The location of six antennas at six stations is called an array configuration, or often, simply a configuration. The westernmost antenna, CA06, is fixed in position. The other five antennas can be positioned at any of 44 fixed stations. The station posts provide mains power to the antennas and network connections that allow commands from the control building to be sent to the antenna, and monitor data to be received from the antenna. The posts also have high-speed optical links that allow data from the antenna to be transferred to the control building. The smallest baseline increment available is 15.306m, and the shortest physical baseline is 30.612m. Five of the six ATCA antennas are shown in Figure 1.1.

A standard set of 17 configurations has been defined, and a subset of these is offered for each semester. These configurations have been designed to give optimum, minimum-redundancy coverage after a 12 hour observing period. Four configuration sets for the principal arrays (750m, 1.5km and 6km) are offered over several semesters (see Appendix H for details of these sets). Your choice of configuration depends on the extent, brightness and complexity of your source (see below).

The Australia Telescope Compact Array is an earth-rotation aperture synthesis radio interferometer. Earth-rotation aperture synthesis was first used in the 1950s for radio observations of the sun. The technique is comprehensively explained, with a historical perspective, in Interferometry and Synthesis in Radio Astronomy by Thompson, Moran & Swenson (Wiley, 3rd edition, 2017). Essentially, the array of antennas is comprised of a number of two-element interferometers. The visibility (i.e., the fraction of the signal common to both antennas of a pair) is derived by multiplying the (suitably delayed) signals together. By combining the correlated signals obtained over a long period of time and with a large range of spacings between antennas, the Compact Array measures the spatial coherence function:

where $\mathrm{I}_{\nu}(\mathbf{s})$ is the two dimensional intensity distribution on the sky, $\mathbf{s}$ is the unit vector in the direction of the celestial source, $(\mathbf{r_{1}} - \mathbf{r_{2}})$ is the separation vector between antennas 1 and 2 and $d\Omega$ indicates integration over the sphere (or, in practice, solid angle of the antenna beam).

The van Cittert–Zernike theorem states that the Fourier transformation of the spatial coherence function yields the source brightness distribution, i.e., the Fourier Transform of the visibilities produces an image of a radio source. The image is formed with the same angular resolution as for observations with a single antenna with a diameter equal to the largest spacing, however it will be less sensitive due to the interferometer's smaller collecting area.

By plotting the tracks that the baseline vectors trace out (from the source's perspective) as the Earth rotates, astronomers can gauge how good the telescope will be at imaging the source and resolving components in the field of view. This plot is referred to as the (u,v)-coverage as, by convention, the two orthogonal axes of the plot are u and v. These variables have units of the observing wavelength. The (u,v)-coverage shows where on the Fourier plane the image has been sampled.

To obtain the fullest (u,v) coverage with the ATCA would require observations with multiple different configurations, and for twelve hours with each configuration. However, almost any program can be successfully carried out with less than complete (u,v) coverage, and the individual configurations are chosen so as to optimise single configuration imaging characteristics. Sophisticated off-line image processing techniques minimise the effect of missing (u,v) coverage and allow reasonable images to be made with one configuration. For most programs, one to four configurations provide the best compromise between dynamic range, (u,v) coverage and time.

The smallest synthesized beamwidths in Right Ascension for each observing wavelength are shown in Table 1.1, but bear in mind that, for east-west arrays, the beamwidth in Declination is greater by a factor cosec(Dec). At high angular resolution, the telescope is only useful for observing southern objects. North of Declination -24°, full (u,v)-coverage is unobtainable; near Declination zero the beam is highly elongated north-south; and north of Declination +48° sources are below the telescope's +12° elevation limit and are inaccessible. For lower angular resolution, the north-south and hybrid arrays improve the (u,v)-coverage for equatorial sources, but only up to a maximum north-south baseline of 214m.

Figure 1.1. Five of the six ATCA antennas in the H75 array.

The rest of this chapter outlines the structure and operation of the ATCA. The path of the radio signal is traced through the system from the dish to the final data recording. This section is designed to help you understand how the Compact Array operates and provide all the information required to design an appropriate experiment. Detailed technical information about the ATCA can be found in the Journal of Electrical and Electronics Engineering, Australia, Special Issue, Vol. 12, No. 2, June 1992, a copy of which is available in the control room at the Narrabri Observatory.

The antennas have a Cassegrain design, i.e., the receivers are located in a turret that protrudes through the main reflector surface — see Figure 1.1. The antennas have an altitude-azimuth mount with wrap limits as shown in Figure 2.1. The shaped (i.e., non-parabolic) dish and subreflector surfaces are designed to maximise the gain to antenna noise ratio. The reflecting surface of antennas 1–5 are solid panels that allow observations up to 116 GHz. This is also true of the inner 15.3m of antenna 6. The outer reflecting surface of antenna 6 is made of perforated panels which permit observations at frequencies up to 50 GHz.

The subreflector mounted near the prime focus has a specially designed shape, differing slightly from the hyperbolic secondary of standard Cassegrain optics. The feedhorns are mounted on the optic axis — this allows polarisation measurements to be made, with very low ( $\sim 1\%$) instrumental polarisation. Subreflector height can be adjusted to achieve optimal focus for the different frequency bands. A major feature of the ATCA is its wide bandwidth operation. The feedhorns and front-end electronics operate over a very large range of frequencies, thus allowing (u,v) coverage to be increased by multi-frequency synthesis and dual frequency observations. The feedhorns are compact, with a corrugated interior surface and are designed for maximum frequency coverage, low noise, low spillover, low reflection and low cross-polarisation sidelobe levels. The feedhorns allow two simultaneous orthogonal linear polarisations to be measured (at frequencies almost an octave apart), and have a main lobe with near-constant, symmetrical beamwidth (James, G.L., 1984, IEEE Trans. Antennas Propagat. AP–22, pp 1134–1138; Thomas, B.M., James, G.L., Greene, K.J., 1986, IEEE Trans. Antennas Propagat. AP-34, pp 750–757).

The frequencies at which the Australia Telescope operates are listed in Table 1.1 see also section Section 1.3.1.

Each range of frequencies is referred to as a band. Bands are referred to by the wavelength of the approximate centre of the band or (especially at higher frequencies) by the frequency (and sometimes by the letter-codes shown over the columns of Table 1.1. There are presently five feedhorns mounted on each antenna (four on antenna 6): a large (2m high) feed-horn that operates in the 16cm band, and a somewhat smaller (50cm high) 4cm band feedhorn and three (several cm high) feedhorns mounted on the same dewar for the 15mm, 7mm and 3mm wavelengths. The feedhorns and the receivers are mounted on a rotating turret. The rotating turret design ensures that all feedhorns, when brought on-axis, are aligned with the optic axis of the antenna, allowing a wide field of view and dual-polarisation observations. The turrets are rotated automatically in accordance with the frequencies selected by the observing file. Rotating the turret orients the desired feed-horn and also ensures that the subsequent electronics suit the frequency at which you are observing. Changing the observing band to or from 7mm requires first rotating the turret to the 15mm position, and then translating the whole mm dewar to bring the 7mm feed on axis. The latter stage takes 2~3 minutes, and so changes to or from 7mm are much slower than any other band change.

Both the 16cm and 4cm band feedhorns are fitted with wide-bandwidth receivers that covers the horn's entire usable frequency range. All receivers run continuously and use cooled HEMT (high electron mobility transistor) and FET (field effect transistor) amplifiers that provide wide bandwidths and total system temperatures between 20K to 65K, depending on frequency.

The signals collected by the feedhorns are fed to the receiver systems for amplification and conversion to standard intermediate frequencies. During observations each antenna provides four independent intermediate frequency (IF) outputs (two frequency bands, two polarisations). These channels allow simultaneous observations of both polarisations at the two available frequencies in either dual-receiver system. The frequencies can be anywhere in the range covered by the selected feed-horn except at 15, 7 and 3mm where the two frequencies must be less than 6 GHz apart, and not straddle the 7mm sideband inversion (see Section 2.3.6).

In order to convert the received radio signals to lower frequencies, the radio signals are mixed with a “local oscillator" signal. Four local oscillator signals provide four IF outputs to enable the dual-frequency, dual-polarisation operation. You can also switch frequencies at the end of each integration cycle (typically ten seconds). To change to a pair of frequencies covered by a different feed-horn requires a rotation of the turret and takes about twenty seconds. However, to avoid excessive wear of the turrets, a limit of four rotations per hour is imposed. Thus time sharing between a number of frequencies is limited only by signal to noise ratio, wear and tear on the equipment, your imagination, and the off-line software.

The following sections describe the path of the signal from the initial reflection off the antenna surface, through to the data recorded onto export media, in more detail than the introduction above. Some radio-astronomy jargon is also explained, as is some important mathematics. The details of this section are not required by observers, but are included for completeness. A knowledge of the system is however essential if you want to design an experiment that uses the array in a new or unconventional way. Familiarity with the names and functions of some critical components can also be helpful if you need to diagnose faults. Many system components are housed in modules, and repairs can be made quickly by local staff replacing a module once a correct diagnosis has been made.

Figure 1.2. Overview of signal paths into CABB (courtesy Brett Hiscock).

Radio waves (approximately within a range of 300 MHz to 120 GHz) are accurately reflected from the primary surface of the main parabolic dish and are re-reflected off a secondary reflector into a feed-horn. At the base of the feed-horn is a directional, coupling waveguide. In this coupler (the noise coupler) is a diode that injects a noise signal of known amplitude. This noise signal is used to calibrate the system temperature (Equation 1.1) and gain of the receiving system.

For radiation emitted by a randomly polarised source, the power received by a radio telescope is given by:

where A is the effective area of the antenna, S is the spectral power flux density and $\Delta\nu$ is the range of frequencies observed (the effective bandwidth). The factor of $\frac{1}{2}$ occurs because a detector can only respond to one polarisation component of the randomly polarised wave. The ATCA overcomes this limitation by providing separate detectors and electronics for two orthogonal polarisations, thus allowing all the power in the wave to be detected.

The sensitivity of a radio telescope is the minimum signal power that can be distinguished from the random fluctuations at the output of the receiving system which are caused by noise inherent in the system. The sensitivity is usually defined as the spectral power flux density of a source that would produce the same signal power as the noise power. It is measured in Jansky (Jy), with $1 \mathrm{Jy} = 10^{-26} \mathrm{\,W\,m}^{-2} \mathrm{\,Hz}^{-1}.$

The noise power consists of two main components: the noise power due to the receiving amplifier and other electrical system components, and the noise due to ground radiation, thermal emission from the atmosphere (which varies with elevation, cloud cover, etc.), background radio emission from our Galaxy and other sources detected by the antenna. These noise powers are usually referred to as equivalent temperatures, although at no point is a physical temperature measured. The “temperature” due to the noise power is called the system temperature, $\mathrm{T}_{\mathrm{sys}}$, and is related to the noise power by:

where $\mathrm{P}_{\mathrm{noise}}$ is the noise power and $k$ is Boltzmann's constant. As both the source and noise signals are random in nature, measurements of the power levels made at time intervals separated by $(2 \Delta\nu)^{-1}$ can be considered independent (e.g., Thompson, Moran & Swenson 1986, 2001). If the signal level is averaged for $t$ seconds then $2 \Delta\nu t$ samples have been measured.

The signal to noise ratio is the ratio of the power in the output that is due to the radio source being observed to that caused by the noise, and is given by:

In this expression, $\mathrm{T}_{\mathrm{ant}}$ is the antenna temperature, the equivalent temperature of the radio source being observed. Note that “antenna temperature" is sometimes also taken to mean the contribution to the noise power from radio noise detected by the antenna, as described above. For typical values for bandwidth (2 GHz) and integration time (12 hours), it is possible to detect a signal for which the power level is less than $10^{-6}$ times the noise level.

The noise source in the noise coupler injects a signal that is about 5% of the level of the system temperature at a rate of 8 Hz. This noise signal is synchronously demodulated and the following relationship holds:

Here, $\mathrm{T}_{\mathrm{cal}}$ is the equivalent temperature of the noise source, $\mathrm{P}_{\mathrm{on}}$ is the power received while the noise diode is on, and $\mathrm{P}_{\mathrm{off}}$ is the power received while the noise diode is off. $\mathrm{T}_{\mathrm{sys}}$ is measured accurately once by placing a thermal radiator of known temperature (a microwave absorber at 300K, giving 300K $+ \mathrm{T}_{\mathrm{sys}}$) directly above the feeds. The power received with this load in place is then compared with the power output with the antenna just looking at the sky ( $\sim \mathrm{T}_{\mathrm{sys}} + 3\mathrm{\,K}$). This measurement is used to establish the level of $\mathrm{T}_{\mathrm{cal}}$, which is assumed not to vary with time. The measurement of system and antenna temperature is described in a technical memo by Sinclair and Gough (1991). And despite our assumption that this level should not vary with time, it can be calibrated by the correlator at any time using a point source with a known flux density, with the acal command (Section 3.3.6.4).

The noise coupler is mounted on another coupler that provides a vacuum seal between the noise coupler and the subsequent receiver system, which is housed in an evacuated, cryogenic cylinder. An air gap thermally isolates the cylinder, the contents of which are cooled by a helium pump to 20K. The second coupler is mounted on the polariser. The dual function of the polariser is to select the two linear polarisations and the desired band, so it is also known as a band splitter. The polariser consists of four strips of metal inside a conical waveguide that “guide” two orthogonal linear polarisations into probes at the narrow end of the polariser. The way in which the metal strips select the linear polarisations can be thought of as analogous to the effect of a ridged waveguide, which constrains a particular mode to propagate along the region between the ridge and the roof of the waveguide. The probes are simply short (with respect to the wavelength) lengths of the cores of the coaxial cables that take the signal into the first amplifiers. Thus the polariser converts a wave into a voltage: the impedance of the circuit is effectively the same as a waveguide impedance, so it works like a well-matched load. The polariser's official title is quad-ridged orthomode transducer, or OMT. Separate electronics exist for both orthogonal linear polarisations, which are measured simultaneously. The position angle of the polariser is fixed with respect to the antenna. As the antennas have altitude-azimuth mounts, the position angle of the linear probes rotate on the sky during the course of an observation (imagine a circumpolar line on the sky orbiting around a stationary antenna). Measurement of both polarisations is therefore required for polarised sources. The two linear polarisations can be combined to give an accurate total intensity: the Stokes I parameter. Without additional calibration, the other Stokes parameters (the linear polarisation measures Q, and U, and the circularly polarised component V) can be in error by about 2% of the value of I. Measurements of the phase difference of the two linear polarisations (referred to as the X and Y (or A and B) polarisations) at the receiver by on-line hardware may need further off-line calibration.

For an unpolarised source, observing both orthogonal polarisations offers a $\sqrt{2}$ improvement in signal to noise ratio for an I image over an image generated from only one polarisation.

The signal is taken from the feedhorn into a low noise, broadband amplifier.

The receivers each use two low-noise amplifiers (LNAs) to amplify the signal by 30–40 dB. All six antennas are fitted with wideband, continuously running, receivers. Cooled indium-phosphide high electron mobility transistors (InP HEMTs) are used at 16cm and 4cm, and indium-phosphide monolithic microwave integrated circuit (InP MMIC) devices are used at 15mm, 7mm and 3mm. Only the “inner" five antennas (i.e., excluding CA06) have 3mm receivers. The accessible frequency range is given in Table 1.1, and the average system temperatures are given in Table 1.4 and in Section 1.3.1. Note that frequencies outside these nominal limits may be accessible.

The next step is the conversion to the frequency range that is used by the CABB digitisers. The 4cm signals require no conversion, whereas the mm signals require a down-conversion stage, and the 16cm signals require up-conversion to the 4 to 12 GHz band.

The local oscillator signals have a relatively narrow tuning range, lying in the regions between the radio frequency bands (this reduces the likelihood of self-generated interference).

The final output from the conversion system is sent to the samplers for digitisation. The CABB (Compact Array Broadband Backend) samplers were developed in-house and are capable of 4 gigasamples per second (GS/s), with 10 bit sampling (more information is available in the February 2007 ATNF newsletter and in Wilson et al. (2011, MNRAS, 416, 832)).

