Spiral Overview | Instrument set up | Observing| Nod and Shuffle| Reducing Data |

SPIRAL Observing Manual

Elizabeth Corbett - February 2003

Spiral Overview

SPIRAL is an integral field unit (IFU) which uses fibres to feed a dedicated spectrograph. The IFU consists of 512 lenslets each imaging 0.7 arcsecs of sky giving a total field of view of 22 x 11 arcseconds. The spectrograph is stituated on the dome floor and fed by 18m of optical fibre. The SPIRAL commissioning report gives further information on the instrument design and pictures of the instrument as it was being assembled can be found here. There is also a AAO newsletter article containing a brief description of SPIRAL.

Instrument Setup

Detector: Both the MITLL2A and the EEV detectors have been commissioned with SPIRAL. The chips are binned by 2 in both directions giving a 1Kx2K image when read out and are orientated with the long axis in the spatial direction so that all 512 fibres can fit onto the chips. This means that the spectra are dispersed along the short side of the detector, which limits the wavelength coverage of SPIRAL.

All 512 fibres fit onto the MTLL2A but, since the EEV pixels are 10% smaller, only 460 fibres fit onto the EEV chip. This means that nearly two rows of the IFU fall off the ends of the chip reducing the FoV to 22 X 9.8 arcsec. The spectral coverge is also reduced by 10%.

Gratings: All the AAO's gratings can be used with SPIRAL except the 270R. A list of the AAO's gratings and their effeciencies can be found on the gratings page. The detector has 1024 pixels in the wavelength direction and the wavelength coverage of the various gratings for the MTLL2A detector is shown below.

Grating Dispersion#
(Å/pixel)
Wavelength#
range (Å)
Resolution*
(/)
per degree
change in grating angle (Å)
1200V,R 0.321 330 7572 308
600R 0.642 640 3786 600
316R 1.313 1340 1869 1200
300B 1.387 1460 1753 1260
# Values for the MTLL2A, reduce them by 10% for the EEV detector.
* spectral resolution with measured 2.7 pixel FWHM fibre image size at 6563Å

Changing Gratings:

This should only be done by the support astronomer or afternoon technician.

To change gratings, unscrew the plate at the back of the grating mount and remove the grating. This can be quite tricky as the grating is a tight fit in the mount. Slide the new grating into the mount. The arrow on the back of the grating should be pointing to the right as it goes into the grating mount.

Replace the plate at the back of the grating. The metal spacers go between the plate and the back of the grating. Align the 3 adjustment screws (one on each side at the bottom and one in the middle at the top) so they fit between the small screws on the back of the grating. The front of the grating should just be touching the small screws at the front of the grating mount.

Do not set the central wavelength before obtaining a good focus at zeroth order (see below)

Focussing:

This requires 2 people, one of whom should be the support astronomer.

There are micrometer screw gauges to focus/adjust the image on the detector.

At the moment we are not using a Hartman shutter to focus the spectrograph. Instead we have the following system.

OBSERVING WITH SPIRAL

See the throughput and sensitivity page for guide-line photon rates and formulae for calculating exposure times.

Restrictions on observing:

When Spiral is on the telescope, we can only observe 4-3.5 hrs to the West or East of the Local Sidereal Time . This restriction is imposed becuase the optical fibres go from the bottom of the telescope down to the spectrograph at the telescope base. When the telescope is pointing a long way over, the fibres will be stretched (which is bad) and may break (which is worse). It is thefore important not to exceed this HA limit.

When rotating SPIRAL by 90 degrees or more, someone must be in the cass cage to check that the fibre cable does not catch on anything. If the cable is caught and stretched the fibres will be damaged.

Starting Observations

Log onto the vax (aat40a) on left hand terminal as OBSERVER. At the prompts type the following commands

The correct CCD display panel should now appear. When the prompt changes to "idle" enter the following:

The image below shows how a typical observation of a relatively small object (e.g. a standard star) will appear on the XMEM display. Note how the bad column on the MTLL2A shows up as a straight line from top to bottom on the left hand side of the frame whereas the sky line next to it is curved. Click to display an expanded version.
Other vax commands:

List (not exhaustive) of commands entered into the Vax (LH terminal) during observing:

A typical Spiral observing run will consist of the following:

Data storage:

Run observations go into /vaxinst/ccd_n/YYMMDD/, where n is the number of the CCD (1 or 2) and YYMMDD is the date in figures (e.g. 3rd Jan 2002 would be 020103). The name of the data file is of the form DDmon####.sdf, where DD is the date, mon is the first 3 letters of the month and #### is the number of the run (e.g. 02jan0010.sdf is the 10th run taken on the 2nd of January). The run number is shown on the ccd display.

Dummy observations go into /vaxdata/ccd_n/YYMMDD/ and are given a letter of the alphabet. When z.sdf is reached, the names restart at a.sdf.

Rotation:

Spiral can be rotated through any angle in order to include (or exclude) features around the target object. The schematic below shows the result of rotations on the sky on the data. Click on the image to see the expanded view.

