WIYN IFUs / HexPak & GradPak


       Matthew Bershady - PI
       Scott Buckley
       Arthur Eigenbrot
       Jay Gallagher
       Eric Hooper
       Andy Sheinis
       Mike Smith
       Marsha Wolf
       Corey Wood

   Description       Construction        Performance   
   Phase-In Schedule       First Light        Observing Information   
   Shared Use       Commissioning       Data Processing   
    Online data acquisition and reduction guides:    GradPak     New!
    Links to related sites:

This project was funded by NSF grants ATI-0804576, AST-100941, and the UW-Madison College of Letters & Science.
Figures and documents on this and related web pages may not be reproduced or published without permission of the Principal Investigators.


HexPak and GradPak are two variable-pitch IFUs that merge into a single slit-mount that feeds the WIYN Bench Spectrograph. Each IFU forms a separate (but parallel) fiber pseudo-slit within the common mount. They are designed for only one IFU to be used at a time, while the other is stowed with a light-tight cap. Variabe pitch means that there are varying fiber diameters bundled together in each IFU with different spacing. These are the first variable pitch and first dual-slit IFUs every built. Each fiber size has is own instrumental resolution and grasp, with implications for observations and calibration, discussed here, as well as data reduction (e.g., sky subtraction), discussed here. Fiber sizes given in arcsec below all adopt a focal-plane scale of 9.374 arcsec/mm.

Some initial technical publications include Wood et al. (2012), Eigenbrot et al. (2012), Hopper et al. (2015). A forthcoming paper on GradPak will be available later in 2016.

HexPak (astrometric map at left) is roughly a 41 x 36 arcsec hexagon of 3 arcsec fibers re-purposed from an earlier IFU (DensePak) with a core of 18 1-arcsec fibers in three rings (one fiber in this core broke during fabrication). The purpose of this IFU is to map fairly axisymmetric (e.g., low inclination, low ellipticity) extended objects where there is a premium for high angular and/or high spectral resolution in a high-surface brightness core while at the same time obtaining adequate signal in a lower surface-brightness periphery. In particular, one of the issues this IFU attempts to minimize is the beam-smearing problems inherent to large-fiber IFUs such as SparsePak, while maintaining large grasp in faint, outer regions. Sky fiber locations: There are 7 3-arcsec sky fibers and 2 1-arcsec sky fibers spaced in an L-shaped perimeter (like SparsePak) approximately 43 arcsec from the edge of the hexagon outer edge, or about an arcmin from the hexagon center. Fiber sizes and types: The exact fibers core sizes are 2.812 arcsec (300 microns) and 0.937 arcsec (100 microns). The small fiber is Molex/Polymicro FBP broad-spectrum fused silica with NA~0.22 and core:clad:buffer ratios of 1:1.2:1.4. The larger fibers have uniform 406 micron outer diameters (corresponding to a core-to-core spacing of 3.806 arcsec), but the core material are of mixed provenance as inherited from DensePak, which was believed to be a mix of "wet" and "dry" fibers handed down from the ancient times of Nessie, and not unlike the Hydra red and blue fibers. Some detailed maps of the hexagonal array and core meterology are given in figures, above right.

GradPak (astrometric map left) is roughly a 39 x 55 arcsec rectangular array of 11 rows of fibers comprised of 5 different sizes from 2 to 6 arcsec, in increments of 1 arcsec. The fiber size gets progressively larger from one end of the array to the other. (One 4-arcsec fibers broke during assembly and polishing.) The purpose of this IFU is to map vertical gradients in highly-inclined spiral disks. Sky fiber locations: There are 4 sky fibers of each size located in two groups separated by rougly 26 arcsec from one edge of the array (where the fibers are largest), and from each other by slightly over one arcmin. The detailed arrangement of the sky fibers was driven by the need for a relatively simple packing scheme. Fiber sizes and types: The exact fiber core sizes are 1.875 arcsec (200 microns, row 1), 2.812 arscec (300 microns, rows 2-3), 3.750 arcsec (400 microns, rows 4-5), 4.687 arscec (500 microns, rows 6-8), 5.624 arcsec (600 microns, rows 9-11). All fibers are Molex/Polymicro FBP broad-spectrum fused silica with NA~0.22. The core:clad:buffer ratios are all rougly 1:1.1:1.2.

