ATNF Compact Array Data Acquisition Problems and Information

See also the ATNF Miriad home page.

27. 29-Oct-2004: Error in Miriad handling of mosaics and primary beam correction (some data since 1-Oct-2003)
26. 14-Mar-2004: Non-standard antenna location coordinate system in RPFITS file (until 11-Apr-1991)
25. 8-Oct-2002: ATCA system clock in error by 1 second (20-Sep-2002 to 1-Oct-2002)
24. 22-May-1998: Baseline-based errors in high dynamic range images (data to mid-2002)
23. 30-Jun-1997: Errors in Stokes V for sources near DEC=-30
22. 28-Oct-1996: Intermittent antenna pointing failure (March 1996 to 1997?)
21. 03-Sep-1996: Possible corrupted coordinates (26-Jun to 23-Jul-1996)
20. 02-Jul-1996: Precision of stored coordinates (until 24-Jun-1996)
19. 17-May-1996: 13-cm off-axis polarimetry
18. 29-Mar-1995: Ghost Images from Gibbs Phenomena in 128 MHz bandwidth mode ATCA data
17. 28-Mar-1995: Dealing with 128MHz birdies in ATCA data
16. 30-Oct-1994: Using B1950 coordinates in the scheduling process (until Oct-1994)
15. 10-AUg-1994: Mixing of sources when using correlator averaging in mosaic mode (21-Dec-1993 to 20-Jul-1994)
14. 04-Aug-1994: Change in the spectrum of ATCA primary calibrator 1934-638
13. 10-May-1994: Potential problem causing ATCA data to not be calibratable (21-Dec-1993 to 12-Apr-1994)
12. 19-Jan-1994: Mislabelled coordinates in AIPS (11-Nov-1993 to 18-Jan-1994)
11. 10-Mar-1993: Sky frequency-correlator channel number relationship changed (1989 to 10-Mar-1993)
10. 07-Oct-1992: Missing system calibration information (24-Sep-1992 to 06-Oct-1992)
09. 01-Sep-1992: Faulty antenna tables (May-1992 to Aug-1992)
08. 24-Mar-1992: Wrong(u,v,w) (04-Mar-1992 to 21-Mar-1992)
07. 22-Aug-1991: Wrong RPFITS headers (17-Aug-1991 to 22-Aug-1991)
06. 01-Aug-1991: Wrong (u,v,w) (1989 to 01-Aug-1991)
05. 24-Oct-1990: Inverted spectra in some bands (1989 to 23-Oct-1990)
04. 02-Aug-1990: Phase negation (1989 to 01-Aug-1990)
03. 23-Jun-1990: Phase error after phase CAL operations (1989 to 22-Jun-1990)
02. 16-Jun-1990: Inverted Spectra (1989 to 15-Jun-1990)
01. 14-May-1990: Incorrect Tsys (1989 to 13-May-1990)


Description: Error in Miriad handling of mosaics and primary beam correction

Type of affected data: Data for mosaicing or where primary beam correction is to be applied

Date of affected data: since 1 October 2003 with ``old'' versions of Miriad atlod

Problem in detail: Since October 2003, the RPFITS files written by the ATCA observing system contain the pointing centre. Prior to 24 October 2004, Miriad atlod contained a bug which potentially confused the pointing centres of different sources. The situation was further complicated by Miriad uvsplit discarding all pointing centre information (both the good and erroneous information) up until 16 August 2004. When pointing centre information is missing, Miriad assumes it is the same as the delay tracking centre.

From the users point of view, the erroneous pointing centre information bug is an issue for data where all four of the following conditions are satisfied:

  • The data was observed after 1 October 2003
  • The data that are to be used for mosiacing or the resultant images are to be primary beam corrected.
  • The version date of Miriad atlod used was earlier than 24-Oct-04.
  • The version date of Miriad uvsplit used was 16-Aug-04 or later.

The version dates of atlod and uvsplit used to process data are given in the history file of each dataset. Additionally the version dates of existing installed versions of uvsplit and atlod are printed out whenever these tasks are executed. Users are encouraged to use the normal Miriad procedure to update their Miriad version if neccessary.

Erroneous pointing centre information will manifest itself at the imaging/mosaicing or primary beam correction step (Miriad tasks invert and linmos). It will usually manifest itself by the image being zeros, as the software believes the pointing centre is well away from the delay tracking centre.

Solution: After updating Miriad to a recent version, there are two possible solutions

  • A UNIX C-shell script, pntfix can be used to patch visibility datasets that may be affected by this bug. The script renames pointing centre information so that it is not recognised by the reduction software. The script can be safely applied to any Miriad dataset. To patch the visibility dataset mydata.uv, use
       % pntfix mydata.uv 

    Multiple datasets can be given on the command line. This script is available in recent versions of Miriad.

  • Re-process data from the RPFITS file. This will be needed in the few instances where the pointing centre differs from the delay tracking centre.