Nod and Shuffle:

Hardware setup:

This should only be done by the support astronomer or afternoon technician.

First you need to insert the metal mask on top of the microlens array to block out half of the array. It can usually to be found taped to the metal box containing the microlens and fibres. Remove the metal cover for the fore-optics and position the mask between the top of the microlens array and the lens just above it. Push the mask in, padded side down, until the metal lip hits the front of the box. The mask only just fits between the two screws on either side of the microlens so it automatically sits in the correct poistion. Tape down the front of the mask so it can't slide back out if the telescope is moved. Replace the metal cover over the fore optics.

Software setup:

On the RH terminal:

On the LH terminal:

Observing During a nod and shuffle run you should see the telescope position being updated in the window running the "TEL_CONTROL" task (RH terminal). If, on starting a nod and shuffle run, you just see a line beginning "WHERE ..." come up in the TEL_CONTROL window and no further activity then the nod and shuffle timimg cable is probably not conntected to the detector. Contact the Afternoon Technician for help.

To resume normal observing enter the standard commands in the CCD window (i.e. the first window opened on the LH terminal).

Fields and arcs should be taken as normal, with the mask in place but without charge shuffling. Below is an example of a flat taken in nod and shuffle mode.

Reducing Spiral Data

The 512 spectra from the fibres are projected onto the detector in one long strip starting with fibre 1,1 (block 1, fibre 1) and ending with fibre 16,32 (block 16, fibre 32). Each block corresponds to one row of 32 fibres on the IFU. The gaps between the slit blocks are generally slightly larger than slightly larger than the gaps between adjacent fibres. There are a few other distictive features which can be used to make sure the fibre numbering of the image is correct (e.g. fibre 8, 2 is broken so there should always be a relatively large gap between 8,1, and 8,3). The following figure is an image of the individual slit blocks. On the detector they will be ordered from bottom to top rather than side-by-side as shown here. Click on the image to see an expanded version.
On the XMEM display fibre 1,1 is shown at the top, however when the data is reduced fibre 1,1 is at the bottom of the frame.

The only way to check the position of the target on the Spiral fibre is to reduce the data and reconstruct the image. For this reason it is important that observers become familiar with the Spiral data reduction software as soon as possible, preferably in the afternoon before their observing run begins.

Jeremy Bailey has developed a package to reduce Spiral data called spiral.idx. A full description of the data reduction procedure is given in the SPIRAL data reduction manual but we give a few tips below.

Tramline map: Spiral data cannot be reduced properly if the tramline fitting procedure does not work properly. It is therefore worth spending time on the first stage of the data reduction to ensure that the IFU images are reconstructed correctly so that the observing run is efficient. Occasionally you can encounter problems with Spiral either missing a fibre completely (even though it is not broken) or adding in an extra one (a particular problem with nod and shuffle).

Throughput calibration: Tests were performed by R. Cannon and D. Lee in January 2001. They concluded that twilight flats gave the best throughput calibration maps, as meausred by the accuracy of the sky-subtraction, even when using flats obtained at dusk on object frames obtained at dawn. Using the [OI] 5577 night sky line gave comparable results. Lamp flats gave similar fibre-fibre throughput maps (with the exception of the odd fibre) but there were large scale trends across the array.

Sky subraction: In non-nod-and-shuffle data, a median sky frame is obtained from designated sky fibres in the target spectrum and the throughput maps are used to scale this spectrum for subtraction from the object. Offset skies can also be used to subtract the sky emission from the data but at the moment this operation cannot be performed in the spiral.idx software. For nod and shuffle data, the sky subtraction is performed by simply subtracting the "sky" observation from the object frame.

Nod and Shuffle Data: Nod and shuffle data is recognised by the entry for the "METHOD" keyword in the fits header and can be reduced using the spiral.idx software. The line "SHIFT 63" (69 in the case of the EEV chip) needs to be added to the top of the spiral.dat file. See the "Nod and Shuffle " section in the data reduction manual.

Missing Fibre: Make sure you are running the most recent version of spiral.idx. Make a file called "extras.dat" and put in a the number of the missing fibre and the distance in pixels above the previous fibre where the extra fibre should be inserted. 3.5 pixels seems to work well for the MTLL2A chip, you may need to increas it to 4 pixels for the EEV chip. An example of an extras.dat file would be:
215 3.5
336 3.5
408 3.5

The extras.dat file is automatically picked up when spiral.idx is started up.

Extra Fibre:This is a bit more tricky. Sometimes with Nod and Shuffle fibre the reduction software finds a fibre in a masked off area of the chip. The additional fibre will only be found in some of the flat fields and seems to be due to a few pixles just reaching a threshold level in some frames. Sometimes you can solve the problem by taking a longer flat exposure. This increases the S/N. Alternatively, use a succesful tramline map from a flat observed using the same setup to reduce the flat field which has the extra fibre. The software doesn't go looking for extra fibres in this situation so the flat will reduce without any problems. JAB is working on an "Add Delete" option for the tramline fitting routine.