History of EDI part II

Demonstration on starlight in 1999 at Lick Observatory
Externally dispersed interferometry was first brought to a telescope and demonstrated on starlight
by Erskine and Ge in Dec 1999, at the 1 meter Nickel telescope at Lick Observatory (Ref. 3). During the summer of 1999 they modified the first EDI apparatus by adding cylindrical lenses to the input optics and a cavity laser stabilization system (with LabView code written by Mike Rushford). The cylindrical optics allow more light to pass through the slit from a round fiber, and the cavity stabilization prevents the interferometer delay from wandering more than a quarter wave during the temperature changes of a long time exposure. Jian Ge designed and built the fiberoptic interface for the 1-meter Nickel telescope at Lick Observatory and operated the telescope. Erskine designed the algorithms, wrote the computer codes and did all the velocity data analyses for the project's publications, including the Lick 1999 data and the earlier sunlight 1998 data.

Moving on
With the end of project funding in early 2000, Ge left for Penn. State Univ. to build his own EDI, with which he has successfully detected planets at the Kitt Peak Observatory. In 2003 he moved his group to Univ. of Florida.

Sabbatical year at UC Berkeley Space Sciences
In 2001, Erskine took a year sabbatical hosted by Jerry Edelstein at UC Berkeley's Space Sciences lab. There he developed a first principles theory of the EDI instrument behavior (
Ref. 4, 7 & 11), conducted laboratory demonstrations of the spectral astrometry application (Ref 6.), and developed computer algorithms for implementing the high resolution spectroscopy application of EDI (Ref. 7). This required a solid understanding of the role of heterodyning in the production of moire fringes, and how the instrument modifies the input signal expressed as spatial frequencies along the dispersion axis (the Fourier domain).
He invented and demonstrated new algorithms for processing uniphase phase stepped data, required for echelle spectrographs where the beam height is too small to allow any phase variation of the fringe along the slit, and where the phase steps can be unknown and irregular. Since echelle spectrographs are the most important astronomical type due to their very wide bandwidth, this was a major advance toward practical observatory application.

Erskine's Comprehensive instrument theory
Erskine's instrument theory (
Refs. 4, 7, 11) is currently the only theory that describes both resolution boosting and Doppler velocimetry, and which gives the exact instrument response to an arbitrary input signal. This because it is the only theory that explicitly uses in its model the variable along the dispersion axis. It is also the only theory that describes processing EDI data in a phase stepped manner, which is critical for reduction of fixed pattern artifacts that can generate large velocity errors. It is the only theory that describes how to process fringing spectra taken in the uniphase mode, such as when the beam height is smaller than the transverse fringe height.
Reducing the beam height is important for reducing the readout noise in faint source spectroscopy. Smaller beam heights could become more important in the multi-object mode proposed by Ge's group (Exoplanet Tracker, Ref. C2) when a large number of fiber sources are to be placed side-by-side along the spectrograph slit.

Ge's Approximate Instrument Theory
Jian Ge's theoretical treatment in
Ref. A. is useful for simple estimates of the Doppler noise for an isolated absorption line, but cannot precisely compute the Doppler noise from a given arbitrary stellar spectrum because it is unaware of the exact shape of the spectrum along the dispersion direction-- it only models the instrument's behavior transverse to the dispersion direction. Its only link to the dispersion direction is through an assumed number of absorption lines ("Ni") across the band, supplied by the user. The problem with this is that the user does not know what to accurately use for Ni, since in stellar spectra there are a variety of absorption line depths and spacings with frequent partial overlap between lines.

The Ge theory also cannot describe resolution boosting and the heterodyning effect that creates it, nor the uniphase data taking mode for Doppler velocimetry or spectroscopy, a mode required when the beam height is small. These deficiencies all stem from the lack of explicit use of the dispersion variable.

The first resolution boosting demonstrations
During 2002-3, Erskine together with Jerry Edelstein's group took data at Lick echelle spectrograph demonstrating a 2-fold boost in the spectral resolution (
page_app,page_data, Refs. 5 & 7). For all, this was work done on "hobby" time since there was no significant funding for EDI in California. However in Oct 2003 Erskine was funded 6 months by LLNL and successfully demonstrated a six-fold resolution boost using EDI with multiple delays (page, Ref. 8).

Time resolution boosting applications in other sciences
In the 2002-2005 period Erskine invented methods of applying EDI-like techniques to boost the time resolution of streak cameras and velocity interferometers in shock wave experiments, which are of importance to LLNL programs studying equation of state of materials. The algorithms developed for resolution boosting in the wavelength domain are analogous to those in the time domain if certain elements of the apparatus are present.

www.SpectralFringe.org site maintained by
David Erskine

Back to EDI home