Before observing check the weather at Narrabri. Try these weather stations:

If you are on-site at SUSI check:
  • Is there anything physically in the way of the starlight or the delay cart (tools/telescope/bridge)?
  • Are all the lights out?

For bugshooting (or something other than basic operation), see the bugshooting page .

This assumes that you're observing on a laptop. If going via ROCS, see the bits in [] square brackets.


Select a north-south pair of siderostats from: N4,N3,N1,S1,S2,S3,S4. To check if they are usable see the SUSI baselines page.

[Log-in to rocs (or your laptop!). Best to avoid using a NX client on rocs as it will cause X windows on rocs to crash. No idea why.]

If observing off-site and away from rocs (e.g. on your laptop), run forward_susi_ports.

Start the software with susi_start or the appropriate icon you have made before.Check that there are no errors, and click REOPEN on any clients that seem to not be connected, and press ESC and read any error messages that come up.

susi_start is just a script that loads:
  1. Servers: acqlabjack, ldccon, pavo.
  2. Non-observing GUIs: k108gtk, skymongtk, roofgtk, envirogtk, vidmongtk (north and south)
  3. Observing GUIs: metrologygtk, picogtk, shuttergtk, filtergtk.

Visually check that the sids you intend to use are clear to go:
  • In VIDMONGTK N do 'TV ON' and 'CAM ON'. Select N#_CAM and visually check the siderostats that there is no obvious obstruction.
  • Do the same for South.

Open the SUSI PICOGTK, and click PING to check that it can PING the server on peleas. No idea why this fails... if it fails, see the bugshooting page .

Check the following:
  1. SKYMON (if you want to see the sky, click TVOFF then TVON). SKYMON may have already been running in the background, and if so the server won't start. If so, the server can be turned off by typing speak skymon into a peleas window (sp to get to peleas) and then quit from a terminal. Having turned off the server in background, SKYMON can be restarted from the SERVERS window or by xrun_skymon.
  2. OPCON Check that the "North Laser" and "South Laser" have 3 OKs. Otherwise metrology laser may need resetting, by turning N_MET or S_MET on as appropriate in the K108 gui, then turning it off again after ~3 seconds... then wait for it to stabilise.
  3. SUSI Metrology. The top trace is North and the bottom trace is South. Make sure that the metrology signal is good. If not:
    • open Picomotor GUI (PICOGTK) and carefully align the I (image) then P (pupil) x and y axes of the metrology...after each system is stable. You do this by selecting the mirror to move, then click "MOVE" to bring up a window that enables the mirrors to be moved (with buttons or keyboard arrow keys). Don't click the buttons too fast (about once per second or two maximum) - if you have to move a long way increase the step size with "BIGGER".
    • Next, adjust it perfectly by typing in "zfine". After several seconds, it will come back with "Fine Zeroset Completed". If this doesn't work, you have to do a "zcoarse". See the bugshooting page.
    • Afterwards, type in "cen" so that the "OPD" is set to zero (not -146 or some other large non-zero number!!!). NB the OPD in the display is rounded to the nearest 9.44mm.
  4. OPCON. Check that Zeroset is "0" and click "GO ZERO" if the OPD Fine is not near zero (i.e. if the cart is far (~ several dozens of cms or more) from the visual zero position on the tunnel vidmon). Then click ZFIND. Hopefully, then Zeroset is found. If not, see the troubleshooting page.
  5. SIDCONs: It means if you want to observe with, say the N3 and S1 siderostats, you need to start the server for each siderostat (conveniently named as N? and S?, in this case N3 and S1) from the SERVERS window. The text command is "xrun_sidcon n1" etc. This is not started with susi_start because it depends on which siderostats you are using. Also open the GUIs with sidcongtk s1 and sidcongtk n1.
  6. LDC: Check that crown wedge is realistic (i.e. not ???), status is ok and air offset is -1 (or type aoffset -1 to set this). If there is a problem, index the crown wedge in the "Wedge Control Menu".
  7. PAVO: Firstly, ensure that the cooler is on in the server window. The detector temperature should be about -45 degrees in the PAVO window. If it isn't on, try coolon -45 into the PAVO server window.
  8. Only If not using Taskmaster (i.e. only for expert mode): You have to tell PAVO which siderostats you are using. To do so, type in "nsid n#" and "ssid s#" replacing "#" by the number you're actually using. Finally in ACQLABJACK: type in 'pvopen' to open its connection with PAVO Make sure the AV68Ks (SIDCON N, SIDCON S and OPCON) are on "run" or not in "Standby" mode.

