List of large components:
List of small components:
PC that is used to control CIRSI; if this fails any PC
with pcAnywhere installed and 100Mb ethernet will do.
Linux PC for data storage.
Fibre to 100Mb ethernet convertor
4 port 100Mb ethernet hub
100Mb ethernet switch that isolates CIRSI network traffic
for the rest of INT LAN.
1. Computers, accounts, networks
Accounts We Use:
We use the following accounts
- intobs (for running the observing software ONLY on lpss13. DO not use
elsewhere, or you may have problems)
- intguest (a general account)
- int_login (used ONLY on the alpha TCS display)
What Computers Where
We use the following machines (May 2000 run, ip numbers may
change!). Addresses are: host.roque.ing.iac.es
**The CIRSI Computers:
- guestip0 (aka circc) - Win 95 PC mounted on the telescope. No monitor
connected, but one should be available in the dome in case of
problems. Runs PixCel acquisition software, and is controlled remotely.
- guestip1 (aka cirob) - Win machine running pcANYWHERE to guestip0,
which it controls. THIS IS IT'S ONLY PURPOSE. Run no other software
- guestip2 (aka cirds) - linux machine running 'samba' where data is
written by guestip0 (circc). Run no other software here. 'Samba' may
not start automatically if you reboot (so don't!) but if you do:
(i) Log in as root
(ii) type /etc/rc.d/init.d/smb [status|stop|restart|start]
type /etc/rc.d/init.d/smb start
**The INT computers:
- lpss13 (18.104.22.168)- SparcStation 20 Observing machine, run observing
scripts etc. from here (in a pink xterm). Also has TCS info displays,
and CIRSI/TCS talkers, which displays error messages. DO NOT run
- lpss14 - has all INT data discs mounted on it. It's downstairs.
- lpss8 (22.214.171.124) - ULTRA60, powerful machine for data
reduction. Also use netscape here (and NOT on lpss13). This is for
data reduction and quick look at pointing etc. It's a good idea to log
on as intguest here, but have plenty of cirsi windows as well so that
you can write to both the cirsi homespace and the cirsi space on the
INT data discs.
[ Also, a useful step is to load the Starlink software (e.g. this gets
Gaia, coco). Only use lpss8 for data reduction and running this
source /star/etc/cshrc ]
Gaia can be used to produce CIRSI finding charts, details of which
will be added at the end of these notes soon.
Where to use what account
a) Log into circc (guestip0) as cirsi
b) Log into cirob (guestip1) as cirsi
c) Log into lpss13 as intobs
d) Log into lpss8 as intguest
e) Log into cirds (guestip2) as cirsi
f) (If required) Log onto LPAS2 on the Alpha TCS as int_login
The Current (May 2000) network arrangement
Currently, data taken by guestip0 (circc) is written to the linux
machine guestip2 via a network drive mounted on guestip0. This SHOULD
THEN BE copied at the end of the night to one of the /obsdata/int
discs on lpss14. Use ftp to do this. The next day, make DAT tapes.
You can FTP as needed to guestip2 (from e.g. lpss8) to get the data
when it is written - but don't do lots of ftp-ing, or the network
slows, causing problems with the data taking. (You tend to lose data
by dropping loops, a bad thing...).
2. Typical startup procedure (computers etc.).
IMPORTANT: Restart guestip0 every day. Use pcANYWHERE on guestip1 to
do this. REMEMBER: Close all programs (e.g. the filter controller)
before restarting, or odd things and crashes can occur on
reboot, doing this can avoid problems during the night.
[SEQUENCING] Reboot guestip1.
In the LILO prompt introduce "win95" (without quotes) (push TAB to see options).
- Start pcANYWHERE if not running on guestip1 (cirob). Change the path
destination of the FITS files for the current night in the "pixel.ini" file (see 20).
Also be sure to make this directory on guestip2.
- Double-click on 'Cirsi camera'. (It takes about 2 minutes for guestip0 (circc) to
boot, you cannot connect during this time..). If the windows appears
pink, minimising and then maximising the window should cure it.
- Start the filter controller as well, and check the filter is in
position (more details on this later).
- Now goto lpss13. If the TCS is running (i.e. and xterm with SYS> is
there) type 'shutdownobsys' in the xterm and then logout of
lpss13. This is useful to get a fresh start.
