- Dates: 2/10/00 evening, to 2/14/00 morning
- Observers: Sebastien Morel and Jesse Bregman (Marc Lacasse was there the first night).
- Focal instrument: FLUOR + NICMOS 3
- Baseline: S15-N35 (38 m)

The purpose of this run was twofold: observations of symbiotic binary stars and test of the fringe-tracker algorithm proposed by the group of NASA-Ames.


The 38 meter baseline had never been used since the upgrade of the IOTA's optics last January. By 1:00 am, we finally found fringes on HR1726 at MC=-17.046017 m (thanks Marc!). We noticed an important drift of the Magic Constant (it was -17.03927 m at the end of the night).

We observed Y-Cvn (8 observing blocks of 200 scans each) with HR4909 as a calibrator (10 OB, same number of scans). I later discovered that HR4909 is actually a variable star (TU-Cvn, diameter=7.4 mas +/- 0.6 mas, measured by Dyck et al., AJ 111, 1705 -1996-). Moreover, the data on Y-Cvn are quite
noisy, but we hope to find maybe a qualitative feature (such as a protuberence at the Lagrange point) of Y-Cvn from the visibiities.

We tried to get fringes on SS-Vir, but we found nothing. It is not surprising: SS-Vir was observed with the 21 meter baseline by Cyril Ruillier last May, and no fringes were detected.


We observed UV-Aur (7 OB) with HR1726 as a calibrator (8 OB). As the day before, fringes had a low SNR. The limit magnitude given by FLUOR was about  3.

By midnight, we shifted to pi-Leo for testing the fringe-tracker. I wrote a LabVIEW VI for running the NASA-Ames algorithm and changed the control panel of the FLUOR software to use either this algorithm or the built-in FLUOR fringe-tracker (computing the centroid of the fringe packet). The display of the RMS difference between the measured null-OPD point and the center of the scan window allowed to compare the two methods. We found:
- 14 microns RMS for the centroid method.
- 15 microns RMS for the NASA-Ames algorithm.
Slight advantage for the centroid method. But I still need to improve the NASA-Ames algorithm (especially the convolution by the modelized coherence envelope). It should be useful when the fringes are noisy (pi-Leo gave nice fringes).

We then resumed the "scientific" observations: nothing on R-Leo, fringes on RY-Dra, but the calibrator (9-Dra) was not suitable. Limit magnitude fell to 2.2 and by 4:50 am, the sky became too cloudy.


Very bad sky. We just got some fringes on nu-Hya (a calibrator). We stopped everything at 1:00 am.


We observed V-Hya (3 OB) calibrated by nu-Hya (4 OB), a star of John Monnier's program. By 1:00 am, humidity reached 85% and we had to shut-down IOTA. End of the run.

- Short but good test for the fringe-tracker. I hope to improve it next time by testing linear prediction (Kalman filtering). I have already coded the algorithm in LabVIEW. The OPD drift recorded will be useful to test this algorithm. Jesse thinks about neural networks. We both agree on the need to have an adaptive fringe-tracker following the seeing variations.

- Usually, the flux was more important in the fixed beam of FLUOR than in the delayed beam. Moreover, the flux in the delayed beam vanished during the night (shaky dichroic ?) and we had to re-do the alignment of this beam. Moving the star-tracker mirrors provides a more accurate setting, but
this might mess with the pupil optimization, no ?

- As usual, we had sometimes the G3 frozen. Notice that this happens only at the beginning of an OB, when the shutters open and close, never before, during a "check signal". I suspected the USB/serial interface (we had a  problem with the USB devices at a reboot, the keyboard didn't work), so I unplugged it. Maybe it  works better now.

- Using the FLUOR software is a piece of cake (once you have understood how to use it!), however the control panel is a little messy. I suggest to have the "track zero-OPD" and "record data" widgets at the same place (organized like a tape recorder, for example). Memorizing the OPD increment such that
a single click restores it after a false fringe detection would be great as well.

- The NICMOS behaved as it did the last months: spikes in the signal at 1-read/1-loop. We used the 4-reads/1-loop and 4-reads/2-loops  modes only. These were good.

- The FLUOR data reduction software yielded no visiblity points for our data (except for nu-Hya). This is a little disappointing. Jesse will process the data from the ASCII files that have been generated. Requirements on the fringe SNR are maybe a little drastic. I think there's a room between 4 significant digits visibilities and no visibility at all.  "Coarse" visibilities are still useful, even if they need to be refined by supplementary observations, of course.

- This was actually my first run with FLUOR, I mean as an observer, and I hope it will not be the last one: despite what I said above, it's very pleasant to use, thanks Vincent!


Sebastien Morel, Ph.D.

    Infrared Optical Telescope Array (IOTA), F.L. Whipple Observatory
    670 Mt-Hopkins road, Amado AZ 85645, USA

Back to Engineering Notes Page