WIYN Instrumentation /
SparsePak IFU


       Matthew Bershady - PI
       David Andersen - Co-PI    
       Larry Ramsey - Co-I

       Description       Construction       Performance   
       Phase-In Schedule       First Light       Observing Information   
       Shared Use       Commissioning       Data Processing   
    Links to related sites:

This project was funded by NSF grants AST-9618849, AST-9970780, and the UW Graduate School. The page in maintained under NSF AST-0307417 and AST-0607516.

Figures and documents on this and related web pages may not be reproduced or published without permission of the Principal Investigators.

SparsePak Phase-in Schedule

(i)InstallationMay 2001
(ii)CommissioningMay 2001 through Spring 2002
(iii)Shared-Use  (See memo)Fall 2001 -
(iv)WIYN Institution-class Instrument  Fall 2002 -
(v)Facility-class InstrumentFall 2003 -

+ Commissioning:  availability to PI research team for testing and verification of science readiness.
+ Shared-Use Operations:  availability to researchers in collaboration with instrument PIs / instrument team.
+ WIYN Institution-class Instrument:  open availability to researchers resident at all WIYN institutions.
+ Facility-class Instrument:  open availability to community through NOAO as well as WIYN institutions.

SparsePak Shared Use

If you would like to use SparsePak in "Shared-Use" mode, please follow these directions.

First, read this memo and be sure that you agree with the terms.

If you do, contact the Instrument PI (Matt Bershady, mab@astro.wisc.edu, 608 265 3392) to reach an understanding and agreement on what science program SparsePak will be used.

Finally, email the appropriate letter of request listed below to the WIYN or KPNO Director well before your observations. Email the WIYN Director if you are a WIYN-consortium user. Email the KPNO Director if you are using NOAO time. It is important to cc a copy to the Instrument PI. When he receives this email he will send a similar letter to the appropriate Director, whereupon your SparsePak shared-use will be granted.

Shared-Use Request Letter
WIYN-Institution Time NOAO Time
Science PI letter - you send this
Instrument PI letter - I send this
Science PI letter - you send this
Instrument PI letter - I send this
Instructions: fill in blanks and email to jacoby@wiyn.org
with cc to mab@astro.wisc.edu
Instructions: fill in blanks and email to rgreen@noao.edu
with cc to mab@astro.wisc.edu

SparsePak Installation and First Light

Installation: May 3-6, 2001 (Bershady, Andersen, Bucholtz, Buckley, Corson, and Harmer).

First Light: Night of May 6th, 2001. Tested were:


Commissioning was completed over four runs on (i) May 06-07, 2001; (ii) May 21-23, 2001; (iii) June 7-12, 2001; and (iv) Jan 2004. Calibration data collection yielded:

Other information established is found on this web page: observing quirks, guider and rotator repeatedablity, on-sky alignment and target acquisition procedures, offset patterns, and observing modes.

Commissioning run calibration and characterization are in the following two papers:

SparsePak: A Formatted Fiber Field-Unit for The WIYN Telescope Bench Spectrograph. I. Design, Construction, and Calibration , Bershady, M. A., Andersen, Harker, J., Ramsey, L. W., Verheijen, M. A. W. 2004, to appear in PASP (June) (preprint: [local pdf] [astro-ph/0403456])

SparsePak: A Formatted Fiber Field-Unit for The WIYN Telescope Bench Spectrograph. II. On-Sky Performance, Bershady, M. A., Andersen, D. R., Verheijen, M. A. W., Westfall, K. B., Crawford, S. M., Swaters, R. A. 2003, submitted to ApJSupp

The stellar library is still under development, but will be released upon forthcoming publication.

Observing Information

SparsePak array alignment ``On-Sky:''   With an IAS rotator offset of 0 deg, SparsePak should be close to having its rows and columns aligned EW and NS, respectively. However, a small (1-5 deg) offset may need to be applied to ensure precise alignement.. This should be verified empirically by drifting a star EW across the array (using handpaddle in equatoprial model). We have no convincing evidence that the alignment changes during a run. Moreover, the alignement is not expected to change since SparsePak is firmly attached to WIFOE with a tight, gripping clam-shell and two set screws. So at the beggining of each run (i.e., a new SparsePak install), one should verify the IAS rotation offset needed to make the SparsePak rows and clumns aligned EW and NS (respectively. We've seen variations on the scale of 5 deg from run to run.

Source acquisition using the pellicle and pellicle camera:   This hardware is part of WIFOE - the port to which SparsePak and DensePak connect. There are several issues to be aware of when using the pellicle/pellicle camera to place sources onto the SparsePak array. First, the pellicle does not repeat. This means that a fiber position marked with a grease-pencil on the TV is not necessarily valid from one insertion of the pellicle to another. Second, the focus of the back-illuminated SparsePak array image onto the pellicle camera is soft, and variable (probably correlating with IAS rotation, but this has not been quantified). This must mean that when a star is focused in the pellicle camera, it is not in focus on the SparsePak array. The good news is that the fibers are big!