- Use TCSEXIT to exit the TCS software on the alpha TCS control (where
the TO sits). Restart the TCS software from the alpha TCS display (if
you have a TO, he/she will do this). ** Make sure you use versions
I13-1-2 and s7-1 of the software, or things won't work **. These are
older versions compatible with CIRSI, and are not the default options!
- Once the TCS is loaded, log into lpss13 as 'intobs'. Minimise the
orange window that appears. In the pink window, type 'obssys' and
select the CIRSI option (no.20 in May 2000)
- Check where startobssys is. Type 'alias startobssys' and you should
source $top/$system/bin/startcirsi.csh $OBSSTATION;echo "startobssys
Also check the following responses:
SYS> echo $top
SYS> echo $system
- type 'cleanup' in case the last shutdown was unexpected.
- type 'startobssys' and lots of windows pop-up. You can minimise the
autoguider as we don't use it. arrange the others so you can see them
all, the infodisp in particular.
The observing system should now be running happily on lpss13. Next,
startup of CIRSI for Remote Control
- cd to cirsi directory ('cd cirsi')
- check that the CIRSI send command is in your path
The source for this is in:
see the README and send.c for more details
SYS> where send
- Set the CIRSI environment variable for remote control
SYS> setenv CIRSI 126.96.36.199 (note: May 2000 value...)
- Now check the link to CIRSI using:
SYS> send status [should return BUSY or READY or rejected if PixCel not
started. Also sometimes get UNKNOWN which is a bit weird]
If PixCel is started and you still get a "connection reject"; check
the Pixel.ini file on the windows machine (guestip1) for:
(i) SOCKET 16421
(ii) 188.8.131.52 [if using lpss13.roque.ing.iac.es]
- Check the TCS link that gets the TCS headers
Note: This file is updated by the user 'intobs' and is
currently read by PixCel as 'cirsi'. The file is created
by user 'intobs' by the script 'CIRSIgetfits' from packets
created by the INT command archive. The file is found in:
(i) make sure there is a file header.fit exists. If not
touch it and make it world writable. This will ensure the
protection is always OK. An alternative would be to
get PixCel to point to /home/intobs/header.fit OR
you could use
ln -s /home/cirsi/header.fit /home/intobs/header.fit
BUT you might need to change CIRSIgetfits to write
header.fits to the correct locations.
Test this all as follows:
SYS> archive [and watch the Talker to see if there is any errors. It
should take around 2 seconds. Sometimes we see the error
DMA_time out ... read cryostat ... [This probably means it is looking for
some WFC stuff]
SYS> CIRSIgetfits 0 0 0 0
should return quickly with
- Check the path where the data is written on guestip0. Change using
'send path ' (and put quotes round the path) if necessary. This
path will probably be to guestip2 (cirds), a linux machine, hence
samba must be running there.
- (OPTIONAL) Now try CIRSIgo to take an exposure.
- Print out some logsheets (found in /home/cirsi/public_html/logsheets)
3. Typical Startup Procedures (Telescope)
** NOTE: you only get a TO for the first *2* nights of the run!! MAKE
SURE YOU KNOW HOW TO OPERATE THE TELESCOPE!!!!! get him to show you
early on. You really need to know how to open up and shut down; notes
are given in the CIRSI green folder which should be lying around
somewhere (or more likely in RGM's office at the IoA). **
Full details of the procedures are given on the TO's web page. Read
carefully... but the general order is (skeleton outline):
DURING THE AFTERNOON/DAYLIGHT:
i) Fill the CIRSI dewar and note the time in the observing log on the
ii) Check the vacuum pressure via the HV instrument on the CIRSI
instrument 'cage'. Note this and the temperatures displayed on
guestip1 pcANYWHERE and note in the log AND the web file found in
/home/cirsi/public_html/CirsiDewarStatus/). Channel 5 gives the
temperature inside the dewar, and should be around -184.