The samplers, which are also referred to as the CABB digitisers, do two things. First the analog (continuously variable in time and amplitude) signal is sampled into a discrete-time sequence of values. This process does not degrade the signal as it is sampled at the Nyquist rate (twice the inverse frequency of the bandwidth of the analog signal). Subsequently, each of the discrete-time, continuously variable values are converted to one of a finite set of values: this is called quantising. CABB uses 9-bit sampling (internally, as stated above, 10-bit sampling is used, with 9 bits transferred to the correlator), with approximately 6-bits used in sampling the regular data range, and the additional bits providing extra robustness in the case of interference RFI (radio-frequency interference). The digitised signals from the sampler are sent from each antenna to the correlator along optic fibres at a rate of 160 gigabits per second for each antenna. Optical fibres run from the samplers in each antenna to the correlator, which is housed in the screened room (to prevent RFI generated by the correlator electronics from affecting the observations) of the central control building. Synchronising code is added to the data stream at the start of each integration cycle to allow each bit to be correctly identified at the correlator irrespective of temporal changes in the length of the fibre.

The correlator effectively multiplies simultaneous signals from two samplers: the more similar the signal from both antennas, the larger the degree of correlation. A plane wavefront approaching the ATCA will, in general, arrive at different antennas at different times. The signals from the antennas therefore need to be delayed before being presented to the correlator so as to simulate a wavefront arriving at each antenna simultaneously. This was previously achieved by separate delay units, but is now included in the CABB boards. The delay calibration undertaken at the start of an observation removes other delays caused by return path length differences, instrumental delay variations and the like, so that wavefront samples are presented to the correlator synchronously.

The CABB filterbank/correlator is divided across a number of Digital Signal Processing (DSP) boards. The DSP boards use Virtex-4 Xilinx Field Programmable Gated Arrays (FPGAs) to provide a flexible, programmable hardware correlator. The output from the correlator is averaged for the period of one integration cycle, and sent to the correlator control computer caccc where it is written to hard disk. Integration cycles can be set by the observer, between 2 and 30 seconds, although the default cycle time of 10 seconds is generally recommended for all observers. (For mosaic-mode observations, 6-seconds is the practical minimum cycle time.)

More information on the correlator can be found at the CABB webpage.

The CABB correlator provides a number of observing modes:

This design ensures that the user will always get high continuum sensitivity from the wide-bandwidth IFs, while providing very high spectral resolution for line studies at the same time.

Figure 1.3.  Average Compact Array system temperatures for each observing band at high elevation under reasonable observing conditions. These are based on hot-cold load measurements, and include the atmosphere at the time of observation, thus representing the total system temperature. These measurements were made by J. Stevens with the CABB system, except for the 3mm measurements which were made by T. Wong in September 2004.

You can also switch automatically to other wavelengths within tens of seconds (with the exception of the 7mm band as described previously). Similarly the 16cm receiver has a single feedhorn and observations with this package will cover the entire 1.1 - 3.1 GHz range simultaneously. Observations at two simultaneous frequencies are possible in the 4cm, 15mm, 7mm, or the 3mm bands. With CABB, the centre frequencies of the two IFs need to lie within 6 GHz of each other, and even this might be too much separation depending on the exact frequencies chosen. For CABB, the nominal standard frequencies for continuum observations are

These frequency designations follow the ATCA custom of stating the central frequency of the chosen observing frequency range. (For archival data from the pre-CABB era the standard frequencies were

The 15mm recommended CABB central frequencies have recently been changed from 17000 and 19000 GHz, due to the presence of RFI from a geostationary satellite (see Section 1.3.2.3 for details).

Switching between the different frequency bands involves a changing the feed horns. The different feed horns are mounted on a "turret" which rotates the appropriate feed horn to the on-axis position. This is done automatically under computer control and takes about 20 seconds. Turret rotation should be limited to once every 15 minutes unless a compelling scientific case is made for more frequent rotations. The additional overhead in changing to or from 7mm has been described earlier in this document.

Observing-band centre frequencies may be set to the nearest MHz only and no on-line Doppler tracking is done.

Observations of weak H 90α recombination lines may be affected by a trapped mode in the 6/3cm horn at 8857 ± 18 MHz. This trapped mode may enhance the power received this recombination line at their rest-frequency. There are also notches reported in the passband due to trapped modes in the receiver waveguides at 4550 ± 10 MHz, 5328 ± 10 MHz and 8780 ± 10 MHz which may need to be flagged during data reduction.

Both terrestrial and satellite emitters cause radio frequency interference (RFI) to ATCA observations, with the lower frequency bands being more adversely affected than the higher frequency bands. This section describes the RFI that is known in each of the ATCA bands, and also covers interference caused by the Sun.

1.3.2.1. 16cm band RFI

The interference is worse at lower frequencies with the main offenders being microwave links, microwave TV, microwave ovens, navigation satellites and self-generated interference. There can also be significant interference at 1381 MHz from the GPS L3 beacon on occasions. These channels may have to be removed from the data. Figure 1.4 shows what the ATCA sees in its 16cm band (centred at the recommended continuum central frequency of 2100 MHz) when fringe-rotation is disabled (achieved by setting the phase centre to be the South celestial pole). The black, red and blue lines show the log of the amplitude as observed by surveys in 2011, 2015 and 2018 respectively, and from this we can see that the RFI environment does change slightly over time. No self-generated interference is visible in this plot. The Australian Communications and Media Authority (ACMA) band licenses are shown on the plot as coloured frequency regions. The strongest sources of RFI are mobile phone towers, satellites and TV transmitters, all of which are relatively stable in time.

Figure 1.4. The ATCA 16cm RFI environment. Three surveys are shown, one from 2011, one from 2015 and another from 2018. The licenses applicable to each frequency range are shown as coloured regions, labelled with a similar colour at the top of the plot.

An RFI monitoring antenna sits atop the central control building at the observatory. It monitors the frequency range 700 - 3000 MHz once every 20 seconds, with 2 MHz frequency resolution, and presents its data in near-real time on a web page. Three different plots are presented on the page: a “latest spectra” plot (Figure 1.5), a waterfall plot showing the last hour of data (Figure 1.6), and a plot that may help in determining the direction from which RFI is being observed (Figure 1.7). The latest spectra plot always shows the earliest spectrum obtained by the monitor in the displayed time range, along with a “maximum hold” value for each channel over the last hour and the time-averaged power for each channel. The waterfall plot is useful for seeing emission switch on or off over the last hour. You can also scroll back in time for the past day in half-hour increments for each of the plots.

All data from the RFI monitor is archived. A page showing the RFI waterfalls from ATCA and Parkes each day from 6am to 8pm (local time) is available at this link. If you want to get the output of the monitor for any particular time range since its installation in November 2014, please contact <Jamie.Stevens@csiro.au>.

Figure 1.5. An example of the “latest spectra” plot available from the ATCA RFI weathermap page.

Figure 1.6. An example of the “latest waterfall” plot available from the ATCA RFI weathermap page.

Figure 1.7. An example of the “latest skyplot” plot available from the ATCA RFI weathermap page.

In addition to the “normal” RFI in the 16cm band, as shown in Figure 1.4, you may also see something dubbed “mid-week RFI”. This RFI has frequency peaks around 1265 and 1300-1310 MHz, and can have extremely high amplitudes. The RFI monitor can be used to see when mid-week RFI is present, as shown in Figure 1.8. Because of the strength of this RFI, it can often drive the front-end system of the 16cm receivers, or the CABB digitisers, into saturation, or at least produce very large intermodulation products since each tone will act as another LO; if this occurs, the data becomes unusable. We do now sometimes receive advanced warning on when this RFI may be present, and if this will affect observations, the investigators will be notified so they can plan how to deal with it.

Figure 1.8. A waterfall plot from the RFI monitor, showing what “mid-week RFI” can look like (in the brighter region).

Only the 16cm band is affected by mid-week RFI, so if mid-week RFI is forecast during your observations, you may wish to avoid using the 16cm band during those times, wherever possible. If this is not possible, you may ask for a schedule swap, but this may not always be a possibility. Our observations have also shown that mid-week RFI is present only for a small fraction of the default CABB 10 second cycle time, but is so strong that it ruins the entire integration. You may therefore be able to cope with mid-week RFI by running with a shorter cycle time (say 1 or 2 seconds) and flagging those cycles in which the RFI is present; this will result in a much higher data rate of course, and is only really suitable for continuum experiments that do not need mosaicking.

1.3.2.2. 4cm band RFI

The RFI present in the 4cm band is direction dependent, and can be quite variable over time. Using a similar survey technique to the 16cm band, where the fringe-rotation is disabled, we can determine the static levels of RFI. We also use the antennas to determine how directional the RFI signals are, by doing the survey twice; once while the antennas are pointing at the South Celestial Pole (SCP, Figure 1.9), and again while pointing at the celestial equator (Figure 1.10). We expect that signals coming from satellites will be stronger towards the celestial equator.

Figure 1.9. The 4cm band, as observed during an RFI survey from 2016-09-07, while the antennas were pointing at the South Celestial Pole (SCP). The colour bands show frequency ranges that represent either the CABB recommended continuum bands, or the ACMA licence applicable.

Figure 1.10. The 4cm band, as observed during an RFI survey from 2016-09-07, while the antennas were pointing at the celestial equator at transit. The colour bands show frequency ranges that represent either the CABB recommended continuum bands, or the ACMA licence applicable.

Indeed, we see that in the pink- and orange-highlighted regions of those two figures, the signal intensity is higher while pointing at the equator. That is, in the frequency bands licenced for satellite communications to Earth, the ATCA will see a lot more interfering signal while observing near to the equator. The frequencies of these highlighted regions are given in Table 1.2.

Table 1.2. Frequency ranges of satellite emission bands observed with a 4cm RFI survey. The ranges given are delimited by ACMA allocations, so several entries may be given even if they represent one contiguous frequency range.

Purpose	Frequencies
	Low [GHz]	High [GHz]
Communications to Earth	3.6	4.2
	5.925	6.7
	7.2525	7.3775
	7.375	7.45
	7.45	7.55
	7.55	7.75
	10.7	11.7
Broadcast to Earth	11.7	12.2

The other frequency ranges shown on Figure 1.9 and Figure 1.10 encompass RFI that is not likely to be from satellites, but as can be seen, the intensity of this RFI is not high and should be easily excisable during data reduction.

1.3.2.3. mm band RFI

There appears to be no significant interference within the 7mm and 3mm bands. An RFI survey in the 15mm band showed that interference from the Australian National Broadband Network (NBN) satellite “Sky Muster” is clearly visible when pointing toward the equator. Similar to the 4cm RFI survey technique, Figure 1.11, Figure 1.12 and Figure 1.13 show that the signal received by the antennas from this satellite (soon to be two satellites) is very dependent on the pointing direction.

Figure 1.11. The 15mm band, as observed during an RFI survey from 2016-08-01, while the antennas were pointing at the celestial equator at transit.

Figure 1.12. The 15mm band, as observed during an RFI survey from 2016-08-01, while the antennas were pointing at the South Celestial Pole (SCP) at transit.

Figure 1.13. The 15mm band, as observed during an RFI survey from 2016-08-01, while the antennas were pointing east at an elevation of 85°.

The “Sky Muster” satellite emits in three broad downlink frequency ranges. These ranges are 17.7 - 18.2 GHz (Figure 1.14), 18.8 - 19.3 GHz (Figure 1.15) and 19.7 - 20.2 GHz (Figure 1.16).

Figure 1.14. A subset of the 15mm band, as observed during an RFI survey from 2016-08-01, while the antennas were pointing at the celestial equator. This frequency subset covers the “Sky Muster” downlink range 17.7 - 18.2 GHz.

Figure 1.15. A subset of the 15mm band, as observed during an RFI survey from 2016-08-01, while the antennas were pointing at the celestial equator. This frequency subset covers the “Sky Muster” downlink range 18.8 - 19.3 GHz.

Figure 1.16. A subset of the 15mm band, as observed during an RFI survey from 2016-08-01, while the antennas were pointing at the celestial equator. This frequency subset covers the “Sky Muster” downlink range 19.7 - 20.2 GHz.

This geostationary satellite is at an azimuth of 341.5° and elevation of 53.1°, as seen from the ATCA. This corresponds to a declination of +6° 35′ and hour angle of +1h 14m (west of transit). This position is approximately 18.9° away from pointing position of the antennas during the RFI survey, so the intensity of the interference can be expected to be much higher than shown in Figure 1.11 for pointings closer to the satellite.

The recommended CABB continuum central frequencies for the 15mm band were 17 and 19 GHz, but these may now contain a significant amount of RFI from “Sky Muster”, depending on where your source is. If you are worried about this interference and wish to avoid it, we recommend a new set of continuum central frequencies of 16.7 and 21.2 GHz. These recommended frequency bands are not contiguous, but do avoid all the “Sky Muster” downlink emissions and in fact are slightly more (between 3 and 10%) sensitive than the 17/19 GHz bands in good dry weather where the atmospheric water content is low. If you need to use the 17/19 GHz bands now in order to facilitate comparisons to data obtained in the past, you will likely need to do extensive flagging during data reduction; the reduced bandwidth lost to flagging will also mean that your continuum sensitivity suffers.

A synthesis imaging telescope such as the Compact Array provides a great deal of flexibility when deciding how to image an object. In addition to the well-known trade-off between observing time and sensitivity, we have trade-offs with maximum resolution and with the sampling interval in the visibility plane.

The choice of angular resolution is fairly straightforward. Increased angular resolution (resulting from longer baselines) leaves the point-source sensitivity constant, but decreases brightness sensitivity in proportion to the beam area (see Table 1.4).

The choice of (u,v)-plane coverage is more difficult. The amount of independent information needed to specify the image should not exceed the number of independent (u,v)-plane samples. But unfortunately neither of these “independents” is easily defined. At one extreme, consider making a high resolution image of a complex source filling the entire primary beam. In this case full (u,v)-coverage is obtained by using all baselines between 30m and 6000m in increments of 15m. Although this was theoretically possible (it takes 25 separate array configurations), it was never attempted and, since the decommissioning of the second 6km station, is no longer actually possible!

If on the other hand the image is smaller than the primary beam then the Nyquist sampling interval is larger and the number of (u,v)-samples required is reduced. The observing time can then be reduced either by decreasing the number of configurations or the amount of hour angle coverage, depending on the array configuration. For east-west arrays, reducing the total number of configurations is the more practical option. If the source is large but partly empty it can be considered to have a size corresponding to its area.

For ordinary observations Table 1.4 gives the maximum sizes of structures that can be reliably imaged for typical sets of observing configurations. This is only a rough guide since the actual coverage needed depends on details of the 2-dimensional brightness distribution, on the actual distribution of baselines, and on the type of deconvolution. Additionally, other techniques such as mosaicking and multi-frequency synthesis, can be extremely effective at improving (u,v)-coverage.

A further consideration is the minimum spacing available. This is never less than 30 m and for a given configuration can be much larger. This acts as a high-pass filter removing all Fourier components less than the minimum spacing. If this is a serious problem, short baseline information (e.g., from a single dish) can be added separately during processing.

Antenna 6 sits permanently on station W392. Maximum baselines to this antenna (the last 5 columns) are shown for all configurations in Appendix H, but form part of a designed array only for configurations 6.0A to 6.0D. The antennas can be moved to, and set up on, a limited number of fixed stations. Because of this and other physical restraints, the shortest spacing available is 30 m, the longest is 6 km, and the minimum grating increment is 15 m. From 2006 October, the standard set of configurations has been:

Snapshot observations using the Northern spur in hybrid configurations will generate a two-dimensional sampling of the (u,v)-plane. The Northern spur was installed so that good coverage of the (u,v)-plane was achievable for observations which are limited to hour-angles near transit. Observations at mm wavelengths should be limited to higher elevations to avoid large atmospheric opacities. Hybrid arrays are also useful for observations of northern sources which are similarly restricted in hour-angle range.

The 6 km antenna may also be added to any of the shorter configurations, but in these cases the distribution of array spacings is bi-modal.

