ATCA Current Issues

Unless you have been trained to do this:

Do Not Reprogram the Correlator while in Zoom Configurations

See here for information.

ATCA schedule revised

Version 3 of the ATCA schedule was released on 18-dec-2012.

CAOBS Start command

The syntax of the start command has been extended to allow looping over a range of scans. E.g., the command:
caobs> start 3-7/9
will loop over scan 3 to 7 nine times, ignoring scans before 3 and after 7. This feature allows you to keep your calibration setup scans in the same file as the observation scans.

(04-Mar-2013)

4cm (CX) Band Focus

The 4cm ("C/X") receivers on the ATCA are in the process of being upgraded to a single 4cm band, covering 4GHz to 10.8GHz, with the limitation at the high-frequency end imposed by the feed.

New prototype feeds have been installed on CA03 and CA05 to extend the frequency coverage up to 12 GHz. (With further modifications to CABB, it may be possible in the future to extend the ATCA frequency coverage up to 12.3 GHz.) The new feed, however, has a different focus position to the original feed, and so it is now necessary to change the focus position when moving to or from 4cm band observations.

Best Focus Positions are:

 4cm FocusOther Focus
CA03-4.710
CA057.818

(Focus Command is:
caobs> focus ca0# focus
e.g.
caobs> focus ca03 -4.7)

(19-Dec-2012)

Top of Page

CABB modes

The CABB web page has been updated with information on CABB modes now available. Briefly, the 64MHz mode now has up to 16 zoom bands in each IF band. There is also a 1MHz/64MHz "hybrid" mode available, with 2048 x 1-MHz (but no zooms) in one IF, and a single 64-MHz zoom channel with 2048 zoom channels across it in the other IF.

(02-Nov-2012)

Top of Page

Receiver Attenuators

The attenuators that are set with the:
cabb> atts on
command are attenuators within the CABB system

There are additional attenuators on the mm and 16cm receivers (and will be on the new 4cm receivers)

These can be set with the caobs command:
caobs> set atten ca0# a b
where
# is antenna number
a is the attenuator setting (in the range of 0-15) for the a polarisation
b is the attenuator setting (in the range of 0-15) for the b polarisation

caobs applies the settings to the receiver that is currently being used, so some care needs to be taken to set the frequency before setting the receiver attenuators.

These settings will be remembered if you go to a different frequency, then come back.

To check the current settings use:
caobs> show atten

(17-Feb-2012)

Top of Page

High Tsys on CA06 3cm-A pol

The system temperature on the A-pol of the 3cm receiver on CA06 is about twice that of other receivers (and the B-pol on CA06) In the short term, there probably isn't much that can be done - Investigations are on-going.

(18-Nov-2011)

Top of Page

ATCA Forum

The ATCA Forum is a forum site dedicated to talking about the problems and successes of ATCA users. It was originally started as an externally-hosted website, but is now run entirely within ATNF. Please visit it at:

http://atcaforum.atnf.csiro.au
Top of Page

New calibrator database

The new ATCA Calibrator Database is now searched by default from the calibrator query page. The new database will be kept up-to-date as more C007 observations are carried out, and those of several other projects that have opted to share their calibrated fluxes. The old database can still be searched from the same query page. Reports of any problems with the new calibrator database, or suggestions for features should be sent to observatory staff.

Top of Page

ATCA Users Guide Updated

The ATCA Users Guide was updated on 2011-04-27. New and updated:

  • Chapter 3: Observing has been almost entirely rewritten and reorganised to assist users who are observing with the ATCA.
  • Clarification of polarisation calibration requirements: for most observations, normal observation of the gain calibrator will provide excellent polarisation calibration.
  • Status of CABB zoom configurations updated.
  • More detail on how to avoid problematic frequency configurations, and more information about ATCA performance figures.
  • New web scheduler actions and fields documented, along with new CACOR commands (eg. czabs) and CAOBS capabilities (eg. increased limit of 2048 mosaic fields per file!).

CABB Using CACCC1

The current correlator control computer is CACCC1.

Top of Page

Do Not Reprogram the Correlator while in Zoom Configurations

While commissioning CABB zoom mode features, correlator reprogramming is somewhat fragile.

If you are using a correlator config that is set up for zoom modes (e.g. ca_2048_2048_2f_zm16) it is important that you do not reprogram the correlator. Doing this is likely to cause all sorts of problems which can only be fixed by selectively reprogramming certain boards which requires intimate knowledge of what does what in the correlator.

(Do not try changing the correlator configs either: configurations are chosen in the light of the schedule.)

