AMBER logo AMBER instrument Visit the VLTI website
Site map Come back to home Restricted acces
You are here : Home > Astronomical community > AMBER Documents > Daily commissioning reports (COM3)
Recent Events

Check also the AMBER/ESO site

Astronomical community

Part of the web site dedicated to the astronomical community

AMBER Documents
Observing with AMBER
Data Processing

Public outreach

Part of the web site dedicated to the general audience.

AMBER Consortium
Picture Gallery
Short presentation
VLT Interferometer
Press room


In the same section :
Daily commissioning reports (COM2)
AMBER Commissioning Documents

Daily commissioning reports (COM3)
First publication : 4 February 2006, publicated: Thursday 9 February 2006, by Fabien Malbet, Romain Petrov

Day/Night: Feb 3, Feb 4/5, Feb 5/6, Feb 6/7, Feb 7/8, Feb 8/9, Feb 9/10, Feb 10/11, Feb 11/12

February 3

Fabien and Romain arrived yesterday at Paranal with Fredrik. Gerard Zins was already here working on the LGS. Mario Kiekebush arrived today. Yesterday we did not have access to the focal laboratory because the ITF (Interferometry Task Force Team) was working on Finito, the VCMs, the DL. Some substancial progress is being made on Finito. It seems accepted that when the injections are stabilized (non saturation in MACAO, 5 herz tunel tip-tilt correction). They have been making progress on these points and are testing tonight on sky with 3 UTs.

Today was the first day of the COM3 run.

  • laboratory: all H and K inputs and outputs readjusted. Standard drifts after several weeks break: between 30 to 50% loss in injection and some beams moving slightly more than the specification (up to 0.8 pix inY -spec=0.5- up to 6 pix in X -spec=4-).
  • laboratory: superimpositionbetween HJ and K fiber heads in CAU have been corrected. It seems that this should be repeated on a trimestrial basis (last check was last July and the drift was significant, corresponding to 20% injection loss in both H and K if you guide somewhere in between)
  • laboratory: BCD has been adjusted in injection and OPD.
  • laboratory: the supports of inputs J1 and J3 and the outputs J1, J2 have been removed by the Paranal mechanics before the run. No interferometric tests can be performed in J. The beam input J2 - output J3 can be used for ADC tests.
    • We met with Pedro Martones to discuss the technical part in March. The dichroic supports will be rebuilt and the the output base plates will be cut and reinstalled by March 5. Romain and Fredrik will realign the dichroics on March 6-7-9 (Romain arrives inParanal March 6). This assumes that the couple parabola - fiber head in each plate has not been affected and that efficient marks have been made to put back the parabola-fiber parts in the same poistion has now has discussed between P.M. and P. Antonelli
  • software: it was decided in a meeting today between Mario, Gerard, Fabien, Fredrik and Romain, that we install immediately the new amdlib and (GZ OS updates) (amdlib2 is really ready, the load on Gerard from LGS is moderate, some of the bugs that we are looking for -fringe criteria MR and coherencing- can be affected by the new software, this is the last turno of MK on AMBER) . Installation has been performed. Succesfulls tests (P2VMs) have been made at the end of the day.

Plan for February 4:

  • move on with software. Start investigating the FC bug in MR and the coherencing using lab data and the VLTI with simulated telescopes
  • install new densities in the focal lab and test it.
  • during the night we will have three hours of technical time with three UTs. Goals:
    • validate new software in operation
    • validate new injection approache proposed by Fredrik using IRIS, beacons and pupil visualisation
    • use the opportunity to check VLTI-AMBER non vigneting status (thanks to pupil -scope inARAL)
    • test the FC and look for the MR bug.