Useful offset patterns:
Array fill -- this 3-point pattern will fill in the aray in the sparses 70x70 arcsec region.
Position Relative Offset Absolute Offset
1 0"E 0"N 0"E 0"N
2 0"E 5.6"S 0"E 5.6"S
3 4.9"W 2.8"N 4.9"W 2.8"S
Array sub-sample -- this 3-point pattern will sub-sample the inner region of the array to increase spatial resolution.
Position Relative Offset Absolute Offset
1 0"E 0"N 0"E 0"N
2 3.3"W 0"S 3.3"W 0"S
3 1.6"E 2.8"S 1.6"W 2.8"S
Note - these offsets have been updated for the proper plate scale at the WIYN IAS f/6.3 focus of 9.374 arcsec/mm, and rounded off 0.1 arcsec precision. (See WIYN Facts).

How to make offsets:   Between guided exposures, small offsets should be "unguided" for highest precision. This means turning off the guiding, making the small move (or dither), moving the guide box back onto the guide star, and finally "enabling" guiding with a "zero lock." The last step ensures that the guider keeps the star at the location within the guide box where it is placed, i.e. it does not try and drag the star into the center of the box.

Spectrograph Setups & Calibration:   At the present time we have used SparsePak in three wavelength regimes in five orders with the echelle grating (316@63.4), and 2nd-order at H-alpha for the 860 l/mm grating. Thanks are due to Di Harmer and Daryl Wilmarth for carefully establishing the foci and optimum/comprimise camera-grating distances!

The Bench Spectrograph echelle setup parameters, calibration exposure times, and delivered spectral resolution are summarized below. For the echelle, the camera-collimator angle is 11.0 degrees and the fiber focus is -162 (-0.162 inches). For the 860@30.9 setup, the camera-collimator angle is 30.0 degrees and the fiber focus is -212 (-0.212 inches). In all cases: the collimator focal length is 1021mm. Exposure times are for un-binned data although 2x1 is often used for MgI and CaII IRT setups (binning in spatial diretion). For these cases, halve the exposure times. Exposure times for "ThAr" will change until a new, permanent feed is established. As of Jan 2002, the "ThAr: lamp is "CuAr."

Notes - (1) the spectrograph instrument setup parameters apply to any fiber cable, as do the delivered performance in central wavelength and dispersion. (2) Exposure times should scale with the relative fiber area; spectral resolutions should be fairly constant because of the large anamorphic demagnification. (3) The three "low" resolution echelle setups have the same grating-camera distance. This means that rapid switches can be made during the night by only changing filters (alpha and camera focus too -- but from the GUI). (4) Order 7 setups are problematic because of high grating angles. While these yield incredibly high dispersions and large demagnification (hence high resolution), they are difficult to focus over the full range in both spatial and spectral dimensions. There is strong trade between spatial and spectral foci. The high grating angles lead to substantial light loss because these angles are both far off-blaze and so large the grating does not fill the beam. We estimate that order 7 is down by x2 at the 8600 central-wavelength setting and x3 at the 8750 central-wavelength setting compared to setups at the same central wavelengths in order 6. (5) Dewar azimuth angle optimization was tested for Jan 17 2002 run setups for order 6 and 7 centered at 8750. Substantial improvements in uniformity of spatial focus was achieved. Daryl Wilmarth suggests improvements could be made for other echelle setting.

Bench Spectrograph Setups
 860@30.9316@63.4   (Echelle)
central wave (A) 6645 5131 6619 6625
order 2   (1.74) 11   (10.96) 9   (8.49) 8   (8.41) 7   (6.53) 6   (6.48)
alpha (degrees) 50.988 68.989 76.410 62.641
dewar azimuth
angle (degress)
+0.100 -0.128 -0.128 -0.128 -0.128
distance (inches)
15.2 40.0 33.1 40.0 31.7 40.0
camera focus
(10-3 inches)
-8 -19 -15 or
-16 for ends
-17 or -16 +3 +2
blocking filter
(slot B: echelle)
(slot B: non-echelle)
G5 (GG-495) X14 X19 X19 X23 X23
full coverage (A) 927 262 253 411 294 573
central dispersion
0.453 0.128 0.122 0.201 0.144 0.280
spectral FWHM (pix) 3.0 3.4 2.7 3.3 2.5 3.2
Spectral Resolution
4877 11747 20318 10049 24003 9683
Wavelength calib:
exposures (s)
This has been changing because of modifications of lamp feed.
Dome flats:
(High lamps)
exposures (s)
3 60 20 20 20 7

WARNING:   We have found that some fibers (from as many as 1 to 5) appear to "snuff out" during the initial setup of the spectrograph. THESE FIBERS ARE NOT   BROKEN. The cause is the filter blocking some number of edge fibers (at the top or bottom of the SparsePak slit). This arises because the filter has been inserted manually. Another sign that there is a problem is a marked increase in scattered light as seen in a dome flat. A robust solution is to remove and re-insert the filter via the GUI.   Do this as a matter of course, no matter what anyone else tells you. Then check your dome flats. If less than 82 fibers, remove filter and try again.