The predetermined set of configurations offered for forthcoming observing terms will assist in planning multi-configuration proposals. Details of previous configurations and those being offered in coming semesters can be found on our website, or see Table 1.3.

Note that proposals requiring two or more configurations will usually be allotted two or more widely separated times; therefore, expect to make two or more observing visits, or conduct the second observation using remote observing. If you are not allotted all the configurations requested, you should re-apply in the next term, as the proposal will not be automatically reconsidered.

The overall philosophy is that, in each semester, there will be a 6km, 1.5km, and 750m configuration, and for these baselines, the full set of 4 configurations will generally be covered in three semesters. Each semester will also contain at least one 375m configuration. Analysis of user preferences for the past few years show some configurations to be more generally useful (e.g., providing better single-configuration (u,v)-coverage), and it is desirable to offer these more frequently. Millimetre observing conditions are optimal in the (southern hemisphere) winter, so arrays of 214m or smaller will be offered mainly in the April semester.

Besides the predetermined configurations, you can request any standard or non-standard configuration in any term. When writing your application, your scientific justification should include a convincing argument of why you need to use a special array configuration, rather than one of those offered for the term. If you need such a “wildcard” request for a significant amount of observing time (e.g., more than 5 x 12h), you can enhance the probability of it being scheduled if you contact ATNF well in advance of the deadline (preferably even before the call for proposals announcement). The wildcard can then be advertised as a potential additional configuration for the term, which may then lead to other proposers requesting it, and making its scheduling viable.

For those who wish to improve their (u,v)-coverage by re-observing on different days with different antenna configurations, specific sets of configurations combines well, e.g., 6A, 6C, 1.5B and 1.5D — for more details see Appendix H.

An interactive tool, the Friendly Virtual Radio Interferometer (VRI), is available to assist users in exploring the (u,v)-coverage of standard (and non-standard) configurations.

It is strongly recommended that the proposer gives a clear indication of the maximum extent of their sources. You should also specify the maximum and minimum baselines, and the number of configurations (days) needed.

The CABB now provides a variety of frequency resolutions. The available resolution options are more commonly denoted by the CABB configuration that is used to provide them. Two correlator configurations are available:

It should be noted that the correlator routinely computes double the number of channels as is recorded. This second set of channels is offset from the available set by half the channel width, and are there to provide flexibility in choosing the zoom band frequency, as well as to ensure that it will always be possible to avoid observing a spectral line at the edge of a channel. It is also possible to combine a number of overlapping zoom bands that are separated by half a continuum channel into a seamless single zoom band.

It is crucial to obtain good calibration data during your observations, and you must consider how much of an overhead calibration will represent before you propose an experiment. A detailed description of the calibration required is given in Section 2.2.

The amount of calibration required will also depend on when the experiment is scheduled. Observations requiring maximum phase stability (e.g., at wavelengths less than 6cm) should be made in winter, or at night.

The general expressions for the flux and brightness sensitivities are given in the document AT/01.17/025, and these general expressions have been used in Table 1.1. The observer has control of the integration time ($t$), bandwidth (B), observing wavelength ($\lambda$), number of baselines (N) and synthesized beam size ($\theta$) only, and with these variables and the system sensitivity ($\mathrm{S}_{\mathrm{sys}}$), the expressions reduce approximately to:

and

where $\mathrm{S}_{\mathrm{sys}}$ is in Jy, $t$ in min, B in MHz, $\lambda$ in cm, and $\theta$ is in arcsec. For the full 6 km array, N=15. There is a convenient ATCA sensitivity calculator which takes into account system temperature, observing frequency, array, correlator configuration and (u,v)-weighting. (A pre-CABB version of the calculator is also available on the ATCA web page for use when considering the utility of archival ATCA data.)

All six antennas of the Compact Array are equipped with 15mm receivers covering the frequency range 16–25 GHz. The 15mm system shares a common dewar with the 7mm and 3mm receivers, and the separate feed horns at the top of the dewar are moved to the Cassegrain focus of the Compact Array antennas using the rotator positioning system. Because the 15, 7 and 3 mm receivers have separate feed horns, observations cannot be simultaneous between the 15, 7 and 3 mm bands.

The ATCA can observe within the frequency range 30–50GHz. Feeds and receivers were installed into the existing mm-wave receiver packages on all 6 antennas in 2007, with the 7mm upgrade jointly funded by the ATNF and NASA's Deep Space Network to enable the ATCA to participate in occasional spacecraft tracking.

The wide 7mm band presents difficulties in signal down-conversion, as the aliased signal from the first down conversion stage can also be within the 7mm band. Although image rejection filters are used, variations in receiver gain across the band, combined with the frequency-dependent performance of the filters themselves, can result in appreciable signal levels being added to the observing band. Generally, this aliased signal does not cross-correlate and is present as a contribution to the noise level. However, in some circumstances, notably as the delay rate drops to zero (e.g., around source transit on north-south baselines), the aliased component can cross-correlate and be visible as “beating" on some baselines. This data must be flagged during the data processing.

The inner five ATCA telescopes (i.e., excluding CA06) are outfitted with a 3mm receiver and can observe in the range 83 to 105 GHz. A noise diode has been added to the 3mm system on CA02 (only) to aid with 3mm polarisation calibration. Aliasing effects similar to those described at 7mm can also arise in the 3mm band.

A limited form of flexible scheduling is operational for observations at millimetre wavelengths. Consult the document Flexible Scheduling at ATCA for details.

A quickstart guide for 3mm observing is also available.

Observations in the mm band are run from a schedule file in the same way as cm-band observations. However, due to the effects of atmosphere stability, mm observations require bandpass and phase-calibration checks at a much more frequent rate.

Note that as ATCA does not Doppler track, it is important to have a good estimation of the magnitude of the Doppler shift for the line of interest. This is particularly so for the higher frequencies available to the 3mm system. The sky frequency can be calculated using an online calculator.

A simple phase reference observation takes about an hour to run, including calibration. The following procedure is typical:

Pointing calibration → Paddle calibration (3mm only) → Bandpass/Phase calibrator → Target source → Paddle calibration (3mm only) → Bandpass/Phase calibration → Target observation.

The CABB web scheduler can be used to construct the observing schedule.

You may not wish to make detailed high resolution images but instead, for example, survey a large number of compact or simple sources. Under these circumstances, if the sources are strong enough, short observations will suffice (Table 1.5). In scheduling these, slew time becomes an important consideration; these may be calculated (see Section 1.7.3), or with the use of the CABB scheduler. A single short observation in an E-W array will give a 1-dimensional (u,v)-coverage; two or more short observations will give some 2-dimensional information. In such short observations consequent higher sidelobe levels will exacerbate problems caused by confusing sources in the primary beam. Because the Compact Array has poor (u,v)-coverage, in contrast to its good sensitivity, this confusion makes it difficult to reach the receiver noise limit when observing weak sources or searching for detections, particularly at lower frequencies. Current experience suggests that at 6cm at least 8 short observations well distributed in hour angle are needed to reduce confusion to the noise level. At 16cm, short observations are unlikely to ever reach the noise limit.

Two orthogonal linear polarisations are measured simultaneously. The position angle of the polarisation splitter is stationary with respect to the alt-az–mounted antennas and so rotates on the sky. A suite of tasks in Miriad is available for polarimetric calibration of Compact Array data. No specific calibrators, other than the usual flux and gain calibrators, are needed for polarisation calibration. An observation of B1934-638 is essential when only making short observations. Small drifts with time can be corrected for using the on-line measurements of the XY phase differences to an accuracy of 0.5 degrees at all bands. Best polarimetric calibration results when the gain calibrator is observed at a good sampling of different parallactic angles.

Miriad is routinely used to calibrate data in all bands, with consistent results being achieved over several months in the cm bands. The on-axis instrumental polarisation is typically below 2-3%. After calibrating for instrumental polarisation, we are currently able to reduce on-axis instrumental effects to 0.1%, and better with some care.

The off-axis polarisation increases roughly as the square of the distance from the pointing centre at least up to the half-power point. At 16 and 4cm, the instrumental polarisation is about 3% of the apparent total intensity at the half power point, respectively. This error is almost purely linearly polarised (there is no circularly polarised component). Because the ATCA antennas have an alt-az mount, the off-axis response varies with parallactic angle, and will be smeared out by a factor of a few by a long synthesis. This smearing is a function of declination. The Miriad task offpol can be used to simulate off-axis polarimetric response of a long synthesis observation. Mosaicking smears out the off-axis response still further, by as much as an order of magnitude.

At 16cm, instrumental polarisation is largely independent of frequency. The leakages are now flat to better than 2% across the entire usable band. More detail on the polarisation properties of the ATCA 16cm receivers can be found in the receiver commissioning report.

Objects which are of similar size or larger than the primary beam will need to be mosaicked. In such cases, the recommended spacing of pointing centres is half of the primary beamwidth. The Compact Array antennas and control system allow for rapid switching between pointing centres — as frequently as once per integration cycle. This enables a source to be rapidly mosaicked without necessarily losing (u,v)-coverage. The antenna acceleration limit is 800 deg/min/min, and the slew limit is 38 deg/min in azimuth and 19 deg/min in elevation. Mosaicking mode may also be useful for observing large numbers of nearby sources, as the observing overheads are reduced.

There is also a facility for “on-the-fly" mosaicking, whereby the antennas are continuously driven. This type of mosaicking may be especially useful for large-area mosaics at low frequencies, where the drive times between mosaic pointings are significant.

Section 2.3.4 explains how to set up mosaic files. The Miriad User Guide describes how to reduce a mosaic data set.

As (u,v)-distance is proportional to frequency as well as baseline length, different (u,v)-spacings can be obtained not only by varying the antenna configuration, but also by varying the frequency.

Additional (u,v)-coverage can be obtained in the mm bands by observing at multiple frequencies. Observing at two frequencies has the added advantage of increasing sensitivity, as they can be observed simultaneously. Observing more than two frequencies requires time sharing. While this will not improve the sensitivity further, it can significantly improve the (u,v)-coverage. However, there is a trade off between gaps in the tangential and radial directions in the (u,v)-plane. Typically two or three pairs of frequencies, observing each setting for 10 minutes, is a good compromise.

When you use both bandwidth synthesis and two or three configurations, and require (u,v)-coverage to 6km, the best choice of configurations is not two or three 6km arrays, but a combination of 6km with 1.5km and 750m arrays (all arrays using the 6km antenna). The flux density of most sources will often vary significantly between different frequencies and, furthermore, this variation itself (i.e., the spectral index) will vary across the source. This complicates the task of combining data from the different frequencies when you want high-dynamic-range images. However software is available in Miriad to account for the spectral variations in the imaging, deconvolution and self-calibration steps. These algorithms solve for, or use, both a basic flux-density image and a spectral-index image. For typical spectral indices, they are appropriate for frequency ratios less than about 1.25.

After each antenna reconfiguration a pointing solution is determined. Generally, rms errors of around 5 arcsec are achieved at this time. These solutions degrade with thermal (and other) effects. By the time observations are undertaken, rms errors of about 10 arcsec (or more) are likely. There is a significant offset (of up to 30 arcsecs) between the mm and the cm receivers.

A reference pointing mode is available which is recommended for mm observations. In this mode, a 1 Jy calibrator, about 5° to 10° away from the target will hold the pointing to 10 arcsec rms. A bright (say 5 Jy) calibrator at 2° to 3° from the target will reduce the errors to about 5 arcsec rms. A reference pointing scan generally takes 2-3 minutes to complete. See the Reference Pointing Guide for details.

In pulsar binning mode, the correlator cycle time (normally 10 seconds) is divided into an integer number of pulsar periods. Each pulsar period is further divided into N equal time intervals called bins. The integration in the correlator is done separately for each bin, such that at the end of the cycle, N quantities are produced, each of which represents the integrated value at a particular pulsar phase. The minimum time bin is around 110 micro-seconds, allowing, for example, 32 phase bins for a 3.5 milli-second pulsar. The correlator will also record both the full 2 GHz continuum bands in addition to the pulsar binning data, analagous to how the zoom modes are recorded. However, the binned data has lower frequency resolution (by a factor of 4) than the unbinned continuum data.

For high time resolution observations, a time resolution of order 10 milliseconds is possible, though with a reduced number of channels across each 2 GHz band (achieved by averaging over 1 MHz channels).

The Compact Array can track sources with non-sidereal rates, such as planets or comets. In this case delay tracking is adjusted continuously to account for source proper motion. The antenna pointing also correctly tracks non-sidereally.

JPL ephemerides of the planets are built into the observing program, and a simple mechanism exists to import current JPL ephemerides of other solar system objects (e.g. new comets).

A tied array capability is available. It provides tying of the array at one or two frequencies and, for CABB, at a bandwidth of 64 MHz.

The tied array adder is controlled via a process called catie which runs within cacor. This allows the choice of which antennas are included and whether the adder produces linear or circular polarisation outputs. The array is phased up in the usual way, and an option to allow the insertion of a 90° phase offset between the A and B linear polarisations at each antenna – thereby forming circular polarisation at the tied-array output – is available.

The tied-array adder is fed into the LBA (Long Baseline Array) DAS which provides outputs for the VLBI disk-based recorders at bandwidths between 62.5 kHz and 64 MHz. Simultaneous Compact Array and tied array operation is possible, although currently this is limited to 64 MHz in both IFs, or 64 MHz in one IF and the full 2 GHz (which is not formed into a tied array) in the other.

The presence of field sources in the primary antenna beam can limit continuum image quality. These sources produce unwanted sidelobes in the images and can lead to dynamic range and aliasing problems. The brightest source expected on average in the half-power primary antenna beam (away from the Galactic plane) is listed in Table 1.1. Note that confusion becomes much worse as you go to lower frequencies, and that this can be particularly serious for snapshot observations (See Section 1.7.1). The SUMSS catalogue and database, and the Molonglo Galactic Plane Survey (MGPS-2) catalogue and database are useful references to search for potential confusing sources in the 16cm band, where the effects of confusion are greatest.

At short wavelengths and/or long baselines, atmospheric refraction can cause serious phase errors. The problem becomes progressively less serious at longer wavelengths and shorter baselines. Atmospheric conditions are most favourable during winter and at night. A seeing monitor at Narrabri continuously monitors the phase stability at a frequency of 21 GHz and over a baseline of 200 m. See The ATCA Seeing Monitor for details (though note this paper describes the initial system; in 2008 a new satellite beacon was adopted, resulting in a change in the frequency from 30GHz to 21GHz, and the design was changed again in 2019 when the satellite we were using became non-geosynchronous).

High winds, storms and extreme heat can damage the antennas. See Section 3.5.1 to Section 3.5.3 for further details on automated stowing procedures.

As with other synthesis instruments, system errors such as DC offsets and sampler harmonics can lead to artefacts at the centre of the field. These artefacts have not yet been observed to affect the data from the CABB correlator, but if such an error would severely degrade your science goals (e.g. in a detection experiment), it is preferable to displace the source positions a few (synthesized) beamwidths from the field centre.

A Duty Astronomer will be present at the SOC to help local observers get their observations started, and to help with any problems that arise during the observations. Our web site has the list of duty astronomers for the upcoming semester, as well as for previous semesters. The Duty Astronomer has no obligation to participate in the observing. For remote observers, the DA is primarily there to help as problems arise, as qualified remote observers should not need assistance to start their observations.

The ATCA is responsive to requests for observations of targets of opportunity. There are two paths to such observations: through a NAPA (non-apriori assignable) project, or through a ToO (target of opportunity) request.

A NAPA project is one whose science goals have been set down in a proposal that the TAC has evaluated in the normal way. Such projects can be triggered any time during the semester, whenever the investigator wants to proceed. The project will have requested a specific amount of time for the semester, and this is the amount of time that can be displaced from scheduled observations to accommodate the NAPA project. However, if the NAPA requires further time, it can be granted using unallocated time.

A ToO request can also be made at any time during the semester. Unlike NAPA projects though, a ToO will not generally be allowed to over-ride a scheduled observation.