February 4/5

  1. Gilles Duvert and Andrea Richichi arrived today
  2. Software: in the morning, the new software was tested with lab data, introducing pistons with the piezos and worked succesfully: apparently correct SNR numbers. The centering of fringes needs several iterations to converge. The quick look values look coherent at high SNR with the CAU.
  3. Hardware: afternoom: the densities with holes have been installed instead of the glass ones. the values are wrong. The so called density F/10 is actually an F/100 (same in H and K). The F/100 was impossible to measure but is more than 1/1000. We put back the glass F/10 and the metallic F/10 to serve as an F/100.
  4. Hardware: the flux in H is about 5 times fainter than in K from the CAU. (normal situation H>K). Tomorrow we will change the lamp and look for a problem in the RAS
  5. Night: We had technical time from about 1am to the en of the night.
    • acquisition: reference pixels for IRIS defined and also position of the FINITO/IRIS feeding mirror.
    • fast spiral search after IRIS acquisition seems to work OK, in spite of bad, very strongly fluctuationg seeing (1 to 2.5 arcsec within minutes). It might be usefull to increase box of fast spiral, and force it to a single photocenter. IRIS reference pixels have beem made on sky, the beacon from UT3 does not work.
    • Fringes where found on 61 Leo around 08 at beam1=-2.7; beam2=-5.
    • fringe search: the FC numbers are coherent with the quality of fringes we had. (max of the order of 2 in frame mode, between 3 and 5 in block mode -ndit=10-. The fringe centering is OK but there is a sign error on the baseline 1-3 (delay line 1 is reference): the piston seems more or less good, but the corrections goes in the wrong direction. In the fringe search mode 1 3, things seem OK
    • we recorded data (dit=50 ms, ndit=1000, dark=sky=1, nexp=5 on 61 Leo in MR_K)
      • piston values are coherent (very noisy) in quick look
      • visibility values are absurd when the SNR is faint.
      • It is vital to compute a piston by averaging validated (FC>2?) values
    • we shifted to LR:
      • beam1 and 2 OK, beam 3 weak after recentering of finito mirror, needed full spiral search which worked fine) (had to use new density F/100 for that)
      • fringe search on beam1-2 was very fast (-2.77) with FC>20
      • fringe search on beam 2-3 was more difficult because we had FC>5 far from the fringes. The correct threshold for fringes by eye was FC=10 for this very bright object. (beam2=-5mm; beam3=-0.1 mm).
      • we failed to record LR because it was too late...
      • only K band visible with ADC = 0

Tomorrow (Feb 5):

  • meeting at 15:00 at parlogs
  • remove specral filter in spectral calibration in LR
  • put in the new lamp and update the DITs so that there is sufficient flux and also tweak the IN/OUT dichroics and fix software issues.
  • Open SW issues are:
    • the piston correction is sent with the wrong sign for the 3-1 baseline with DL1 as reference.
    • Coherencing and paralell FS.
    • Validating piston and selection of frames in QL mode.
  • commissioning of HR_K, then MR_H and HR_H, according to COM3 plan.

February 5/6

The report from the third day and first night of our comm3 (February 5-6) will be unfortunately very brief:

  • we could not open the telescopes because the humidity was too high (70-80%) through all the night.
  • during day time:
    • software: we discussed the coherencing methods and we might be moving in the right direction
    • software: it has been decided to make the block mode for fringe search systematic for the general user (who is supposed to be relying on a good DL model), with NDIT=CST/DIT (typical CST=1s) and a very small initial scan stroke ( 5 steps in a coherence length).
    • software: the discrepancies in quick look between large pistons and their measure in Low resolution mode (due to the rapid extenuation of the fringe contrast in the small coherence length) must be treated at amdlib level (today at 5 microns amdlib measures 5, at 10 it measures 8, at 20 it measures 14 etc..). For fringe centering a practical temporary solution is to iterate at least twice, gradually centering the fringe packet.
    • software: an action is started to reduce the overheads in block mode in the fringe search or coherencing.
    • software: it has been requested that all files obtained for, e.g., fringe centering, are archived but not delivered to the astronomers.
    • hardware: the lamp has been changed, a small but insufficient progress made on K/H (3 instead of 5). Pierre A., asked by mail, confirms that when the lamp type is changed, injection in RAS fibers must be reoptimized inside the RAS. We must consider doing that at the end of the afternoom.
    • hardware: fast verification of SF adjustments: minor corrections. As usual, the inputs and outputs relax after a major intervention and need a second, much faster correction, after which they are usually stable enough.
  • night: no telescope observations.
    • HR P2VM tested and corrected (since the spectral calibration plate is inefficient in HR mode, amdlib’s spectral calibration routine has been patched to just return zero offsets).
    • HR templates were not present (only p77 templates were there). Corrected.
    • The HR observation spectral modes templates where only 5, giving non jointive windows (explaining the failure to observe 2.222 microns in Feb 2005). This has been corrected and 14 modes have been defined, with a small overlap between each.
    • Still, the min and max lambda value available in each mode is not visible in the P2PP panel. This will not help the user to choose the right mode. No solution for the moment to that (small) annoyance. We note that since p77, P2PP has a new eature: a set of rules that can dynamically aid the user to avoid mistakes. Fredrik is the responsible for the writing of these rules.
    • Fredrick has very substancially refined the definition of the AMBER pixel on the IRIS sensor, this should speed the injection process very much (ideally suppressing the spiral search if the mirrors feeding IRIS are repeatable enough)
    • Many OBs have been prepared for the commissioning of HR_K and of measurement stability...
    • Some keywords have been changed by Paranal in the headers. this implies adapting the ammyorick software. We request that ESO informs us of such changes in the future.
    • some difficulties to transfer data to Nice (Florentin ansd Armando are waiting for data to reduce) must be solved (see Fabien’s mail)
  • Feb 6 afternoon: RAS injection in the HJ fiber ? (one persone active in RAS room -storage room-, the other monitoring numbers in front of a screen in control room and communicationby phone). TBC.
  • Feb 6 night, if we are lucky enough to open: HR_K, MR_H, first ADC test, HR_H.