In case of a block failure,

  • boot the block (see here for info on doing this)
  • in corr gui, use the edit button, then
    prog #
    where # is the block in question

Then look at all the 2GHz autocorrelations (i.e. frequencies 1 and 2) and make sure they look OK. If not, try a prog # again. Even this does not guarantee that the zooms will be OK.

Top of Page

VNC Issues

Virtual Network Computing (VNC) is a graphical desktop sharing system that allows a user to remotely control another computer. It transmits the keyboard and mouse events from one computer to another, relaying the graphical screen updates back in the other direction, over a network. (Adapted from the wikipedia entry for VNC)

While VNC generally works well for remote observing, allowing handovers without the previous requirement to quit caobs, and allowing the correlator status to be monitored, from time to time things freeze up.

The <cntl> <c> (simultaneously holding down the control and c keys) may not work within a vnc session if your vncviewer does not handle the <cntl> properly. In this case, use the <alt> in combination with <cntl> character

If you appear to have lost the ability to type in a window, try the following:

  • Check that you are not connected 'view only'
    (I think the only way to do this is to kill the viewer and start again - rmw)
  • Check you are not battling for control of the mouse with another user
  • The (somewhat non-intuitive) sequence of <alt><Left Mouse Button> will often allow control to be regained.
  • Other combinations, such as <alt><enter>, <alt><space bar>, <alt>[some typing], or <alt>[other mouse buttons] may work.
  • If using a Mac, the control keys to move between open spaces may be interpreted by the VNC as a <cntl>s (the ascii no scroll character). Simply trying <cntl>q may be sufficient to regain the ability to type into the VNC window.
    (As a general rule, it is best to move the focus out of the VNC window (by, e.g., clicking the mouse on another part of the screen) before using any special Mac sequences of keys.)
  • Again, on a Mac running Chicken of the VNC, if you find yourself in full-screen mode, use the key combination <cntl><alt><command><`> to recover.
Experience suggests issues are occasionally caused by someone joining the VNC session. If you are monitoring observing that someone else (in your team) is doing, it is recommended to start your VNC session as a "viewonly" session. If you are joining to take over from the previous observer, don't start your session well in advance of your start time in case you cause problems for the preceding observer.

If the windows have lost their header bars, then it may be that the window manager has crashed.

Vince and Jamie have created a script to restart this, which is reached by right clicking in the VNC window manager. Click on "Stop Window Manager", wait a few seconds, then click "Start Window Manager"

If you are completely stuck, you may need to kill the window manager from outside the vnc session.

ssh into xbones:
> ssh atcaobs@xbones

kill the xfwm window manager:
> killall xfwm4

You should now have control of the vnc session, albeit with no window manager running.

Restart the window manager, either from the window manager script (right click in the VNC window, and click "Start Window Manager" or execute the command
>xfwm4 &
in a terminal in the vncsession. The session should now be restored to a normal state

If you find other faults or features, please pass them on.

Top of Page

7mm polarization data (09 Mar 2010)

If you have processed 7mm polarization observations with a miriad version older than 4 Mar 2010, the polarization angles will be off by 90 degrees. You can either flip the sign of Q and U in your final images or reload the data with the latest version of atlod and redo the reduction. [If you loaded 7mm data with the miriad version from 18 Feb 2010, you are advised to reload it with the latest version, as it contained an incorrect fix which made the problem worse.]

Top of Page

Saving tvchannels

tvchan settings in CABB define the channel range over which data is averaged and passed to caobs. This range is also used to generate the Tsys values in the RPFITS file headers.

Thus, if these channels are not changed every time there is a change that would need different channels, the Tsys values will not be correct.

Mark Wieringa has implemented a new field in ATCAsched which can define a channel range. The parameters are four, comma separated, values representing (in order) the first and last tvchan for IF 1, and the first and last tvchan for IF 2. If it is left as blank (or null) then the correlator tvchan settings are kept, otherwise this field will override the correlator settings. These tvchan settings are settable for each scan.

Top of Page

Block errors in CABB

If a section of the SPD amplitude plot has approximately every second channel set to zero, a cabb slot has died and needs attention

To solve,

  • Work out one channel which is bad: The best way to do this is to look at just 20 channels In this case channel 552 (or 553 or 557 or ...)
  • Subtract 1 (i.e. 551)
  • Work out what slot this is you need to grep /home/corr/cor/cordb for this channel: e.g.

    giving slot 7 as the one with the problems
  • If the problem is in the second frequency, add 20 to this slot number (e.g. 27)
  • Try reprogramming this slot (stop obs, go into the cab edit, prog 27)
  • If this doesn't work, reprogram the block (stop observing, use spray, remove and install the block from the correlator)

  • If the problem persists, try a full reprogram

Top of Page

C26 Power Levels

It has become aparent that it is important to work hard to ensure that the CABB sampler values are as close to 20 as possible (i.e. 10 is not close enough). At mm frequencies this generally means that the mm attenuators have to be set to low numbers. It is no longer necessary to ensure that the power levels into the C26 modules are less than 20V.