Signed: Bob Sault (29-Oct-2004)


Description: Non-standard antenna location coordinate system in RPFITS file

Type of affected data: All ATCA data

Date of affected data: until 11 April 1991

Problem in detail: Sometime shortly before 11 April 1991 it was realised that the X-Y-Z coordinates used to describe the ATCA station locations were in the "AGD84" system or "Australian Geodetic Datum 1984" system, although they purported to be in WGS84 (World Geodetic System 1984). The original site survey had confused the two systems, giving the coordinates in AGD84 but describing them as WGS84.

Thus in 1991 the tabulated coordinates used by the ATCA were converted to WGS84, using a standard transformation. The coordinates stored in the RPFITS file would have jumped by a few hundred metres at this stage. This change should have had almost no effect on the operation of the instrument, and would have been transparent to most users, but brought the data headers into line with standard international practice of using an earth-centred XYZ reference system such as WGS84 or similar.

The AGD84 system embodies a spheroidal approximation to the local geoid over the Australasian region (the Australian National Spheroid: ANS84) which is not earth-centred. (That is, the point X=Y=Z=0 does not coincide with the geometric centre of the earth, but to a point some hundreds of metres away). It was designed as a convenient reference frame for Australian surveyors and cartographers in the days before GPS, but is rather inconvenient for astronomers and geodesists who require a global reference frame.

The AGD84 system and its close relatives are now obsolete, and the standard Australian system is now GDA2000, which is earth-centred and closely tied to ITRF and its related frames. The advantage of having a localised fit to the geoid over a region of the earth's surface (e.g. AGD84/ANS84) has largely disappeared owing to the advent of GPS

Signed: Bob Sault for John Reynolds (11-Mar-2004)


Description: System clock has 1 second error

Type of affected data: All ATCA data

Date of affected data: 20-Sep-2002 to 1 Oct-2002.

Problem in detail: From 20 September to 1 October, the main clock at the ATCA has had a 1 sec error. This happened when we had a clock-related hardware failure. When the clock was reset after this, it was apparently reset 1 second off.

The net effect on observations is described below. Centimetre observers are largely unaffected - mainly because we were in a compact configuration. Millimetre observers will need to do an extra step in the reduction.


Assuming that you did regular phase calibration, there would be a resultant positional error in RA of the program source after calibration of
   15*(cos(dec1)-cos(dec2)) arcsec
where ``dec1'' and ``dec2'' are the declinations of calibrator and program source.

In the EW367 array (excluding CA06) this error is not particularly significant at 3cm, and quite insignificant at 20cm.

You can correct this problem by using the Miriad task uvedit, which can correct phases for a clock error. Use
   uvedit vis=in.uv out=out.uv time=1
uvedit must be used before calibration, and on both calibrator and program source.


There would have been a 15*cos(dec) arcsec error in the pointing. If you were using reference pointing, the pointing would have had an error of
   15*(cos(dec1)-cos(dec2)) arcsec
in the pointing, where ``dec1'' and ``dec2'' are as before.

Signed: Bob Sault (8-Oct-2002)


Description: Baseline-base errors in high dynamic range images.

Type of affected data: All ATCA data

Date of affected data: All ATCA data until mid-2002.

Problem in detail: The ATCA has non-closing amplitude errors, which you will be able to see by doing a closure amplitude plot of a point-source calibrator. Amplitude closure errors of 0.2-0.5% should be apparent. This problem appears to be related to phase noise in the local oscillator system, in part induced by the round-trip phase measurement hardware. It is worst on baselines involving antenna 6. The errors become noticable in only very high dynamic range images (better than 5000:1), and mainfest themselves as rings around point sources that will not clean out (i.e. classic amplitude errors).

Solution: No software solution exists. The problem can be partially avoided by turning off the round-trip phase measurement hardware during the observation. Note that it is the act of making the measurement (not applying the measurement) that is causing the error. Simply disabling the correction in CAOBS does not help. You must switch off the cycling of the round-trip phase measurement hardware in the screened room. Note by doing this, you will suffer additional phase errors. But these are closing errors, that can be removed with self-calibration. You should not switch off the round-trip phase measurement system if you do not intend to do self-calibration. Discuss this with Narrabri staff before the observation. The MNRF-1997 upgrade to the LO phase distribution system has cured this problem.

Signed: Bob Sault (22-May-1998)


Description: Errors in Stokes-V for sources near DEC=-30.

Type of affected data: All ATCA Stokes V data near DEC=-30

Date of affected data: All data

Problem in detail: When observing near transit for sources near DEC=-30, a time-varying leakage, which is not corrected by standard ATCA polarisation calibration, can corrupt the Stokes-V data. The error is antenna based.

The problem arises in calculating the parallactic angle. In doing so the orientation of the antennas is assumed to be "perfect", e.g. the azimuth axis is at right angles to the Earths surface, and the focus of the antennas is at the centre of the dish. These assumptions are in error by of order 1 arcmin, and the error varies between antennas (e.g. antenna 4 mounted at station 30 is tilted at 2 arcmin to the south). This small error can lead to significant variation from nominal (~1 degree) in the parallactic angle near transit for DEC \approx -30 sources. The result is that the antennas have unequal errors in the computed parallactic angle, which results in a time-varying polarisation leakage. The severity of the problem goes roughly as (DEC+30)**(-2), with the error being more narrowly confined to transit the closer you get to DEC=-30 (indeed for DEC=-30 sources, you may not see it, as you will not be able to track that close to transit).