February 6/7

  • second UT night (out of three) without opening the telescopes because of excessive humidity in the dryest desert in the word.
  • good progress on software (important improvement of overheads in block mode) and other things but I am a little bit depressed and will report better tommorrow (I hope)
  • hardware: twicking the HJ fiber injection in the RAS produced substancial gain (K/H now 2). The installed lamp was 12V/20W so probably 10W and redish with the 6V power supply. No need to twick experiment, bof...
  • some BCD/lab data produced, Martin please could you please transfer it to Nice, so they can look at it...
  • For Nice "Data Reduction Task Force": sorry guys.... However stay tuned for tommorrow because if we manage to observe the decision to offer MR_H or HR_K must be taken very rapidly...

February 7/8

Night 3 with UTs:

  • We finally had one observing night, almost full with some losses due to UT1 and MACAO (UT1 andUT2) crashes
  • Our first goal was to offer the mode MR_H. We observed gamma vel (well known bianry in the continuumand with interesting features in lines) and a calibrator. The problem is that these K=2 and K=2.5 objects seemed relatively low SNR ( 3 per channel per 100 ms at MR). Compared with K there is at least a 2 magnitudes deficit. This leads to the:
  • MR_H absolute mystery: when you observe a star (calibrator) with K=2 and H=2.1 in low resolution, the ratio between the max of the bands is Kmax/Hmax 1. That means the strehl, transmission, fibers, dichroics behave similarly in H and K as demonstrated in lab (actually H transmission is slightly better and H strehl should be slightly lower) When you do the same in MR you get Kmax/Hmax 6. So we suspected that the MR gratting was very unefficient in H. But, when you observe the CAU, the ratio Kmax/Hmax is 5 in Low Resolution and is 4.5 in Medium resolution. We tested very carefully and many times that all other things were equal. I repeat: with the CAU: the ratio K/H efficiency is the same in Low Resolution and in Medium resolution, on the star the K/H 1 inLR and K/H 6 in MR. Got an idea...
  • Then we made observations in HR_K. ESO might offer it with limiting magnitude K=2 after some firther spectroscopic tests with the ATs. Again, the magnitude is disapointing but the conditions (seeing were not excellent, but not ugly either between 0.9 and 1.3, more often 1.3). We need to reduce the data to be more precise. However, the full potential of this HR (12000) mode assumes that FINITO is working. There might be hope. As written before, as soon as the injections are activelly stabilized,it should work.
  • We used IRIS for image injection. Currently IRIS is fed by a dichroic taking all H away from AMBER. When working in K, the uses of IRIS represents a very huge gain in time (20 minutes par star) in the acquisition, fiber injection phase.
  • In H, we work like with ARAL before: acquire the image with IRIS, remove IRIS and observe in H blind. We had not enough time to find out if it is necessary to spiral again in H (i.e. does the removal of IRIS dichroic move the image on AMBER). We should try to assess that in the next nights with the ATs.
  • A very substancial improvement in the speed of the fringe search have been confirmed in LR.
  • It seems that there is a persistent difference between the fringe criteria values compute by amdlib and the same FC values in the OS, inspite of the facts that it uses the same amdlib. I leave it to Gerard and Gilles to comment on that if they wish.

