change the sky position angle of the slit by 180-degrees if the mount position angle would otherwise enter the problematic range during their integrations.
• In the blue or red arms, a faint, ghost image of the pupil, caused by reflections between the cryostat window and the surface of the CCD or its surroundings in the cryostat. This is usually caused by strong emission from a wavelength region just off the CCD. If the originating source is a continuum source, then the ghost, pupil image is smeared in wavelength accordingly.
• So-called Rowland ghosts in the blue and red arms, which are caused by periodic errors in the ruling of the gratings, are manifest as very faint (~0.01%) ghost lines centred around strong target emission and comparison-arc lines.


• From time-to-time slightly diffuse 'streaks' running in the spectral direction will be evident in flat-field images. These features are caused by particles of dust on the slit, and not by some inherent design aspect of ISIS. Dust on the slit of course affects both the blue and red arms. In this context note that for the default EEV12 and RED+ detectors, the spatial directions in blue-arm and red-arm images are reversed, e.g. a dust particle that is evident in the 'left' of a red-arm flat-field will be evident in the 'right' of a blue-arm flat-field, and vice versa. When dust features are easily evident over the slit, and especially if they move with telescope slews, submit a fault report requesting that the slit is cleaned.

Most of the scratches described below are very shallow except for the gratings R1200R and R1200B. R1200B has few small scratches located close to the edges that are deeper than for other gratings, but shallower than for R1200R. R1200R has three places 0.3 cm long located close to the edges where the scratch is very deep so that the grating surface is missing.

R158R: clean, only few very tiny stains
R158B: one scratch 10 cm long, one scratch 5 cm long, both in central area, two finger prints at edge, few small stains
R316R: two parallel scratches 0.3 cm long, one scratch 0.5 cm long, one small stain
R300B: one scratch 1.5 cm long, one scratch 2 cm long, one scratch 1 cm long, all close to one of the edges, two parallel scratches 3 cm long in the central area, few small stains
R600R: few small stains
R600B: few small stains
R1200R: four scratches 0.3-2 cm long close to shorter edge of grating, one scratch 0.5 cm long on the opposite shorter edge, about 14 scratches 0.1-1 cm long close to longer edge of grating
R1200B: three scratches 0.5-3 cm long close to shorter edge of grating, four scratches 0.2-1 cm long close to longer edge of grating, four shallow scratches 6 cm long close to central area, one small gray stain
H2400B: one small stain, area very slightly dimmer covering around 40 % of the grating surface When a dichroic is put in the beam to allow simultaneous red and blue arm observations then light to the red channel passes through an extra glass layer. This introduces a focus offset at the detector that can be corrected with the following movements of the red collimator.

Dichroic   Corrective Coll. Movement (µ)
----------------------------------------
5300              +860
5400              +1270
5700              +1350
6100              +1340
7500              +1320

Although light to the blue arm reflects off the front surface of the dichroic (bfold 2) and mirror (bfold 1) the two optics are not necessarily coplanar in the dichroic slide. For example with the D5300 dichroic, switching from mirror (blue arm only) to dichroic (both arms) requires a positive adjustment of the blue collimator by a ~+100-500µ, dependent on grating used. This switch is rare in practice, but it's important to check for such focus offsets in the blue arm when using any of the dichroics.

Observers should note that the dichroics introduce 'ripples' into the blue arm light with a specific wavelength profile. These can be removed by careful flat-fielding techniques when reducing the data, but observers may wish to avoid the problem altogether. Each dichroic has a characteristic reflection profile, following plots will effect your dichroic choice (plots for each dichroic linked from here).

Several colour filters are available for use in either the red or blue arms separately. If a colour filter is placed in the light beam (e.g. for order sorting purposes) then a focus offset is again introduced due to the optical thickness of the material. A collimator movement is again necessary to compensate for this focus chanage.

The following values have been measured for the red arm filters:

Filter     Focus shift (pix)      Corrective Coll. Movement
-------------------------------------------------------------
GG495         0.58                    +800
RG630         0.54                    +750
GG395                             
OG550
NG4
BG28
BG39
OG515

Values for the blue arm filters will appear here:

Filter     Focus shift (pix)      Corrective Coll. Movement
-------------------------------------------------------------
NG4
BG28
GG495            0.92                       1038
RG630            1.26                       1413 
BG39
Og550

If an observer is in doubt about which filters to use, if any, they should ask advice from their support astronomer. SA's should note that the filter slides can only be mounted in the appropriate locations, i.e. RfiltB slides can only go in the red arm in position B, RfiltA slides can only go in red arm position A. The filter slides are all labelled correctly (as of 22/2/99), and their contents are given below aswell as written on the slides themselves.

The diagram below shows the filter slides schematically, in each case the filter written uppermost is in Position 1. The positions 1 and 2 are the same for each filter and the corresponding ICL commands (rfilta, rfiltb bfilta and bfiltb for positions 1 and 2) correspond to these positions.