All the servers should now be ready and communicating with each other. Check that the following GTK clients are running - if not open them from the SUSI GTK Clients window.
  1. K108GTK
  3. ROOFS controller to open roofs at N# (say, N3) and S# (say, S1)
  4. SIDCON GTKs (N# or S#)
  8. TASKMASTER (cannot be run in the background, reason unclear)

ALIGNING (For MUSCA observation only)

This alignment is critical for MUSCA, but does help with PAVO too if you are on-site and about to do a standard laser alignment (especially step #1-4). If there was a recent alignment and you are only using PAVO you can skip this whole section.

  1. Turn on the IR LED or the white-light (WL) source (which is brighter and easier). IR LED can be turned on from SHUTTERGTK by un-clicking the LED and clicking the IR buttons. WL can be turned on from K108GTK by clicking the WHITE_LIGHT button.
  2. Align the pupil on PAVO using "PUP". You will need to have a PAVO integration time of ~50 ms and gain 150, with a dispavg ("Average" on the GUI) of 1. You want the "XYP" numbers to be 0.0 0.0 and thousands on the PAVO server. (NB: On 24 Oct 2010, it was pretty clear that the XYP number wasn't centroiding the stripes in an x-y direction properly... so we aligned for even illumination by eye).
  3. If on-site, do the standard alignment here as written up in the ATNF folder.
  4. Start the PAVO server going with a gain of 0. Turn on the laser with ND3, and open the NTT and STT shutters. Move the NTT and STT mirrors with the picogtk in order to center the flux.
  5. Maximize the flux in the pupil (e.g. looking at the WL source) by moving the "N-MAC" and "S-MAC" mirrors one at a time. When moving the "NAC" mirror, ensure that the south shutter is closed, and vice versa. This adjustment seems to be very stable, rarely needing movement.
  6. Use the NAC-Z mirror to maximize the PAVO fringes and make them as straight as possible. A scale 0 1000 in PAVO server can be helpful.
  7. Switch on the CCD camera from K108GTK and launch its software control, SBIGGTK. Click "CAM ON".
  8. Launch also SLIDEGTK, click the "CAMERA" button to slide it the penta-prism in MUSCA into the main optical path and use it SBIGGTK to align:
    • the pupil of the light source (WL, lasers or LEDs from the siderostat by adjusting "MBS" in PICOGTK.
    • the image of the light source (by clicking the "IMAGE" button in SLIDEGTK first and then) by adjusting "N-MUS" in PICOGTK.
  9. Aligning the WL source requires about 8000ms of exposure time and about 40ms for the LED with the SBIG camera. If the alignment of WL was carried out a day earlier, just align the LEDs from the siderostats (see step #10 onwards).
  10. If the positions of the North and South laser images and pupils are aligned (overlapping each other) to within a few pixels you are done. Otherwise continue with #8.
  11. Launch the server for the siderostats which will be used during observation and point the mirror to the opposite periscope intended to be used. For example, if N4 north periscope is to be used, point the siderostat to azimuth=180 and elevation=0.
  12. Adjust S14 from PICOGTK to maximize the IR flux in PAVO (look at the 3rd number besides the XYP label, first mentioned in step #2). Mark south pupil position in SBIG camera.
  13. Align the north LED pupil to overlap with the south on the SBIG camera by adjusting N14 from PICOGTK.
  14. Launch APGTK and slide in the aperture for the north beam. With the apertures in place, adjust the N14 to maximize the IR flux in PAVO (refer step #12). Slide out the aperture from the beam and repeat step #13 until both north and south pupils are overlapping each other.