Solution: The pointing solutions (pparams table) contains the information needed to fix/prevent this. It gives the relationship between the encoders and the sky. Ignoring some terms (the presumed encoder errors) leaves us with parameters which give the deviation of antennas from their nominal orientation.

A task has been developed in Miriad, ``transfix'', to correct the problem. Given the pointing solution, transfix subtracts off the extra leakage. In this process, there is no fitting to the observed error -- it simply uses the Narrabri-deduced pointing solution to predict the extra leakage. The user has to give transfix the Narrabri pointing parameters -- a table of 6x12 numbers extracted from leon::at$log:pparams.log.

Signed: Bob Sault, Michael Kesteven (30-Jun-1997)


Description: Intermittent antenna pointing failure

Type of affected data: All ATCA data

Date of affected data: from approx March 1996 to 1997?

Problem in detail: Over the past months there have been numerous reported cases of ATCA antennas suffering pointing errors. The pointing errors have been noted to have the following properties:

  • It occurs about once per week (on average)
  • It only occurs in azimuth
  • Errors are typically of the order of one degree
  • It may occur on any antenna
  • The ACC does not normally recognise the error and, as a result, on-line monitoring software may not alert the observer to the problem.

Solution: As the cause of the problem is unknown, no permanent fix is available at this stage. If detected while actually observing, a reboot of the relevant ACC will fix the problems. Observers are asked not to reboot until monitor data has been recorded by the Duty Astronomer or local staff. With the possibility of no errors being reported, astronomers should be vigilant in checking to make sure that visibilities displayed on-line look reasonable, especially for calibrators (which should have constant amplitude). The pointing error would show up as all baselines with the bad antenna suddenly decreasing in amplitude.

If the error is not detected and corrected on-line then the antenna in question should be flagged out when reducing the data.

Signed: Derek McKay (28-Oct-1996)


Description: The RA and DEC coordinates can be corrupted when using Miriad ATLOD and when only one IF was observed.

Date of affected data: Observations between 26 June and 23 July 1996.

Problem in Detail: There was error in the information written into the RPFITS file when only a single IF was being observed. The problem only affects data loaded with Miriad ATLOD (AIPS ATLOD does not attempt to interpret the information which is erroneous). The result is erroneous RA and DEC coordinates. The coordinates are very obviously wrong (e.g. declinations exceeding -90, hour angles greater than 24 hours).

Solution: Either use AIPS ATLOD or see Bob Sault.

Signed: Bob Sault (2-Sep-1996)


Description: Precision errors can occur in the RA and DEC coordinates when the highest astrometric precision is required.

Date of affected data: Observations before 24 June 1996, and particularly before 27 November 1995.

Problem in Detail: Prior to 27 November 1995, programs SCHED and CAOBS contained precision errors (rounding and truncation) which were causing the RA and DEC coordinates given by the users in SCHED not to be honoured. The error could be as much as about 100 milliarcsec. The RA and DEC given by the user, could vary from that used by CAOBS as the phase centre, which again could vary from the value stored in the RPFITS header.

Between 27 November 1995 and 24 June 1996 the situtation was improved to at least be predictable. The SCHED file now contains the coordinates given by the user (to real*8 accuracy). However CAOBS rounds this position to the nearest 1 millisec of RA and 10 milliarcsec of DEC. It is these rounded values that are used in phasing the array. These are also the values stored in the RPFITS header.

Since 24 June 1996, full (REAL*8) precision is maintained throughout.

Solution: There is no solution for data observed before 27-Nov-95. Be aware that SCHED files created before 27-Nov-1995 (or derived from those created before then) may suffer from the precision problems. SCHED files created from scratch after then do not suffer problems. Since 27-Nov-95, for those interested in the highest astrometric precision, the effect is correctable. This is left as an exercise to the reader.

Signed: Bob Sault and John Reynolds (2-Jul-96)


Description: At 13 cm, the off-axis polarimetric response of the ATCA is poor, possibly leading to erroneous interpretation.

Date of affected data: All 13-cm data.

Problem in Detail: A flaw in the design of the 22/13 cm horn means that the off-axis polarimetric purity of the ATCA dishes is very poor at 13 cm. This can lead to appearent percentage polarisation levels of 10% linear and 5% circular at about the half-power point of the primary beam. A detailed description of this problem can be found in

Solution: The Miriad task, "offaxis", can be used to subtract the off-axis response. This may reduce the problem by a factor of 3 or so in linearly polarised emission, and by a larger factor in circularly polarised emmision.

Signed: Bob Sault (17-May-96)


Description: When observing in continuum mode, each real source produces a ghost source at the 0.1% -- 0.5% level positioned diametrically opposite the phase centre.