The ATCA antennas have a hard elevation limit of 12°, and for observations in the millimetre bands it is recommended that you stay above an elevation of 30° in order to avoid the thickest parts of the atmosphere.

A quick estimate of how long a particular source will be observable on a single day with the ATCA can be made using Table 2.1. It shows, for a source at a particular declination, the LST range before/past transit it is above elevations of 12° and 30°.

Example: The source 0823-500 (coordinates RA=08:25:26.869, Dec=-50:10:38.4) will rise above the 12° elevation limit at an LST of 01:05 (take 07:20 from 08:25), and it will fall below 30° at an LST of 13:42 (add 05:17 to 08:25). It will be above the 12° elevation limit for approximately 14 h 40 m each day.

To calculate precisely how long your source is observable each day with the ATCA you can use the Interactive Observability Chart Plotter.

CABB makes available two 2048 MHz wide bands, and the centre frequency of one IF must be within 6 GHz of the centre frequency of the other. Generally speaking however, the closer the two centre frequencies are to each other, the more sensitive the observations will be.

This is due to the design of the CABB system. All frequency bands observable by the ATCA are first mixed to fit into the primary CABB frequency window of 4 to 12 GHz (except for the 4cm band, which doesn't require any mixing). By default, closely-spaced IF pairs are mixed so that they lie at the top of this frequency window because the lower frequency end can suffer from poor image rejection. If two widely-spaced IFs are selected however, the observational sensitivity for one of the IFs may suffer because of this.

It is important to note that the ATCA sensitivity calculator does not account for this effect.

To counter this problem, you may wish instead to use frequency switching techniques; you can read about how to do this in Section 2.3.6.

For normal continuum observations we recommend that you use our recommended frequencies, shown in Table 2.2.

If you are using the 3mm system, the choice of simultaneous frequencies can affect the observational sensitivity in another way, due to restrictions on the LO frequencies:

RFI at the ATCA falls into two groups: external, and self-generated. External RFI is worst at lower frequencies, and can significantly reduce the bandwith usable in the centimetre bands. Self-generated RFI is less destructive, but also mostly unavoidable, and can be present regardless of which band you're observing in.

During a normal observing run you will have to do calibration before you can start your observations, and then you will need to schedule time on one or more calibration sources.

At the beginning of your observing run you may find that while observing a strong point source, the phases will not be constant as a function of frequency across the band; this indicates that the array requires delay calibration. To successfully perform delay calibration will require a point source of at least moderate flux density ($\mathrm{S} \gtrsim 1\mathrm{\,Jy}$). This is so the correlator can accurately determine the phase slope with frequency; the accuracy improves with the signal-to-noise ratio.

The response of each antenna to the incoming radiation must also be determined, and this is comprised of bandpass and flux density calibration. This calibration can be done with only one source if the flux density of that source is known, it is a point source, and has sufficient flux density. Otherwise a point source with a high flux density can be observed as the bandpass calibrator (the source used for the delay calibration can usually be used for this purpose), and a source with a known flux density, or a planet can be used for flux density calibration.

Over the duration of the experiment, several factors that affect the antenna response will change. These include atmospheric phase, opacity, elevation-dependent dish surface distortions and focus changes, aperture illumination and spillover. To monitor these changes you will need to observe a point source nearby to your science target. This point source will need to have a relatively high flux density, although averaging over a short period of time (~1 minute) if this is not possible.

If you need to measure the polarisation of your science target, you will want to determine the leakage parameters: the fraction of linearly-polarised photons that are detected by the receptor that should not have been.

If you are observing at high frequency, where the primary beam of the antenna is small, it is important that the maximum response of the beam is directed toward the source you're observing. To ensure this is the case, you should do periodic pointing-correction scans on a nearby point source with reasonable flux density (>300 mJy).

Finally, if you're observing in the 3mm band, you will need to use the receiver's absorber paddle to keep track of how the system temperature varies over time. This is done periodically on-line using “paddle” scans.

A summary of what type of calibration you will require for each observing band is given in Table 2.3.

For more detail on each of these types of calibration, read the sections below. It is important to note that you must repeat all the calibration for every frequency configuration that you observe, as it is not generally possible to interpolate or extrapolate the calibration parameters determined from one frequency configuration to another.

The calibrator you use to calibrate the delays is largely dependent on the band you are observing in, and the LST when you are setting up.

For the centimetre bands (16cm and 4cm), delay calibration is best done with one of two sources. When the LST is between 11:00 and 04:18, you should use PKS 1934-638 to calibrate the delays, because it is has a large flux density and you'll need to observe it anyway for flux density calibration purposes. At all other times you should use 0823-500.

For the higher frequency bands, you will need to use one of the many available sources with a large flux density in the ATCA calibrator database, such as:

You should choose whichever is at the highest elevation at the time, or nearest to your targets. There are many more sources that have large flux densities in the millimetre bands, and you can search for those calibrators using the ATCA calibrator database search interface.

The bandpass calibration is best done with a source with a large flux density, since you're trying to precisely determine how each frequency channel responds to a known amount of flux density.

To do this ideally, the response should be dominated by the signal rather than the thermal noise in each channel. Since this is also a very useful property for a delay calibrator to have, the same calibrators that can be used for delay calibration (Section 2.2.2) are ideal for bandpass calibration as well. If the bandpass calibrators is used as the delay calibrator, then the bandpass calibration observation can be made immediately following the completion of the setup without further slewing overhead. If the observations are at 3mm, a paddle measurement should be made immediately before the bandpass observation.

As the receiver response does not change greatly as a function of time, for a continuum observation it is usually sufficient to make an observation of a bandpass calibrator at the very start of the allocated time. The amount of time spent integrating on the bandpass calibrator depends on its flux density, but generally 5-10 minutes is sufficient at all frequencies.

For spectral line observations though, it has been noted that the CABB bandpass is not perfectly stable over the course of a normal 12 hour observation, and for this reason it is advisable to observe a bandpass calibrator more than once per observation. Visiting it three times (once at the start, once at the end, and once sometime near the middle) during a 12 hour observation should be sufficient to obtain a reliable bandpass solution. Alternatively, provided the gain calibrator has a large flux density, the periodic visits to it may provide accurate enough information to determine how the bandpass solution changes over time.

The integration time spent on the bandpass calibrator will have an effect on the accuracy of the flux density that is measured after reduction. It is necessary to obtain each correlator channel's response to the same precision as required for the flux density calibrator, ie. obtain a signal-to-noise ratio of 100 per channel if the required flux density precision is 1%. Since any point source can be a bandpass calibrator, the sensitivity calculator should be used to find out the RMS noise per channel for your observations, and then a calibrator found from the calibrator database that provides enough flux density to meet the signal-to-noise ratio requirements. For example, a typical CABB observation has 2048 channels, and at 50 GHz, the RMS noise per channel is about 35 mJy for a 5 minute observation. For a signal-to-noise ratio of 100, a calibrator with flux density greater than 3.5 Jy is required, and a search through the calibrator database identifies 19 such sources. Of course, if a flux density calibrator is strong enough to be a bandpass calibrator as well, then observing time can be conserved by not having a separate bandpass calibrator.

Flux density calibration is required to translate the arbitrary gain scaling that is produced by an observation to an absolute flux density scale. The most effective way of doing this is by observing a calibrator that has a known flux density (on the absolute flux density scale) and comparing it to the sources that you are observing that have unknown flux densities.

For the ATCA, there are currently only four flux density calibrators, and of those only two are regularly used.

For frequencies between 1 GHz and 50 GHz, the preferred flux density calibrator is PKS 1934-638. It has a known, stable flux density, and conveniently has no linear or circular polarisation, at least at low frequencies. The flux density model for 1934-638 that is installed in Miriad are based on the models described in the memo

and the paper

The model is separated into two frequency ranges. Below 11 GHz, the Reynolds model is used, which is:

where $\mathrm{S}$ is the flux density in Jy and $\nu$ is the frequency in MHz. For frequencies above 11 GHzMiriad uses the model from the Planck study:

The high-frequency model has been shown by Partridge et al. (2016) to match quite well with the VLA, but unfortunately the low-frequency model does not.

For frequencies higher than 50 GHz you can use 1934-638 as the flux density calibrator, but the planet Uranus is brighter. Its flux density is known to vary with time, but it does so in a way that is understood and can be modelled. The planets Mars and Neptune can also be used, but Uranus is preferred because its angular size is smaller than Mars (making it easier to observe with typical Compact Array baselines), and it is brighter than Neptune (which would require a longer scan to provide the same signal-to-noise level).

To be as effective as possible, the flux density calibrator should be observed when it is at the same elevation as the target source, and at as high an elevation as possible. Doing this means that any gain-elevation dependence is reduced, and the effect of airmass is also reduced. At low frequencies (below 7 GHz), these requirements are not as important as they are for higher frequencies, where the atmospheric effects become a significant factor. Indeed, many (if not most) centimetre observers simply make a scan on 1934-638 at the beginning of their observations, and this is usually good enough to get a flux density uncertainty of only a few percent.

Because 1934-638 is a point source (at least at the resolution of the Compact Array), it can be observed the same way as any other calibrator. That is, a scan on 1934-638 should be included in the schedule, and should be preceded by a pointing scan when it is observed with the 15mm or 7mm receivers.

Observing the planets is a little more complicated. Because they are not point sources on even the most compact of ATCA baselines, they cannot be used as a pointing reference. A nearby source must therefore be used to determine the pointing offsets before observing the planet. Also, because the planets will substantially fill the primary beam of the ATCA antenna, the apparent system temperature of the antenna will be increased while a planet is being observed. Because of this, if a paddle scan is made while observing a planet, the system temperature will not be correctly determined, and the flux scaling will be in error. When observing with the 3mm system therefore, the paddle scan should be made while observing the same nearby pointing reference calibrator.

To make a good observation of Uranus, follow the procedure:

The procedure is similar for the other planets. Note that because the phases of the nearby calibrator are not used by the Miriad flux calibration task mfboot, it is not absolutely necessary to bracket the planet with a calibrator observation. It can however be useful to have that data to diagnose potential problems, should they arise.

The integration time spent on the flux density calibrator is largely governed by the precision required for the observations to be successful, and the strength of the primary flux density calibrator at the observing frequency. If a flux density precision of 1% is required, then the aim should be to make an observation with a signal-to-noise ratio of 100. For a observation with 2 GHz bandwidth, this is relatively easy, as even a 5 minute observation of 1934-638 at 50 GHz gives a signal-to-noise ratio in excess of 1000.

The sensitivity calculator should be used to estimate the observing time required to achieve the desired flux density precision. However, it is never a bad idea to pad the required time out a little to ensure that enough good data will remain after flagging.

The gain parameters that vary significantly over time include:

As these factors vary, the path lengths between the antennas will vary (which varies the phase), and the system temperatures may change, thus changing the sensitivity. To accurately correct for these changes, they must be regularly monitored.

The best way to monitor these changing parameters is to observe a simple source (a point source in an otherwise empty field is the ideal case) every time the observing configuration (elevation, weather, etc.) changes significantly. The definition of “significant” change depends on the observing frequency and the observing conditions (such as array configuration, atmospheric seeing, temperature, precipitation etc.). More precisely, by observing a source that looks the same irrespective of time or (u,v) coordinates, changes in gain can be determined directly. Therefore, gain calibrators should be unresolved, with a stable flux density over the observation timescale.

The ATCA provides a calculator to advise observers on how often to visit their phase calibrator. For this calculator:

For example, for an observation at 19 GHz, with a seeing RMS of 300 microns, in a 750m array, the calculator determines that if the gain calibrator was visited every 2 minutes, the observations would have an RMS phase of 15°, be decorrelated by 3%, and have a maximum dynamic range of just 282. However it also determines that this is as bad as it gets, so that a 10 minute visit interval would give the same results. Since 2 minutes is a very frequent visit schedule, it would seem that self-calibration would be preferable for high dynamic range observations.

Dropping the seeing RMS to 100 microns (which would be a good winter day) improves matters significantly, allowing a dynamic range of 874 with no decorrelation and a 5° RMS phase for any visit interval greater than 2 minutes.

It is important however not to extend the interval too far, as then the atmosphere being sampled by the gain calibrator will be different to that the source has sampled during that time. For cm observations, one might need only visit the gain calibrator once or twice per hour in good conditions, while for mm observations, it is unwise to visit the calibrator less frequently than once every 15 minutes.

The ATCA has a large list of gain calibrators, spread over the viewable sky. This list can be queried using a web interface. This interface queries the database in one of three ways:

Selection of an appropriate gain calibrator can also be done while you are preparing the schedule in the CABB web scheduler (see Section 2.3.12 for an example of how this works). Ideally, choose the nearest possible point source calibrator to your target. Sometimes that calibrator may not have a high flux density though, and you may wish to choose a calibrator that is further away in order to lessen your overhead: a calibrator with a higher flux density can give a good measurement of the gain in a shorter time. For continuum observations this is largely irrelevant since the large CABB bandwidths allow for a very accurate measure of gain in just a few cycles. For spectral line observations though, the more flux density the source has the better.

The ATCA uses dual linear orthogonal feeds to measure the incident radiation. Ideally, these two feeds would be independent: in reality this is not the case, and a fraction of the radiation measured by one feed is also received by the other; this effect is called “leakage”. Polarisation calibration is primarily concerned with measuring the leakage fraction and the possibly non-zero signal phase (the so-called XY-phase) between the two feeds.

A paper by Sault and Cornwell (1999, in “Synthesis Imaging in Radio Astronomy II”, ASP Conf. 180, p657) showed that for an observation where a calibrator with apriori unknown polarisation properties can be observed with at least three different parallactic angles (what they call the “parallactic angle rotation trick”), polarisation calibration is required to determine the absolute alignment of linear polarisation angle and the term which causes leakage between Stokes I and V.

The absolute alignment of linear polarisations is achieved by injecting a linearly polarised signal at 45° to the linear orthogonal feeds several times per second: this is known as the “noise diode”. This signal is measured by the correlator on each antenna, and the XY-phase is used later in data reduction to determine the absolute linear polarisation angle. All the receivers except for the 3mm system have a noise diode on each antenna. For 3mm observations, CA02 has got the Mopra spare receiver installed, which has a noise diode, while all other antennas do not. Should Mopra require the spare at any point however, this capability may be removed from the ATCA; in this case you would need to use a different observational strategy to recover the absolute alignment information.

This means that to complete the linear polaristion calibration, any of the point source calibrators in the ATCA calibrator database can be observed at least a few times over the course of an Earth-rotation synthesis observation. Since this is likely being done for normal gain calibration, most observations will contain enough information to calibrate the linear polarisation.

If you can't play the parallactic angle rotation trick (if for instance you are observing for only a short time), then you will need to know the polarisation parameters for the calibrators you are observing. At low frequencies (in the 16cm band), it is known that PKS B1934-638 has zero linear polarisation. For short observations then, you can use the flux density calibration observations to calibrate linear polarisation as well.

Dave Rayner's “Circular Polarization User's Guide” is a useful guide on how to accurately measure circular polarisation with the ATCA.

The global pointing model for the ATCA antennas is usually good to a few arcseconds under ideal conditions, and the tracking accuracy is better than an arcsecond. But changes in temperature can make the pointing model incorrect by a few 10s of arcseconds, and this would mean that for the millimetre bands the pointing may be incorrect by a significant fraction of the primary beam size (see Table 1.1). This would adversely affect the sensitivity of the observations of your targets.

While observing, you can correct the pointing model to ensure that the peak response of the primary beam is pointing directly at your target. A pointing correction scan is made by observing a point source directly and at the half-power point of the primary beam in each of the azimuth and elevation axes, and comparing the amplitude seen at each of these points.

With 2 GHz of bandwidth, almost any point source in the ATCA calibrator database has enough flux density to allow determination of the pointing corrections, however the pointing corrections only really improve the global pointing model within 20° of the azimuth/elevation at which the corrections were determined. The natural choice of calibrator for this purpose is therefore the same as for the gain calibration, and in general you should use your gain calibrator for pointing calibration.