Program with the ATs:

  • check alignment, field and vigneting with the improved DL and now working VCMs. This is currently in progress (we are at the beginning of night 8-9) and the improvement seems spectacular: 6 or 7 arcsec field and nicely focussed pupil, very marginally vigneted in one beam (some optical element in the train).
  • we will observe in MR_K (to find the fringes and set the DL model offsets), assess limiting magnides and vibrations (visibility improvement).
  • Then go to LR_HK, find limits in that cases
  • Use it to fully spectrally calibrate HR modes
  • Assess the MR_H issue again...
JPEG - 409.7 kb
ITF Master
H. Bonnet at the VLTI console (ITF=Interferometry Task Force)
JPEG - 404.9 kb
Pupil image
One of the advantage of IRIS: we can see the pupil and make the required optical adjustment.
JPEG - 429.2 kb
Stellar iamge with AT
Another advantage of IRIS, we can see the star spot just before AMBER and identify the field of view with ATS (=7 arcsec).

February 8/9

Night #1 with ATs.

We don’t have the UTs anymore so the air is dry, the sky clear and the seeing good (0.6 to 0.8 with 5 to 9 ms coherence time. This gave us a good night with ATs.

  • The first part of the night was used by H. Bonnet, from the ITF team and Fredrick to check the alignment with the ATs, the set up of the VCMs and the use of IRIS. As reported previously, there is a spectacular improvement in field and pupil quality. As we will see below, that corresponds to a massive improvement in limiting magnitude.
  • It is confirmed that IRIS fantastically boosts the acquisition speed for observations in K. However because of some problems of communication between IRIS and AT3 (to be solved tomorrow hopefully) we had no IRIS guiding.
  • We observed several stars at MR, from K=-1.3 (Sirius) to K=3. we detected and recorded fringes on all of them. Data processing will confirm it, but the limiting magnitude for MR_K with the ATs might be K=3 in good conditions (especially when IRIS guiding will be re established). Remember that in July the limiting magnitude with ATs in Low Resolution was 1 !
  • The VLTI contrast seems quite better than with UTs, with a V2 max of 0.68. When we change exposure time from 32 to 87 ms, V2 max changed from 0.68 to 0.55. Of course further data processing is necessary to assess all this.
  • During the first observations, the DL model was very bad, with fringe drift of 300 microns/min (5.3microns/s) at very high z. Then a second model was installed and now the drift is less than 50 microns in 10 minutes, so we can consider low resolution observations.
  • We agreed with Fredrik on the specifications of the coherencing mode to be implemented by Gerard, now that we have all the pieces apparently well tested. This is a critical step for low HK


  • low resolution HK limiting magnitude and MR_K with resolved objects...
  • test the coherencing to be finalized tomorrow by Gilles and Gerard.

Note: All this is nice news with ATs, but may be I forgot to tell you the sad thing: there are only two of them and the availability of the third (with the corresponding corrected Delay Line and VCM) is not foreseen that soon given the VLTI and ITF teams work load.

February 9/10

Night #2 with ATs.

(report made by FMa)

  • Day work: GZi & GDu checked why we do not obtain the same FC numbers with the OS and with off-line data reduction. They found this comes from different flat-fields. It has been deceided to use the flat field measured intially in Bonn
  • second night with the ATs: planning includes test of coherencing algorythm in MR in LR, test of limiting magnitudes with the ATs in
  • MR and LR, check of flux between H and K coherencing: GZi has implemented the fringe centering algorithm and it was successfully tested both in LR and in MR. It lasted until 2am and it was successful.
  • IRIS did not work on AT3. There must be a sign error and a PPRS has been issued.
  • Test of limiting magnitude. We acquired and recorded fringes in LR-K on:
    • HD 91120 (V=5.6, H=5.5, K=5.5)
    • HD 116918 (V=9.1, H=6.5, K=6.4)
    • We did not get fringes yet on HD 117862 (V=9.0, H=7.4, K=7.3) even if we got some signal because of IRIS problems
  • Tests done on Theta Cen (K0III bright calibrator) to see if we have a problem with in MR-H:
    • in LR, we have a ratio K/H of approx 4 (see file H-band-test0001.fits)
    • in MR, we have ADU at lamb=1650 around 600 and at 2130 between 1500 and 3000 so a K/H approx between 3 and 5
    • more test should be done with IRIS working or better sky conditions


    • check beampos and eventually realign
    • update the DIT for each P2VM
    • do cycles on several calibrators both faint and bright, resolved and unresolved.