Date of affected data: Potentially all 33-channel/128-MHz data

Problem in Detail: XF correlators, such as the ATCA's, inherently suffer from a problem known as the Gibbs phenomena (see Albert Bos, in "Image Formation from Coherence Functions in Astronomy"). This is caused by a discontinuity in the visibility spectra at zero frequency in the baseband signal. As this discontinuity is different for the real and imaginary parts of the signal, the resultant spectral sidelobes in the real and imaginary parts of a spectrum are also different. This results in an effective difference in the gain between the real and imaginary parts of a correlation. The triangular lag weighting function used for the continuum system has the undesirable property that this gain error does not average out across the visibility spectrum (for spectral line correlator configurations it will tend to). In the image plane, the error manifests itself as ghost sources, which are located on the opposite side of the phase centre to true sources. The error also is a non-closing one.

At the centre of the observing spectrum, the real/imaginary gain difference is small (~1%), with the error increasing (slowly at first) towards the edges of the band. Like the true source, antenna phase errors will cause ghost sources to decorrelate. Unlike true sources, ghost sources are decorrelated further by the calibration process (primary calibration or self-calibration). The net result is that the ghost sources are only apparent in high dynamic range images, where observation conditions have been moderately good (i.e. no antenna phase to cause decorrelation). Typically the ghost sources will not be apparent in images with dynamic ranges less than a few hundred to a thousand. The problem is likely to be more pronounced in more recent data, because of general improvements in telescope stability (e.g. LO phase stabilisation).

Solution: Although solvable off-line, the correction must be performed at the very start of the reduction process (whereas the problem will not normally become apparent until the end). Using the option `reweight' with Miriad's atlod task reweights the visibility spectrum in the lag domain in a fashion which eliminates the problem. This option also prevents a given narrowband interfering signal corrupting more than one channel on each side (because the new lag weighting function is no longer triangular and its Fourier Transform has much better sidelobe characteristics than the sinc squared function -- see the previous note on self-interfernce caused by 128~MHz birdies). However the option also causes extra bandwidth smearing across a correlator channel -- this will be a consideration for wide-field 20cm imaging. For data already loaded, Miriad's atxy task also has a reweight option. This correction must be applied before any calibration (including XY phase calibration), frequency averaging or channel selection in performed.

Signed: Bob Sault (29-Mar-95)


Description: ATCA data suffers from self-interference at integer multiples of 128MHz.

Date of affected data: Potentially all data

Problem in detail: ATCA data suffers from self-interference at integer multiples of 128 MHZ; these signals (which are generated in the samplers) are very narrow. However, the frequency spectrum is convolved by the Fourier transform of the lag weighting function, so, in certain correlator configurations these "birdies" ring through the spectrum. Because the signal is quite phase-stable, it will cause phase-centre artifacts if not eradicated. A particularly annoying birdie occurs at 1408 MHz, as it is in the external interference-free part of the 20cm (L band) spectrum that we all like to use.

Ideally, we would like to eradicate these birdies by hardware changes. But, this is a non-trivial exercise. In the meantime, it is possible to deal with the birdies reasonably successfully during the data reduction in the most common observing configurations.

The character of the ringing that occurs depends upon what spectral setup the correlator is configured to. That is, the lag-weighting function changes with the correlator setup.

In standard 128 MHz 32-channel mode, the lag-weighting function is a triangle, and its Fourier transform is a sinc-squared function. This function has a straightforward character though. Firstly, a birdie will occur in the middle of integer channel N provided the central frequency is a multiple of 4 MHz (the channel increment); this is the case for the most commonly chosen frequencies. In this case, the sinc-squared function convolving the birdie will have zeros at channels N+/- 2, N+/-4, N+/-6 and so on. All other channels are corrupted by the birdie because of its non-zero sinc-squared signal. Thus, if we flag out or drop channels N, N+/-1, N+/-3, N+/-5 and so on, the birdie will be fully eliminated. This causes virtually no loss in sensitivity because adjacent channels are not independent in this correlator configuration anyway. The miriad task "atlod" has been changed to offer a new option to eliminate the corrupted channels in this way. Alternatively, you can flag out the bad channels yourself. Additionally, the ATCA scheduling program "SCHED" now warns you if a birdie is expected to be in your band.

At other bandwidths, the situation is not so straightforward. As we decrease from 128 MHz to 4 MHz bandwidths in steps of 2, the shape of the lag-weighting function changes gradually from a triangle to a rectangle (top hat).

The Fourier transform of the top hat (appropriate to a 4 MHz bandwidth) is a sinc function, rather than sinc squared. If it has a peak in channel N, then all other channels align with zeros of the sinc function. Thus, with the 4 MHz bandwidth, the birdie will not ring at all, but will simply show up in one channel that needs to be flagged out as bad. At the other bandwidths between 4 and 128, the situation is somewhere in between, where the zeros of the Fourier Transform of the lag-weighting function will not occur at integer channels. In these cases, the birdie will ring through the spectrum in such a away that it cannot be flagged out. This would also be the case in 128 MHz mode if the birdie did not fall exactly in the middle of a channel.

In summary, at 128 (with the central frequency set to a multiple of 4 MHz) and 4 MHz bandwidths, any birdie can be fully eliminated by eradicating the appropriate channels with no significant loss of sensitivity. At 8 MHz bandwidth, the lag-weighting function is sufficiently close to a top hat that a birdie will generally not ring substantially. At other intermediate bandwidths, it is not possible to eliminate fully birdies from the spectrum by post processing.

