Training requirements for INT service/queue observing
This data is redundant, in the sense that the TO's most likely are not going
to act as queue observers after all. However, this information still deserves
it's place on the web as we *will* encounter many of these situations
during the night, queue observers or not
Access atmospheric conditions
Read proposals and make the selection
Starting up the observing system
Quality control of detector performance
CCD read out speeds, gain, linearity
Taking bias frames
Internal & Sky flats with IDS
Dome & Sky flats with WFC
Taking comparison arcs with IDS
Taking flux standards with IDS
Taking Velocity standards with IDS
Taking photometric standards with WFC
Checking rotation, tilt and focus using ING scripts for IDS
Checking tilt and focus using ING scripts for WFC
Adjusting the cryostat micrometer settings
Checking the central wavelength of IDS by arcline plots
Setting up the CCD window
Changing grating and collimator in IDS
Changing filter of the WFC, updating the ING filter database
Analysing the data using IRAF
Making the data available to the observers via the ftp area
Checking diskspace, deleting files
Writing data to D-tape
Control instrument from the engineering PC
Recover from faults on the obs/data systems, including shutdown
Getting finding charts, positions, magnitudes using the SIMBAD account
As experienced TO's, we know the several ways to judge the weather
conditions :
- MET Display in the control room
- Weather Map/Movie from IAC
- Handhelp Vaisala, in case of MET displaying crap
- And of course a VISUAL inspection, which can prove particulary useful in the
vinther when there is a risk of Riff-clouds comming over
Description of safety limits can be found here
How to record weather downtime can be found here
The Service Proposals are stored in folders at the control room, one for
IDS and one for WFC. Making the decision on which proposals to do should be done
according to the following guidelines:
- Accessibility: The proposals are ordered with respect to RA, which
is displayed on the first page as e.g. 'RA range=18-19'. The first rough
selection is always to choose out those proposals which are at all possible
to do that night on basis of the Siderial Time.
- Instrumental setup: Selecting proposals which requires widely
diferent setups (like different cameras, collimators, gratings etc. for IDS)
only makes the night much tougher! If possible, avoid such combinations
and try to settle for one particular 'base' setup, like 235 camera and
AG-wide collimator. None of us really want to mess around with the WFC filter
wheel in the middle of the night, so the same goes for this instrument; Settle
on one particular choice of filters that suite all the nights proposals.
- Rating: The proposals have been rated by outside referees into three
classes: 'alpha', 'beta' and 'rejected', alpha proposals having the highest
priority. Only go for beta proposals if no alphas are available (only
happens rarely).
- Nationality: UK, NL and spanish proposals should each get their
overall share of the service nights. If no NL proposals, for
example, have been carried out for a long time, these proposals
will take priority over the others. To find out whether this is a concern,
talk to Steve Smartt.
Once having settled on a set of proposals and ordered them throughout
the night, it is important to get an idea of the calibration needed.
Almost all proposals require at least flats (being sky or lamp), and
specially the IDS lamp flats for all proposals should be done
before starting observations of obvious reasons (waisting 2 hours dark
time doing lamps doesn't look good in the papers). This can mean a
lot of work: changing gratings, wavelenght regions etc., doing
at least 5 flats at each particular setup. This is really the reason
why the proposals should be selected with the least amount of instrumental
changes; Each setup requires calibration, and the more setups you work
with, the greater is the risk also that returning to a particular
setup later that night has suffered from small changes due to
grating/collimator flexing etc.
The most up-to-date information on how to start up the
current observing system is kept elsewhere on the TOweb, click
here
to view
This information is public available on the ING web pages
This information is public available on the ING web pages
Introduction: To prevent a low level signal to transform into a
negative value after being converted to ADU (by the Analog-to-Digital
Converter in the CCD controller), a constant bias offset of typically
500-1000 ADU is introduced to the value of each pixel by means of CCD
hardware setup. Ideally, subtracting two bias frames should not yield anything
but the random pixel-to-pixel noise, gaussian in distribution with a mean value
of zero. The bias frames can however sometimes suffer from:
- Temporary variation of the mean bias level caused by drifts in CCD/
Electonics temperature or CCD offset voltage.