February 10/11

Night #3 with ATs.

Tonight, we continued our pleasant tests with two ATs:

  • Low contrast observations on very bright stars (Betelgeuse calibrated by Sirius). It is the first time that we get an absolute calibrated visibility with better than 1% accuracy ‚€œat the telescope‚€ .
  • We iterated many times between two calibrators (K=3 and K=4) and a double star (K=4.5) to assess contrast stability. We will need to reduce the data to give answers but:
    • The division of the two calibrators one by the other gives something fairly flat and close to 1.
    • We still have biases connected to image selection by the SNR criteria when science and calibrators have substantially different fluxes, but we should know now how to overcome this.
    • We confirm that the VLTI/AMBER visibilities are quite higher with the ATs
  • We have been testing all night the ‚€œcoherencing‚€ procedure installed by Gerard with full success. Now the fringes are automatically centered at the end of each exposure. We cannot wait to see that working on three T: it will change low resolution observations.
  • We have been able to make a measure (star, fringe acquisition and data recording) in 20-25 minutes repeatedly, in spite of many things (incomplete pointing and focusing tables on the ATs) which need now manual interventions but will soon be updated and re-automated, allowing to be even faster.
  • So working with the ATs might result fairly efficient for stellar physics in Low Resolution, as soon as we have 3ATs with DLs, VCMs, etc... And it will also be a breakthrough in Medium Resolution and High Resolution has soon has we have fringe tracking. It is already partially a breakthrough in medium resolution: for example gamma-vel, eta-car, alfa-ara would be well observable.

February 11/12

Night #4 with ATs: our last commissionning night.


  • align the remaining J fiber for later test on ADC


  • The first part of the night was dedicated observing Betelgeuse in Low Resolution HK, with strong density filters, to test a known object with very low contrast. The filters were necessary to avoid saturating the detector.
    • We observed very interesting visibility as a function of lambda curves (slight raise from 0.05 to about 0.10 in H and decrease from 0.15 to about 0.12 in K, all this with calibrated visibility accuracy much better than 1%. The differential and the absolute visibility curve are very close in shape, except for the reference channel shift. This illustrates the major progress in quality with the ATs.
    • The fringe search procedure and the fringe coherencing procedure behaved quite well in this very low contrast conditions.
    • The differential phases are stable but show patterns which need calibration. A general statement about this run is that we have been more "socks" sensitive than usual with IRIS and the ATs (The socks are the spectral and phase patterns which appear when injection far from the center introduces imperfect spatial filtering).
  • We made some unconclusive tests of the ADCs: we don’t really know what is the J fiber status in the single fiber left on the table (J2 input - J3 output).
  • A systematic spectral calibration of the HR_K mode has been performed without fringes on L Car. Many absorption lines have been detected.
  • Limiting magnitude tests in LR_HK in relatively good seeing (0.8 as, 4 ms) conditions. HD 91120 (K=5.5) was detected easily and show a constant un resolved visibility in 50ms and lost some contrsat with DIT=200ms. HD 116918 (K=6.4) shows fringes and a decent unresolved visibility. HD 117862 (K=7.4) was gave up but we think that we did not fine the fringe position
  • BCD OB was tempted just before Romain left, leaving the end of it to Sebastien and Fabien. A bug was identified that the fringe coherencing is not sending the correct delay to the delay line (the BCD is always IN when the delay is sent). Probably a problem of sign. We lost fringes in the first observing blocks and then we put it back by hand and did not request any centering. We did not acquire a new OB because one of the telecope failed.

Romain, Gilles, and Frederik left at 6:15 in order to catch their bus at 7:30 (Romain and Gilles fly back at 18:30 from Santiago tonight). Fabien and Sebastien stay for the end of the night.

Site map | Home | Admin