Date: Tue, 06 May 2003 11:22:46 +0200 Subject: IONIC-3T chromatic response From: Jean Philippe Berger To: Wes Traub , Sam Ragland , John Monnier , Rafael Millan-Gabet , Pierre Kern , Fabien Malbet , Frederic Rooms , Karine Perraut , Laure Lagny , Peter Schloerb , Nat Carleton , Ettore Pedretti Dear Wes, I send this answer and your message to a wider audience since I believe we are all interested by what you describe: I have also noticed the chromatic dependence of the closure phase (broadband observations on stars of different colors) although Sam's use of narrow filter is probably better to describe it. As I mentioned it to Sam I have no immediate help on that. This is something we are discovering now with the IOTA experiment and have started thinking of. - We have no capability of measuring spectrallly dependant transfer functions, it requires to couple a spectrometer to our lab interferometer or to use calibrated spectral sources. The only thing we could have done was to measure the spectral throughput which we had not time to do. From previous experiments we are hoping the spectral response is fairly flat in the H band. - We have been discussing this issue and have solutions to implement lab testing equipment but as you say it is a project by itself and I don't expect any progress until the end of the summer. However I agree this will be one of our lab priorities. - I don't believe calculation would be extremely useful since what we are observing is probably a departure from what was designed due to technology parameters. This is a known discrepancy between simulation and technology that has not been solved so-far (expecially in the broadband case). There is probably unwanted coupling before the couplers (when the two waveguides get closer together) itself which depends on the wavelength (I don't believe the chromatic coupling at the fibers entrance is to be blamed here). Also I think that the intrinsic asymmetry of the coupler (used to flatten its spectral response) must have a phase influence that we did not evaluate. I have been discussing recently this issue with people more aware than me about software that could simulate that and it turns out that most of the commercial products do no take into account phase properly. This is something we must solve also for our VLTI 8 beam combiner project. THEREFORE: --------- Until we setup a proper calibration instrument here (which will require to bring back to France the combiner at some point) I think we have several things we can do: - observing with narrow band filters, which requires calibration each time we are changing filter, it lowers the limiting magnitude but lowers the chromatic dependance. - measuring on site IONIC transfer function with a list of stars of known spectra with broad and narrowband filters. If all the observing teams share the work we could have a calibration by summer. - estimating a rough transfer function by an analytical computation: I do not expect a lot of that but this could provide at least a first order trend of the closure phase chromatic dependance -I wonder if the lab closure phase measurements we have done before shipping the combiner could be use to evaluate the chromatic dependance. I am thinking estimating the closure phase at different points of the fringe peak in the Fourier Space hoping the scanning piezo were good enough for rough Fourier transform spectroscopy. This is the best I can think of right now. I think it is worth recalling that IONIC is still a work in progress and I hope people are not disappointed by that. We'll solve it. I agree it deserves to be worked out soon. Using a small spectrometer at the output of the beam combiner would eventually be helpful for calibration and science. I'll keep you informed of the progress we are making here and let me know, all of you, if you have some comments. Best wishes JP PS: We have also nice restaurants, and nice mountains to look at Grenoble's roofs. Winter time is the best time to come (skying is the main research activity at that time). ******************************************************************** JP: We need some help from you and Pierre and Fabien, regarding the complex transfer function of the integrated optic component currently at IOTA. Sam has been using some narrow-band filters, each about 1/3 the width of the standard H-band filter, centered in the short-wavelength 1/3 of H, the center 1/3, and the long-wave 1/3. In other words, we are resolving the H band by a factor of about 3. On a given star, Sam notices that the closure phases are repeatable to about 1 degree or better, which is good. In the short wavelength band, the closure phases from the "left" and "right" outputs of the beam combiners (by analogy with a classical Michelson combiner) are the same to about 1-2 degrees. The same is true of the middle wavelength band, but the closure phase value here is different by roughly 10 degrees. For the long wavelength band the "left" and "right" outputs are not the same; they differ by about 10 degrees, and are also different than the closure phases of the other beams by about the same amount. Graphically, and schematically, it looks like this: 200 deg L c R l o s R u 190 deg r LR e L p h 180 deg a s short middle long e lambda lambda lambda <-----------H-band---------------------> Also, the magnitudes of the raw visibilities vary systematically, roughly as V(short) ~ 40%, V(middle) ~ 60%, V(long) ~ 40%. None of this surprises me, because single-mode fiber couplers are well-known to have strong chromatic properties, due to the fiber geometry itself, ane also due to the dispersion of the materials. However we are having a hard time calibrating the data, because the color of the star seems to have a large effect. Again, this is to be expected because of the wavelength-variation of the coupling efficiency and phase shift. The problem is that we cannot easily find calibrators of the same color as the target stars. *** Therefore we would like to have measured (and theoretical) values of the complex transfer function as a function of wavelength across the entire H band.*** Can you give us some guidance on this last point? In principle I could calculate this myself, but it is a big effort, and I probably would not have the right parameters to use anyhow. I assume that you have done all this, or that it is in a thesis or paper somewhere. We would very much appreciate guidance on this issue. We have a lot of data now, and need the guidance rather soon. The best would be if we could hear from you before Sam starts his observing run on 7 May, so he can use the result to guide his choice of targets and filters. Electronic or faxed information would be most welcome. We could measure these properties ourselves, but it will take some time and effort to set up the optics. It will be a big project. By the way, I spent a very pleasant three days in Paris this week, visiting Meudon, playing in the coronagraph lab, giving a talk, and dining (Le Procope is not a bad place!). Jean Schneider and his wife took me to the roof of Paris Observatory, and the old refractor. Very pleasant. Someday I hope to get as far as Grenoble to visit your labs (and restaurants)! Thank you for any advice you can offer on our IO problem Best wishes, Wes