PPRuNe Forums - View Single Post - ATPL theory questions
View Single Post
Old 6th Aug 2012, 15:46
  #134 (permalink)  
lasseb
 
Join Date: Aug 2009
Location: Copenhagen
Posts: 150
Likes: 0
Received 0 Likes on 0 Posts
Hi Dick
This is actually harder to describe en words than I thought :-D.

What I'm trying to emphasize here is that the schuler error signal is not read from the platform, or memorized, or anything, it is strictly calculated/generated in the computer (based on a math model) and deducted from the incoming signals from the accelerometers almost at the very start of the calculations.

So the signal going to the integrators is = (SENSOR_DATA - SCHULER_DATA_MODEL.)

Regarding initial oscillation you are kind of correct. In theory if the platform is perfectly aligned no error should ever be present, and the platform should never oscillate. There are 2 problems with this. The first is that there is no such thing as perfect, and even small noise levels in the signal wiring will introduce this error. Also mathematical noise in the computer will do this, since we are not dealing with infinite decimals.

Secondly, because we have introduced the schuler compensations, that will actually make the platform oscillate if starts with no oscillation :-D.

Consider the following:
Lets say that the platform is perfectly aligned, dead-level, perfect wires, perfect computer, no oscillation.
The signal input to the integrators, are now the signal from the accelerometers - (which is zero), but deducted for the current schuler period value. So if the platform is perfect level, the input to the integrators would give us the 84 min oscillation signal (or rather the inverted value). This would then lead to the integrators giving us speed, which leads to the tie-mechanism trying to tilt the platform, and then we have the oscillation...

You could say, that because we have introduced the schuler compensation, the platform must oscillate all the time,
If we where only flying around in a 200NM radius (or so) from home base all the time, we could skip the entire schuler compensation mechanism, and use the same tie-mechanism as an attitude indicator. We would then only need to compensate for heading, and it would be much easier.

Hope this answers the question :-)

Last edited by lasseb; 6th Aug 2012 at 15:53.
lasseb is offline