Because the pointing corrections are only valid with 20°, a pointing correction scan will be required approximately every hour while you're observing if you're tracking the same source as it crosses the sky, or more frequently if you're observing multiple sources across the sky.

The “paddle” is a mechanical arm that holds an absorbing material large enough to cover the 3mm feed horn. By observing the paddle, the temperature of which is assumed to be the ambient temperature in the antenna vertex room (which is monitored), and comparing it to the assumed temperature of the mostly empty sky, you get a system temperature that includes the effect of the atmosphere. This is sometimes referred to as the “above-the-atmosphere” system temperature, because in this case the system includes the atmosphere.

Because changes to the system temperature affect the response of the system to an otherwise equivalent amount of radiation, it is important to keep track of these changes as your observations progress. With the other receivers, the injected noise signal (the temperature of which is assumed to stay constant) is monitored constantly, and no overhead is incurred for sytem temperature monitoring. For observations with the 3mm receiver however, you will need to observe the paddle reasonably frequently and incur a couple of minutes overhead each time you do.

How often the paddle is used depends on the conditions. Since in principle you are using the paddle scans to interpolate the actual system temperature at any particular point in time during your observations, you will need to observe the paddle frequently enough to catch any departures from a linear change to the system temperature. If you are observing in good and stable weather, you can go 20 minutes between paddle scans. If the atmosphere is very turbulent though you may have to use the paddle more often.

It is important to remember that the system temperature calculation depends on the assumption that it will be looking at mostly empty sky after it looks at the paddle. For most point source calibrators this assumption will be valid, but if you are pointing at a planet, the system response will include its temperature, and this will affect the system temperature calculations.

Observations on the ATCA normally consist of a sequence of scans: a scan is a period of observing where the telescope configuration stays the same. A complete observation is made up of a number of scans. Observations typically alternate between scans on the scientifically interesting sources and scans on one or more calibrators. The configuration of the scans and their order are kept in a schedule file.

The main observing task, caobs, reads the schedule file to determine what sources are to be observed, for how long, and in what order. The schedule file also defines the frequencies and correlator setup for each scan and can also be used to make the system automatically execute commands. The ATCA web scheduler is the tool to use to create ATCA schedule files in the CABB era. Schedules made with the old sched program are no longer understood by the CABB version of caobs and thus cannot be used. However, it is possible for the web scheduler to read in old schedules and save them for use with the new system. Please also note that an active ATNF username and password is required to gain access to the web scheduler page.

Generally, users will want to prepare at least two schedule files:

You may also want to prepare contingency schedule files in case your preferred frequency is seriously affected by interference, or the selected calibrator proves to be resolved.

If your goal is to make a high-quality image of a source that has an angular extent much smaller than the size of the primary beam of the ATCA antennas, then you should observe your source in “snapshots”. The same strategy can be applied if you want to image a small number of sources that are separated in the sky.

Snapshot observations will allow you to obtain a wide range of uv coverage without having to spend all your observing time on source. This is necessary because you will need to obtain calibration data within your time allocation. Generally, a snapshot consists of one or more calibration scans, followed by some integration time on the science target, followed by a gain calibration scan. These snapshots are repeated until your observations are complete.

Snapshot observing is best illustrated with examples.

If the source you will image is bigger than a small fraction of the ATCA primary beam, or the region you want an image for is considerably larger than the primary beam, you will need to use the ATCA's mosaicking functionality.

In interferometry adjacent pointings are not independent and so we can get fundamentally better images by processing the different pointings together. Necessary information about the pointing grid pattern to use for proper Nyquist sampling and the time to be spent on each position to optimise tangential uv coverage can be found in Chapter 21 of the Miriad Users Guide.

For close-packed mosaicking (pointing centres every half-beamwidth), drive times are determined by the acceleration limit of 800°/min². Half the drive time is spent accelerating and half decelerating. For drive times much further than 108′ in azimuth or 27′ in elevation, the drive times are dominated by the slew rate limits of 38°/min in elevation and 19°/min in azimuth. For example, a 16′ drive for a 16cm mosaic will take, depending on the elevation and azimuth of the source, approximately 2.2 seconds. Data taken during the off-source period is blanked in the correlator using a predicted drive time. As drive times can be a significant overhead, care should be taken to not slew the antennas more than necessary.

Mosaicking is less useful if frequency switching is desired, as no frequency changes can occur during the mosaic scan. Since slew times are often much greater than the time it takes for the correlator to switch frequencies, the overheads involved in repeating a mosaic scan for each separate frequency might quickly negate the normal benefits of mosaicking.

Mosaicking mode can also be useful when a large number of nearby sources are to be observed, as observing overheads are reduced. Other applications include holography and source surveys using one dimensional cuts.

When scheduling a mosaic observation, the “snapshot” strategy should be used in order to get calibration data. That is, a point source calibrator should be observed periodically between observations of the mosaic pointings.

If you are mapping a very large, regularly-shaped region at low frequency, where the slew time between each pointing centre is dominated by the maximum speed of the antenna rather than the acceleration/deceleration rate, then you may get a significant reduction in overhead by using on-the-fly mosaicking.

On-the-fly (OTF) mosaicking has a couple of important differences to traditional mosaicking. Most importantly, although the mosaic pattern should be broadly the same – the hexagonal pattern is still the optimal sampling – the order of observations should be such that the telescope scans in a straight line for as long a distance as possible.

During an OTF mosaic, the phase center follows the mosaic points as specified, and the pointing center moves continuously across the pointings. In effect the primary beam of the antennas becomes elongated (to 1.5 beams) in the scanning direction.

To satisfy proper sampling of the sky, the antennas should not move by more than half a beam per integration cycle, so the spacing of pointings in the mosaic file should be the same as before. Because caobs assumes you are scanning in rows or columns, the direction of scanning is inferred from the next point. At the last point of a row or column the previous direction is maintained to complete the scan. One extra point, marked with the name “TURN”, needs to be specified at the start of the next row/column, to allow the antennas to change direction and scan at the correct speed again.

The primary advantage of OTF mosaicking over regular mosaicking is its speed in covering a large area of sky, making it possible to get good uv coverage of a large field in a single observing run. It will not usually provide any better sensitivity than a carefully-crafted regular mosaic.

Although OTF mosaicking is most useful at low frequencies where the slew time between mosaicking points is a significant fraction of the total integration time, this mode could also be used for mm observations by using bin mode and moving the antennas by e.g., 16 beams in a single integration with 32 time bins. The Miriad program atlod will need some work to properly record the phase center and pointing center in this case (ie. such a mode is not currently supported, but may be if demand for it is high).

Mosaic observations use a standard schedule file and an additional file, called the mosaic file, which lists the positions of the mosaic field centres. Up to 2048 fields can be specified in a single file. This mosaic file must consist of one line per pointing centre in the mosaic pattern. Lines beginning with the “#” character are ignored and can be used for comments.

The general mosaic file format is:

# Comments  .... ignored by the software
#
d(RA)  d(DEC)  INT  $FIELD_1
d(RA)  d(DEC)  INT  $FIELD_2
d(RA)  d(DEC)  INT  $FIELD_3
d(RA)  d(DEC)  INT  $FIELD_4

The items in each line are:

This file should be named “mosnam.mos”. The “.mos” suffix is required, or caobs will not find the file. The “mosnam” part of the name must match the source name given in the schedule file.

To include a mosaic in a schedule file, insert a scan and set:

All other scan settings have their usual effect, including the “Pointing” mode, and the observing frequencies. The mosaic file must be placed in the /atca/mosaic directory on xbones so that caobs can locate it during the observations. The mosaic file name must be completely in lowercase letters, even if the corresponding source name in the schedule has uppercase letters; caobs will report that the mosaic file could not be found unless it has an all-lowercase file name. The “.mos” suffix is required, or caobs will not find the file.

The file can be copied into position manually, but there is also a web interface that will ensure your mosaic files are put in the correct location.

The time specified in the “ScanLength” field in the schedule file will determine how many times the mosaic is looped through in a single scan. caobs calculates this by dividing the “ScanLength” time by the total length of the mosaic and rounding this to the nearest integer; this will be the number of times the mosaic is repeated during this scan. The mosaic is always observed at least once, regardless of the “ScanLength”.

It should be noted, in particular for mosaic files containing sources spread over a range of RAs, that if caobs encounters a source in that has set (or not risen), it will skip to the next scan in the schedule file, not to the next source in the mosaic file.

0.0000    0.00    2  $@1934-on C
-0.1812  -0.16    1  $TURN-10
-0.1812  -0.08    1  $1934-11
-0.1812   0.00    1  $1934-12
-0.1812   0.08    1  $1934-13
0.0000    0.16    1  $TURN-24
0.0000    0.08    1  $1934-23
0.0000    0.00    1  $1934-22
0.0000   -0.08    1  $1934-21
0.1812   -0.16    1  $TURN-30
0.1812   -0.08    1  $1934-31
0.1812    0.00    1  $1934-32
0.1812    0.08    1  $1934-33

How large the mapping region is will determine whether you can image using standard snapshot observations, or if you will need to use mosaicking. In either case, the imaging strategy is no different to that of a continuum observation, but you may need more frequent calibration to ensure you get a sufficient signal to noise ratio for your lines; CABB's extra bandwidth doesn't help you observe a single line at known redshift!

If you're looking for your spectral line in one of the CABB continuum IFs, one of the most important things to consider is where the line to be observed will fall within the bandpass, ie. which channel(s) it will appear in. There are several channels in the CABB continuum IFs that are always flagged, and several more that are often contaminated with self-generated interference, and these should be avoided when observing spectral lines.

The CABB correlator always flags the centre continuum channel and each of the quarter-band channels. If you select a CABB mode that provides 1 MHz continuum channels, the flagged channels will be 513, 1025 and 1537. For CABB modes that provide 64 MHz continuum channels, the flagged channels will be 9, 17 and 25. You should avoid making the central CABB continuum frequency the same as the expected frequency of the line.

Self-generated interference will likely be present in the 1 MHz continuum channels 129, 157, 257, 641, 769, 1153, 1177, 1281, 1409, 1793 and 1921. All these channels can be easily flagged with the online correlator software (and they routinely are) and the interference does not leak into the adjacent channels.

If you are using the CABB zoom bands to detect your spectral line, you should again avoid using one of the continuum channels listed above, especially the quarter- and half-band channels. The zoom bands derived from those channels will usually have a very large spike at the centre of the zoom, and this spike may leak into adjacent channels since the zoom channels are not perfectly square. However, since the zooms come from the interleaved channels, the zoom channel number is derived from the continuum channel number by multiplying by 2 and subtracting 1. For example, channel 17 in the 64 MHz continuum band is actually zoom channel $17\times 2 - 1 = 33$.

It is important to note that ATCA is not able to Doppler track. You will need to ask for enough frequency coverage to ensure that your line will not drift out of the band due to observatory velocity changes.

Calibration of spectral-line observations is done with the same “snapshot” strategy that was described in Section 2.3.2. It is important to note however that the calibration itself is slightly different. Because accurate measurements of a spectral line requires good bandpass characterisation, but in general you are dealing with much smaller bandwidths than for continuum observations, you will need to observe calibrators with higher flux densities, observe the calibrators for longer, or observe the calibrators more often in order to get a sufficient signal-to-noise ratio.

There are many restrictions on the frequencies that can be observed with the CABB system. The most basic restriction is that only two 2048 MHz IFs can be observed simultaneously. All of the additional zoom bands must lie within these IFs, so they do not count as “extra” bandwidth. You can only specify the central continuum frequencies to MHz precision.

If you want to use the entire array for a single band, then there are band-specific restrictions on the frequencies that can be observed.

If you don't need all the baselines in the same band, it is possible to split the array up so you can observe in two bands simultaneously. For each of the two CABB continuum IFs you can select a set of antennas to use with the caobs command “set band1_antennas”. In this way you can specify a frequency in IF1 that is in a different band to the frequency specified in IF2. The antennas specified in the “band1_antennas” would use the IF1 band, while the rest of the array would use the IF2 band.

The primary restriction in this mode is the central site oscillator. Since there is only one central site oscillator that feeds the entire array's L86 modules, if you want to observe in more than one mm band simultaneously, both the frequencies must be observable with the same central site frequency.

Since neither the 16cm or 4cm bands need the central site synthesiser, there are no special frequency restrictions when one of the simultaneous bands is 16cm or 4cm.

It isn't possible with CABB to observe more than 4096 MHz of bandwidth simultaneously, but it is possible to quickly switch between frequencies within a single band, and also to swap between different bands with only a moderate loss of simultaneity.

If you want to change frequencies, you will need to start a new scan in the schedule. This restriction makes it impossible to change frequencies within a mosaic. Whenever you change scans you will experience two integration cycles of overhead, and if you are changing frequencies within a band, this is the only applicable overhead. For example, if you changed your central continuum frequencies from 5500 & 7500 MHz to 6500 & 8500 MHz, you will only need to incur two cycles of overhead as the scan is changed.

There is also no extra overhead when changing the zoom band configuration, which is simply viewed as a change of frequency within the same band.

If you are changing bands, you will likely incur more overhead. In this case, the turret on the antennas will need to rotate to put the new receiver at the right position, and if you are changing to or from the 7mm band the mm receiver package will need to be moved using the translator as well. Turret rotations will usually add only a few cycles of overhead at most, since the turret can be rotating during the normal two cycle overhead that occurs at every change of cycle. Translating the receiver to go to or from the 7mm band takes much longer: between 2 and 3 minutes.

When you change bands you may also have to refocus the antenna by moving the subreflector. When going to or from the 4cm band, all antennas need to be refocused, and this takes approximately 90 seconds. When going to or from the 3mm band, antenna CA01 needs to be refocused, and will take approximately 40 seconds.

WARNING: Movement of the turret does produce wear on the bearing, and it would be extremely difficult to replace it should it fail. We therefore ask that observers not change bands more than once every 15 minutes.

To schedule an observation of a major planetary object, simply put its name in the “Source” field of a scan. During the observations, caobs will interpret this name and calculate where the planet, Sun or Moon is and point the array towards it.

The scheduler will not automatically calculate where the planet will be during the scheduling process, unless you enter the appropriate date and LST or UTC and then click on the “Listing” tab. At this point, the scheduler will determine the planet's position and fill the “RA” and “Dec” fields. Alternatively, you could use the Miriad task planets.

It is useful to note again that the “RA” and “Dec” fields are ignored by caobs when the source name is recognised as a planet, the Sun or the Moon, as caobs will compute the appropriate position for the body in real time. There is no reason therefore to prefer absolute time scheduling when observing planets.

Solar system objects differ from most astronomical sources in that the proper motions are large enough that the object position may change in the course of an observation. caobs has the ability to phase track an object with a non-sidereal rate, and command the antennas to correctly track it as well. If the object moves across the sky too quickly however, the antenna speed limits may prevent you from tracking the object.

For minor bodies, caobs cannot compute a position, so the correct position must be provided by the user. The scheduler treats source names beginning with the @ symbol (eg. a source name of @hbopp) as pointing to an external ephemeris file (eg. a file $ATCA_EPHEM:hbopp.eph).

Ephemerides for minor solar system bodies can be computed before scheduling using JPL's Horizons on-line system. Only the telnet (terminal-based) access method is described here, but there is also a web-interface method available from the web page that is self-explanatory.

Horizons asks the user a series of questions. To generate an ephemeris file for ATCA observations:

The ephemeris file produced by Horizons needs to be massaged into the format that the scheduler requires. This is best achieved with the ephformat command available on xbones.

To use the ephemeris you just created in the schedule, enter the name of the ephemeris (the “file” name described in the last dot point above), preceded by the “@” symbol, into the “Source” field of the web scheduler.

There is a limit to how far the antennas can slew in azimuth. The limit arises because of all the cabling inside the antenna which passes up the centre of the structure to the receiving and cryogenic equipment in the vertex room. There is a limit to how far these cables can wrap around in either direction (similar to the head of an owl).

The wrap limits for the ATCA antennas are CCW=152.5° and CW=327.5° as shown in Figure 2.1.

Figure 2.1. Wrap limits for the Compact Array and Mopra antennas.

