Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.mrao.cam.ac.uk/projects/OAS/pmwiki/pmwiki.php/MROIFastTipTilt/TechMeet9?action=print
Дата изменения: Unknown Дата индексирования: Sat Mar 1 21:49:57 2014 Кодировка: Поисковые слова: trifid nebula |
A quick calculation to indicate what the thermal drift of the table tilt might be. Assume that the table is supported by a vertical support (representing the side of the telescope) and a diagonal support running from the telescope to the end of the table furthest from the table. Collapsed into 2 dimensions three items together form a right-angle triangle, and for simplicity we will assume it is equilateral, i.e. the diagonal support is at 45 degrees (the final numbers change according to the tan of this angle, so in this range it does not make too much difference exactly what angle this is). We will assume that all three edges of the triangle are perfectly rigid, but that the joints are flexible, and that the vertical support is perfectly vertical at all times, i.e. forms a reference surface.
First let us assume that the table and the supports are of carbon steel. When the system is heated or cooled (uniformly), all the sides of the triangle change proportionately, so the table remains perfectly horizontal. Now assume that the table is stainless but the supports remain as carbon steel. There is approximately 5x10^{-6} difference in CTE, so the table will not "fit" into the triangle in a perfectly horizontal position, there being a mismatch of 5 microns per meter of table per degree C. The only way the system can remain rigidly attached together is for the triangle to deform. Some simple geometry will show that the deformation is such that one end of the table will, for this geometry, rise or fall by the same 5 microns per meter per C compared to the other end, which means that the table tilts away from the horizontal at just over 1 arcsecond per degree C.
Over 5 degrees C temperature change the table will tilt by 5 arcsec, which is way outside the error budget, which requires of order 0.1 arcsec stability. We can compensate for this by measuring the temperature, this would require knowing the mean temperature of every member of the structure to 0.1C.
These numbers will change by small factors if we introduce a more realistic geometry, but unlikely to change the general conclusion...