Project "TEDI"
Funded by the Nat. Science Foundation 2006-2010

A collaboration between UC Berkeley - Cornell - LLNL
NSF awards AST-0504874 and AST-0505366

Updated August 2011

Photo of comet Hyakutake at Mt. Palomar by Don Bartletti, 12:30 am March 25, 1996.

In the TEDI project, an EDI interferometer was built to fit inside the Cassegrain output hole of the Mt. Palomar 200 inch telescope to demonstrate a novel technique for a Doppler planet search in the near infrared (0.9-2.5 micron), for planets around cool stars, which are the majority star type. Instead of measuring Doppler wavelength shifts with a spectrograph alone, which requires a very high resolution spectrograph, the shifts are observed in fringes in a small interferometer. The interferometer is used in series with a spectrograph, but this spectrograph is just a "helper" device that separates the many wavlength channels so that their fringes don't overlap-- it doesn't make the fundamental measurement. Thus the EDI technique enables the spectrograph to be of lower resolution and thus smaller size. Since the spectrograph is usually the largest and most expensive component of the system, this allows the whole instrument to be 10x or more compact.

Our EDI is in series with a near-infrared spectrograph (Triplespec), built by a previous project by Cornell University that mounts to the telescope mirror at the Cassegrain output. This spectrograph only has a native resolution of 2700, which by itself is too low to do precision Doppler planet searching (30,000 to 50,000 is usually required). But now by combining with the EDI, this spectrograph can indeed do precision Doppler measurments.

Additionally, the EDI has boosted the spectral resolution of this spectrograph 10x, from its native 2700 to effectively 30,000 or more, enabling science studies of cool stars through broad bandwidth high resolution spectroscopy. Note that, normally, increasing a spectrograph's resolution demands increasing the length of the spectrograph by the same factor. Thus this resolution boost of 10x saves about 1000x in mass. Conventional Doppler spectrographs are kitchen-sized! which clearly could not be mounted to the back of the telescope mirror. (Triplespec is only 1.2 m tall.) Thus EDI is an enabling technology.

A typical visible light high resolution (~50k) spectrograph used for most Doppler planet searches are very large, such as the kitchen-sized Keck HiRes spectrograph shown here. Since spectrograph linear size scales with wavelength (and telescope aperture and resolution), moving to a several times longer wavelength (eg. 2 microns vs 0.5 microns) means a near-infrared spectrograph of similar resolution would be enormous and prohibitively costly. The EDI technique offers a solution for using much lower resolution and hence smaller spectrographs.

The EDI technique allows use of a much smaller lower resolution spectrograph, because the fine measurement of the Doppler wavelength change is done by the interferometer not the spectrograph. The spectrograph TEDI uses is in blue, being attached to the Cassegrain output hole in the 200 inch mirror. In the photo to the right it's in the cage area at the bottom of the mirror.

The TEDI interferometer sits on top of the blue cylinder and appears black and silver in these photos. It is constrained to intercept the starlight beam, process it, and then return it to its original path and cone angle so that it enters the spectrograph as if nothing had happened. This explains why there are so many mirrors. The interferometer portion is in the top angled plate. It creates a path length difference (delay) selected from 8 choices in a "filter wheel", of up to 4.5 cm,

For Doppler radial velocimetry work, only a single delay is needed, usually the 4.5 cm delay. For interferometric spectral reconstruction (resolution boosting) 7 or 8 delays are used in succession, and the data later combined in analysis to reconstruct the original spectrum to 10x or more higher resolution than Triplespec used alone. The Triplespec native resolution of 2700 has been boosted to 30,000.

Example reconstructed spectrum of starlight (kappa CrB) to a resolution 10x higher than the Triplespec spectrograph used without the EDI. The green dashed curve is the Triplespec spectrum without benefit of the interferometer, at a resolution of 2700. It cannot resolve many of the narrow features caused by atmospheric absorption (telluric lines). The red curve is the spectrum reconstructed by combining 7 delays worth of fringing data and special Fourier processing, in a technique called ISR (interferometric spectral reconstruction) or resolution boosting. It has an effective resolution of 27,000 and agrees well with a model of telluric features calculated by Henry Roe and artificially blurred by us to that resolution. A ThAr spectral lamp provides calibration emission lines. This graph shows only a small (110 cm-1) section of the reconstructed bandwidth which extends 6000 cm-1 from 4000 cm-1 to 10000 cm-1. The ISR technique is described in these 3 papers: link2, Scot16Bppr2.pdf, BoostApJ1693b.pdf.

TEDI measurement of the velocity of star GJ 699 over 3 months shows the velocity measuring capability, since this star is known not to have exoplanets and thus can be used as a velocity reference. Several velocity components can be seen in the data (left). The 3 month downward trend is due to the motion of Earth around the sun. The daily cycle is due to Earth's rotation (inset). After subtracting these known components the residual is obtain (bottom graph), which hovers around zero velocity with a scatter that is due to a random component from photon noise, plus a systematic error component of about 15 m/s.

The amount of systematic noise was determined in the graph to right by combining larger and larger residual datasets together (N Bins plotted on horiz axis) and plotting the net scatter, which should behave in a square root behavior if the scatter was purely random. A model which includes a 15 m/s systematic component (dashed curve) best fits the data.

The systematic noise may be due to presence of telluric lines in data and reflect our current ability to accurately model them to remove them from data. (The telluric line depth may change with pointing of telescope in sky, etc., as these can be due to water vapor.)

Both random and systematic noise components could be reduced in the future by straightforward means. Thus the EDI technique holds promise. All Doppler planet search techniques working in the near-infrared will have to solve the issue of better modeling of telluric lines.

Team TEDI page


             Contact info:
Jerry Edelstein......jerrye@ssl.berkeley.edu  
David Erskine ......erskine1@llnl.gov
James Lloyd......jpl@astro.cornell.edu  
Terry Herter......herter@astro.cornell.edu
Phil Muirhead......philm@astro.caltech.edu

Back to EDI home

www.SpectralFringe.org site maintained by
David Erskine
erskine1@llnl.gov