It is a good idea to check what wrap the telescope is in at the start of your observations, as “unwrapping” can take up to 10 minutes. To determine the wrap you are in, consult MoniCA or atdrivemon. MoniCA shows the wrap explicitly, on the “ATCA Summary” page. atdrivemon shows the azimuth using the ATCA convention — while in the South wrap azimuths are expressed as positive angles, and negative angles in the North wrap.

The CABB web scheduler is used to prepare schedule files. The web scheduler checks the schedule for completeness and, where possible, checks choices such as observing frequency for hardware compatibility. Drive times, and source azimuths and elevations can also be computed to enable you to estimate overheads and observability.

Three steps are involved in using the web scheduler:

When constructing a schedule, consider:

The web scheduler can produce a listing to help answer these questions. It is a good idea to study these listings very carefully before observing commences. The scheduler considers drive times: all coordinate entries are automatically followed by a calculation of the azimuth and elevation of the source, as well as the drive time from the previous source, using the entered value for LST. When making a schedule file, the integration times are specified for each source (and calibrator source). Note that, except in the case of mosaic schedules and Dwell scan types, these integration times include drive times, so the specified time must be somewhat longer than the amount of integration on source required. This is particularly important for gain calibrators, when the drive time (up to a minute or so) may be comparable to the integration time (one to three minutes). Alternatively, the ScanType field can be set to Dwell to ensure a source or calibrator is tracked for the specified scan duration.

The schedule file submitted to caobs (the online Compact Array control task) need have no specific start time. Rather, the schedule file asks caobs to start the first scan as soon as possible. The sequence of scans will be exactly as specified in the schedule file. However, when using the web scheduler, the calculation of azimuths, elevations and drive times is possible only if the starting sidereal time is known (at least approximately). The sidereal time for the first scan in the schedule can be set in the web scheduler by giving the time in the Time field, and setting TimeCode to LST. Alternatively, the starting UT may be specified by setting TimeCode to UT, and by setting the Date to the appropriate start date by clicking on it and using the calendar that appears. Note that when viewing any scan other than the first in the schedule, the fields Time, TimeCode and Date all become read-only.

The environment flag can be used to instruct the online software to remember certain settings, such as attenuators, and to recall them whenever a scan with the same environment flag is encountered. However, this is generally not required, as the online software will remember user settings, and recall them when the same conditions are encountered again. There are at least 128 slots for user settings, so for a normal schedule, the environment flag will probably not need to be used.

See Appendix F for a complete guide to all the functionality available in the web scheduler.

Yes! In this section a simple continuum schedule file is constructed. It will contain three sources: a program source and two calibration sources. The following observing sequence will be specified:

These sources will be observed at the standard 4cm band frequencies of 5500/9000 MHz. To create this schedule file, perform the following steps (instructions are given in bold, and the results of the instruction, or important caveats, and described below it):

To produce a listing of the schedule, in order to check that the schedule proceeds as expected, click on the Listing tab.

Yes! This example adds to the example from the previous section; ensure that the schedule created in the previous section is in the scheduler before continuing.

Now the aim is to observe 1 zoom band in each IF, each covering at least 70 km/s. The galaxy “test” is actually at an LSR recessional velocity of 25000 km/s, and these observations will look for CH₃OH methanol and HC₃N cyanoacetylene lines.

To create a schedule that can perform suitable observations for this situation, follow the steps:

At this point, the schedule would correctly describe a zoom mode observation.

The data rate during CABB observations can be calculated with the formula:

where

The data-rate per second is thus:

where $t_{\mathrm{cycle}}$ the cycle time of the correlator.

A few example calculations are given below. In all cases it is assumed that $\mathrm{b}_{\mathrm{corr}}=8$ bytes (in all modes, 4 bytes are recorded for each of the complex visibilities, and 4 bytes are recorded for the weight component), and $t_{\mathrm{cycle}}=10$ seconds.

Since all modes give you the continuum IFs, you start off with one of two base data rates, per cycle. For the 1 MHz modes, the base data rate is:

bytes per IF. For the 64 MHz modes, the base data rate is:

bytes per IF.

For each pulsar-binned IF you have (only available in 1 MHz mode) you add:

For each single zoom you have (i.e. zooms with a width of 1) you add:

For each consolidated zoom you have (i.e. zooms with a width of 2 or more) you begin by adding:

Then, for each zoom band in the consolidated zoom, you add:

For example, if you had a consolidated zoom with a width of 5, it would produce

bytes per cycle.

Example: If you were using the 1 MHz/64 MHz hybrid CABB configuration, and had one zoom of width 1 and one consolidated zoom of width 12 in the 64 MHz IF your total data usage would be:

Doing the sum gives $\mathrm{b}_{\mathrm{cycle}} = 1.192224\times10^{7}$ bytes per cycle. Assuming a 10 second cycle time, this would generate 3.997 GiB per hour, which would give you roughly one new CABB data file every hour.

Users of the ATCA are expected to organise the execution of the observations. Any member of the team that proposed the experiment is eligible to observe from the Marsfield SOC, or any other researcher that the primary investigator might like to nominate. If the researcher has not previously observed with the ATCA and CABB, or has not observed within the past 12 months, they will have to travel to the SOC to be trained.

Remote observing is available for experienced ATCA observers. We will consider granting a request for remote observations if the observer:

Applications to do a remote observation must be made at least 2 weeks before the observations. This is so observatory staff can check for possible network disruptions during your observations, and advise you if remote observing will not be possible. The later you leave your remote observing application, the less time you will have to change your plans should your request for remote observations be refused.

Permission to do your observations remotely is given at the discretion of science operations staff, and you should not presume that permission will be forthcoming. Should permission be refused and you are not able to come to the observatory, you may need to find a local collaborator to do the observations for you. The duty astronomer is under no obligation to observe for your project if you are unable to.

To apply to make an observation remotely, use the “Booking” tab on the ATCA observing PORTAL, which will look something like Figure 2.2. To book for a slot, click on the start time and then drag out a box to the end time of the booking. A booking box will pop up, and you can adjust the start and stop times if you need to.

Figure 2.2. The “Booking” tab on the ATCA observing PORTAL.

Remote observers must have their own ATNF UNIX account and remember their password for this account.

To be sure of a successful remote observing session, all observers are required to test their remote observing setup at least a week before their observations.

At a minimum, you will require the following for remote observing:

Remote observations will be easier if you also have:

To test your setup, you should contact the duty astronomer and ask to be allowed to login and perform the test. If the test will not disturb current observations, you should follow the instructions on remote observing found in Chapter 3 to ensure that you can monitor and control all the required VNC sessions. If problems are encountered here, and the duty astronomer can't help you resolve them, contact the observatory staff with a description of the problems and ask for advice.

The ATCA rapid response mode is only available to proposers who have submitted a NAPA proposal for the current semester. If you are interested in being granted permission to use the rapid response mode for your NAPA proposal, please talk to Jamie Stevens (<Jamie.Stevens@csiro.au>).

The rapid response mode is designed to take a user-made schedule, and get it onto the telescope as soon as possible after the request has been made. Getting the telescope to stop one schedule and start another is very easy; determining when "as soon as possible" actually starts is the tricky part. The rest of this section describes the process which takes place to perform an over-ride observation.

All scheduled slots that have a project code starting with “C” can be over-ridden by the rapid response mode, provided that your TAC score is sufficiently high and CABB is in the correct mode.

Blocks of time scheduled for reconfigures, maintenance and VLBI cannot be over-ridden.

Any over-ride project that has a score higher than or equal to the lowest score given to a scheduled project may over-ride any other scheduled project. The theory for this is that if we knew at the beginning of the semester when a NAPA would trigger, and that NAPA had a score about the lowest score that was actually scheduled, we would have attempted to schedule it at the appropriate time and scheduled the displaced project at another time, regardless of the score of the displaced project.

As part of the project to develop this rapid response mode, a library for creating and manipulating CABB schedules was developed. It can be found on GitHub, and is free for use.

Because many of the existing libraries that astronomers use to deal with transient event streams are written in Python, the choice was made to focus first on providing a Python library for schedule manipulation. This library is targetted toward Python 2.7, but will work with 3.8.

Documentation on how to use the library in your own code is part of the GitHub repository.

To ensure that only those projects that have been pre-approved can perform over-rides on the ATCA, any request for over-ride time needs to be accompanied by a valid authorisation token. This can only be obtained by emailing <Jamie.Stevens@csiro.au>.

There are two types of authorisation token available. The first is a test token, which allows you to submit a request, and go through the entire process of the over-ride, but not actually get time on the telescope. This is designed to be used while you are trying to develop and perfect your automated trigger routines. You may issue as many requests as you like with a test token.

The second token will allow your request to allocate time on the telescope. This type of token will not be issued until you have demonstrated that your code can properly submit a test request. You will need to work with Jamie Stevens during this process.

Authorisation tokens are valid only for a specific project and semester, and will be rejected if used outside those bounds.

The authorisation token itself is a JSON Web Token. You will be supplied with a normal ASCII file that will need to be passed back to the service whenever you make a request.

All requests to trigger an over-ride observation are submitted through a web service. The exact format of the request won't be documented here, and we suggest that you use the Python code we provide when you are building your trigger response software.

The library which makes it easy to submit over-ride requests can be found on GitHub, along with documentation and examples.

If your request is unable to be scheduled given the bounds you provided, you will receive your rejection in the response from the web service.

If the web service doesn't reject your request, your schedule will enter the queue to be executed. At this point, two emails will be sent to this effect. As with all emails sent by the rapid response service, they will be from Jamie.Stevens@csiro.au and will go to all the members of the NAPA team (as specified on its OPAL coversheet), along with all the members of each project that will need to be displaced.

The first email will have the subject “ATCA rapid response mode triggered for Cxxxx ”, where of course the real NAPA project code will be specified instead of “ Cxxxx”. The second email will have the subject “Your ATCA rapid response over-ride request has been queued for execution.”.

The first email will look something like this.

As you can see, the time of the request, and the expected start and end times are all listed, along with the unique ID assigned to the over-ride, and a list of all the projects that will be displaced.

The next email will look similar to the following.

This second email introduces the time-stamped log of the progress of the over-ride. All subsequent email communication will contain this log, as it was when the email was sent. The log will have entries for all the important actions made by the rapid response service, and entries that will be useful for later data reduction.

At the start time, the service will either start or delay the schedule, and in either case an email will be sent. If the schedule is started, the email will have the subject “An ATCA rapid response over-ride is now being observed”, while if the schedule is delayed the subject will be “Your ATCA rapid response over-ride request has been delayed”; the reasons for the delay will be included in the log. If the schedule is started, the name of the schedule will be given in the email, and you may load it in the CABB web scheduler to see how the service has altered it.

No emails will be sent while the schedule is running, except if the service encounters a problem. In that case, you will get an email with the subject “There has been a problem while executing your ATCA rapid response over-ride”.

When the schedule finishes, the service will send out one final email, with the subject “ATCA rapid response over-ride is now complete”. This email contains the full log, which will include the name of the RPFITS files that contain the data.

If for some reason the service is unable to perform the scheduled over-ride (for example, if the antennas are weather stowed for an extended period of time), then the schedule might expire and be removed from the queue. If this happens you will get an email with the subject “Your ATCA rapid response over-ride request has expired”.

The schedule you make depends on what type of observation you are trying to perform. What follows are some general recommendations for rapid response schedules.

The service shortens a schedule in four stages.

If at the end of this process, the schedule length falls below the specified minimum useful time required, the start time will be delayed and the schedule restored to its original specification.

The subreflector on each of the ATCA antennas can be raised and lowered to allow the signal to be focussed at the receiver feed. However, changing the subreflector position will change the path length (obviously) which will change the geometric phases and delays, so, it should only be done when required, usually when the frequency changes. Extra care is needed with calibration with programs that change focus during the observation.

A good time to do this is at the start of the observing and it can be done while driving to source.

Default focus positions have been determined for each band.

Note: As the antenna tips over, it deforms slightly, and the detected focus position changes by up to 1mm. This is normal and will not significantly affect the data.

The rms digitiser quantisation levels (as displayed in the sampler tab of cacor; see Figure 3.2) should be close to 20 (Range: Approximately 15-25). To set the levels, attenuation can be added or removed. The values in the Atten section of the window show the settings of the cacor attenuators. They have a range of 0-15 dB.

Figure 3.2. cacor sampler level display

For 16cm observations, strong RFI in the band will often make the sampler levels vary significantly from cycle to cycle, making it difficult (and often impossible) to set the attenuators appropriately. It is therefore recommended to skip this step when observing at 16cm. However, in rare cases the attenuator settings shown in the Atten section could be quite low (all values far below 7), and at those times it could be useful to reset them either automatically or manually as described below.

For all other bands, if the levels are not close to 20, the ATTS button in the cacor GUI can be activated to enable cacor to automatically add or remove attenuation.

This can also be done in the cacor interface:

Wait for the levels to settle to around 20, then turn the adjustment off again with:

Fairly obviously, the algorithm is unable to “improve” the levels if the attenuators are at their limits (0 or 15).

As noted above, strong RFI in the 16cm band makes it difficult to set the attenuators appropriately. High values of attenuator settings, even up to the maximum value of 15, are not harmful. However, you can also set the attenuators manually to values that are known to be good, as listed in Table 3.2. To set the attenuators manually, you will need to use the att command. For example, to set the CA03 attenuators as per Table 3.2, you would need to give the commands:

See attf ca0# n m for more complete information about how to use this command.

If the sampler levels are very low (approx 0), there is probably a problem which will need to be checked.

There are another set of attenuators in the receivers (as opposed to the correlator) which can also be used to set the levels.

These are controlled from caobs.

cacor generates an average visibility (either the mean or the median across the spectrum) that is used for T_sys and array calibration. In general, the median is much more stable and allows for a more reliable calibration. Otherwise, care needs to be taken to ensure that the range of cacor channels used for calculating the average is as wide as possible within the useable passband, but also excludes interference.

The default behaviour is to compute the average across the spectrum. To ensure that the more stable and reliable median is being used, it is recommended to enable the tvmedian option:

At higher frequencies and when the tvmedian option is enabled, it is usually appropriate to use the central fifty percent of channels. The easiest way to do this is to type

Alternatively if tvmedian is disabled (not recommended), the information below can be referred to.

At cm frequencies, interference means that it is often better to select non-default channels, when tvmedian is disabled (not recommended). See Table 3.3 for information on recommended channel ranges for 2048 MHz continuum bands. If these channel ranges are not generating the anticipated visibility in vis, inspect Appendix C to determine the cause of the problem.

The tvchan command sets the range of channels that are used for calibrating the array and for T_sys calculation. The same range is used to compute the average (or median) visibility for display in vis. All unflagged channels are written to the data (RPFITS) file regardless of what the tvchan setting is.

(Note: Channels that are flagged are written to the file as zeroes. See Section 3.3.5 for more information).

Current settings are displayed in the cacor GUI, as shown in Figure 3.3.

Figure 3.3. cacor display showing the currently used tvchannels.

It is also possible to check the current setting with the cacor command

Setting the range is done with the cacor command:

Table 3.3 lists the suggested tvchan ranges for the standard continuum bands. On any given day (or interference environment) other ranges may be more appropriate.

(There is a wide range of valid frequency settings at mm frequencies).

To calculate the delay of an individual IF, cacor checks the phase in each channel in the tvchan range (just defined) and calculates an average linear slope from the phase measurements as a function of frequency; this slope is the delay.

If the noise on the phase is high (due to a faint source or poor weather, for example) or there is some strong unavoidable RFI in the tvchan range, then the average slope may not accurately represent the real slope due to the delay.

To reduce the noise, or to average out the effect of the RFI, setting delavg in the correlator averages together (or takes the median of) the corresponding number of channels in the tvchan range before calculating the slope to determine the delay.

Here too it is useful to use the median option rather than the default mean (computed within each group of delavg channels):

Higher values of delavg can be useful for determining the delays. However, care needs to be taken when selecting this value! If the phases are wrapping quickly, averaging over a large number of channels will give a meaningless average phase, and therefore incorrect delays will be determined.

The current value of delavg is in the status bar of the cacor interface.