Slit design: The GradPak slit has fiber sorted by increasing size, while the HexPak slit has the small fibers (centered in the IFU hexagon) located also in the center of the slit. An actual image of the slit is shown at above right, where both IFUs are simultaneously illuminated. The fiber numbering in the maps indicate the location in the slit. These are given in the two schematics shown below, although there are some inconsistencies in the numbering due to the fact that there are a few broken fibers and the numbering in the IFU plan has been forced to be consequitive. [TODO] Sky-fiber locations: For GradPak there are 4 sky fibers for every fiber size. They are mapped into the slit such that they bracket and span the block of fibers of their corresponding size, spaced uniformly along the block. This means that there are two sky fibers adjacent to each other at the junction of every fiber-size block. For HexPak, the seven sky fibers for the larger fibers are spaced in the upper and lower blocks in the same was as for GradPak; they bracket and uniformly span each block except...[CONFIRM. Also figure out where small sky-fibers are.]

IFU Phase-in Schedule

(i)InstallationNov 2013
(ii)CommissioningNov 2013 - Mar 2014
(iii)Shared-Use  (See memo)2014A or 2014B
(iv)WIYN Institution-class Instrument  TBD
(v)Facility-class InstrumentTBD

+ 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.

IFU Shared Use

If you would like to use HexPak and/or GradPak 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 (Matthew Bershady, mab@astro.wisc.edu) to reach an understanding on the scope of the agreement.

Next, 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.

Note: 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 so that your HexPak / GradPak shared-use request will be granted should you receive observing time from your TAC.

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

Construction to Commissioning

A description of construction installation, first-light, and commmissioning will be forthcoming; please refer to Wood et al. (2012), Eigenbrot et al. (2012), Hopper et al. (2015). Some pictures illustrating the trials and tribulations of stringing 25m of optical fibers in a busy hallway during term will be compiled here later after we recover.

o Design & Construction:

Head termination: These CAD drawings show the fiber layout for HexPak (left) and GradPak (right) as given for fabrication. The HexPak IFU head was made in much the same way as SparsePak, while GradPak was made of stacks of layers, each formed inside precision cut channels. Green fibers are object fibers; red fibers are sky fibers; the remainder are for mechanical packing. Note in HexPak that the small fibers are placed inside a fused-silica capillary with required ID. The capillary OD was slightly under-sized compared to the larger fibers and had to be coated.

Rotating cable, designed by Mike Smith, aims to minimize torque from the very stiff fiber cable external conduit (necessary to protect the fibers) on the Nasmyth telescope rotator, as well as ease handling, while at the same time ensure that the fibers do not twist over too short a cable length. In addition, the termination of the teflon fiber sheathing ensures that the bare fibers do not twist to prevent damage and wear during use (this was a major problem in the DensePak design). Images and schematics will be forthcoming.

Cable bifurcation uses a beautiful argon-weld exhaust manifold merge-collectors from SPD that splits from the large OD, 15m-length SS flex-tube conduit feeding the spectrograph foot (slit) to two small OD, 11m-length Al flex-tubes going to the two IFU heads. The junction is shown in the image at right. Large flex-tube is epoxy-bonded to the merge-collector OD, while the smaller flex-tubes are epoxy bonded to the merge-collector ID, using standard automative adhesive. Shrink-wrap is applied for strain-relief and protective sealant.

Dual slit block based on wire-EDM cut V-groove clam-shell with standard outer-dimensions and mounting holes for the Bench foot design. The clam-shell V-groove pattern is asymmetric, as show in the schematic and image to the right. Each half is built separately in a bonding jig with thin mylar as a glue buffer. The clamshell (including mylar) is assembled and polished as a unit.

o First Light: Observations were taken on the night of 14-Nov-2013. Examples of initial data will be forthcoming. More details can be found under Performance and Commissioning sections, and forthcoming papers.

o Commissioning: This will be a link to a document reporting the performance of the IFUs. Information for GradPak will soon be available in a paper by Eigenbrot & Bershady (2016).


