From: henry@spsystems.net   
      
   In article <28o2e0lcbokfvmcu7c06p4hap8m9lqbdr7@4ax.com>,   
     wrote:   
   >The mechanical gyro hroizion...   
   >...These devices have weighted vanes that try to erect the gyro to   
   >the local 1G vector. The erection time period is long, but it is there.   
   >Make a standard rate turn (3 degrees per second) and hold that for 720   
   >degrees, when you return to level you will notice a pronounced   
   >horizion tilt.   
      
   Which means that it's *not* erecting to the local gravity vector, it's   
   erecting to the local lift vector.  The assumption is that over a   
   sufficiently long period of time, the aircraft will on average be flying   
   straight and level, i.e. with those two vectors aligned, and so a heavily   
   filtered version of the lift vector can be used as a proxy for the gravity   
   vector.   
      
   Remember, gravity pulls on all parts of the aircraft equally.  It doesn't   
   pull on the gyro-horizon vanes any harder (per kilogram) than it pulls on   
   the wing spar.  Such an instrument *can't* sense gravity directly.  It   
   gets erected by torque on those vanes, and a uniform gravity field can't   
   produce a torque -- it accelerates everything equally.  (Very small   
   torques *are* produced by real gravity fields because they aren't quite   
   uniform, but those are too small to be an issue for this.)   
      
   Consider the aircraft sitting on the ground.  Gravity is pulling on it;   
   why doesn't it fall downward?  Because the ground is pushing up on its   
   tires.  *That* is the force such an instrument senses when the aircraft is   
   on the ground -- not the force of gravity -- because *that* force is   
   applied locally, only to the tires, which then transmit it to other parts.   
   It's the transmission of a locally-applied force that can create torques,   
   and those torques are what erect the gyro horizon.   
      
   In flight, lift takes over the same role... but there's nothing that   
   inherently constrains lift to be (on average) vertical, except the pilot's   
   desire to survive. :-)   
      
   >As for my earlier  statement about MEMS devices being good enough for   
   >atmospheric flight, It would be very difficult to maintain a steady 1G   
   >acceleration with steady state turn rates less than the MEMS drift   
   >rates of ~1deg/sec.  Combine this information with 3D sensing of the   
   >eaths magnetic field, and GPS postion and it is possible to have an   
   >excelent AHRS (attitude, heading refernce system) using MEMS devices.   
      
   Yes, I quite agree -- what I was disputing was the statement that the   
   gravity vector itself is a useful reference, which it's not.  It's the   
   MEMS gyros and the other added sensors, plus the constraints imposed by   
   the aircraft aerodynamics, which make such combinations workable.  Alas,   
   those don't always read over to other applications.   
      
   (Consider putting the aircraft into a very slight bank, while adjusting   
   pitch etc. to maintain 1G of lift.  Because of the bank, the aircraft will   
   be turning slowly; because it's very slow, the MEMS gyros can't reliably   
   pick it up.  Because of the bank, that 1G of lift is not pointed exactly   
   upward, so its vertical component is less than 1G... and so the aircraft   
   is accelerating downward.  GPS position change or magnetic field may   
   reveal the turn, and aerodynamics will try to point the aircraft into the   
   wind and thus reveal the growing sink rate, but the MEMS gear itself won't   
   notice unless cued by one of those things.)   
   --   
   "Think outside the box -- the box isn't our friend."    |   Henry Spencer   
                                   -- George Herbert       | henry@spsystems.net   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)