Figure 3.4. cacor showing current delavg setting

Setting delavg is done with the cacor command:

Figure 3.5.		Figure 3.6.
An spd plot demonstrating slow changes in phase. The phases are not wrapping at all in the A pol, and wrapping approximately every 250 channels in the B pol. Averaging over 32 channels would still provide enough points across the band to achieve a useful linear fit to the data, so delavg can be set to such a large value.		An spd plot demonstrating rapidly wrapping phases. The phases in both pols of this graph are wrapping across a small number of channels. It is possible to check just how quickly the phases are wrapping by viewing a smaller number of channels: spd> chan f# 400 500 The value of delavg will need to be set to a small number in this case.

Occasionally, there will be channels that, because of interference (or other issues), you may wish to completely excise from the data (i.e. not write those channels to the data file).

To check the extent of currently flagged channels:

Figure 3.7. cacor display showing the current number of channels that are unflagged.

There is no easy way to discover which channels have been flagged, and flagging will remain even with a change of project (or frequency etc.) so flagging should be checked at the start of observing. If there are more channels flagged than anticipated, flagging should be cleared then reflagged if necessary.

This is done by:

To flag the set of known bad continuum channels

This needs to be done at both f1 and f2.

To flag explicit channels, use the command:

Again, this needs to be done at both frequencies.

If you know exactly which channels have been flagged, they can be unflagged with the funflag command: formats are similar to the fflag command i.e.

Once the array is working, the system can be calibrated.

Calibration will not fix a broken system.

The calibration involves calculating and applying the delay offsets, checking the pointing (for mm observations), setting the phases to zero degrees and correcting the visibility amplitudes to the appropriate flux levels.

Figure 3.8. vis display showing output during ATCA calibration

3.3.6.1. Calibrate the Array Delays

Delay offsets need to be determined and corrected at the start of the observation. Failure to correct for delay offsets may result in decorrelation, and loss of data.

In the 64MHz zoom configuration, delay calibration is more complex as it generally needs to be done in two steps. (First delay calibrate a zoom channel, then in the continuum. See Section 3.3.6.1.1 for more information)

To calibrate the delays, track an unresolved source with large flux density (e.g. PKS1934-638 or PKS0823-500 for cm observations; for mm observations use PKS1253-055, PKS1921-293, PKS2223-052 or PKS0537-441). PKS1934-638 and PKS0823-500 do not have enough flux density to calibrate at mm frequencies.

Correlation should be apparent before you proceed. Phases in spd should be stable, though they may be wrapping rapidly and delays should be flat in vis. Check that you have correlation in both frequencies and both polarisations.

If there is significant interference (see the amplitude plot in spd), you will need to carefully craft the tvchan range (or, if bad enough, actually flag interfering channels) to avoid the interference in the calibration data.

If working at 3mm or with significant interference, it may be necessary to average over several channels (to improve the signal/noise ratio) with the command:

caobs > corr delavg #

where 8 is a good initial value for # - however, if the delays are large (>50nsec), you should try a smaller value to start with.

If you still don't have correlation, check Section 3.5.6 at the end of this chapter, and if this doesn't resolve your problems get help.

Assuming that the phases in spd are stable, enter the command:

caobs > corr dcal

dcal calculates and applies the required correction.

This command should correct the delays so that the phases across the band on the spd plot are flat (or fairly close to flat) and the delay offsets as seen on vis are close to 0 nsec. It may be necessary to do this again.

(Early versions of the CABB software required an “a” switch at the end of the dcal command for the correction to actually be applied. The delays are now applied automatically by the command. Any switch other than an a will result in the delays not being applied.)

3.3.6.1.1.  64MHz Delay Calibration

Delay calibration is slightly more complex in 64MHz mode.

The issue is that typical delays can cause the phases to wrap more than once within a single (64MHz wide) continuum channel. As is the case when delavg is set too high in 1MHz mode, a valid delay solution cannot be determined in this circumstance. The solution for this is to obtain an initial solution for the delays using zooms, which will unwrap the phases sufficiently to be able to determine the delays using the continuum channels. The process is illustrated in Figure 3.9 to Figure 3.13.

To use the zooms for delay calibration, set calband to z z:

cacor > calband z z

Uses zoom data in both frequencies. For a hybrid correlator config, use f z

Do the delay calibration as normal. At this stage a large value of delavg is often acceptable.

caobs > corr dcal

As always, ensure that there is a working system before doing the dcal

Figure 3.9.		Figure 3.10.
An spd plot demonstrating typical phase variations across a zoom band. In this case the delays are large enough that the phases wrap multiple times across the zoom band, and thus are also wrapping multiple times within the corresponding 64MHz-wide continuum channel.		An spd plot showing the zoom band phases after delay calibration. After performing delay calibration on this zoom band, the phases have been flattened across the 64MHz range.

Once the majority of the delay error has been corrected for, the whole continuum band needs to be used for a second delay calibration:

	cacor > calband f f
	caobs > corr dcal

Pay attention to the value of delavg at this step to ensure that it is sufficiently low. There are only a handful of channels to work with in the continuum bands in this mode!

Figure 3.11.		Figure 3.12.		Figure 3.13.
Continuum band phase variations before delay calibration. This is the continuum band corresponding to the zoom band shown in Figure 3.9. It is clear (especially for the A polarization) that before any delay calibration, the average phases within each 64MHz continuum channel are not useful to determine a delay solution.		Continuum band phase variations after zoom-band delay calibration. After performing delay calibration on the zoom bands, the phases have been largely flattened. The delays determined across a 64MHz frequency range are not sufficient to fully calibrate the continuum band, but they are sufficient to allow a refined delay calibration over the full continuum frequency range.		Continuum band phase variations after final delay calibration. After performing a further delay calibration on the continuum bands, an excellent solution has now been obtained. This is apparent from the flatness of the phases across the full frequency range.

It is important that sufficient calibration data is taken to characterise the array. This means taking flux density and bandpass calibration data for all observations, and leakage calibration data for polarisation observations.

This does not have to be done at the start of the observation, but should be done at some point during the observing session. If appropriate calibration sources are not above the horizon during the observing session, contact the scheduler (Jamie Stevens) to discuss options.

For cm, 15mm and 7mm observations, PKS1934-638 is the appropriate flux calibrator and can also usually be used as a bandpass calibrator (although spectral line observations at 15mm or 7mm will ordinarily require a stronger source). For 3mm observations, a planet is needed for flux calibration, and a source with large flux density (e.g. PKS1921-293 or PKS1253-055) is required to calibrate the bandpass.

Bandpass calibrators should ideally be observed at high elevation (above 40 degrees) if possible. This is especially important for excellent polarization calibration.

The anemometer at the observatory measures local wind speeds every 2.5 sec. There are two (independent) systems that stow the array in the face of high winds:

Observers should respect this and not restart observing until the wind speeds have died down. In fact, if PMon has issued the stow you will have to wait 15 minutes until after the stow condition is no longer present before the antenna will be allowed to drive again.

In addition to high winds, electrical storms present a real danger to the ATCA. Storms and lightning strikes within a 56 km radius of the array are closely monitored. Threat levels (from 0 - no threat, to 4 - high threat) are assigned based on the number of approaching lightning strikes and those within 9 km of the array. When the threat level reaches 3, the back up generators are turned on. At a threat level of 4, xbones stows the array.

On very hot days (above 40°C), especially with a strong wind, the cryogenic compressors can become overwhelmed. In order to protect the receivers, if the ambient temperature in the internal system of the scroll compressors on the antennas (mm systems) reaches 91°C, xbones stows the array and turns the antennas to shadow the compressors from the wind.

Weather monitors (for wind, storms and heat) can be accessed in MoniCA. Alternately, the ATCA Weather Station also provides information about the weather.

If observing in the ATCA control building, you may hear alarms when certain problems occur. If you are observing remotely, you may ignore this section.

The first action in the face of an observing problem should always be to wait for at least a minute. (There are obvious exceptions to this rule, e.g. fires and similar catastrophic events that require quick action.) If waiting doesn't fix the problem, stop, restart and wait for at least another minute.

The following flowcharts provide details on how to recover from some of the most common observing problems. They explain what to do in the case of No Correlation (or funny correlation), Antenna(s) not 'on source', VNC Problems, and Rebooting a CABB block.

If you are unable to fix or diagnose your problem after following the flowchart you should call your Duty Astronomer. Sometimes you may be able to diagnose your problem but not be qualified to execute the required fix and in this case you should also call your Duty Astronomer. This may be the case if you need to reboot a CABB block, for example.

When we talk about “unattended observing”, we mean any period during your experiment where you are not immediately aware of the state of the telescope. You may be any distance away from your observing computer, or asleep, or otherwise engaged in another activity, but you are listed as the “observer in charge” on the ATCA Portal.

The observatory does not require you to be responsible for the safety of the telescope equipment or of personnel. We have sophisticated systems and procedures which should handle any unsafe situation that might arise during a normal observation. For example, the automated systems will turn on the generators when a storm is approaching, and turn them off when it passes. It will also automatically handle stowing in periods of high wind, or when there is a high chance of lightning strike.

We do however expect that you are solely responsible for monitoring the quality of your data, and we do expect to be able to contact you via the phone number you supplied when you registered as the observer in charge. You should consider how often you need to check your data quality, given how much time you could lose if something goes wrong.

For example, if an antenna stops tracking due to a drive fault, you may call the duty astronomer and have them call the on-call person to remedy the situation; it might take only 10 minutes to rectify the problem if you were immediately aware of the fault. If you were asleep at the time of the fault however, you may lose a lot more than 10 minutes, and you would be ineligible to be compensated for that time. We recommend that even while sleeping through an observation, a periodic check is made every couple of hours.

We have developed several systems to proactively reach out to observers when something goes wrong that might need urgent action. This section describes the unattended observing alarm notification system available for the ATCA.

3.6.3.3. Twitter feed

The same alarms that will be displayed in the MoniCA alarm manager will also be turned into tweets on the ATCA Twitter feed. This should allow you to be a long way from your observing computer and still be alerted to problems with the telescope. Whenever an alarm is generated by MoniCA, or an alarm status goes away, a message is posted to the Twitter feed describing the problem.

Figure 3.14. Some example tweets from the ATCA alarm Twitter feed.

A couple of example tweets are shown in Figure 3.14. Each tweet starts with the UTC date and time that the alarm state changed. The date is listed as year-month-day, but where the century part is omitted, to save characters. All alarms are given the hashtag #ALARM; this may currently seem redundant, but this Twitter feed will likely be expanded to include problems that do not generate MoniCA alarm states.

If an alarm has been triggered, then the next part of the tweet will be a hashtag describing how important the alarm is. An #INFORMATION alarm simply presents a condition report, but likely does not require any action. #MINOR alarms can be ignored until it is convenient, where you should probably investigate #MAJOR and #SEVERE alarms as quickly as possible.

If an alarm state has gone away, there will be no severity hashtag.

The rest of the tweet is then some guidance text about which system has caused the alarm. The guidance text is not designed to help you respond to or correct the problem; training on how to deal with problems that arise with the ATCA is still required. Rather, the tweet is there to quickly alert you to the problem.

The Twitter feed is still under active development, to make it as useful for observers as possible. All feedback about your experiences with the Twitter feed is welcome, as are suggestions on ways in which it can be made more useful.

You can copy your data to your laptop from the Marsfield SOC. When your laptop is plugged in to a pink cable, you will have to use an SSH tunnel through venice to access the area on cetus where your data resides.

To set up the SSH tunnel, type the following in a terminal on your computer:

$ ssh -L 2222:cetus:22 usr123@venice.atnf.csiro.au

This makes all traffic directed at port 2222 on your computer get transported through venice to cetus via the usual SSH/SCP port of 22. This tunnel will exist as long as you remain logged into venice in this terminal. Thus to copy your data across, type the following in another terminal:

$ scp -P 2222 “usr123@localhost:/DATA/NEWDATA_1/ATCA/archive/*.CX167” .

which would copy all project CX167 data, and where usr123 is your ATNF username.

If you are a CSIRO staff member and can connect your laptop directly to the ATNF network via a blue cable, you may copy the data directly from cetus:

$ scp “usr123@cetus:/DATA/NEWDATA_1/ATCA/archive/*.CX167” .

Your CABB data will appear on the Australia Telescope Online Archive (ATOA) within a few days of your observations. Using the OPAL account you used to submit your successful proposal, you will have access to your data. Full instructions for how to use the archive interface are available on the website.

If you are unable to get your data from the ATOA as soon as you would like, then you may get it directly from the disk on cetus that buffers the data.

To do this, you must set up an SSH tunnel from your local machine to cetus via the ATNF gateway machine venice. This can be set up with the command:

$ ssh -L 2222:cetus:22 usr123@venice.atnf.csiro.au

where usr123 is your ATNF username. This will log you into venice and set up the SSH tunnel, which will remain until you log out. To copy data from cetus you should go to another terminal and do something like:

$ scp -P 2222 “usr123@localhost:/DATA/NEWDATA_1/ATCA/archive/*.CX167” .

There are now several software packages that are capable of reducing CABB data. For the purposes of calibration, the Miriad software package is still probably the best choice. It is possible though to do a full reduction without really using Miriad, if you prefer to use the CASA package. Below is a high-level comparison of what each package is capable of doing, and this may help you decide on how to proceed.

This section will describe how to get started on reducing CABB data using the Miriad software suite. More information can be found in the Miriad manual. The manual's cookbook should be read by all users before starting their first data reduction. This section will summarise how reduction differs for CABB data, and give a reasonable guide on how to do a simple continuum reduction. The Miriad manual has been updated to incorporate information on CABB data reduction.

$ mirsync -u usr123

Task:   atfix
vis      = 3mm_data.uv
select   =
out      = 3mm_data_fixed.uv
dantpos  = 
tsyscal  =
options  =
array    =

Task:   uvsplit
vis      = c007.uv
select   = -shadow(22.5)
options  =
maxwidth =

4.3.2.5. Flagging

In most cases, if the birdie and rfiflag options are given to atlod, the data will be mostly free from the most destructive RFI at this point. However, you should always examine your data with the uvplt and uvspec tasks to make sure.

Using uvplt, examining amplitude and phase versus time is a good way to find time ranges that have been affected by RFI. Ensure that you specify options=nofqav to prevent uvplt from averaging your data over frequency before plotting. Any RFI present in your observations may be restricted to only a few channels, and these channels can be identified by using uvspec.

To flag this data, you can use the following tasks.

uvflag

Use the select and line keywords to select the bad data and set flagval=flag to flag the data as bad. uvflag can be used to flag any set of bad data, but can be tediously repetitive.

blflag

This task can replicate many of the plots that uvplt generates, which may allow you to directly select the outliers representing bad data.

pgflag

This task replaces the old tvflag task that no longer works on displays with more than an 8-bit colour depth. pgflag displays data as a channel-time waterfall plot and allows direct selection of ranges of bad data and flexible flagging options. It also includes an automatic flagging routine that uses the SumThreshold method that is available in the AOFlagger, and is highly recommended for use during continuum reductions.

One trap that you may find is that if you are observing at 3mm and some other band in the same run, and you elected to keep CA06 tracking for the 3mm observations, the data from CA06 will not automatically be flagged. This is because CA06 has almost all the systems required for 3mm observing (save for the electronics), and so the antenna really is “on source” according to the telescope. You should use uvflag to flag out CA06 data in this case.

Automatic flagging with pgflag can be very effective if it is done in the right way. It is most useful for continuum observations in any band, and for spectral line observations when the line is not detectable in a single integration. If your spectral line is bright enough to be detected in a single integration, then the flagging routine will likely identify it as RFI.

The pgflag task accepts a number of parameters that control the automatic flagging process. Craig Anderson identified, through trial and error, a recipe that works quite well for 16cm data; this recipe is described in an ATCA Forum post Craig started. We describe this recipe here as well, and we recommend this recipe for data in any band.