- Patterns/Structures in the bias frames as a result from electonic
interference. These structures normally varies with time as well.
Application: As one of the very first things to check when getting to
the telescope is whether the bias suffers from any of these effects. If
structures are present, call the DT. It might be that the problem is wrong
cabling. If the bias looks ok, there is really no reason to waste time during
the night to take biases unless the observer specifically ask for it. Any
changes of the mean bias level during the night is then typically taken into
account by the observer by using the CCD overscan region(s) to estimtate
the actual bias level (descriped in more detail in the image/spectra reduction
courses notes). Examining a bias frame can tell you whether there is an
offset present between the mean 'real' bias level and the mean overscan level.
This is a subtle effect, an worries the fewest observer. In summary: Take some
3 bias'es in the evening, get the mean. If everything looks ok, take a couple
more in the morning and everybody will be happy.
Introduction: The light reaching the CCD pixels through a spectrograph
is basically affected by three distinct sources of disturbance which can be
removed using flatfield images:
1. Dirt/dust particles on the slit and CCD window. Grains on the slit will
show up as lines of lower intensity along the spectral direction, whereas
unwanted objects on the CCD window/filters will show up as classic dougnut
shaped blobs on the frames.
2. Fringe patterns. These are mostly seen in the red, and can unfortunately
vary in time.
3. Pixel-to-pixel variations. These variations are basically the result of
small structural variations of the individual pixels. Some pixels can turn
out to be nearly 'dead', i.e. with almost no light response. Dead columns
and other cosmetic effects (caused by bad chip manufaction) will also be mapped
by the flatfield image.
The basic idea is to extract these spatial structures from the flatfield image
and use this information to remove the same structures on the object frames.
As decriped in the Spectral Reduction Notes, you will need to do a
proper normalization of the flatfield image before applying it to the object
frame(s). Because each arithmetic operation performed between two CCD frames
will increase the noise (and therefore decrease the quality), the more
flatfields the better! More counts means higher Signal-to-Noise ratio and
therefore less noise. Due to CCD/grating efficiency dependency of wavelength,
you will typically get much less counts in the blue than in the red part of
the spectrum (easily a factor of 10). To accumulate enough signal in the blue, it
is therefore necessary to take many flatfields, which in practice means of the
order of 10.
Application: If the observers program deals with extended objects (galaxies,
clouds,..), he most likely requires sky flats to ensure a flat illumination along
the slit (spatial direction). Sky flats should be taken at parallactic angle to minimize the
effects of atmospheric refraction. The slitwidth when taking the flats (sky or lamp)
should be kept as close to the actual slitwidth used during observations.
Internal tungsten flats should be taken before beginning the actual observations
in order not to waste dark time. If several programs are planned
for the night, tungsten flats should be taken for each setup, which means
for each grating/central wavelength required by the observer(s). As at least some
6-10 exposures should be taken for each setup, you can easily end up with 50
flatfields! If you dont want to end up in the morning having to spend hours doing
this, it is a very good idea to start out early in the afternoon by carefully
reading the proposals to figure out which flatfields you need.
Introduction:Pixel-to-pixel variations, dead columns and other cosmetic
effects plus vignetting at the edge of the field are all effects that can be
corrected by taking appropriate flat fields. As with spectroscopy, all imaging
applications will benefit from a good set of flatfields.