 ___________   
|  _______  |   B filt A    B filt A1     B filt B
| |___1___| |    GG 495       NG 4          BG 39
|  _______  |    RG 630       BG 28         OG 550
| |___2___| |
|           |   R filt B    R filt B1   R filt B2     R filt A     
 \         /    RG 630       BG 39        NG 4         OG 550 
  \       /     GG 495       OG 515       BG 28        GG 395 
   \     /
    |   |
    |   |
    _____

We have found that some of the dichroics in use with ISIS cause an image rotation on the Blue arm detector with respect to the mirror which sits in the same slide, and which is used for blue only observations. A summary follows:

Dichroic     Image rotation in pixels top to bottom  
-----------------------------------------------------
5400         1.00
5700         2.86
6100         1.33
7500         1.20

Arc lines were aligned along the detector columns with the mirrors, so that the "Image rotation" was effectively zero. The dichroics were then put in and the values for "Image rotation" in the table above refers to the offset between the arc lines at the top of the CCD and at the bottom. These values are in EEV12 pixels (i.e. 13.5 microns), the slit extent was 400 pixels, and the 1200B was used. In all cases no image rotation was found between the mirrors in different slides, only the dichroics with respect to the mirrors.

For the 5300 dichroic used with the R600B grating and EEV12 detector image rotation is 0.8 pixel over a slit extent of 966 pixels.

Some observers may want to work out the direction that the slit was pointing, and which end of the slit corresponds to which direction spatially on the particular CCDS. To help to determine which direction on the CCD corresponds to which direction on the sky, the following information is needed which links the ROTSKYPA angle in the image header to the slit view orientation. The ROTSKYPA is the Parallactic Angle of the Cassegrain Turntable on the sky. See also the information on the CCD orientations at the WHT home page.

                                             East
                                            /|\
                                             |            
                                             |
                                             |
For PA=0:                                    |
                                              ------------> North


For PA=90:                                   ------------> East
                                             |
                                             |
                                             |
                                             |
                                             |
                                            \|/
                                             North

The slit is ALWAYS viewed lying left-right on the acquisition camera no matter what the PA of the rotator is i.e. the acquisition camera is permanently fixed with respect to the slit.

An astigmatism problem in the ISIS red arm was reported in February 2000 with the D6100 dichroic. Fault report #12788 found that the origin of the astigmatism could be the dicroics.

During the 2002-01-03 night, the spatial focus was determined using different ISIS dicroics. The spectral focus was fixed in the afternoon for the Red and Blue arms of ISIS without any dichroic. During the night, we corrected the spectral focus offset introduced in the Red arm for every dichroic placed in the beam. The general set-up of the instrument was:

ISIS Arm    Grating  Central wavelength    Collimator Position      
--------------------------------------------------------------------------
Red          R316R            7500                        8784      
Blue         R600B            4500                        6279      

The results obtained observing a bright star close to zenith and using a slit width of 1 arcsec are summarized in the following table.

  ISIS arm  |  BFOLD position | Best telescope focus  
________________________________________________________
     Red              Clear                     97.75
     Blue             Mirror                    97.75

     Red              6100                      97.70
     Blue             6100                      97.75

     Red              5700                      97.70
     Blue             5700                      97.70

     Red              4500                      97.75
     Blue             4500                      97.75
________________________________________________________

We have found that some of the gratings in use with ISIS cause a focus offset when the focus is fixed for a specific grating. The origin of this is currently unknown, but the results are repeatable. Support Astronomers and observers should pay attention to this fact; it is good practice to check the spectroscopic focus following a grating change.

ISIS focus offsets caused by gratings were re-measured in 2012 using the dichroic 5300. The RED+ CCD was used on the red arm and EEV12 CCD on the blue arm of ISIS. The test was performed on 7th June 2012 and repeated on 28th October 2012. In the first occasion, the focus was fixed using R316R (red arm) and R300B (blue arm) gratings, on the latter occasion, the focus was fixed using R1200R (red arm) and R1200B (blue arm) gratings. The focus offsets from the measurements on 7th June 2012 were recalculated to correspond to fixed focus using R1200R (red arm) and R1200B (blue arm) gratings. The final results are averages of the two values, with an uncertainty corresponding to a standard deviation. The results are useful, for example, for support astronomers when using different gratings during one service night. Optimally, a support astronomer should take into account focus offsets of all the gratings planned to be used in one service night, and finish the afternoon set-up with a collimator value suitable for all of them. For example, if in one service night the gratings to be used on the red arm are R1200R and R158R, with the default D5300 dichroic and order-sorting filter GG495: the optimal collimator value for R1200R grating alone is 10960μ (see here) but now that also R158R grating is in use, the optimal collimator value for both is 1163/2=582μ less than 10900μ as the grating R158R will introduce a positive offset of 1163μ.

ISIS Focus               R158R/B   R316R/R300B     R600R/B    R1200R/B     H2400B
-----------------------------------------------------------------------------------
Red Arm (Dichroic in)  1163 ± 149   801 ± 175     502 ± 258     0 ± 0                                         
Red Arm (Dichroic out)  157 ± 202  -113 ± 225    -330 ± 283  -691 ± 42
Blue Arm (Dichroic in)  184 ± 113   412 ± 28       10 ± 4       0 ± 0   -1095 ± 110
Blue Arm (Mirror)       -48 ± 45    306 ± 199    -564 ± 8    -515 ± 26  -1435 ± 184

The plot below shows how spectral resolution degrades as a function of collimator offset from the optimal collimator focus, 8837μ in this example (not mm as wrongly indicated in the gif image). These measurements were performed using Tek4 CCD instead of currently used RED+ CCD.