This is something do do if you aren't confident of alignment to the siderostats. Unfortunately, there isn't much you can do if alignment is poor...
  1. Put the siderostat in an auto-collimation position, or near-auto-collimation position. This is most easily done by typing "elabs 0" and "azabs 180" into the siderostat server window.
  2. Put the periscope mirror opposite to the autocolimation position you are in. e.g. if you did "azabs 180", then put in the north periscope with a "nin".
  3. View the LED on the acquisition system and on PAVO. It should be very out of focus on acquisition especially, but should be a round circle centered on the fiducial position.


  1. Make sure the alignment laser is off, or with at least ND3 in it. The laser on/off is part of the K108GTK, and the filter wheel is part of FILTER. Make sure with the help of SKYMON that the sky is clear and that no clouds can be seen. Make sure that the "Relative Humidity" (RH) is less than 100% in the SUSI Environment Monitor window. If it doesn't work, make sure the ATCA weather station prints a RH < 90 %.
  2. Turn off DEHUMID in K108GTK.
  3. Open the roofs. Make sure to keep an eye on the weather in SKYMON; clouds aren't good but rain definitely is something to avoid! Now with the help of VIDMON and ROOF GTK, open the roofs of the siderostats you're using.
  4. Using PAVOGTK, open the display ("DISPLAY" button) and the waterfal diagram ("PLOTS" button). In Display: Order is Combined (fringes)-North-South. You then have to set the gain for the pavo camera. For bright stars, a gain of 150 is good, for 1-4 mag stars, it should be at ~170, and for fainter stars try 190, or 210 with photon counting. Higher gains age the camera if they are used at ~>10 photons/pix. Click on GO/STOP to start the camera and click BACK to take a background measurement for some time (~10 secs is enough). Then turn the background acquisition off by toggling BACK.
  5. On TASKMASTER, type the following commands:
    • sid n? s? ?
    • reopen, to make all connections
    • pin, to put the periscopes in
    • fs -1000 1000 6, eg. to search from -1000um to 1000um at a rate of 6um/s
    • hr 1234, e.g. to observe HR 1234.
  6. PLC Tell OPCON to start tracking the star, either by going in the submenu "4 Tracking menu" then clicking on "1" or by typing the command "start_tracking" (the opposite being "stop_tracking").
  7. SIDCON Start tracking the star (click "Track"). Make sure no other periscopes are in the vacuum pipe and blocking the light. This can be seen by looking if any other green lights are on on the control room panel (these can be seen from the Metrol camera if not in Narrabri - start this on NX to peleas).
  8. ACQLABJACK With the client, turn the camera ON and move the pellicle (to north, then south). Once the siderostats are tracking the star you should normally see the star (make sure the intensifier is on "2/3"). Then click "Track ON" and if everything is aligned and working properly the star light should make it to PAVO. On the PAVO server, look for the North and South servo indicator to turn ON. You can then proceed to the south beam or park the pellicle.
  9. LDC Do the same with LDC by typing "track" (stop by typing "stop").
  10. For the first bright star of the night, you need to adjust dispersion compensation by servoing. With good fringes, type "tg" into PAVO server. This switches on the LDC servoing - wait until "Disp" in PAVO settles down at zero and then switch off servoing by typing "tg" again. Then type "dispersion" to set this as zero.
  11. Turn on autosave by typing "autosave on" in the PAVO server window - this will save data automatically (100 seconds of fringes) when good S/N is detected. To adjust the S/N thresholds to, say, switch on at SN of 2.0 and off at SN of 1.5, type "snon 2" and "snoff 1.5" respectively. Switch of autosave by typing "autosave off".
  12. Alternatively, saving can be initiated manually by pushing the "Save" button in the PAVO GUI.
  13. If for some reason things fail and you want to acquire manually, then...

