> This $30 PICAXE board can be used to build a cheap digital auto
         > pilot. You can use an RS232 input from a GPS
         > with a NEMA 1083 output and control a model airplane
         > servo connected to an aileron trim tab.
         > The electrical connection from the PICAXE to the
         > servo could not be easier.
         >
         > The servo shown puts out 20 inch pounds of torque.
         >
         > GPS NEMA 0183 data out looks like this second jpg.
         > This ASCII data is sent out ever 2 seconds from the GPS.
         > Any cheap $89 dollar GPS with NEMA interface will work.
         > Most have the interface.
         >
         > The job of the PICAXE is to read the heading in from the
         > GPS and compare it to a heading entered into the PIXAXE by
         > you using a single push button. Fly a steady heading and push
         > the button to record that heading in the PICAXE. From now on
   the
         > PICAXE will hold the heading. The numbers in this case are
         > 181.9 degrees. (Field 8) These ASCII character codes are
   converted
         > to a binary number and this is added to or subtracted from
         > the preset initial heading binary number. The difference
         > forms a pulse width sent to the model airplane servo on
         > the aileron trim tab. A simple programming job. It does
         > this every two seconds. The rest of the GPS data is ignored.
         >
         > You could use a separate PICAXE for altitude hold and direction
         > because they are so cheap. The altitude hold would work much
         > the same way only it would read a Free Scale pressure sensor
         > every few seconds and control a servo on the elevator trim tab.
         >
         > --
         > Paul Lamar ...No rotor no motor.
         >

        Neat! I have to get used to the idea of a single-job computer. :)

        Doc Custer
        Retired Software Engineer and sometime Audiologist



   I generally agree that this is a slick idea, but I think your AP really
   also needs instantaneous turn rate (and probably bank angle)
   information.  I think otherwise it will either have such low gain/weak
   response as to seem ponderous, and unable to deal with turbulence, or it
   will have weird non linear instability problems..  GPS data by
   itself is
   too slow to provide information that will make an AP work.

   Bob Nuckolls has mentioned that what most GA airplanes need is a GPS
   enhanced wing leveler.  I think I agree with him.  The GPS portion will
   work essentially as you describe, but will serve another useful purpose
   - to correct for drift in an accelerometer which will provide turn/bank
   rate information.  The accelerometer replaces the gyros which drive most
   AP's.

   The problem with accelerometers is that the cheap ones have somewhat
   high rates of drift.  Drift is the accelerometer telling you that you're
   turning when you actually aren't (or the opposite).  Drift values change
   relatively slowly - change with temperature and other environmental
   effects.

   GPS provides a means to monitor and correct for how much the
   accelerometer is drifting..

   The picaxe can read the output from a chip like this:

   http://www.sparkfun.com/commerce/product_info.php?products_id=700

   And know roughly what the turn rate and bank angle are.

   To start, the picaxe reads the GPS at least twice - two track datapoints
   to see what the long term turn rate is ( track1 - track2 )/time = turn
   rate.  Between GPS samples, the picaxe also takes A/D measurements from
   the digital gyro (dg) chip.  The dg might say you were in a 3deg/sec
   turn (standard rate), but if you weren't actually turning at all, the
   GPS data would show that.  When you engaged the AP, it would know to
   offset the dg data by 3deg/sec.  Assuming you didn't start with any
   skid/slip the picaxe could derive roll rate/bank angle too - useful for
   generating the signal for the servo and counteracting turbulence.

   I'd have 3 modes of operation.  Wings level/no turn, standard rate left
   and standard rate right turns.  A plain switch on the panel enables the
   AP.  A momentary switch on the stick/yoke overrides the AP allowing easy
   maneuvering.  The turn switch could be on the panel or on the stick.

   This would be a fun project...


   Regards,

   Matt-

   I am not so sure it would not work Matt. The GPS puts out
   heading to .1 degree. Some do it once a second. A rate gyro as
   you say can drift but the GPS system always knows exactly how far
   off course it is. I think it is worth a try with out the rate gyro's.

   An airspeed sensor might help to change the gain as a function
   of airspeed when diving or climbing. Of course those solid
   state rate gyros in model helicopters work amazingly well.
   Model helicopters were unflyable until those came along.

   You will probably need the gyros for standard rate turns if
   your going to use something so cheap for IFR :)

   Paul Lamar




Maybe Paul..  It's probably worth a shot.  I agree that high precision
track data will help some.  I suspect it won't be possible to make the
system have very graceful response in turbulence however..  A lot can
happen in 1sec.  I don't need an AP when the air is smooth - the trim
wheels are pretty easy to set..

If GPS data by itself doesn't provide adequate performance, later adding
a gyro ic wouldn't be too tough (in terms of cost or time).  Doing so
would guarantee good behavior.


Matt-

Mine has been broken so long in my C182 I forgot how it handled
turbulence. I hit my head on the roof with my seal belt
flying through the Banning pass one time. I can't imagine
any auto pilot handling that. People have reported a 90
sudden bank angles :)