This flagging recipe uses the concept that the polarisation properties of interference are quite different to that of the signals we're interested in. For the most part, the radio sky is free of discrete sources of circular polarisation, and thus most measurements of Stokes V from our interferometer should be thermal noise only. This assumption holds over all frequencies as well. So if we see a large deviation from thermal noise in Stokes V, and this deviation is confined to a narrow range of frequencies, the most likely culprit is RFI. The first step of the flagging recipe then is to look for such signals in Stokes V and flag all the visibilities over the frequencies and times they are seen. To do this with pgflag:

$ pgflag stokes=i,q,u,v flagpar=8,5,5,3,6,3 "command=<b"

This flagging can be done on each individual source after it has been split out. Of course, because the output of the interferometer should not vary for Stokes V regardless of the field being observed, you can also flag a single IF all together. That is, you can take a dataset output by uvsplit with options=nosource and flag it directly without further splitting it into individual sources. You can also do the flagging before and after calibration. Indeed, you may have success in doing an initial round of Stokes V flagging before trying to mfcal the bandpass calibrator, and another round of flagging with the same command after calibration.

As an illustration, Figure 4.1 and Figure 4.2 show what pgflag will display when told to show Stokes V from a dataset centred at 2100 MHz (a typical 16cm dataset) with fourteen scans on nine different sources, obtained in October 2015. Channel number is shown along the horizontal access, and as the 16cm band is presented to the correlator as a lower sideband, higher channel numbers correspond to lower frequencies. The time of the observation increases along the vertical axis towards the top of the image and the amplitude of each channel is represented as a grey scale which is outlined in purple toward the right of the image. Flagged regions are shown as completely black. Figure 4.1 shows a 60m baseline and Figure 4.2 shows a 4469m baseline. The only flagging done to this dataset is at the edges to remove the roll-offs, which removes 10.5% of the data on each baseline, and the data are uncalibrated. These two figures clearly show how much more sensitive short baselines are to RFI than long baselines.

Figure 4.1. A pgflag display showing Stokes V data on a short baseline in a 16cm dataset before flagging.

Figure 4.2. A pgflag display showing Stokes V data on a long baseline in a 16cm dataset before flagging.

After using pgflag as described above on the uncalibrated data, then bandpass calibrating and flagging with pgflag in the same way again, the data on the same baselines look like Figure 4.3 and Figure 4.4. Approximately 40% of the short baseline data is now flagged, and approximately 25% of the long baseline data.

Figure 4.3. A pgflag display showing Stokes V data on a short baseline in a 16cm dataset after flagging and bandpass calibration.

Figure 4.4. A pgflag display showing Stokes V data on a long baseline in a 16cm dataset after flagging and bandpass calibration.

As you can see, no obviously wrong signals can be identified in Stokes V after this process, but that does not necessarily mean that all RFI has been deleted from the data. Figure 4.5 and Figure 4.6 show Stokes Q on the same baselines as before, after the Stokes V flagging and bandpass calibration, and further RFI is obvious in those figures. We now flag using the linear-polarisation Stokes parameters Q and U.

Figure 4.5. A pgflag display showing Stokes Q data on a short baseline in a 16cm dataset after flagging in Stokes V and bandpass calibration.

Figure 4.6. A pgflag display showing Stokes Q data on a long baseline in a 16cm dataset after flagging in Stokes V and bandpass calibration.

To do this we use pgflag as follows. First, flag Stokes Q:

$ pgflag stokes=i,v,u,q flagpar=8,2,2,3,6,3 "command=<b"

And then Stokes U:

$ pgflag stokes=i,v,q,u flagpar=8,2,2,3,6,3 "command=<b"

Again, these flagging commands can be done on each source individually, or on the dataset as a whole.

An example of how effective this flagging recipe can be is shown in Figure 4.7 and Figure 4.8. Using the calibration solutions derived from the flagged dataset, we can compare the 16cm spectrum of the calibrator 1953-325 before any flagging (Figure 4.7) to the spectrum after flagging (Figure 4.8). After the flagging process, only a small amount of RFI remains, and this is not likely to cause problems when measuring the continuum spectrum. On the shortest baselines, between 45-50% of the data has been flagged, but on the baselines to antenna 6 less than 30% of the data needed to be flagged. In terms of channels, 1625 of the channels have usable data out of the 2049 total channels, representing a spectral coverage loss of approximately 21%.

Figure 4.7. The 16cm spectrum of the source 1953-325 without any flagging.

Figure 4.8. The 16cm spectrum of the source 1953-325 after automatic flagging with pgflag.

4.3.2.6. Calibration

The calibration strategy for ATCA data has been refined for the CABB era, to ensure accurate calibration of wide bandwidths. This strategy also has the benefit of being applicable to the vast majority of ATCA data with little alteration. It is described best in the Miriad cookbook, but it will also be summarised here (although not explained in the same depth).

4.3.2.6.1. Bandpass calibration

Calibration begins by determining the bandpass correction, using mfcal with your bandpass calibrator. To prevent rapid gain variations from being entangled with your bandpass solution, the interval parameter should be set to the cycle time (which is usually 0.1 minutes). Since your bandpass calibrator should be very bright, such a short interval should still have plenty of signal-to-noise. By default, mfcal uses CA03 as the reference antenna, but if that antenna wasn't available during the observations, ensure that you set refant to be an antenna that was available.

If you are reducing 3mm data, and CA02 was available, you should set it as the reference antenna. You should do this because it has the only noise diode in the array at 3mm, and setting it as the reference antenna may make it easier to do polarisation calibration.

cm reduction	mm reduction
Task: mfcal vis = 1934-638.2100/ line = stokes = edge = select = flux = refant = minants = interval = 0.1 options = tol =	Task: mfcal vis = 1921-293.17000/ line = stokes = edge = select = flux = refant = minants = interval = 0.1 options = tol =

Bandpass calibration in the cm bands (16 & 4) is usually done with the ATCA primary flux density calibrator 1934-638, as it has a large flux density at these frequencies. Since the flux density of 1934-638 is known as a function of frequency, the bandpass solution obtained in these bands is very good and takes into account how 1934-638 varies across the band. But if you use another calibrator for which Miriad doesn't have a flux density model, then the bandpass solution will be dependent on how that calibrator's flux density varies with frequency, which is a bad situation. You will need to correct for this, and how to do this correction will be described later on.

4.3.2.6.2. Gain and polarisation calibration

After bandpass calibration, the next step is to do some time-variable gain and polarisation calibration. If the bandpass calibrator is 1934-638, or if the bandpass calibrator has been observed several times during the run over a large range of parallactic angle, then you can use it for this step. Otherwise you should use the gain calibrator you observed.

cm reduction	mm reduction
Task: gpcal vis = 1934-638.2100/ select = line = flux = refant = minants = interval = 0.1 nfbin = 2 tol = xyphase = options = xyvary	Task: gpcopy vis = 1921-293.17000 out = 2243-123.17000 mode = options = Task: gpcal vis = 2243-123.17000/ select = line = flux = refant = minants = interval = 0.1 nfbin = 2 tol = xyphase = options = xyvary,qusolve

You should use an nfbin of at least 2 for all bands. This will allow gpboot to correct any linear slope errors present in the bandpass table. This also makes it unnecessary to use mfboot unless your flux density calibrator is a planet. You should also increase the interval parameter if the calibrator is not bright enough to give enough signal-to-noise with 0.1 minutes of integration time.

4.3.2.6.3. Flux density calibration

4.3.2.6.3.1. Calibration in the cm bands

If you used 1934-638 in the mfcal and gpcal steps, then you already have a flux calibrated dataset that can be used to fix the flux density scale on the other datasets.

You should now copy the solutions from the 1934-638 dataset to the gain calibrator dataset using gpcopy, and determine the time-variable gains with gpcal.

Task:   gpcopy
vis      = 1934-638.2100/
out      = 2243-123.2100/
mode     =  
options  =

Task:   gpcal
vis      = 2243-123.2100/
select   =  
line     =  
flux     =  
refant   =
minants  =  
interval = 0.1
nfbin    = 2  
tol      =  
xyphase  =  
options  = xyvary,qusolve

The gpcal step is likely to make the flux density scale on your gain calibrator slightly different to the true scale. To fix this you should use gpboot.

Task:   gpboot
vis      = 2243-123.2100/
cal      = 1934-638.2100/
select   =  

You can now be confident that your gain calibrator is properly flux calibrated, and can now proceed to use it to analyse your science target.

4.3.2.6.3.2. Calibration in the mm bands

At this point in the mm reduction, you need to do the flux density calibration. How this happens depends on whether you're using a planet as the flux density calibrator, or 1934-638. After copying the calibration tables from the calibrator used in the gpcal step to the flux density calibrator, you use gpcal followed by gpboot if you're using 1934-638, or mfboot for planets.

1934-638 flux calibrator	planetary calibrator
Task: gpcopy vis = 2243-123.17000/ out = 1934-638.17000/ mode = options = Task: gpcal vis = 1934-638.17000/ select = line = flux = refant = minants = interval = 0.1 nfbin = 2 tol = xyphase = options = xyvary,qusolve,nopol Task: gpboot vis = 2243-123.17000/ cal = 1934-638.17000/ select =	Task: gpcopy vis = 2243-123.33000/ out = uranus.33000/ mode = options = Task: mfboot vis = 2243-123.33000,uranus.33000/ line = select = source(uranus) flux = mode = clip = device = /xs options =

Note the use of the nopol option for the options parameter in gpcal, when 1934-638 is being used as the flux density calibrator. This option tells gpcal that it should accept the leakage solution obtained from the gain calibrator, which should be good since the gain calibrator is likely to have much greater hour-angle coverage than the flux density calibrator.

In May 2016, Miriad's high-frequency flux density scale for 1934-638 was changed to align the ATCA's scale to that of the VLA and the absolute scale used by Planck. See Section 2.2.4 for details about this new scale. By default, after May 2016, Miriad uses this new scale when setting the flux density of 1934-638 for frequencies above 11 GHz. The flux densities measured from reductions using pre-May 2016 and post-May 2016 Miriad will therefore differ for observations above 11 GHz. If you would like to keep using the pre-May 2016 flux density scale for 1934-638, you should use the oldflux option that is available in the mfcal, and gpcal commands.

At this point the gain calibrator will be reasonably well calibrated. If you are not especially concerned about flux density calibration accuracy, you can proceed to analyse your science target.

If you want to flux calibrate your data as well as possible, then you can use data from the other simultaneously observed band to improve your solution. Basically, the bandpass solution you made earlier assumed that the bandpass calibrator had a constant flux density with frequency (i.e. a flat-spectrum source). Because of this, the bandpass solution will likely be in error. Our gain solutions with two bins will have corrected this somewhat, but we can also try to emulate the situation in the cm band, and provide mfcal with a model of the flux density of the bandpass calibrator.

To do this, we copy the gain solutions from the gain calibrator back to the bandpass calibrator, recalibrate, and bootstrap the flux density scale.

Task:   gpcopy
vis      = 2243-123.17000/
out      = 1921-293.17000/
mode     =  
options  =

Task:   gpcal
vis      = 1921-293.17000/
select   =  
line     =  
flux     =  
refant   =
minants  =  
interval = 0.1
nfbin    = 2  
tol      =  
xyphase  =  
options  = xyvary,qusolve,nopol

Task:   gpboot
vis      = 1921-293.17000/
cal      = 2243-123.17000/
select   =  

Note that these steps don't change regardless of the type of flux density calibrator you are using.

You should do the same calibration process with the other simultaneously observed band. At the end of the process you should have two bandpass calibrator datasets, each being flux density calibrated via the gain calibrator. You can now measure the flux density of the bandpass calibrator over the entire range of observed frequencies, using the uvfmeas task. The principle here is that even if the slope over a single band is wrong due to the incorrect bandpass calibration assumption, fitting a flux density model using two independently-calibrated bands should lower the uncertainties and provide a very good model of the actual flux density of the source.

Task:   uvfmeas
vis      = 1921-293.17000/,1921-293.19000/
select   =  
line     =  
stokes   = i
hann     =  
offset   =  
order    = 1
options  = plotvec,log,mfflux
yrange   =  
device   = /xs
nxy      =  
log      =  
fitp     =  
feval    =  

The uvfmeas task does a linear least-squares fit to the log of the vector-averaged Stokes I flux density as a function of the log of the frequency. The plot it makes will show the averaged data in both bands, and the line of best fit (Figure 4.9). It will also output a string like:

MFCAL flux= 9.4126, 17.0,-0.3801

Figure 4.9.  A plot made using uvfmeas, showing the line of best fit for the flux density model.

The string is usable as the flux in mfcal, which will fit the bandpass solution while assuming the flux density model we just determined. The image in Figure 4.9 is actually the result of this two stage process, and you can see the excellent alignment between the two bands, which are independently processed.

Once you have measured the flux density model for the bandpass calibrator, you must redo the calibration, but the bandpass calibrator becomes the flux density calibrator. The process is slightly different as well, and is summarised below.

Task:   mfcal
vis      = 1921-293.17000/
line     =  
stokes   =  
edge     =  
select   =  
flux     = 9.4126,17.0,-0.3801
refant   =  
minants  =  
interval = 0.1
options  =  
tol      =  

Task:   gpcopy
vis      = 1921-293.17000/
out      = 2243-123.17000/
mode     =  
options  =

Task:   gpcal
vis      = 2243-123.17000/
select   =  
line     =  
flux     =  
refant   =
minants  =  
interval = 0.1
nfbin    = 2  
tol      =  
xyphase  =  
options  = xyvary,qusolve

Task:   gpcopy
vis      = 2243-123.17000/
out      = 1921-293.17000/
mode     =  
options  =

Task:   gpcal
vis      = 1921-293.17000/
select   =  
line     =  
flux     =  
refant   =
minants  =  
interval = 0.1
nfbin    = 2  
tol      =  
xyphase  =  
options  = xyvary,qusolve,nopol

Task:   gpboot
vis      = 2243-123.17000/
cal      = 1921-293.17000/
select   =  

Task:   mfboot
vis      = 2243-123.17000,1921-293.17000/
line     =  
select   = source(1921-293)
flux     = 9.4126,17.0,-0.3801
mode     =  
clip     =  
device   = /xs
options  =  

At this point both the bandpass and gain calibrator datasets will be very accurately flux density calibrated. Simulations suggest that this two-stage process can improve the flux density calibration accuracy to the 1% level for the bandpass calibrator, and to around the 3% level for any other calibrators referenced to it. Observations have confirmed this in the 15mm band, while for the 7mm band the flux calibration accuracy for an arbitrary gain calibrator is around 10%.

4.3.2.6.4. General advice

The calibration technique outlined above is robust and generally applicable, but may not work adequately for your data without alteration, and will almost certainly not work in an automated fashion. You should always be prepared to iterate between flagging and calibration, and you should use the visualisation tasks uvplt, uvspec and closure to check that the calibrator visibilities look as they should. You can also look at the calibration solutions directly using the gpplt task. Using uvfmeas to view the calibrators in both simultaneously-observed bands may also give you confidence that your calibration is good, or otherwise.

Task:   invert
vis      = obs.uv
map      = obs.map
beam     = obs.beam
imsize   = 3,3,beam
cell     = 
offset   =
fwhm     = 
sup      = 
robust   = 0.5
line     =
ref      =
select   = 
stokes   = i
options  = mfs,sdb
mode     =
slop     =

Task:   mfclean
map      = obs.map
beam     = obs.beam
model    =
out      = obs.clean
gain     =
cutoff   =
niters   =
region   = percentage(33)
minpatch =
speed    =
mode     =
log      =

Observers are requested to acknowledge the ATNF in any publications resulting from the use of the ATCA as follows:

We also encourage an acknowledgement of our Indigenous community as follows:

The global research community is increasingly adopting persistent identifiers to improve how resource use is cited and tracked. CSIRO has adopted the Global Research Identifier Database (GRID) for its National Facilities and Collections, and the ATNF acknowledgement statements now include these. The GRID code for all ATNF telescopes is grid.421683.a. The use of a persistent identifier ensures our facilities are consistently referenced and described globally, and individual Digital Object Identifiers (DOIs) such as ORCIDs support the inclusion of GRID codes.

Information about how to acknowledge other ATNF facilities is available on our website.