It's a good idea to fill the dewar again now as well.
i) Turn off all the stairwell lights and ensure the blinds are pulled
down. Close all the doors to the dome and turn out ALL the lights in
the corridors on the observing floor, as some light leakage can occur.
ii) If you have a TO, he/she will do most of the following, if not,
it's down to you!
iii) Turn on the dome lights if not on (on the TCS console)
iv) ** Turn on the oil pump (on engineering rack) **
v) Move the telescope to zenith (either use 'park zenith' in computer
mode or do it by hand in engineering mode).
vi) Zeroset the telescope. The HA position should be near 12 O'clock
position, and the DEC near 9:50.
vii) On the TCS console type 'zeroset HA target' and 'zeroset DEC
viii) Use HA and DEC [+/-] buttons and in 'SM' mode move the telescope
through the zeroset positions stated above. the TCS will beep and the
zeroset light flash when this is done. The telescope is now zeroset.
ix) Open the dome. Use RAISE MICRO until the OVERTRAVEL indicator goes
off, then RAISE MAIN until open. Repeat for the lower shutter if you
need to use it (only for objects lower then 57 degrees zenith
x) Open the primary mirror cover using the engineering rack.
xi) Turn on the dome vent fan if you want to use it ** NOTE don't do
this in humid conditions, or you'll bring the humidity inside the
xii) Turn off the dome lights and pull the blind down.
xiii) Put the telescope in COMP mode by turning the key switch on the
TCS console. Both SM buttons must be lit for this to happen. The
telescope is now under computer control.
ixv) Ensure that none of the manual control buttons are lit on the TCS
console, or likely the telescope won't track or the dome won't move or
something like that.
IF YOU HAVE PROBLEMS WITH THE TELESCOPE: Call the Duty Engineer (name
& number on whiteboard, or use the intercom. Don't worry about
dragging them out of bed if you have a problem). If you have concerns
about the weather and whether to open, try the WHT TO.
4. LOADING CATALOGUES
The INT catalogues are in /int/cat which is writable by 'intobs'. So
if you prepare a catlogue as user 'cirsi' then you can only put it in
the write place as 'intobs'.
To load a catalogue type
SYS> cmd TCS USER "include sa57_tcs.dat"
There is also a script which will load some default catalogues you
might use every night. It is run by:
and it will load a list of specific catalogues, which you can
update. These catalogues will also show you the correct format, which
is important or the catalogues will not be loaded. Update this script
as you add more object catalogues. NOTE: Catalogue entries are
The following apertures are defined:
- Aperture 0 - Centre of CIRSI, i.e. not on any chip!
- Aperture 1 - Centre of chip 1
- Aperture 2 - Centre of chip 2 etc.
- Some default apertures are loaded on startup; if you change them,
and update this file as you change them at the start of a
run or night.
6. Checking the pointing
(i) 'gocat' to a very bright star from the system bright star
There is a copy of this on the CIRSI ING WWW homepage
The telescope will move there, and place the star on aperture 0.
(ii) type 'aperture 1' (moves the telescope so that the star is
centered on chip 1.
(iii) type 'send chip 1' [to select chip 1 only to be read out]
take a 10sec 1 loop exposure (use 'send loops 1' and 'send exptime 10')
(iv) measure the x, y cordinates of this star with imexamine
WRITE THE NUMBERS ON THE LOGSHEET AS A PERMANENT RECORD
(e.g. 384, 524)
(v) X-offset: Subtract 512 from X and multiple by 0.457" => -58
Add this to the existing Aperture 0 offset.
(vi) Y-offset: Subtract Y from 512 [YES, THE SENSE IS DIFFERENT]
Add this to the existing Aperture 0 offset.
(vii) type 'cmd TCS USER "ENTER APERTURE 0 x y" where x & y are the new
aperture 0 values.
(viii) ** VERY VERY IMPORTANT **
(i) 'gocat' back to the target.
(ii) 'aperture 1'
(iii) Repeat the exposure and CHECK that the star is now centred.
(iv) If it isn't: You've made a mistake (easy to do...) OR the
above senses are wrong (they are from May 2000)
The boring command in section (vii) could be put into a
script [rgm 20000614]
(ix) NOW edit the CIRSILoadApertures script to have the
new value of aperture 0. This will mean that you can quickly
reload it later in the night OR the next night. Spending a minute
doing this now will saave you pain later.
This can be quite time-consuming...
