NavList:

That's interesting, I wonder why only the American ones have a one minute averager. I have attached the operating manual and paragraph 2-12 clearly states that it only runs for one minute.

gl


Werner wrote:

All my Kollsman periscopic sextants with the pendulous mirror (made in Germany
and UK, under Kollsman licence) have the two minutes averager.

Werner


Am Mittwoch, 25. November 2009 20:24:22 schrieb Gary LaPook:
  

 Even if the individual sights within the average are more accurate with
the MA-1 than with the bubble instruments the mirror is still subject to
the same accelerations as the bubble. These are of two types, random,
caused by turbulence, and periodic, caused by the normal oscillations of
the aircraft about its three axes such as phugoid and  dutch roll, with
are sinusoidal in nature.  To eliminate or minimize the random errors
"more is better." Taking a greater number of sights will result in a
better average so taking more sights during the two minute period works
for both bubble and mirror artificial horizons. The natural oscillation
period of large aircraft is about 40 seconds and to minimize the errors
caused by them it is best if the averaging period matches the period of
the aircraft or exact multiples of it.  Two minutes matches three such
periods while one minute is only one and a half periods and so will
leave an inherent error in the data and the bias will depend upon where
in the cycle the observation starts.

gl

douglas.denny@btopenworld.com wrote:
    

May I offer a possible explanation?  It is only a guess however.
The pendulous reference is easier to use than a bubble, being an horizon
indicator rather than a circle in which the star is placed.  This I
believe gives simpler and greater accuracy of observation than a bubble
where the observer has to estimate the centre placement of the star in
the bubble, and hence would require less time to complete an assessment
of the star alignment with the reference.

The oscillation period of a large aircraft I would have expected to be
within a minute anyway,  so anything above one minute is not necessarily
going to increase accuracy on this count. The only increase in accuracy
would be if there are longer term accelerations present (such as an
unwanted inadvertent side-slip) corrected within the time period of
observation.

The only answer must be ultimately that tests would have been carried out
by Kollsman and they probably found little difference in results for the
one minute as opposed to two minute observation periods.

Douglas Denny.
Chichester.  England.