Finally, please see the next note in this series on Ghost images from Gibbs phenomena in 128 MHz bandwidth mode. The solution to this problem involves reweighting the lag function so that it is no longer a triangle. With regards birdies, this has the desirable effect of minimizing their ringing into the spectrum, because the Fourier Transform of the reweighted-lag function is no longer a sinc-squared function, but something with much weaker (less than 1~percent) and alternately positive and negative spectral sidelobes. The penalty is a modest loss of spectral resolution.

Signed: Neil Killeen (28-Mar-1995)


Description: When using B1950 coordinates in scheduling, the conversion to J2000 coordinates was mildly defective, and introduced a position error of order 0.35 arcsec.

Date of affected data: up until October 1994

Problem in detail: The ATCA on-line system works in J2000 coordinates, and produces data in the J2000 frame. If B1950 positions are used in the observing schedule, CAOBS converts these to the J2000 frame as one of its first steps. In this conversion, CAOBS was failing to correct for the elliptic terms of aberation (the E-terms) which are normally included in B1950/FK4 catalog positions. These terms are of order 0.35 arcsec.

If a J2000 position was used for the phase calibrator, but the program source was in B1950 coordinates, then the resultant J2000 coordinates of the program source are correct. No action is required. However the actual observing centre (which is correctly recorded in the J2000 frame) will be offset from the intended position by the E-terms.

If a B1950 position was used for the phase calibrator, then there is a problem: the subsequent calibration operations will correct the observed phases to put the calibrator on axis -- in effect shifting the array by the E-terms.

Solution: A correction is needed only in this latter case. The E-terms are

 Delta RA = -(C*cos(ra) + D*sin(ra)) / (15*cos(dec)) Delta DEC = -(D*cos(ra) - C*sin(ra)) * sin(dec) - C*tan(eps)*cos(dec) 


 C = -0.065838 arcsec D =  0.335299 arcsec C*tan(eps) = -0.028553 arcsec. 

These should have been added to the B1950 catalog positions.

Reduced images can be corrected by adding the above terms to the RA and DEC values -- stored in the header as CRVAL1 and CRVAL2 -- using PUTHEAD (in AIPS) or puthd (in Miriad). These are stored as degrees in AIPS, and radians in Miriad.

For visibility data before the calibration stage, correct the phase calibrators phase and position using uvedit (in Miriad) and UVFIX (in AIPS). Process only the B1950 calibrators. Note that UVFIX does not handle a multi-source file, and so you will need to split, UVFIX, and then reconstitute the multi-source file for calibration.

Signed: Mike Kesteven and Bob Sault (27-Oct-1994)


Description: Mixing of sources when using correlator averaging in mosaic and frequency switched mode

Type of data affected: Mosaic and frequency switched ATCA data

Date of affected data: 21-Dec-1993 to 20-Jul-1994

Problem in detail: It is possible to average N cycles together in the correlator. However, if you were observing in mosaic mode (SCTYPE=MOSAIC) or rapid frequency-switched mode (SCTYPE=FREQSW) then the correlator was not resetting the accumulators at source or frequency boundaries if the averaging period crossed that boundary. It was correctly resetting the accumulators every N cycles and also at scan boundaries as expected.

In addition, if you spent, say, 3 cycles per mosaic field and averaged N=3 cycles in the correlator, there is still the strong possibility that your data would be corrupted. This is because until now, there are generally approximately 2 extra cycles in the beginning of a mosaic pattern so that the averaging process across source boundaries would still cause your data to be mixed. Note also that those extra cycles were labelled as coming from the first source in the mosaic pattern, whereas they were in fact off source. In addition, those extra cycles were unflagged. As of now, the extra cycles are flagged by the on-line system and should no longer cause problems.

Solution: You must reobserve.

Signed: Warwick Wilson and Neil Killeen (10-aug-1994)


Description: Change in the spectrum of ATCA primary calibrator 1934-638

Type of data affected: All ATCA data

Date of affected data: 04-Aug-1994 to ...

John Reynolds has finished his analysis of the 1934-638 spectrum and has produced a new polynomial fit. This new fit has been incorporated into ATNF-AIPS/SETJY (we will send it to NRAO) and Miriad/gpcal and Miriad/mfcal.

In all cases, the new polynomial is now the default. To obtain the old polynomial, use APARM(2)=1 in AIPS/SETJY and options=oldflux in Miriad/gpcal,mfcal

Details of the change to the flux density scale, can be found in John's ATNF techincal memorandum AT/39.3/040. You should be careful when combining new data with already calibrated old data as the change in the flux density scale is upto 10% (depending on the frequency) as is shown by the plot (dashed line is old spectrum, solid line is new spectrum).

Signed: Neil Killeen and Bob Sault (04-aug-1994)


Description: Potential problem causing ATCA data to not be calibratable

Type of data affected: All ATCA data

Date of affected data: 21-Dec-1993 to 12-Apr-1994

Problem in detail: There has been, since the new correlator computer ATRIA replaced the old correlator computer SANCHO, the potential for ATCA data to be unrecoverably corrupted. Owing to a hardware flaw, it is possible for the lag spectra to be fractionally shifted. This causes shifts in delay and phase on all baselines, and also results in closure offsets.