Application: If the weather permits, always go for sky flats instead
of dome flats. Due to the enormous field of view of WFC (30 arcmin squared), none
of the cataloged ING flatfield areas will be free of stars. To get a 'clean'
flatfield, the procedure is to step the telescope some few arcseconds
between each flatfield exposure. This enables the observer the filter out the
stars when doing the datareduction (For example by using the IRAF task 'flat
combine', as described in the Image Reduction I notes. The four EEV chips
are linear up to some 50k ADU, and observers typically aim for some 20-30k ADU
counts in the flatfields. The read-out time for the 4 CCD's is around 160 seconds
which means that in order to get the same amount of counts in the next
flatfield exposure, you will need to
double the exposure time. One way to estimate when to start exposing is to
use the autoguider readings. When the autoguider loral-chip no longer saturates
in 1 second read-outs, you may consider beginning. On a typical twillight, you
have time enough to get some 5 flatfields in one filter before it gets to dark.
Remeber that the sky is much brighter in R than in B, so if the observer requests
flatfields in both filters, start out with B in the evening and R in the morning.
Introduction: Spectra of an arc lamp is used to relate the pixel values
along the dispersion direction with wavelengths, as the lines produced by the
arc lamp are well-known from the laboratory. As IDS suffers from a certain
amount of flexure (due to rotator/telescope motion and temperature variations),
the wavelength-to-pixel correlation will change during the night. The flexure
is of the order of few tenth of a pixel, translating into a wavelength displacement
of some few tenth of an Angstrom.
Application: Depending on the actual program, the observers might request
only few (or none) arc calibration spectra, one for each object or even two for
each object, one before and one after the object exposure. Which lamp(s) to use
depends mainly on the observed wavelength region as the different arc lamps (cupper,
neon and argon) produces different amount of lines as a function of the wavelength.
In the blue region (3000-5000A), a combination of CuAr is a good choice. In the red,
CuNe is best. For very high dispersion, ThAr can be used. Exposure
times for arcs are typically some 10-60 seconds. As the sensitivity of the CCD's
drops of rapidly in the blue, it is important to check that there are enough
counts in the blue arclines. It might be necessary to use a blue filter to
prevent the red arclines from saturating.
Introduction: In order to derive accurate effective temperatures, bolometric
luminosities, and surface gravities for astronomical objects from their energy
distribution, it is neccessary to calibrate spectroscopic observations with sources
of known spectral energy distribution. These sources are known as flux standards.
The difference between the true (tabulated) energy distribution and the observed
energy distribution of a flux standard is mainly due to the finite grating effeciency
and CCD quantum effeciency, although all light reducing elements throughout the
atmosphere/telescope/instrument is included. Deriving a function which multiplied with
the observed spectrum yields the tabulated values (response function?) can then
be applied on the target objects to give the true energy distribution.
Application: It is essential that all light from the flux standard enters
through the slit. To achive this, a wide slit of typically 7-9" should be used.
For good policy, observe when possible at parallactic angle to reduce the
effect of atmospheric refraction. If the observing program extends over several hours,
the flux std. observations should be taken as far apart as possible to take into account variations
of the atmospheric conditions, like dust. No need to say, of course, that flux standard
observations in cloudy conditions are pretty useless if the application calls for absolute
flux calibraion! A list of suitable flux standards can be found in the
blue folder labelled "Spectral Flux Standards", situated in the INT control room. Typical
magnitudes of these flux standards are v=11-13, so an exposure time of 100-300s should be expected.
Each flux standard should be accompanied by a comparison spectrum (arc lamp).
Introduction: Velocity standards have a well-known radial velocity, and a descrepency between
this velocity and the velocity deduced from an observation is the indication of an instrumental zero-point
offset.
Application: A list of suitable velocity standards can be found in the astronomical almanak.
These stars are fairly bright, and typical exposures are only some few seconds. It is essential that
an arc is taken immediately before or after the observation so that the effect of flexture is
supressed.
Introduction:The observers require photometric standard star observations if
the program deal with photometry, that is: absolute flux measurements to estimate
magintudes and/or surface brightnesses. Almost all programs will require such observations.
Application: In the INT control room there should be a red folder labelled
'Landolt 1992: Photometric standard fields', which is a commonly used list of UBVRI
standard stars. Each field (plate) contains a certain number of standard stars, which in the
case of WFC sums up to quite a lot of stars due to the big field of view. If the observer
do not explicitly tell you the field to observe, pick an appropriate one from the list
on basis of ra/dec. Observe, if possible, the same field at different airmasses so the observer
can calculate the atmospheric extintion curve.
