2dF Frequently Made Mistakes

It was suggested that a potted summary of the most frequent howlers be prepared, to assist observers planning for a run.

Summary of FMMs.

Not really reading the FAQ and FMM

Guide stars not from the same catalogue as the targets

Guide stars are too bright.

Guide stars have unreliable magnitudes.

No sky positions in input file

Bright stars in the configuration or the range of target brightnesses to large

Blue central wavelength to short, requiring offset sky observations

Spectrograph arms don't overlap, difficult to splice spectra

Requested single configuration exposure to long

Not allowing time for reconfiguration

Not enough guide stars.

Forgetting to put enough sky fibres in.

Not allowing enough time for standard stars.

Turning up at 4pm on the first night of 2dF observing.

Not having run configure before arriving to check your proposed setups look reasonable.

Putting TOO much effort into CONFIGUREation before the run.

Common gotcha's running 2dFDR

Nod and Shuffle observations

Complex configuration strategies for multiple re-observations of a field

The full story on the FMMs.

Not really reading the FAQ or FMM
Most people, on their first AAOmega run, make one of the mistakes outlined here or in the FAQ. If one reads these instructions carefully, and consults with the support astronomer early, most problems have simple solutions. This will improve your data quality.

Guide stars not from the same catalogue as the targets
The choice of guide stars is critical to the success of AAOmega. The most fundamental point to make is that the guide stars simply must be on the same astrometric system as the target sources. Simply using two independent catalogues which claim to be J2000 will not give good results. If they are not then one may get excellent acquisition of the guide stars, but miss the science targets. The astrometric accuracy has to be good to ~0.3arcseconds. Most people use HST Guide Star Catalogue, or the APM catalogues. UCAC-2 or 2MASS have worked well in recent years, as proper motions are not yet a big issue. However, it is simply not good enough to just pick stars at random from these catalogues. The star coordinates need to be as measured from the same data from which you pick your targets. With SDSS this is rather easy, but do watch out for errors in the data in SDSS for bright stars. While this may all sound rather harsh, it is a condition of observations with AAOmega and complaints such as "but I don't have any way of getting better astrometry for my guide stars" will fall on deaf ears and will probable see your program rejected at the proposal stage.

Guide stars are too bright.

Bright guide stars (<13th mag) have high proper motions and should not be used unless they have had their positions measured in the last few years, or are PM corrected. They also suffer from bad centroids due to halo and diffraction spike effects which cause errors if positions are measured from sky survey plates. DON'T USE THEM. A case in point are UKST Bj plates some of which date from the 1970's. Better to use more recent R plates and to go fainter. AAOmega can work to 14-14.5th mag for guiding in dark of moon. In bright of moon we need stars in the range 12-13th mag and so you will have to put some work into guide star selection. Again, accurate astrometry with appropriate guide stars is a requirement for AAOmega observations.

Guide stars have unreliable magnitudes.

You may think it is OK to have dodgy magnitudes for your stars - after all they are only guide stars aren't they? WRONG. If your guide stars are too bright they will have bad positions and guiding will be difficult. We have frequently had people turn up with guide star magnitudes 1-2 mags out. How did this happen? They applied their galaxy photographic calibration to their stars. Additionally, all guide stars should have a small range of magnitudes (<0.5mag). A large range of guide star magnitudes will cause a dynamic range problem with the guide camera meaning that only some of the stars from a given set can be used, the brightest or faintest being rejected.

No sky positions in input file

While it is possible to add a grid of sky positions to your configuration using the CONFIGURE software, if you don't add sky positions to your input file then it is remarkably easy for the field to be configured WITHOUT any sky fibres. This makes data reduction next to impossible. Sky positions should be added to your .fld file. If you really want the standard uniform grid then it can be saved from configure (using File->lists) and pasted into the .fld file. However it is far better to add real blank sky positions, which you have checked are blank using your input imaging data. It only takes one 1st magnitude star in a sky fibre to really ruin your data.

Bright stars in the configuration or the range of target brightnesses to large

With 392 fibres all projecting onto the CCDs, inevitably there is some light from each target scattered across the whole CCD. AAOmega has been designed to minimize the effects of scattered light and reduce this crosstalk between different targets to an absolute minimum. However, it is not a good idea to allow the range of targets magnitudes in any given configuration to grow to large. A range of <3mags is typically best. While it is tempting to include some bright stars in any configuration, to allow simultaneous calibration of data, the magnitudes of such objects should be kept close to those of the targets to avoid scattered light compromising the science data. There can be no hard and fast rule on what magnitude range is acceptable, it depends on what information is to be extracted form the spectra.

Blue central wavelength to short, requiring offset sky observations

The dichroic wavelength has been selected to place the strong airglow line at 570nm on the red end of the blue arm. If the central wavelength of the blue are is set to short, particularly at high resolution, then one will loose the sky line in the blue spectra. This may well be what is required for the science program, particularly at high resolutions in the blue, however, it means that offset sky observations will be required to allow fibre relative throughput calibration and sky subtraction. The user should make sure that the choice of central wavelength is really what is required.

Spectrograph arms don't overlap, difficult to splice spectra
AAOmega is a dual beam system, with red and blue spectrograph arms. The standard dichroic change over is at ~5700A. In most default modes of operation at low and medium resolutions, the system set to allow a small overlap between he two arms to splice the full spectrum together. The user should be certain that splicing is not requirer before requesting a central wavelength which does not give an overlap between the two arms. See also the note on the 5577A sky line in the blue arm.