The net result is that the data will not calibrate and cannot be recovered.

The probability of this ocurring to your data is thought to be low, but it is known to have occurred in the week 01-Mar-94 to 08-Mar-94. Specifically, to data taken on 04-Mar-94 to 08-Mar-94.

Do my data suffer from this ? To detect whether you have this problem or not you should look at the closure phase on your calibrators. Normally, this should be zero plus or minus noise. The symptom of the problem is a constant closure offset on all baselines. It is of the order of 20 degrees for 128MHz bandwidth data and is expected to be greater for narrower bandwidth data.

In AIPS, you can use the tasks CLPLT or SHOUV to examine the closure phase. In Miriad you can use the task uvtriple.

This closure offset will prevent your data from calibrating correctly. That is, when you apply the calibrator solutions to the calibrator itself, you will not get zero mean phase (self calibration will fail also for the same reasons).

Solution. No software solution exists, you will have to scrap the data and apply to reobserve. Mention of the hardware flaw will be looked upon favourably by the Time Assignment Committee.

All observers in the above period will be contacted by email, and our apologies to those of you who have been affected by this problem.

Signed: Warwick Wilson and Neil Killeen (10-may-1994)


Description: Mislabelled Coordinate System in AIPS Headers

Type of data affected: All ATCA data

Date of affected data: 11-Nov-1993 to 18-Jan-1994

Problem in detail: Between 11-Nov-1993 and 18-Jan-1994, ATLOD at ATNF Epping was labelling AIPS files as having the coordinate system type UU-L-NCP, VV-L-NCP, WW-L-NCP, but FITTP was, as always, writing files with UU-L etc. so that the labels were lost. Trying to DBCON data with NCP strings with data without NCP strings resulted in failure

The NCP change will be implemented in the future but without making a mess. In the meantime we are changing back to the original labelling without the NCP strings. If you are having trouble and need to remove the NCO strings, you can do this with PUTHEAD according to

 keyword='ptype1'; keystrng='UU-L'; puthead keyword='ptype2'; keystrng='VV-L'; puthead keyword='ptype3'; keystrng='WW-L'; puthead 

Signed: Henrietta May


Description: Sky frequency-correlator channel number relationship changed

Type of data affected: All ATCA data from 5300-6060 MHz

Date of affected data: 1989 to 10-Mar-1993

Problem in detail: Because of hardware constraints, there is a frequency in the 6 cm band at which the frequency-correlator channel number relationship changes. That frequency used to be at 6060 MHz, but when observations were made at 6050 MHz, there was a local oscillator within the 3 cm band at 8310 MHz.

The change-over frequency has now been moved to 5300 MHz so that:

  * Sky frequency DECREASES with increasing channel number in the 20 cm and 3 cm    bands and the lower part of the 6 cm band ( below 5301 MHz ), and  * sky frequency INCREASES with increasing channel number in the 13 cm band and    the upper part of the 6 cm band ( above 5300 MHz, including 6670 MHz ). 

Note that this may cause problems if you are trying to combine data in that frequency range taken before 10/3/93 with data taken after 10/3/93)

Signed: Russell Gough (10-mar-1993)


Description: Missing system calibration information in RPFITS files

Type of data affected: All ATCA data

Date of affected data: 24-Sep-1992 to 06-Oct-1992

Problem in detail: Data taken with the Compact Array between 24-Sep-1992 and 06-Oct-1992, 1600 AEST have parts of the system calibration information missing. A software bug in a new version of CACOR was the cause.

Missing are quantities such as Tsys, the sampler statistics and the polarisation calibration (XY) phases for all except antenna 1.

The interferometer data is present and correct.

Strategies for processing this data are discussed below.

 Solutions   ------------- 

The absence of the system calibration information will only affect you if you are going to make a full polarimetric calibration in Miriad.

Regardless of whether you are making a polarization calibration or not, you need to ensure that ATLOD does not look for any of the missing system calibration information. There are a number of switches that you can turn on in ATLOD which access the SYSCAL info, here is how to make sure they are off. Set

 ATLOD ----- APARM(8)  = 0 APARM(9)  = 0 APARM(10) = 0 CPARM(1)  = 0 CPARM(2)  = 0 CPARM(6)  = 0 CPARM(9)  = 0 CPARM(10) = 0 

set all your other favourite adverbs as usual.

If you plan to make a polarimetric calibration, then read on, otherwise proceed as you would normally and read no more here.

What you do next depends upon whether you made an on-line phase calibration or not. Generally observers do this after making a delay calibration while setting up.

 CASE 1 -- On-line phase calibration done ---------------------------------------- 

If you did this, then the XY-phase differences on each antenna will come out to be close to zero. In this case, you can proceed as if you had followed the path where you plotted up the XY phases, chose a representative value for each antenna and applied it with ATLOD. The representative value should be close to zero, and in the absence of more knowledge, we will assume it is precisely zero. After editing, you then transfer the data to Miriad as usual and solve for residual XY phase differences on each antenna except the reference antenna for which we assume the XY phase difference was truly zero. If you have an observation of 3C138 or 3C286 you can work out the value for the reference antenna as well in the usual way.

