PFIS Home > Documents > Code 3170: Operations

Southern African Large Telescope

Prime Focus Imaging Spectrograph

PFIS PI Planning Tool User's Guide


Ken Nordsieck

Modification Record
Version Date Comment
1.0 14-Feb-2005 Initial Release
2.0 27-Sep-2005 Major Upgrade

Click this to start the PFIS PI Planning Tool.

1. Scope of this document

This document describes the use of the PFIS PI Planning Tool (PIPT) for the Prime Focus Imaging Spectrograph (PFIS)at the Southern African Large Telescope (SALT).

2. The PFIS PI Planning Tool is an interactive application that allows a user to select a PFIS instrument configuration for an observation based on a throughput model of the instrument and simulated data. It has the ability to simulate the input spectrum for a target and the sky, propagate them through the instrument in imaging, spectroscopic, or Fabry-Perot modes, and to calculate the signal/noise per resolution element at the detector given a choice of detector readout parameters. The tool will ultimately be integrated into the SALT PI tool, with telescope and environment parameters supplied by the telescope tool and the PFIS tool responding automatically with an XML description of the configuration to be loaded into the SALT Science Database. For now, the XML description is shown in a frame, allowing a cut-and-paste into a proposal.

2.1 System Requirements

The PI Planning Tool is a 100% Java application, and should therefore run on all systems where Java is available. Java 1.4.2 or higher is required. We use Java Web Start to distribute the tool.

2.2 Starting the PI Planning Tool

To start the PI Planning Tool, click here. If you have Java Web Start installed on your computer, the download of the components will begin. Every time the tool is started, Web Start will check to see whether a new version is available and, if so, will download it. If not, or if you are not plugged into the internet, it will use the version of the tool currently on your computer.

NOTE: The tool will have to read and write data from your hard drive, and so it will need special permissions from your system. To get these permissions, the Java library files have to be digitally signed. We don't own an official signature key at Wisconsin. So Java Web Start will warn you about a self-signed program that wants all permissions, and it will recommend you not to start the application. Until we get an official signing key at Wisconsin or find better ways for distribution, we ask you to ignore this message and to start the application.

3.0 Using the PI Planning Tool

After the start of the tool the main window will appear. There are three main tabs:

These set up the simulated input spectra, put PFIS into one of its possible configurations, and generate a signal/ noise estimate based on input detector parameters.

3.1 Generate Spectra

The first five sections create a simulated F_lambda target spectrum (above the atmosphere):

Clicking the "Use? checkbox" will add the listed option into the target spectrum. All except "emission line" are normalized by the V magnitude; the emission line flux is given in absolute units of erg/s-cm^2.

The next two sections, Solar and Lunar Items, allow you to model the sky spectrum. The Solar items define the airglow (depends on time in the solar cycle) and zodiacal light (solar elongation and ecliptic latitude). Taking the defaults in Solar Items is probably fine for initial planning. The moonlight spectrum is defined by the lunar zenith distance, phase, and target - moon distance (elongation). For initial planning purposes, a "quick select" option is given, typifying dark, gray, and bright conditions.

The final section, "Earthly Items," defines what the atmosphere and telescope does to these spectra. The default zenith distance (the center of the SALT target availability annulus) and effective area (the mean effective area across the whole annulus) is adequate for initial planning, but choice of seeing (good, median, poor) may be important for your slit throughput.

Clicking the "Display" bar will calculate the extincted target and sky spectrum at the telescope focal plane (erg/sec-Angstrom, and erg/sec-Ang-arcsec^2, respectively), display them, and list the UBVRI magnitudes of target and sky. A future release will support diffuse targets.

3.2 Configure PFIS

Now we do the configuration. The PFIS configurations are not described in detail here. For a description, see the PFIS Observer's Guide.

The configuration tab has three sub-tabs, representing the three major PFIS configurations, Imaging, Grating Spectroscopy, and Fabry-Perot Spectroscopy. Each configuration tab also has a "polarimetry" checkbox. A "Repeat Count" selector causes the observation to be divided up into a number of identical blocks, performed in succession.

Under "Imaging", one has a choice of one of the 40 interference filters, which are labeled by their central wavelength and FWHM (in Angstroms). One would not normally select one of the spectroscopy order blocking filters, except for Polarimetric Imaging, because of the PFIS lateral chromatic aberration. Wide band imaging should be done with SALTICAM.

Under "Spectroscopy," one can choose from four standard slit widths, which affect the throughput (depending on the seeing) and the spectral resolution (depending on the grating settings). If you change the slit width, you will see a change in the the slit throughput and the resolving power R = lambda/delta-lambda. A "Mask Type" selector specifies whether the observation will require a multi-object slitmask, or can be done with a reflective longslit. This selection does not affect the throughput, but does allow sorting of proposals according to the need to cut new slits, to be fully defined at Phase II.

Next, choose the grating parameters: the grating (six options) and camera angle will suffice for initial planning. The grating angle automatically defaults to Littrow (half the camera angle), and the order blocking filter is chosen to block second order light over the first order free spectral range. Choosing the grating limits the possible camera angles to the usable ones for this grating, and the camera angle defaults to the angle of peak efficiency for the VPH gratings.