Rotation - Cambridge link, parts out of date
Tilt - Cambridge link, parts out of date
Focus - Cambridge link, parts out of date
There are at the moment no specific scripts available to calculate tilt for the WFC.
If you want to check that the ccd's are aligned properly (PA=180 -> North is up), step
the telescope across a starfield and measure manually the X & Y positions. Also check
the actual kapstain readings agree with what is recorded in the logbook. If the readings
are correct, the WFC should always be optimal aligned. Stars should come
in focus around 46mm (telescope focus).
Cambride link - some parts out of date.
Introduction: Although I've never seen IDS fail to center on the demanded center
wavelength, it is quite re-assuring to quickly check by using an arcline spectrum.
Who would like to suddenly realize that the last 2 hours of observation has been
wasted due to wrong wavelength region?
Application: After having acquired an comparison arc,
we need to be able to identify some lines in the spectrum. At the moment,
we have at least two ways of doing that:
- Using actual IDS line plots. These plots should be placed in a folder in the
INT control room. These plots gives identifications of the strongest CuArNe lines for
different wavelength regions and resolution power (grating settings).
- Using crb's interactive arclines program. This program uses generic CuArNe etc
datafiles, and applies wavelength cuts and resolution depending on the chosen IDS
configuration. Quite cool! To run the program, simply type ~crb/ing/arclines
Note that the tabulated relative line intensities not necessarily corresponds to what is
measured in the observed spectrum. This is due to lamp temperature dependence among other
things.
Having identified some arclines in the spectrum , it is straightforward to do a quick
pixel-to-wavelength correlation using IRAF (for full documentation, see notes from
the spectra reduction course). This allows us to determine the central wavelength
exactly and at the same time determine the spectral resolution (or dispersion) in
A/pixel:
1. run task identify (in noao/onedspec) directly on the 2D frame. The parameter nsum
determines the number of lines/columns to sum in the 2D image.
2. Mark identified lines with 'm' in interactive mode, when 3 has been marked, use 'l'
to look up the rest of the features from the coordinate list (use the right coordinate
list!)
3. Check fit with the 'h,j,k,l' keys, and if it looks reasonable, use the colon command
':show' to list the fit parameters. Here you find wavelength of the first pixel and the
dispersion in Angstroms per pixel.
Introduction: The WFC do not allow any windowing; All four chips or nothing!
Windowing the CCD when using IDS is on the hand very important, as i) it decreases
use of diskspace and ii) decreases the read-out time. Windowing the CCD basically
decreases the amount pixels in the spatial direction. Depending on the program, this
may be a concern or not (dealing with extended objects or point sources). In any case
it is completely safe to window out the pixels outside the length of the slit (30
arcseconds), as no light enters there anyway. (For the new EEV12 chip, this means
almost 1000 pixels in x)
Application: With no windowing (SYS> window 1 0 0 0 0), take an arc or lamp.
Looking at the image, identify the edges of the slit roughly where the light level
falls off. If left and right borders were determined to be at x=700, x=1300, then
it would make sense windowing the chip as SYS> window 1 1000 0 600 4150 (x centered
at 1000, 600 wide and full range in y). Further windowing should only be done if
observer agrees.
Grating change: Gratings are placed in the metal shelf on the observing floor.
Before you can change the grating, type SYS>Change_IDS to unlock the grating door
and position the grating for removal. Once you have unclamped the grating, carefully
take it out and store it in it's box. When putting in the new grating, make sure you have
the direction right (marked on the grating)! Otherwise the reversed blaze angle will result
in a huge loss of throughput. After clamping the grating and closing of the grating door, lock
it from the SYS> prompt. Remember to identify the new grating in the menu!