The easiest way to count the number of fibers is to run noao.twodspec.apextract.apfind on a dome flat, as follows, with these parameters:

Aperture "1" should be on the right side of the spatial profile in the graphics window. It should be obvious from an immediate visual inspection if there are indeed 82 fibers. Note that the vignetting for the edge fibers should put their amplitude no lower than 50-60% of the central, peak fibers.

Otherwise, counting fibers by hand is painful. If you must: Using an "l" (spatial) cut in imexamine fiber 1 is at high pixel x values and fiber 82 is at low pixel x values. Fiber 37 can be identified as the anamalously low fiber found near the middle of the slit. If xstart=450 and xsize=100, cutting near the middle of the dispersion direction of a dome flat (y = 1024) puts fibers 1, 2, and 3 at pixels (x) 159, 169, and 180 respectively.

Data Processing

  • Data Reduction:   SparsePak.iraf file for dohydra
  • Astrometry:   table (postscript), map (jpg)
  • Quick-look software: Repacking .ms files
Repacking raw 2D spectral images   Strong spatial vignetting pattern and the complex nature of the mapping of the fibers from the telescope focal plane to the slit make it difficult to interpret the raw data. The best route to determining your data quality is to reduce your data and process it through dohydra. There are some repacking tools below that can be used with these data. For a first-look, however, a simple iraf script and support files can be found here as part of an IRAF package named "ifupkg." Down-load this package and install it via an entry in your loginuser.cl. The script "rawrepak.cl" allows you to operate on the raw image, and requires only In addition to resorting the fiber spectra, this tool bias-subtracts and field-flattens, but does not not wavelength rectify the data. (Wavelength rectification is currently the principal advantage of using the .ms files.) For this reason, the primary utility of the raw repacking tool is to repack radially, but re-packing at different position angles is also an option (as described below for the repacking of ms files). While literally the raw images may be used, we recommend CR cleaning your images if possible. Also note that the code is currently hardwired to extract the central 4 pixels of each fiber spectrum (roughly the FWHM in the spatial dimension) in un-binned mode. An example is shown below.

``Raw'' spectrum showing 82 spectra, left to right. Shorter wavelengths
are towards the top. Ha and [NII] emission are evident, as is a weak sky-line
blue-wards of [NII]
Radially ``re-Packed'' spectrum over
the same wavelength range;
the seven sky fibers are at the left edge.
spectrum showing
strong sky lines

Repacking .ms files   Because of both the of the complex nature of the mapping of the fibers from the telescope focal plane to the slit, it is difficult to interpret the extracted spectra, as output from, e.g., dohydra. This output is an "ms" file, i.e. an 82x2067 image, where each row is the extracted spectrum corresponding to fibers 1 to 82, respectively numbered according to the fiber position in the slit-block. One solution, suitable for a quick-look at the data, is to re-stack the spectra into meaningful orders. A simple iraf script and support files can be found here as part of an IRAF package named "ifupkg." Down-load this package and install it via an entry in your loginuser.cl. The script "repak.cl" allows you to take the output ms file from dohydra and repak the fibers so that they are ordered either radially (rad) outwards from the central fiber (#52), or stacked by rows oriented at a variety of position angles (pa[x]). The packing symmetry of the fibers allows for PAs of 0, 30, 60, 90, 120, and 150 degrees. The radial repacking is useful for inspection of S/N as a function of radius. The PA packing is useful for making a quick check on, e.g., whether the source has rotation, and a rough estimation of the position angle. Here's an example:

     rad      pa0      pa30      pa60    pa90    pa120    pa150  



The "raw" and "rad" ms spectra may be directly compared to the 2D spectra above using rawrepak.cl

For this source you can plainly see there is the least coherent spatial gradient is for PA near 90d (pa90), perhaps slightly offset towards PA of 120d. The kinematic major axis is there 90d offset from this PA, or roughly 10d. This agrees to within 10d of the value determined from a tilted-ring fit to the velocity field.

Research Home Education Outreach

last updated: Jun 08 2004 (mab at astro.wisc.edu)