FLOYDS is a cross-dispersed, low resolution spectrograph. There are two (nearly identical) copies installed on the 2m telescopes on Haleakala (FTN) and in Australia (FTS). The wavelength coverage is extremely broad, with 540-1000 nm in first order and 320-570 nm in second order simultaneously projected onto the CCD. 

The resolution varies from R=400 (at the blue end of each order) to R=700 (at the red end of each order) with the narrow slits, and the dispersion produces 3.51 Å/pixel in 1st order and 1.74 Å/pixel in second order. The four available slit widths, 1.2, 1.6, 2.0, and 6.0 arcseconds, project to 3.5, 4.4, 4.9, and 14.4 pixels on the detector.  

Fig. 1   A 2D FLOYDS spectrum illustrating the spectral format on the chip. The lower (red) spectrum is 1st order, while the blue is second order. The CCD is a 512x2048 E2V with 13.5 micron pixels.

Slit Position Angle

There are two slit angle positioning ("rotator") modes available for FLOYDS. The Parallactic mode determines the parallactic angle at the beginning of an exposure and then maintains the slit at that angle for the duration of the exposure. The angle is measured in degrees East of North and is recorded in the header of the raw image file with the ROTSKYPA keyword. The Sky mode allows the user to specify the position angle of the slit, in degrees East of North. The slit is fixed at this position angle for the duration of the exposure. The angle is stored in the raw image header with the ROTSKYPA keyword.

Target Acquisition

FLOYDS acquisition (aligning the target onto the center of the slit) is done robotically with an acquisition/guider camera which images a 4' x 6' field around the slit. First, the local pointing error is calculated by matching the objects in the acquisition image (detected by SExtractor) with a list of catalog objects. (The 2MASS catalog is the current default, but if no match is found, the USNO-B catalog is used.) The most common reason for the procedure to fail is if too few sources are in the acquisition image. After the local pointing error is evaluated, the final slit position is determined by the mode that the user selected when making the request. The On Coordinates mode simply places the designated target coordinates on the slit center by removing the pointing error. The On Brightest mode takes the additional step of placing the brightest object in a user-specified radius around the target coordinates on the slit center. In this case, a small offset is introduced (before acquisition starts) to bump the target away from the slit so that it can be identified. When acquisition is complete, the artificial offset is removed, and autoguiding keeps the field stable for the rest of the observation. The user-specified radius should only reflect the uncertainty in the source coordinates; any pointing offset of the telescope is corrected by the acquisition process.

Fig. 2   A FLOYDS acquisition camera image. Note the slit in the center of the image. World coordinate system (WCS) solutions to images like these are used to place objects onto the slit.

If initial acquisition fails, the telescope is offset by 10 arcminutes, and another acquisition is attempted. If the second attempt is successful, the large offset is removed, and autoguiding is started on the original field.

After the gross acquisition error has been corrected, the target is held in position by using a pattern matching strategy. A comparison of new acquisition camera images with the pattern of objects in the solved (initial) image is used to generate guide corrections, until the patterns match (within a certain tolerance). The center of the slit is also determined dynamically from the initial aquisition image. The "found" slit position is used as the pointing center (header keywords CRPIX1, CRPIX2) for the next acquisition image.

Calibration Lamps

There are a set of HgAr and Zn lamps for wavelength calibration and a Tungsten-Halogen + Xenon lamp for flat fielding. It is strongly recommended that these calibrations be taken at the position of your target.


Observations of standard stars were made in February 2017 through the 2" slit with the FLOYDS spectrograph on FTN (Hawaii). In 1st order, a r'=15.0 star produces 2.12 e-/s/pixel at 6200Å. This is 10.4 e-/s/resolution element (for the 2" slit).  In 2nd order, a g'=15.0 star produces 0.66 e-/s/pixel at 4800Å. This is 3.25 e-/s/resolution element (for the 2" slit).

FLOYDS exposure times can be estimated from the large sample of spectra and photometry assembled for the 2014-2017 Supernova Key Project. Average S/N ratios were fit to a wide range of target brightnesses. The results are plotted below.

Fig. 3   Signal-to-noise per resolution element for FLOYDS on FTN in Hawaii (top) and FTS in Australia (bottom). The left, center, and right columns correspond to the SDSS g', r', and i' filters, respectively. The magnitudes and average S/N ratios are calculated over the wavelength ranges shown in the panels. The g-band data are from the second order (blue) spectra; the r-band and i-band data are from the first order (red) spectra. The curves are derived from spectra made with the 2" slit but can be scaled to other slit sizes. For these fits, we used a signal-to-noise model that included the Poisson contribution from the source, a sky noise term, and the read noise. FTN is at a better site than FTS, so the FTN S/N curves are less affected by sky noise than FTS.

Data Processing

Data returned from the FLOYDS spectrographs are automatically processed to produce 2D rectified images and extracted 1D spectra, with wavelength scale and flux calibration applied.  The FLOYDS pipeline runs at the end of each local night, and the final and intermediate reduced data products can be retrieved from the science archive.  A description of the FLOYDS Python/Pyraf pipeline is available in the documentation section of the LCO website. This description is derived from a paper by Valenti, Sand, and Howell that includes instructions for downloading and running the pipeline. 

Guidelines for writing proposals involving FLOYDS

The readout time is 26sec, and typical exposures for arcs and flat fields are 60-80s and 25s respectively. We recommend that you allow no more than 2min for target acquisition. Although the FLOYDS slit can be oriented at any angle, because of the instrument's broad wavelength coverage, users are strongly encouraged to use the Parallactic angle mode.

FLOYDS specifications
Spectrograph Detector Andor Newton DU940P-BU
Format 2048x512, 13.5 micron
Spatial pixel scale 0.34"
Acquisition/Guide Detector SBIG ST-1603 1530x1020, 9 micron
Acquisition/Guide FOV 6.4x4.3 arcmin
FLOYDS cycle time (readout + write) 26s
Gain 2.0 electrons/count
Read Noise 3.3 electrons
Dark Current negligible
Slits on FTN (30" long) [1.2,1.6,2.0,6.0] arcsec
Slits on FTS (30" long) [1.2,1.6,2.0,6.0] arcsec