With first light approaching quickly (15th May, 1998 for the internal technical first light and 25th May for the official scientific first light) the activities in Paranal for the final integration of the first Unit Telescope have been frenetical.
The telescope has been used almost 24 hours per day and the great amount of people involved made necessary careful planning and shifts for telescope time.
I have worked mainly at the base camp during late morning and afternoon and at the telescope during the night, up to about 03 in the morning, using the early morning for sleeping. In this way I have accumulated more than 12 working hours per day.
During this mission, the activities on Paranal (for what concerns Telescope Control Software) have been focused on two main aspects:
During this mission we have done the first tests on the sky, using a small 13 inches finder telescope installed on the main structure of UT1 and equipped with a technical CCD (the VLT 8 meters mirror will be installed at the end of april, together with the secondary mirror).
Obviously a lot of time has also been spent in discussing and solving general problems, emergencies and bugs that do not fit directly into these categories.
A number of important pending points or things to be discussed are listed at the end of this document.
A big part of the Telescope Control Software has been extensively tested and integrated on the NTT in La Silla and on the VLT Control Model in Garching, but a relevant part of the software contracted out could not be tested in this way, as well as the interactions with the real hardware.
For many modules, this test period actually provided the first possibility of testing the interaction of the software with the hardware and with other modules.
The system was proven very stable and no major problems have been found. Obviously a lot of minor bugs had to be fixed and many small changes have been necessary due to interface incompatibilities between modules.
Most of the problems have been found in the startup and shutdown phases. Typically, after a painful startup (that sometimes required a couple of hours to identify every night new unexpected behaviours), the telescope was working very fine and without problems for the rest of the night. Must be also taken into account that the startup was very often affected by changes in software and hardware done during the day and impossible to be tested beforehand.
We also run all the test procedures foreseen for the Milano test period and that we never had a chance to run because of Ansaldo delays. We just found minor bugs.
A few man days were also spent to improve the VLT user interface panels, adding functionality, improving usability and reducing CPU load.
We also made some preliminary Auto Guiding tests, mainly to verify CPU load on the alt and az LCUs when error corrections are sent.
In this period the following main subsystems have been integrated (at
hardware and software level):
In order to avoid any possibility of damaging the expensive 8 meters Zerodur mirror, all first tests on telescope mechanics have been performed by using a dummy concrete replacement of approximately the same weight.
This has allowed a proper alignment of the telescope structure, calibration of the altitude and azimuth encoders and basic tuning of the control loops for the altitude and azimuth axes.
After this first phase, a small 13 inches guide telescope has been installed on the structure and equipped with a standard Technical CCD. This has allowed to point for the first time the telescope to the real sky and to make tracking and pointing measures.
The first night with the guide telescope has been, and in particular the search for the first star, have been really thrilling, with a brave colleague (Martin Cullum) hanging on the telescope structure at about 5 meters from the ground. With the help of the classical piece of white paper, he looked through the objective of the guide scope, while we were moving the telescope trying to find our first target: Sirius, the brightest star in the sky.
When this appeared in the field of view of the small telescope, he plugged in the CCD camera and we got a first, blurred image. Then he focused the camera rotating by hand the mounting screw and we finally got the first real image of Sirius.
After this "quest for the first star", looking for other stars has been much easier and we needed just to look around the preset telescope position with small offset steps.
I have been really surprised and impressed by the fact that already with the first 6 stars the Pointing Modelling Software has been able to produce a pointing model that was always putting the searched star within the field of view. This despite the fact that the guide scope was not aligned accurately and was installed on one side, off the telescope center axis. (The Pointing Modelling Software takes as input the theoretical and measured position of a set of stars in the sky. It produces as output a set of correction coefficients that, applied to the theoretical altitude and azimuth of a target object, return the actual values to be passed to the telescope to point to the object. The pointing model corrects for many typical mechanical errors, like axes misalignments, rotation, tube flexure and so on).
During that and all the following night we have build more sophisticated pointing model and measured tracking performances by making long exposures with the guide scope camera and by "writing VLT" in the sky moving the telescope with predefined offset commands and making "star trails" using additional velocity commands.
This has clearly demonstrated (as was already known by analysis of the measured tracking position error), that the telescope was suffering from a major altitude oscillation problem (+- 2 arcsec).
We spent a lot of time trying to track the source of the problem, both in hardware and in software, but without success (the problem was tracked down to a defective electronic board a couple of weeks after the end of my mission). For more details on the testing activities, look at the VLT log book and at the daily logs (not available online).
The more frequent problems we had were with the hydraulic bearing system and with unexpected interlocks in the az and altitude axes control.
We also took some very nice images of sky objects: even such a small telescope, affordable for any amateur astronomer, can produce beautiful images (probably even artistically better than the once produced by a big telescope, thanks to wide field of view).
Last modified: Thu Dec 30 15:00:54 MET 1999