There are two approaches, either focusing on a standard star (which
appears not to be as reliable, as you only have one measurement per
focus value), or focusing on a field (good in more crowded
fields).. Either way, use the following approach:
i) Use 'send chip 1', 'send loops 1' and 'send exptime 10'
i) type 'focus xxxx' where xxxx is the just below the lower end of
where you wish to start the focus test from. [note the command will
not execute if value is the actual current value]
ii) type 'CIRSIfocus value_start value_step nsteps' (a script)
e.g. CIRSIfocus 48.5 0.1 7
USE 0.2 AT THE START OF A RUN OR AT THE START OF THE NIGHT
NOTE: 'value_start' cannot be the same as the current focus value, or
the script will stick (a TCS problem we believe). Also watch for
sticking, and use to break out of script if necessary.
The frames will be taken, with an offset between the first and
subsequent frames. This allows the first frame to be subtracted from
the others to remove sky. Use 'imexamine' in IRAF to determine the
FWHM. As a guide, the focus value in Hx May 2000 was around 49.2mm, and
the focus should not change drastically during a run as the focus is
NOTE: THE J FILTER OFFSETS HAVE NOT BEEN DETERMINED RELIABLEY YET
A list of suitable standards is linked from the CIRSI web page, and is
also one of the default catalogues loaded in when using CIRSILoadCats
- these are from Persson et al. (1998). Select a suitable standard
from there (they all have numbers like 9148). If you choose a Persson
standard, the script 'CIRSIstandard' will do the observation. Usage
CIRSIstandard StandardNumber RunNumber
<> values are optional. The script will place the standard on aperture
1, with default observation parameters of 'loops' = 3 (or the value of
'loops' on the command line) and 'exptime' = 10s (or the 'ExpTime'
value on the command line). It's a good idea to defocus slightly
beforehand. CIRSIstandard uses a simple 4 x 100" dither script to take
the observations, which places the standard on each quadrant of chip
one. If the run crashes, use CIRSIstandard again, but add the dither
position of the crash.
Some standards from May/June 2000 run were:
LCIRS 1441-0031 9155 11.7
LCIRS 1221-0010 9149 11.9
Lockman 1006+4101 9142 11.7
SA57 1214+3535 9148 11.4
ELAIS-N1 1559+4736 9162 11.9
ELAIS-N2 1631+3009 9166 11.5
By hand type
9. PROGRAM objects
Make sure the object is in the catalogue, then:
Example dither patterns can be found in /home/intobs/cirsi/
By far the most efficient way to observe! All the CIRSI observing
scripts are in /home/intobs/cirsi.
If you write a script:
(i) put #!/bin/csh at the top
(ii) chmod gou+x to make it executable
(iii) if you are in a hurry source it
Here are some examples. The most commonly typed ones are marked with
'**', the others are usually called by other scripts. Arguments:
** CIRSIdither (dither position)
The general observing script. Example patterns can be found in
/home/intobs/cirsi, the offsets given are in arcseconds. The script
will start observing at run number RunNumber. It will also start at an
arbitary dither position (useful if the script crashes halfway
** CIRSIstandard (loops) (ExpTime) (DitherPosition)
Described in the 'Standards' section. Performs a 4x100"
dithered observation of standard star 'StandardNumber' with 'loops'
loops and 'ExpTime' seconds starting at run 'RunNumber'. 'loops' and
'ExpTime' default to 3 and 10s respectively. The standard must be
listed in the Persson et al (1998) paper, and this catalogue must have
been previously loaded (using CIRSILoadCats for example).
Starts an exposure using the current settings as displayed in
** CIRSIrun (also CIRSIdark )
Runs CIRSIgo n times
Generates a fits header from
2, Current offset
3, Local text file
Best not to edit it unless you know what you're doing.
Get base position of current object from tcs header
Gets current telescope position from tcs.1
Loads in and defines the apertures. Add or edit as necessary,
Loads in catalogues. Add your own catalogues in here.
(and many others...)
11. The Filter Wheel
Controlled remotely from guestip1 (cirob) (via pcANYWHERE to
guestip0). Obvious to use; select filter and click 'move'. BUT: always
exit cleanly if you reboot guestip0 (circc), or the software 'forgets'
where it is. If this happens, click 'setup' and move it to reference
The filter namining convention is:
Pixcel Filter Gui
0 Dk Blank
1 Jc Z [New Carnegie replacing Z]
2 Jg J J(Gemini consortium specification)
3 Hs Hshort H short side
4 Hx H H(Geminini consortium spec)
5 Hl Hlong H long side
6 Jn JOH narrow band in OH gap in J
7 Hn HOH " " " "
the names are hardwired in the filter gui.
