Binospec FAQ

Overview

Binospec is a multi-object wide field optical spectrograph for the 6.5-m MMT telescope, with two rectangular 8×15 arcmin fields of view, each with its own spectrograph. It is run in a queue-observed mode. For basic properties of Binospec see the Binospec information page.

General questions
Spectroscopy
Imaging
Data: formats, pipelines, etc
Other topics

General

Strengths

Binospec has a wide field of view and is very efficient across most of the optical window (roughly 3600-9200 A). It produces data with a clean reduction free of most artifacts and sky subtraction problems. A data delivery system and pipeline give quick access to raw data, and data can be reduced by users or by SAO staff. There is a tool for measuring galaxy redshifts in reduced data. What this adds up to is that Binospec is good at getting sensitive spectra of largish numbers of objects in fields of order 15 arcmin.

Binospec has acquisition and calibration overheads that add up to nominally 30 minutes per spectroscopic target (often a bit less in practice). This makes it less efficient at acquiring spectra of large numbers of single bright objects needing eg 10-15 minute exposures, or at survey tiling fields of a degree scale or more.

Overheads and observing block length

The queue catalog will assign each observing block a fiducial overhead, currently set at 30 min for spectroscopy (either longslit or slit mask) and 10 min for imaging. This includes target acquisition, instrument configuration, calibration exposures, and CCD readout times. The actual overhead is often a little shorter for spectroscopy (like 25 min) and we only charge programs the actual overhead. Some of the overhead is due to night time calibrations (flats, arc lamp) that take several minutes. Staff are exploring taking some calibrations during the day to save time, but this will only knock off ~10 minutes at most and could affect reduction quality. What this means is that taking a lot of spectra of objects that only need 10-15 minute exposures is not an efficient use of Binospec time.

Additionally, the catalog software breaks requested observations into observing blocks of at most 2 hours exposure time, for scheduling practicality. This means that for example, if you assign a lot of targets each 2.5 hours of exposure time, your program will look inefficient. You could, for example, give some of them 2 hours and some of them 3 hours, and lose less to overhead. In practice, if we do manage to integrate for over 2 hours consecutively on a target, your program would only be charged for one acquisition.

Guide and offset stars

You must supply accurate coordinates that are in the Gaia DR2 reference frame (SDSS is good enough). Binospec’s guide cameras can then acquire guide stars with Gaia positions and set your target directly onto the slit. There is not currently a slit with a mirrored surface, so it’s not possible or necessary to use a slit viewing camera to set the pointing on an offset star. Just supply good coordinates.

Spectroscopy

Common grating setups

The default setup for large wavelength coverage is the 270 lpmm grating and central wavelength 6500 A. If you need higher resolution, the 1000 lpmm grating is better at the bluest wavelengths < 5000 A, and the 600 lpmm grating is better at redder wavelengths. The central wavelengths are continuously tunable but some values are not allowed, see the table on the Binospec info page.

Longslit choice

The 1.0 arcsec wide slit is always in the instrument. You should use this unless you have a strong reason not to. It is well matched to the typical seeing, and any other longslit has to be installed, which means competing for space with all the slitmasks in a given queue (ie your program could only be observed on nights when your longslit is in).

Spectroscopic exposure lengths

We strongly recommend a maximum of 20 minutes per single exposure due to build up of cosmic rays. We also recommend to get at least 3 exposures per target to remove cosmic rays; the pipeline will run more easily and data will be better.

Exposures shorter than 60 sec or so are inefficient due to readout and only recommended for bright objects like standard stars. If you are doing a long time series like a planetary transit over several hours, individual exposures of 120 sec have been used in the past.

Magnitude range for targets

If you put too-bright objects on a slitmask along with faint objects, they can bleed into neighboring slits and/or cause problems in the pipeline reduction. It is best to keep a magnitude range no greater than about 6-7 mag. For example, if you are trying to take spectra of faint stars or galaxies at magnitude 22-24 or fainter, you should also cut your catalog to remove stars/galaxies at about mag 16 or brighter.

Blue blocking filters

Binospec has two spectroscopic blocking filters, LP3500 and LP3800. These are long-pass, blocking blue light. They have about 98-99% throughput longward of 3500 A and 3800 A respectively. They prevent 2nd order blue light from contaminating the 1st order spectrum, which can be a problem especially for intrinsically blue objects. Generally, you should use LP3800 unless you are observing the 3500-3800 A range. Binospec does have some throughput blueward of 3800 A, so you can reach that region by using LP3500.