The following values are given as representative starting values for the mm attenuators

Freqenciesmm Attenuator Levels
17GHz/19GHz3 3
33GHz/35GHz0 0
43GHz/45GHz3 3
93GHz/95GHz0 0

Top of Page

Leap Second Problems (21-Jan-2009)

A leap second added to UTC at UT 00:00 on 1 January.

We had thought we were on top of handling this, however it turns out that the ephemeris handling is more complex than originally recognised. The net result was that there is a 1 second timing error in data observed between UT 00:00 on 1 January and 13th January (up to the reconfig). It has been fixed now.

The error is not substantial, and it might be that it will not be noticeable for low frequency observations. However the effect becomes more significant at higher frequencies.

To correct the data use miriad "uvedit":
uvedit time=-1 vis=in-file out=out-file

This corrects the input data-set "in-file", and produces output "out-file". You should do this before any calibration.

Top of Page

Pointing Offset Between mm and cm Receiver Packages

There is a pointing offset between cm and mm receiver packages.

It is of order 20-30'' and is most obvious on antennas 2 and 3.

When mm observations are scheduled we generally provide a cm and mm pointing model - please contact local staff for current information.

Top of Page

Accidental online selfcal in CAOBS (10-Sep-2008)

In a small number of observations between April 2007 and September 2008, the selfcal function in CAOBS (developed as part of the array's satellite tracking capability) was inadvertently switched on. We have identified these occurrences and notified affected observers. To restore the phases in data with selfcal switched on in CAOBS see this link.

Top of Page

New mosaic mode

There is a new scantype in sched, OTFMOS, which is like MOSAIC, but instead of moving from point to point, the antennas scan continuously across the mosaic points. For more details see On-the-fly Mosaicing Strategy.

CloudSat

Cloudsat is a NASA satellite mission providing observations to advance our understanding of cloud abundance, distribution, structure, and radiative properties. CloudSat has the first satellite-based millimeter-wavelength cloud radar, allowing detection of small particles of liquid water and ice that constitute the large cloud masses that make our weather.

So much for the advertising - It's a nice experiment, but that radar has the potential to destroy our 15-/7-/3-mm receivers.

Operationally, we need to ensure that we do not have the mm receiver on axis, at an elevation of greater than 88 degrees when there is a Cloudsat pass over the ATCA or Mopra. To this end, we have modified the ACC software so we can't park the antennas with the mm receivers on axis and we send emails warning ATNF staff and visitors of Cloudsat passes greater than 85 deg for ATCA and Mopra. For more information see the Satellite Overpass Closest Approach Predictor

If you are observing with the mm receiver, please check that your observations will not risk the receiver and be prepared to do something else for a few minutes if there is a risk of this.

Top of Page

Linux CAOBS

Schedules

Both schedule files and mosaic files are stored on xbones in the following directory paths:

Linux Name Linux Environment Variable
/atca/sched $ATCA_SCHED
/atca/mosaic $ATCA_MOSAIC
Task: print a schedule listing
  • while logged in as atcaobs on xbones
  • print the file with a2ps:
    xbones$ a2ps -Pcacolour -r --columns=1 -l 132 /atca/sched/sched.lis
    or lpr:
    xbones$ lpr -Pcacolour /atca/sched/sched.lis
Task: upload a new mosaic file from your account (abc123) on, e.g., kaputar
  • log in to xbones as atcaobs
  • copy the file across
    atcaobs@xbones$ scp abc123@kaputar:mysource.mos /atca/mosaic/
    Note that filenames are case sensitive on Linux, see below.
Task: copy your sched files for safekeeping
  • log in to, e.g., kaputar
  • copy the file across, e.g. to your schedules directory:
    kaputar$ scp atcaobs@xbones:/atca/sched/yourschedfile.sch schedules/
Top of Page

Known Linux CAOBS Problems (and solutions)

The 'Stop/Start' Problem

The first scan after CAOBS starts is always bad: CACOR will give blue "DELBAT" error messages.

These errors happen on the first scan after CAOBS is re-started, but may sometimes also appear after typing start later in the run. Simply stop the scan and start it again and the errors should go away.

Top of Page

Data disks and computing

The computer ``kaputar'' is the main compute server. To find an area with a large amount of disk space to work within, use /DATA/KAPUTAR_3.

Top of Page

 

Original: Robin Wark (19-Jun-2003)
Modified: Mark Wieringa (20-Jan-2011)