 CASE 2 -- On-line phase calibration not done -------------------------------------------- 

The system calibration information is missing except for antenna 1, and then it is only recoverable if you run a special version of ATLOD that I can provide. You can then fish out the XY phase difference for antenna 1. When loading the data with ATLOD, you would not apply this single XY phase difference. When running GPCAL in Miriad, you must choose antenna 1 to be the reference antenna, tell GPCAL the va;ue for the XY phase difference for antenna 1 and solve for the residuals XY phase differences on the other antennas. As the measurement for antenna 1 will not be very good, you should then solve for the residual XY phase difference on antenna 1 as well provided you have an observation of 3C286 or 3C138. Thus, all is recoverable in principle.

I suggest you come and talk to me (nebk) if you are in either of these latter two categories to make sure all is clear.

Signed: Warwick Wilson and Neil Killeen (07-oct-1992)


Description: Faulty Antenna Tables

Type of data affected: All ATCA data

Date of affected data: May-1992 to Aug-1992

Problem in detail: Because of a CAOBS problem between May and August 1992, ATLOD has been creating antenna tables with antenna names A01 - A06 instead of CA01 - CA06. This will be a problem if the data is to be combined with data with the correct antenna names. A new task, ANFIX, has been installed to modify these faulty tables.

Signed: M. Wieringa


Description: Data from compact array with incorrect u,v,w

Type of data affected: All ATCA data

Date of affected data: 04-Mar-1992 to 21-Mar-1992

Problem in detail: Compact Array data taken in the period 4-Mar-92 to 21-Mar-92 were inadvertently labeled with incorrect baseline coordinates v and w (they reflected a pseudo-polar projection). These data cannot be combined with data from earlier periods as is.

The most straighforward remedy, which returns the data to the standard projection, is to run the AIPS task UVFIX with UVFIXPRM(1) = 2 and UVFIXPRM(12) = 0.1 and the rest at zero. This will cause UVFIX to recompute u,v and w in the standard projection.

It is important that UVFIX be run on single-source files (after SPLIT) as it uses header information to get the field centre. After SPLITing and UVFIXing, you could reform the multi-source data base with the task MULTI, if desired.

Signed: M. Kesteven, M. Wieringa, N. Killeen (24-mar-1992)


Description: CAOBS bug affecting RPFITS headers

Type of data affected: All ATCA data

Date of affected data: : 17-Aug-1991 to 22-Aug-1991

Problem in detail: A CAOBS bug has caused data taken during the period 17-Aug-1991 to 22-Aug-1991 (12:noon) to have a problem with the RPFITS header: the frequency increment and spectrum flag were not written.

The consequences: in attempting to reduce your data in AIPS you will likely find that AVSPC will fail with an arithmetic error.

If you overcome that, and get to making a map, you will then find your map inverted.


 1.  Use ATLOD to invert the phase of 20 cm, 6 cm and 3 cm band data.     by setting CPARM(3) = 1 when you load your data.       With 13 cm data you should NOT do this.   2.  After loading your data, use PUTHEAD ( keyword='CDELT3';  keyval=-4.e6)     to correct the frequency increment in the aips uvdata header.  If you     were at 13 cm then the increment should be +4E6.  Spectral line     observers should set the increment with the same signs as these, but     with the correct magnitude reflecting their setup.   3.  If you plan to use bandwidth gridding in MX or to make spectral line     cubes, you should also use TABED to correct the FQ table.   Set the     sideband to +1 and set the frequency increment to the correct signed     value.  The first column in the FQ table is the offset from the frequency     in the header and should be 0.0 for one FREQID, and something else     for the second FREQID if there is one (these should be correct as is).  

Signed: Apologies, M. Kesteven (22-aug-1991)


Description: CAOBS bug causing wrong (u,v,w)

Type of data affected: All ATCA data

Date of affected data: 1989 to 01-Aug-1991

Problem in detail: A CAOBS bug has caused the visibility coordinates (u,v,w) to have been incorrectly computed. They were calculated for the START of the integration period, whereas the UT associated with the visibility is (as it should) the mid-point.

This means that maps will be rotated by 1.25 arcminutes * sin(dec) for a 10 second integration.

The bug has been present since the start of observations,

The bug will be removed for the august (1991) sesion.

All credit to Ravi for finding the problem and its cause.

Signed : M. Kesteven (01-aug-1991)


Description: Inverted spectra in some bands

Type of data affected: All ATCA data

Date of affected data: 1989 to 23-Oct-1990

Problem in detail: ATCA observations produce spectra that can have frequency increasing or decreasing with channel number, depending on the precise LO setup that was used. ATLOD did not correctly assign this direction sign to the frequency increment when the spectra were inverted. Inverted means frequency decreasing with increasing channel number, it does not reflect "right" or "wrong", it is just a convention. Here follows a summary of band directions for the 128 and 64 MHz bandwidth AT bands.