Requested single configuration exposure to long

The atmosphere acts as a giant, time variable (due to changing Hour Angle, HA) chromatic lens. The 2dF top end is equipped with an Atmospheric Dispersion Corrector (ADC) which corrects for the chromatic component, but atmospheric refraction changes the plate scale of the 2dF field plates as a function of HA (or rather Zenith Distance, ZD) and there is no way to account for this stretch of the field (Differential Atmospheric Refraction, DAR) during an observations. Each 2dF configuration has a specified mid point for which the field is setup, accounting for the effects of DAR at that time. For a full 2degree field, a configuration is typical valid for +/- 1hour either side of this mid point. Smaller fields are valid for longer, fields at high airmass are valid for short time periods. Losses due to fibre position mismatchs (and note this is in no way an error within 2dF, it's is the earths atmosphere which is at fault) can be significant outside of this time window. It is for this very reason that 2dF was originally designed to be reconfigured every hour. An excellent paper discussing the full horrors of the effect is Newman 2002 PASP 114 918.

Not allowing time for reconfiguration

For a full field reconfiguration (i.e. to remove the old configuration and replace it with a new one) for relatively uniformly sampled fields (i.e. not to strongly clustered) the reconfiguration time is of the order 60-70mins.

Not enough guide stars.

The 8 guide fibres can not access the full field. If your stars are of uniform density you need about 20-30 per field to ensure all guide fibres can reach stars. While we can guide with 2 stars, we prefer 3, and there is little reason not to use all 8. Don't deallocate important science targets in order to get every last guide star, but equally, don't only use 3 guide fibres to avoid having to do a bit of extra work. Good field acquisition is the single most important factor when observing with AAOmega.

Forgetting to put enough sky fibres in.

If you don't put sky fibres in your configuration drcontrol can not do sky-subtraction! (Don't laugh, observers HAVE done this.) Also you need about 20-30 allocated sky fibres in order to get a good median sky spectrum. This means you need MORE than 20-30 BLANK sky positions in order to get good sky subtraction without impacting the allocation of science fibres.

Not allowing enough time for standard stars.

Allow 15 minutes per standard, and even longer if you want to observe the standard in multiple fibres. Surprisingly this is how long it takes in practice with all the slewing and offsetting to get the star down a fibre and observe it. Also note that while we can defocus the stars a little (and in fact we probably need to for bright stars to avoid saturation in the red before giving good blue counts) we cannot defocus over a 2degree field plate and so standards are typically done with a special configuration which will take 10-20mins to set up on the field plate. Standards should typically be fainter than 4th mag (and note that some of the Lick standards are not) and brighter than 12mag for low res work. At higher resolutions we have little experience with this yet. Much fainter and you need a rather long exposure time to get high count rates. Also, not that while it is possible to do good relative flux calibrations (spectral shape) it is not possible to do absolute flux calibration with fibres due to unknown variables such as fibre placement or seeing losses.

Turning up at 4pm on the first night of AAOmega observing.

We generally start setting up fields for the night at 2pm in the afternoon, especially on the first night of a new project. So all the field configurations must be sorted out prior to this time. As field configurations must be tweaked for the current astrometry model it is strong advised that observers turn up the night before.

Not having run configure before arriving to check your proposed setups look reasonable.

It can take some considerable time to prepare good 2dF configurations. For example having 400 targets in the central arcminute will not work due to crowding! It is best to download configure from the 2dF ftp site and prepare in advance.

Putting TOO much effort into CONFIGURE-ation before the run.

Indeed. The 2dF astrometric model on the telescope will differ ever so slightly from that before the run. You will no doubt lose 5-10 fibre placements due to collisions. It is not worth worrying about the finer details of configuration until the correct astrometric models are available. i.e. don't run 'Allocate' and then do manual tweaking too early! As long as we have rough guidelines in advance of numbers of objects configurable that usually suffices. Contact your Support Astronomer or 2dF observer if you have any questions.

Common gotcha's running 2dFDR.

DON'T run 2dfDR in it's own software directory!!! When 2dFDR start's up and is used to reduce data it creates scratch files based on templates in it's software directory. If you run 2dFDR in it's software directory you will probably corrupt it and have to re-install it from the AAO ftp site.

Make sure you have plenty of disk space both in the data directory where you are planning to run 2dfDR and in your home directory - 2dfDR needs to create scratch files in ~/imp_scratch. If it runs out of disk space 2dFDR will die, and the error message may not reveal the true cause of the error.

Nod and Shuffle observations

By using Nod and Shuffle observations, where by the telescope is nodded rapidly between the target and the night sky while at the same time shuffling the charge around on the CCD to retain the integrity of the observation, AAOmega has demonstrated Poission limited sky subtraction. However, with dedicated sky fibres it is possible to obtain 1% sky subtraction accuracies with AAOmega. Due to the way Nod and Shuffle must be implemented, there is an increase in read noise and background sky noise plys usual a reduction in the number of targets that can be observed. The upshot is that N+S observation will tend to give lower S/N spectra for a given exposure time when compared to dedicated sky fibre observations (by a factor of between sqrt(2) and 2). For short exposure (less than 4 hours) there is little to be gained form N+S observations. N+S is ideal multi-night observations when simple stacking of many dedicated sky fibre observations has generally been shown to be limited in ultimate depth by systematic error in the reduction process. Current investigations suggest that one could in principle integrate for ever using N+S and increase the sensitivity according to Poission statistics.

Complex configuration strategies for multiple re-observations of a field
Some observing programs require repeat observations of the same field in order to observe the required number of target objects. In such cases it is often desirable to pad the target list with lower priority targets as filler for individual observations, and then to cull higher priority targets form the list once they have been successfully observed. The best approach for setting this up can depend on the nature of the observing program. Rather than trying to engineer a complex set of setup files unaided, contact your support astronomer and explain your requirements. The chances are we have run such a program before and can offer advice on the best way to construct the input files.

Rob Sharp (rgs@aao.gov.au)