12. OBSERVING TIPS and MISCELLANEOUS STUFF
1. Sometimes Pixcel crashes OR the observing script hangs (the frequency of
which varies from night to night). We think this happens most frequently when:
- The script attempts to get the TCS information (at the beginning of
an exposure). Error messages appear in the 'Talker' window.
- On moving the telescope during a dither, sometimes the script never
returns from issuing the command to the TCS, though the move is
- On changing focus during a focus run, control is often not returned,
especially if you try to move to the current position.
THERE IS NO AUDIO WARNING OF A CRASH, so stay awake :).
RECOVERY: The best thing is to crtl-c in the whiteish pink
xterm. Restart PixCel, and check that the observing parameters are
correct.If part way through a dither, gocat back to the object (to
return to offset 0 0) and then use:
to restart the dither at the crashed pointing (in this case number 4).
2. To avoid crashes, after each dither exit and restart PixCel while
you are moving the telescope (actually, I've no evidence that this
3. Keep a good log - although header files with TCS info are supposed
to be written, don't rely on this! (it didn't work for some of the
dithers this evening for example.)
4. Check that the data looks sensible from time to time by ftp-ing it
to lpss8. Also check the pointing using your finding chart. **
Especially monitor the sky brightness level ** (see later section).
5. PixCel occassionally drops the last few loops in an exposure. It's
always obvious when this happens in the PixCel display; the positions
of the quadrants of thw chips appear in a different place in the
readout display. PixCel may also hang. We think this is due to excess
network traffic, so when exposing keep the network traffic to the
computer that the data is written to to a minimum.
6. You MUST close the dome if humidity rises above 90%, and it can't
be reopened until it drops below 80%. There is an alarm on the MET
console to warn you of this, ensure that it is switched on. The limits
of other weather conditions (e.g. wind, dewpoint) are given on the
white board in the observing control room.
7. If you make a lot of changes to the settings in PixCel (e.g. loops,
exposure time, and especially read rates) then EXIT and restart. This
will change the default settings, which is useful because if PixCel
crashes it 'forgets' any changes that you have made since the last
exit'. It's very easy to implement the wrong observing strategy if you
don't do this.
8. If you need dome lights during the night (ie to see if the
telescope and dome shutter are aligned use the dimmer lights, not the
main lights, or you may be unpopular with other telescopes. These are
located near the terminal area door.
9. To make an arbitary offset, use 'offset arc x y' where 'x' & 'y'
are in arcseconds.
13. Calibration Images
It's a good idea to take every day:
- Dark frames. Take with the 'Blank' filter position! Also take with
same number of loops and exposure times as program objects.
- Dome Flats. Important to get to avoid fringing effects. We
stol.... borrowed an anglepoise lamp from the library as the dome
lights are too bright (and red). Also do dome darks (ie turn off the
- Linearity tests. These appear essential. May 2000: very strange
effects above 12000 counts (in 428 kb/s, gain~10 read out mode). To do
these, put the dome flat light on (anglepoise lamp) and point the
telescope away (eg axis park). Adjust the lamp so that the counts in
30s are around what your lowest sky level is in your shortest
exposures (probably 10s standards). Then adjust the exposure time
(reading all 4 chips) so that the count levels increase by ~10% in
each expsure, but do a reference one in between. E.G. 30s, 34s, 30s,
38s, 30s, 42s, 30s, 46s worked well for us; though levels may
vary. The 30s expoures will account for variation in light levels in
the dome. SEE RGM or CS for further advice.
14. The Night Sky Brightness & Readout Rates
Nominal values in H would be around 14 mag/arcsec^2 or 50,000 electrons
per pixel in 30seconds.
However, in May
2000 it was seen that the night sky brightness can vary a lot. This
can lead to saturation occuring, so it is important to check images as
they are taken to ensure that this is not happening.
If the sky goes above 10000 counts in 30secs you need to shorten the
exposure time. Change the read rate as described below.
