Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.arcetri.astro.it/irlab/fasti/progress/time/dit_error.php
Дата изменения: Unknown
Дата индексирования: Sat Dec 22 10:30:35 2007
Кодировка:

Поисковые слова: annular solar eclipse
Fasti Noise in the Lab



A More Flexible DIT Engine

2 Apr 2004
Until few day ago, the Fasti way to program the detector effective integration time (named DIT or TINT) was minimal. We waere able to program DIT with a granularity of .1s up to 22s ant with a 1s granularity up to 220s. No longer integration times were available.
Now we implemented a 3 nested delay loops of delay. The granularity of inner delay is 4ms. The delay is now performed using a three level delay loop. This nested loops structure requires three integer numbers, each less than 255, with a total product equal the desired time, each unit worth 4ms. Of course is not always possible to get such a decomposition. We have written a mathematical procedure which tries to get the decomposition for the number requested, and, if this is not possible, add or subtract a small correction until find a suitable triplet of factors.
This approach then can generate delay times which can contain small errors. We have tested all possibilities up to 5000s (the maximum DIT programmable is 125000s). The relative error is plotted in the next figure.

Relative error up to 5000s

Next figure show an enlarged view of the 1-60s area, were we get the larger errors. The higest error value (0.4% of error) is at at a time of 28.1s, not a probable one.

Relative error up to 60s

Last figure show an enlarged view of the 1-600s area, but only for integer values of DIT. The error values are sensibly less (at most 0.08% of error).

Relative error up to 600s for integer values



Last Modified:
28/Jul/2005


Back one page Labir Home