Collimator change: As with grating change, type Change_IDS to unlock the collimator
door. The collimators (UV, Red, Wide) are stored in the metal shelf. Three mechanical clamps
are holding the collimator in place. Once released, take out the collimator carefully
and replace with new. The correct orientation of the collimator is given by the clamping
positions. Once in place, close the door, make the apropriate changes in the menu and
save & exit to lock the door. An adjustment of the collimator focus is most likely needed
after a change.
The procedure for changing WFC filters are described in the WFC notes found elsewhere on
the TO webpage.
See notes from the TO image- and spectra reduction courses.
In order to deposit data in the service FTP area simply copy the file
(or write from IRAF or whatever) to
/obsdata/serviceinta(b)
/obsdata/servicejkt
/obsdata/servicewhta(b)
Write access depends on being a holder of the group anftpint/jkt/wht.
To confirm for any acocunt type 'groups'
To see who HAS got write access use niscat group.org_dir | grep ftp
If you mistakenly cd to /obsdata/service/int you will see a reminder
that at least until the migration is complete you must use
/obsdata/serviceint and not /obsdata/service/int.
Diskspace on the three obsdata disks (inta,b,c) can always be checked using unix 'df'
command. The diskspace on the currently selected disk is also displayed at the bottom
of the log window (as 'enough space for xxx runs..'). If diskspace during the night
becomes low, you can either i) Delete verified data on that disk (follow link below),
or ii) Change the active disk to one of the others. This is done by the command
'SYS>obsdata /obsdata/intc 68', which tells the system to start writing the data
to /obsdata/intc/yymmdd, and that each run is 68Mb large (to calculate remaining
diskspace only). Dont forget you changed disk during the night when it comes to
writing the data to tape!
1. Insert the DDS3 (125m) tape in the drive /dev/rmt/0n or /dev/rmt/1n.
A DDS3 tape (12GB) should be able to store 150 runs in fits format, or about
300 using tar.
2. Open a new xterm on the Data Reduction machine and change directory to
where the data are stored.
3. prompt>tar cvf /dev/rmt/0n *.fit - writes all fit files to tape.
To be processed under 'TO technical training'
Check TO notes on the specific instrument for fault finding/recovering. These change constantly!
The shutdown account should only be used if there is no other way to get the machine up running.
When used (directly on the console or through a telnet session to the machine), a proper shutdown
will be performed. In cases it is necessary to type 'boot' in the startup process to actually
reboot the machine.
Introduction: It happens unfortunately quite often that the observers come up
with wrong coordinates for their objects. Therefore, before beginning to examining
the telescope encoders in greater detail to find the reason why the object did
not turn up on the tv, it is quite useful to first check that the target coordinates are
correct. This is where the SIMBAD database comes in. This WWW based database contains
coordinates, identifiers, basic data and bibliography of a HUGE amount of objects, and
it is more than likely you will find the data you are interested in. SIMBAD is fast,
and every TO should have access to the database with a simple left-click on the mouse!
Application:A link to the main SIMBAD search window is put on the TO webpage.
At the moment, no userid/password is required to logon, but in case that changes, a valid
userid/password has kindly been provided by Peter (to be found on the TO webpage). Once
logged on, simply type in the object name and hit the submit button. The only tricky thing
is to get the identification right. For Messier and NGC objects it's straight forward,
simply type 'm31', 'ngc6058', etc. For other objects, however, it can be a bit more tricky.
Abell objects for example (planetary nebulae) are normally called e.g. A39 by observers. For
SIMBAD to get the right object, however, you'll have to type 'PN A66 37' ! You will need to
have a bit patience to get this right, and read the on-line helpfiles. If this leads nowhere,
you can as well directly type in the coordinates in the query field as '12 30 45 +30 00 00'
which will return a list of objects around a specified radius you select. As a new feature
, plots of specific fields can as well be produced. Using the Digitized Sky Survey (DSS) is another
option for getting target coordinates (actually it uses the SIMAD database). In addition, DSS
can produce fits or gif files of specific fields from the survey. This is properly the closest
you get to real finding charts. A link to DSS is also available on the TO webpage.