Introduction ============ It is widely accepted that time domain astronomy, particularly the sub-fields of transient, variable star, and asteroid astronomy, is undergoing exponential growth. Indeed, the selection of the Large Synoptic Survey Telescope (LSST) by Astro-2010 explicitly recognizes that time domain astronomy will enjoy a golden phase during this decade. The resurgence and revitalization of this field is due to technological advances in sensors as well as computing. We argue that the present approach of using "low" resolution spectrographs (:math:`R=\lambda/\Delta\lambda\sim 1,000`) is overkill. Furthermore, the traditional slit-spectrograph inevitably requires many minutes for object acquisition, leading to a less than perfect "on source" efficiency. Here, we present the design for a very low resolution (:math:`R\sim 100`) Integral Field Unit (IFU) spectrograph (hereafter SED Machine or Spectral Energy Distribution Machine, see :ref:`the figure ` below) which is sufficient for the classification of transient candidates, and when mounted on a 1-m to 3-m class telescope, will perform this task more efficiently than current instruments on larger aperture telescopes. The greater availability of observing time on small aperture telescopes further means that more candidates can be classified, and the classification can occur closer to the discovery epoch. .. _fig-model: .. figure:: Layout.jpg SED Machine spectrograph instrument layout. History ------- The SED Machine project began in December 2009 at the Palomar Observatory holiday party. There Shri Kulkarni, Robert Quimby, and Nick Konidaris sketched out the idea for a very low cost (~$20 K in parts) classification spectrograph. Since the holiday party, the SED Machine has evolved from a super-low-cost spectrograph made with Nikon lenses, to a full professional development effort. We thank the NSF ATI program for providing us with grant \#1106171. Scientific Goals ---------------- The goal of the SED-Machine is to be able to classify objects brighter than 20.5 mag with exposure times of close to an hour or less. A further goal is to perform this classification in under an hour, such that no backlog of processing will arise. Currently, the turn-around time from shutter close to output spectrum is 20 minutes or less. Our classification efficiency at 20.5 mag is roughly 50% and rises sharply with brightness. The telescope is robotically scheduled and controlled as of May, 2016. Instrument Architecture: Imager + Spectrograph ---------------------------------------------- The instrument is designed to allow IFU target acquisition with the larger-field Rainbow Camera (RC). The reference pixel used for centering is on the most sensitive filter (r-band) and once a target is placed there a standard offset will place it on the IFU. Once the object is exposing on the IFU, the RC is then used to provide guiding offsets for the telescope. In addition, the RC can provide ugri photometry through simple offsets once a target is acquired. Last updated on |version|