Parallactic angle

Binospec is behind an ADC (atmospheric dispersion compensator), so you don’t need to observe at the parallactic angle, although the observers will often set the longslit at a sky position angle close to parallactic anyway.

Same mask with different gratings or central wavelengths

You can use the same multislit mask with two different wavelength tunings. This is a common way to get large wavelength coverage at moderately high resolution with the 600 line grating, for example (of course it takes twice as much time, but you also resolve the sky better). It’s also possible to observe the same mask with two gratings, eg the 1000 and 600 lpmm gratings.

Using a previously designed mask

Currently we are storing all the previously designed masks, so if you want to observe a mask from a previous semester, that’s possible. The queue catalog will let you select one of your previous masks. You should also be able to find masks designed by one of your Co-Is. If you have any problems finding an existing mask, contact your MMT instrument scientist.

Adding spectrophotometric standards to your catalog

If you need standards for flux calibration or telluric correction, put a note in the comments field of your object. If you’re especially concerned about tellurics you may want to note that so we know to take the standard reasonably close. You can add standard stars to your catalog. This is especially helpful if you are using a grating/wavelength combination different from the typical G270/6500 A.

Add your standard stars as priority 3 so the queue scheduler doesn’t try to schedule them as normal observations. You must update standard star coordinates to the current epoch using Gaia. Many standard stars are white dwarfs with high proper motions, and many popular lists of standards are horribly out of date. One way to add standards to your catalog is to update their RA/Dec coordinates to the current epoch (eg 2021.8), enter this RA/Dec and zero for the proper motions, and set the epoch/equinox field to “2000”. You can in principle enter the Gaia coordinates and proper motions, but you must 1) convert the PMs from mas/yr to the units MMT uses of sec of time or arc per century; and 2) enter the correct epoch for the coordinates (2015.5 for Gaia DR2), and we find that this doesn’t work well because the telescope mount code wants to use the epoch 2015.5 as the equinox of the coordinates (ie it precesses them wrongly). For more information see the MMT spectro standards page .

Spectrophotometric standards for current or past observations

We frequently take standard stars during a run as general observatory calibrations. These are most commonly in the G270/6500 setting but others can be done. These observations are taken in an observatory (director’s time) catalog, not charged to an individual program. So they don’t appear in your catalog, but we can give individual PIs access to the data. If you need standard star data, contact Ben Weiner and specify the grating, central wavelength, and date. We don’t have data in all settings from all runs but the throughput is generally stable.

Pseudo-IFU mode

It’s possible to build up an integral-field-like spatially resolved spectrum by taking a series of exposures stepping the longslit (or slitmask) in the perpendicular to slit direction. This step is controlled by the observer in instrument X-Y coordinates, so you need to specify the step size across the slit, not in compass directions. For example, say “use PA=45 and step by 2 arcsec perpendicular to the slit = instrument X-direction, doing 5 positions at -4, -2, 0, +2, +4 arcsec.” Don’t say “use PA=45 and do 5 positions stepping by 2.8 arcsec north,” because the observers have to set the position in the instrument frame, not the sky frame.

Afternoon “sky” flats for spectroscopy

Sky flats aren’t part of the normal Binospec reduction process. However, some programs that need very good knowledge of the instrumental profile can benefit from afternoon spectroscopic flats using the solar spectrum. These are used to measure the instrumental profile by comparison with model spectra. You probably don’t need these unless you are measuring velocity dispersions or doing precision radial velocity work. They are time-consuming to take and require coordination between TO and queue observer, so please contact your MMT instrument scientist if you think you might need them.

Avoiding rotator limits for slitmasks

The instrument rotator has limits near +180/-180 degrees. These can prevent observing slitmasks at certain PAs and certain times. In practice, the limits mean that to avoid hitting the limit at transit, for masks that are north of the MMT (Dec > +33), you should avoid designing masks with position angles near 0, and for masks south of the MMT (Dec < +33), you should avoid PAs near +180 or -180.

Imaging

Imaging exposure lengths

Individual direct imaging exposures should be kept pretty short, especially if you are working in a field with a lot of stars (lower Galactic latitude) or near the peak of detector sensitivity (r or i band). If so, direct imaging exposures should be no more than 60 seconds, or you will saturate a lot of stars. Generally in all fields, imaging exposures should be less than 120 seconds in r and i filters, and no more than 180 sec in g and z filters. Saturated stars cause persistence problems that can damage data on faint targets taken after your field.