For the 300 l/mm surface relief grating there is only one standard camera and grating angle, and the simulator does not model the small blaze changes that result from going off-nominal. Choosing a grating and altering the camera angle will result in the spectral coverage indicators on the right changing: these indicate the blue and red end of the spectrum, the location of the chip mosaic gaps, and the central wavelength.

The grating angle can be altered from its default if desired: for the VPH gratings this will move the blaze peak in the direction of the slider; this is not advised unless the altered blaze function is really required, since it will result in an off-nominal configuration that may not be adequately supported by calibrations. Similarly, we advise going with the default order separator, unless you choose to do without order separation entirely by choosing "clear."

NOTE: The following things are not modeled in the current version:

  1. Second order contamination.
  2. The effect of off-axis field angles on the spectral coverage, efficiency, or resolution. Thus the "chip edge", "gap", and "central wavelength" are valid only on-axis. The effects of being off-axis along the dispersion are modeled with the VPH simulator at The effects of being off-axis perpendicular to the dispersion (e.g. line curvature) should be calculated by hand.
  3. The 900 line/mm grating is not correctly modeled, especially being low for UV wavelengths, owing to its departure from the simple "Kogelnik" approximation used here. It is correctly modeled in the VPH simulator mentioned above.

Clicking the "Display" bar will calculate a spectrograph "loss function" for this configuration, including slit throughput, grating and filter efficiency, and detector QE.

Under "Fabry-Perot", one can choose from the four Fabry Perot Configurations, TF and LR (single etalon configurations) and MR and HR (dual etalon configurations). Adjusting the "Central Wavelength" of the Etalon scan will automatically select the appropriate order-blocking interference filter. "Number of Steps" specifies the number of etalon settings that will be performed around the central wavelength. The exact scan pattern will be supplied at Phase II. It is required that a single PFIS configuration consist of one interference filter only. If a scan is known to cross filter bandpasses, one will need to specify multiple configurations.

If "Polarimetry" is selected, one can choose one of four standard waveplate angle sequences, for linear polarization (4 halfwave angles), high-precision linear polarization (8 halfwave angles), circular (8 half and quarterwave angles), and all-stokes (likewise 8 half and quarterwave angles). It is also possible to user-specify the waveplate sequence: if this is desired, note it in the proposal and define the sequence at Phase II. The "Beamsplitter Configuration" defines whether the beamsplitter is to be oriented with its splitting perpendicular (nominal) or parallel to the direction of the spectroscopy dispersion The parallel configuration is used in high-speed and drift-scan polarimetry.

NOTE: The current Imaging Polarimetry mode does not allow for the dispersion of the beamsplitter, which disperses each image into a 20-arcsec spectrum, and does not allow for introduction of a slitmask to allow for imaging spectropolarimetry. It should however be correct for Imaging Polarimetry through interference filters.

3.3 Make an Exposure

On this tab we run the detector and calculate the resultant spectral Signal/Noise. We choose the total exposure time for the observation (including readout overhead), the number of frames this will be divided into, the detector binning, readout speed (low, high), and number of rows to be read out. That latter four parameters affect the amount of overhead readout time, and are only important for low exposure time or high number of frames. The number of frames will generally be at least two, to allow for cosmic ray removal. The readout speed will also affect the readout noise, which will be important only for high spectral resolution, dark skies, or many frames. Binning will generally be 2x2, except for observations with the smallest slit, where a 1x2 binning will improve wavelength sampling (the focal plane scale at the detector is 7.8 arcsec/ unbinned pixel), and 3x3 or 4x4, for high dispersion Fabry-Perot. "Readout window" refers to the number of unbinned rows to be read out (centered on the central row). This may be reduced for on-axis spectroscopy to reduce the readout time. Note that "Readouts per Exposure" specifies readouts to be performed in succession, while "Repeat Count" on the configuration tab refers to repeating a whole block. For normal imaging, these are redundant: use one but not the other in this case. The actual integration time is shown, accounting for all the overheads.

Clicking the "Expose" bar will calculate a plot of the net signal/ noise per resolution element, and a mean SNR for this observation. The SNR calculation includes target and sky photon noise and CCD readout noise.

The Signal to Noise is the total over the observation: for imaging it is the total for an image the size of the seeing disk; for spectroscopy, the total over one spectral resolution element (the value on the exposure tab is the mean of the values in the plot), and for Fabry-Perot, the total over the central wavelength etalon setting. For polarimetry it is the signal/noise for each Stokes parameter being sampled (ie, Q and U for linear, V for circular, and Q,U, and V for all-stokes). The saturation is the percentage of CCD saturation (full-well or A/D saturation, whichever is lowest) for the brightest bin (target plus sky). This should be kept below about 33% to avoid loss of data, unless bright spectral features are to be allowed to saturate. If the saturation is only a few hundredths of a percent, one is approaching readout noise limit; one should try to lengthen the exposures or increase the binning.

NOTE: The detector model does not yet support shuffle, drift, or high-speed modes.