 1) an inverted spectrum for X band observations. 2) an inverted spectrum for L band observations. 3) an inverted spectrum for most C band observations. 4) a non-inverted spectrum for some C band (those at the top end of the band) observations 5) a non-inverted spectrum for S band 

As of 23-Oct-1990, ATLOD assigns the correct sign to the frequency increment. This date reflects a change to the software, not the actual data recorded. Bands other than 128 and 64 MHz wide should also now be correctly dealt with by ATLOD.

For older data that you have already loaded, you can find out what sign the frequency increment should have been as ATLOD writes the "inverse video" value to the terminal in both its "LOAD" and "LIST" options;

-1 means inverted, +1 means non-inverted

You can change the sign of the frequency increment in the AIPS uv data base if necessary, with the AIPS verb PUTHEAD

Note however, that the inverse video indicator is incorrect for some old data, and ATLOD will be misled as follows.

The indicator was stuck at +1 until 19-Jun-1990, whereupon it held fast at -1 until 01-Aug-1990. Then the glue came loose and it reflected the true state of affairs. If you have data in this category, and you need to know what direction frequency goes, consult with Mike Kesteven or Warwick Wilson. Good luck !

Signed: Neil Killeen


Description: Phase Negation

Type of data affected: ATCA data as described below

Date of affected data: 1989 to 01-Aug-1990

Problem in detail :

Under certain circumstances, the phase of the visibilities in the RPFITS data from the ATCA must be negated to produce images which are the right way up.

The visibility phases can be negated either by using the appropriate option in ATLOD ( CPARM(3) ), or by using the local AIPS task NEGPH.

A certain, highly technical, and therefore unnamed, criterion divides the data from the ATCA into two distinct types, which we will call TYPE A and TYPE B. You must first determine which type of data you have.

 TYPE A data includes the following :   All 3cm data    All 6cm data - EXCEPT that observed during the period 09-May-1990, 14:59    to 20-May-1990, 12:55 (AEST) AND in the frequency range 5.500 to 6.100 GHz.    All 20cm data  TYPE B data includes the following :   All 13 cm data    6cm data observed during the period 09-May-1990, 14:59 to 20-May-1990,    12:55 (AEST) AND in the frequency range 5.500 to 6.100 GHz.  

If you have any doubts about which type of data you have, then ask an expert.

Whether the data requires a phase negation depends on four factors :

 1. The data type 2. Where the data is/was loaded into AIPS (i.e. Epping or Narrabri) 3. When the data is/was loaded into AIPS 4. When the data was observed 

  LOADED at EPPING :    Loaded before 2 July 1990      if data type A - - ok as is      if data type B - - negate      Loaded after 2 July 1990 :      Observed prior to 19 June        if data type A - - ok as is        if data type B -- negate      Observed 19 June to 19 July        if data type A -- negate        if data type B -- ok as is      Observed 19 July to 1 August        if data type A -- ok as is        if data type B -- negate  LOADED at NARRABRI :   Loaded before 1 August 1990     Observed prior to 19 July       if data type A -- ok as is       if data type B -- negate     Observed 19 July to 1 August       if data type A -- negate       if data type B -- ok as is 

Signed : Warwick Wilson


Description: phase error after phase CAL operations.

Type of data affected: All ATCA data

Date of affected data: 1989 to 22-Jun-1990

Problem in detail: phase referencing can be compromised if the field centre and the phase reference source are too far apart on the sky. The problem is that the UHF synthesiser can be set to different frequencies for the field and the reference - eg, 605 and 606 MHz. This is because at present we attempt to fix the spectra in frequency in the solar barycentre frame - thus a line observer should see a spectral feature in trhe same channel irrespective of when his observation was made. But this means that the LO settings will depend on the time of year and source location, even if the one frequency was requested.

Cure: for data already taken, there is no simple cure - I doubt that a calibration of the synthesisers after the event would be of much use. For the present, prevention is the best measure - make sure that both field and reference are close together. There is a utility available: "check_scan" which will compute the LO settings for each source in a schedule, and issue a warning if these change from one source to the next. If necessary, you can shift the observing frequency by a fraction of a MHz to avoid the problem.

I will try (for the July observing session) to activate a line/continuum switch.

Signed: M. Kesteven


Description: non-inverted spectra labelled inverted

Type of data affected: All ATCA data

Date of affected data: 1989 to 15-Jun-1990

Problem in detail: All 13cm data have non-inverted spectra, and these will have been consistently mis-labelled (in RPFITS). 6cm data in the range 5.5 to 6.0 GHz for the period 11-May-1990 (to 12:55 [aest]), 20-May-1990 are also affected.

Cure: relabel the spectral increment in AIPS with PUTHEAD

Signed: M. Kesteven


Description: Incorrect Tsys scaling after a frequency band change.

Type of data affected: All ATCA data

Date of affected data: 1989 to 13-May-1990

Problem in detail: The first scan after a band change (e.g. 20cm to 13cm; 6cm to 3cm) uses an interpolation based on the previous band, and is thus quite wrong.

Cure: none simple. (Is anyone seriously affected?)

signed: M. Kesteven