The minimum exposure times for CIRSI is set by the readout time plus
the write to disc time. The default observing mode was 428kb/s with a
gain of ~10. The minimum expsoure time then was 30s. However, part way
through the run, images became saturated, and the readout time was
changed to 750 Kb/s, gain ~18. Times of 22s were then used. The images
were never saturated under these values, but there were light
gradients seen in the images. (We haven't reduced these yet to assess
how bad they are).
If you do change readout rates, remember to observe standards with the
new rates, and also perform linearity checks (or at least plenty of
dome flats) and lots of darks.
15. GAIA Finding Charts
GAIA is a starlink program which can be used to produce CIRSI finding
charts. To get starlinkprograms to work, use 'source /star/etc/login'
and 'source /star/etc/cshrc'
Gaia is run using 'gaia &'
Download a FITS image of your area with the width and height of the
full moasic (~30') from the ESO DSS. This is done on GAIA using the
'data server' then 'image server' menu options.
A perl script 'cirsi_grid.pl' in /home/cirsi/ms will produce the
correct overlay files for use in GAIA, as well as suitable TCS
coordinates for the observing.
Prepare a catalogue with the centres of our 30'x30' mosaics in the form:
RA DEC name (must be J2000 currently)
00:00:00 00:00:00 test (must have colons)
then use cirsi_grid.pl
Files will be created of the form name.gaia (ie test.gaia for the
example above) which will have the 16 positions of the chips. load
this into GAIA using 'data servers' and then 'local catalogues' to see
The file will have a catalogue suitable for use with
the INT TCS system. Place this in /int/cat (must be intobs to do this)
and load it in (see above or CIRSILoadCats for an example of how to do
this). There are 4 positions per mosaic in the file.
16. Catalogue format
ObjectName hh mm ss.s dd mm ss.s A|Byyyy|Jyyyy [pR pD [epoch [Px [RV]]]]
name: simple string, no spaces allowed.
RA: int hours, int minutues, real seconds.
Dec: int degrees, int minutes, real seconds
Equinox: A for apparent, B1950 for Besselian or J2000 for Julian
Proper-motions: real seconds of time/year, real seconds arc/year
epoch: real year, epoch of proper motions
parallax: seconds of arc
radial velocity: km/s
proper motions: 0
radial velocity: 0
object NGC3628 11 20 8 13 36 30 J1997.5 ! comment
Comments are preceeded by !
type 'shutdownobsys' in the xterm
control c is safe
20. Annotated sample pixel.ini file
Most of the parameters in the pixel.ini file are not changed
! path for FITs data
21. Writing tapes
CIRSI uses DDS-3 DAT tapes. We write the data using tar.
Write 1 D-tape and 1 C-tape. You should verify that the
tapes are OK but reading the tape back on a different tape
cd to the directory above the data; then type
tar -cvf /dev/rmt/0n 200000613 to write the tape
tar -xvf /dev/rmt/1n to read it back [do this on a different drive]
If you want to write two or three nights of data on the same tape:
mt -f /dev/rmt/ fsf n where n=1, 2 3 etc is the number of nights
of data you want to skip.
BE VERY VERY CAREFUL WHEN WRITING ONTO A TAPE THAT ALREADY HAS
DATA ON IT.
For each night of INT data. A copy for the ING Archive should
be made using the Observatory provided DDS-3 tape. This is called
a D-tape. The observatory on labels the DDS-1 tapes you need to steal the
number and label from the DDS-1 D tape. Fill in the tape sheet to
prevent the D-tape number being duplicated.
Also make a C-tape for IOA. Use an ING DDS-3 DAT for this. If you run out
of ING tapes ask for more using the fault database. As a fall back use
some of the IOA tapes which are in the box on the Linux PC trolley.
Leave the D-tape in the tape tray on in the INT control room. This
is on the shelf above the music system.
If there is a collaborator who needs a C-tape, they shoudl make one
Bring the C-tapes back to IOA.
It is important to verify the tapes by swapping them in the
two DDS-3 DAT drives. Then use tar -xvf or -tvf to verify the contents.
-xvf is prefered but -tvf may be OK.. Dump it to a tmp directory.
Since the ING network is isolated from the CIRSI network (we hope) by
a ethernet switch it is safe to write data from /obsdata which are on
lpss4 to the DATs that are mounted on lpss8.