Imaging flatfields

Flatfields for imaging are taken in the afternoon with an internal lamp, and we have to use one of the arc comparison lamps because the spectroscopic continuum lamp is too bright. This means that imaging flatfields will show up in your raw data package with image type ‘comp’. Flatfields can be shared between targets, so you might get a notification about data being available for a target, even though only the afternoon flats have been taken so far.

Imaging dithers

It’s not currently possible to make really large dithers while imaging because the guide stars would go off the subarray (region of interest) of the guide cameras. The detectors are pretty clean so modest dithers should take care of most detector artifacts.

Data products

Raw and reduced data

Raw data will typically show up in your catalog the next morning, copied over around 9 am MST. These include *.fits and *proc.fits files – the proc files have been overscan and bias subtracted and trimmed. Spectroscopic data can be reduced by the Binospec IDL pipeline. SAO staff will reduce the data when time is available. The data package also includes a file with an auto-generated reduction script, so you can download the pipeline and try reducing the raw data yourself. Several of our users have reduced the longslit data with the conventional IRAF cookbook workflow (Massey et al) that they’ve used for other longslit instruments.

The Binospec pipeline doesn’t reduce imaging data, so you should download the raw *proc.fits data, and reduce those with IRAF, ccdproc, or any other typical method for reducing imaging. You may want to look at the CIERA/POTPyRI imaging pipeline (K. Paterson et al 2022 in prep). Note that the imaging internal flatfields will show in your catalog as type “comp” and sometimes are filed under a different object in your catalog than the nighttime science data.

How do I look at the data?

The reduced data files include a 2-D FITS image that has 1-d spectra, one per row; and a FITS file with many extensions, each the 2-d spectrum of one slit. See a description at the Binospec pipeline repository wiki. To inspect spectra and fit redshifts, you can try the Specpro software in the version adapted for Binospec, see the Using Specpro to inspect Binospec data page.

Where’s my longslit object in the data?

The longslit masks have a slit on each of Side A and Side B, and the data are stored in extensions 1 and 2 of a fits file. Your object is in one side, and the other slit is on some random part of sky about 10 arcmin away. Typically longslit targets are observed on side B. So if you don’t see the object’s trace, look in the second extension of the FITS file. Try loading the file with ds9 > File > Open As… > Multiple Extension Frames.

We usually apply a small offset (about 6 arcsec = 25 pixels) along the slit to move the catalog target away from the boundary between readout amplifier quadrants. So if you look at the raw data sci_img_*_proc.fits image, where the images have been trimmed and re-assembled into one fits extension per side of Binospec, the trace of your object will usually be in FITS extension 2 (side B) at about pixel y=2070 to 2080.

Where’s my imaging object in the data?

Similarly to the longslit, your catalog target coordinates will usually be in Side B, so look in the second FITS extension of the sci_img_*_proc.fits file. The WCS should be approximately correct. The target will usually be offset from center to avoid the amplifier quadrant boundaries. If you just load the fits file into ds9 as a normal image, it will show the extension 1 image, which is about 10-11 arcmin away. Try loading the file with ds9 > File > Open As… > Multiple Extension Frames.

Slit and image orientation on sky

For a sky position angle of 0, the along-slit direction of North is “down” on a Binospec sci_img_*_proc.fits image; North is towards low-y in pixels. Similarly for other sky PAs, the low-y direction is along the requested PA, so the y-axis increases in the direction opposite the PA. This holds for both side A and B of Binospec.

The spectrum dispersion direction is redder towards high x on side B, and redder towards low x on side A.

For Binospec imaging, on both side A and B, in the sci_img_*_proc.fits image, North is down (to low-y) and East is to the left (low-x). Binospec direct images have a WCS written into the header that is approximately correct, so loading an image in ds9 or a similar viewer should show a correct compass rose and coordinate readout. The coordinates should be good to a few arcsec or better, so you can use them to locate objects, but for precise astrometry you should align the images to Gaia stars, as with any raw telescope/instrument WCS.

Further questions?

If you have a question not covered here or on the Binospec information page, please contact the instrument scientist, Ben Weiner, at bjw @ mmto.org.

Benjamin Weiner, bjw@mmto.org