Now that the two (south & north) images can be seen in the left and middle target in the display, you can do a fringe search with the "Start" and "Rate" entries on the PAVOGTK. Typical value are -200 for the starting offset and 10 for the rate. If all goes right, you should get some fringes and PAVO should take a dataset (if autosave is on - otherwise click the "Save" button in the PAVO GUI) and initiate the shutter sequence afterwards. If you haven't been able to get a dataset, cry and go read the 50 pages PAVO manual. Once you're done with that star, repeat the process (stop both the SIDCONs and OPCON from tracking before) at step 3 or close everything up.


If PAVO is the only beam combiner in use, you can do the following two steps to tweak-up the alignment. This is necessary if an on-site alignment hasn't been done in a while.
  1. Close the shutters one at a time. Click "FLUX" on the PAVO GUI. For each beam adjust the NTT and STT mirrors respectively to maximise the flux shown as the green line.
  2. Again, close the shutters one at a time. Use the mirror N15 and S15 to center the pupil in the PAVO fringe box. This should only have to move a few 10s of steps. At the time of writing, I *think* that left/right and up/down were reversed.
  3. If the images are horrible, you can further optimise by tweaking focus. Use the "zaber" server, and the menu items like "Bump NFOCUS In". The values of the stage positions at 27 April 2012 for good focus were 227608 for North and 131327 for South. The South stage has an adjustment range of 0 to 130000, so this is a little suspect. The North stage has an adjustment range of 0 to 260000. One step is 0.1 microns in separation between a +2000 and a -2000mm focal length lens, spaced to give a total focal length of 35m.


If Taskmaster is working, you should just have to type park into taskmaster... but still check that everything is OK. When observing is over, the servers can technically be closed in any order, but first make sure that:
  1. The Siderostats are parked (click on "STOW" on both clients).
  2. The roofs are closed. Ensure the siderostat mirrors are pointing down (to stow) before moving roofs. To make sure they are closed, see if the "N/S#_CL" sensor on K108 displays "OFF". Also make sure the DEHUMID are switched back on.
  3. The cameras (PAVO and ACQLABJACK) are switched off and parked. OPCON and LDC are not tracking a star anymore.
  4. The PLC cart is back to its zero position. This can be done by pressing the GO ZERO button. Check that it really is at zero on the camera, and do a ZFIND to double-check.
  • This is optional. In a terminal on peleas, type speak peleas 3055, and then tvoff, and then tvon, to send the SKYMON images to an intranet webpage for display.

Now complete the end of night logs: fill in the SUSI Observation Report at the log page. Previous logs can be found at elogs.

Set up for a MAC

Firstly - you'll want to set up an easy way to bring up peleas windows. e.g. in your .cshrc file:

alias sp 'ssh -tX ssh -X observer@peleas'
alias cvs_tunnel 'ssh -N -L3001:peleas:22 &'
setenv CVSROOT :ext:peleas_cvs:/usr/local/susi/cvsroot
setenv CVS_RSH ssh
setenv PATH /usr/local/susi/bin:$PATH

To get this fancy CVS thing to work, you also need to set up the "peleas_cvs" fake host in a file .ssh/config:

host peleas_cvs
hostname localhost
port 3001
user observer (only if you do not have personal network account with SUSI)

An example file to go in /usr/local/susi is .
Next - you need to set up the /usr/local/susi/ directory structure, and in /usr/local/susi/bin, add and add in /usr/local/susi/etc. Run /usr/local/susi/bin/forward_susi_ports.

Then when you manage to compile the gtk clients, they will connect to local sockets on your machine, then this will be forwarded to peleas.

To update software:
run cvs update in the cliserv directory
run cvs update in the src directory
run cvs update in the lib directory
Do make and then make install in each of these directories.