Both IFUs perform very well with mean throughput above 80%.

The following information will be forthcoming on this site, and for GradPak will soon be found in publication by Eigenbrot & Bershady (2016).

o Examples of dome flats matched to slit image: To show both the unextracted 2D frames as well as some representative spatial cuts across the pseudo slit to illustrate varying intensity and width of trace with fiber size.

o Throughput estimates: Note--relative spectral response in the blue for HexPak fibers needs to be quantified for large fibers since the fiber types are different. What we establish here is the relative throughput and slit function for the extracted spectra, as well as estimates for total throughput of the larger fibers. This will combine dome-flats, twilight flats (for the far-blue) and standard-star observations.

Observing Information

Spectrograph configurations:

WIYN Bench Spectrograph Manual - see this document for information on available gratings and filters. It is misnamed Hydra, since it also addresses the HYDRA fiber feed. This is an excellent manual by Sam Barden and Taft Armandroff, but it is quite dated and does not include information about the upgraded Bench Spectrograph. A nice, revised version by Daryl Willmarth and John Glaspey can be found here. A description of the concluded Bench Upgrade Project can be found here, with some higher-level information here and links therein. [TODO: re-order new and old links; add references and links to two Bench Upgrade SPIE papers].

The three salient features of the upgraded system are (i) a shorter, all transmissive collimator (reduced from 1023mm to 800mm) that properly places the spectrograph pupil to increase throughput, flattens the slit function, while not degrading spectral resolution by virtue of better control of aberrations and improved detector sampling; (ii) the addition of two VPH gratings (740 and 3300 l/mm) that are optimized for use in the CaII-triplet (860nm; 740 l/mm grating, order 1) and MgI (513nm) regions (740 l/mm grating, order 2 and 3300 l/mm grating, order 1); and (iii) a new CCD with smaller pixels and lower detector-noise. As a result of the shorter collimator and smaller CCD foot-print, the Simmons catadioptric camera no longer fits the full fiber complement and wavelength range onto the detector.

Some common survey configurations that we have used with SparsePak are given on this site.

The key issue to consider is the range of instrumental resolution and detector sampling that will be obtained given your choice of spectrograph configuration (grating choice, grating and camera-collimator angles) and detector binning. The current STA1 CCD has a pixel size of 12 um, un-binned. The spatial demagnification factor of the all-refractive collimator and camera is a factor of 0.357. There is additional anamorphic demagnification of 0.5 (for extreme off-order echelle configurations) to 0.7 for on-order echelle and most low-order echelletes (note the VPH gratings are Littrow so there is no anamorphism). This means that the smallest (100 um) fibers have diameters of roughly 3 un-binned pixels with Littrow gratings, and more typically only 2 un-binned pixels with most other gratings, while the larger fibers will be significantly over-sampled. With HexPak it is important to decide if you want to undersample the smallest fibers (thereby degrading spectral resolution) versus potentially taking a detector-noise penalty with un-binned data with the larger fibers if the sample low light-levels. For GradPak, the smallest fibers typically stay above 4 un-binned pixels so it is possible to consider binning with little loss in spectral resolution.

Run preparation

The follow two links give ds9 overlays for HexPak and GradPak to draw fibers in the right location and with the right size.

Calibration data-1: Daytime

Calibration data-1: On sky

Observing tools

For convenience we have linked some observing tools also found at the WIYN site here. There should be a link to a description of the Bench setup GUI...this is the interface that allows you to twirl the grating around, focus (piston) the camera, insert filters and mirror for rear-illumin(iz)ation...if you find it, let me (mab) know.

Data Processing

last updated: 10 Jul 2016 (mab at astro.wisc.edu)