I had a bunch of time over the holidays at my parents house to think, and my mind drifted to adding pitch axis autopilot functions. There are several add-on pitch axis options on the market, but I'm interested in something cheaper and lighter than a traditional pitch servo.

An E-Racer builder came up with an altitude hold solution that used a pressure transducer feeding info to a little microprocessor that then fed commands to the pitch trim to hold altitude. At one point he made and sold the EZ-Trim, but now he has put the plans up for download so you can make your own. More info on how it works.

At first glance, a system that uses pitch trim shouldn't work as well as a system that has a pitch servo driving the elevator. But, there have been quite a few larger aircraft over the years that used flight control systems where the cockpit controls were connected to tabs on the elevator, aileron and rudder, and the main control surfaces were free floating. The pilot's inputs would move the tab on the back of the surface, and the force created by the tab would cause the whole control surface to move. I've never flown such an aircraft, but this type of control system was moderately successful on larger aircraft where the forces to move the control surfaces would be too high for conventional flight controls, and the designer didn't want the complication of hydraulically boosted flight controls.

I realized that I might be able to use the altitude serial data output by the Dynon EFIS instead of using a separate pressure transducer. This would help keep the cost down. Then I realized that the Dynon EFIS also transmitted airspeed and pitch attitude data, so this opened the door to possibly creating a more sophisticated pitch axis autopilot with several possible modes:

  • altitude hold,
  • airspeed hold,
  • vertical speed hold, and
  • pitch attitude hold.

The pitch attitude info could be used in a couple of ways to improve performance. I could use it to avoid large pitch attitude changes when trying to chase airspeed or altitude. Or I could actually make the basic inner loop a pitch attitude hold function, and then have the other modes drive an outer loop which would simply adjust the current pitch attitude target as required to follow the outer loop's airspeed or altitude target.

If I add some means to input target values, I could also have:

  • selected altitude capture,
  • selected airspeed capture,
  • selected vertical speed capture, and
  • selected pitch attitude capture.

Of these additional possible modes, the only one of great interest would be selected altitude capture.

Developing the control laws would be a major effort. I would probably start by creating a simplified pitch axis model to run on a computer, so I could mess around with different types of control laws. Once I had learned what types of control laws were the most robust, I would then move to flight testing to establish the control law gains. Hopefully I could find a set of gains that will work at both forward and aft CG - otherwise I would need to add a CG switch on the unit to allow me to select between two different gain schedules. Or perhaps a potentiometer to allow the most important gain to be varied in flight. I would need a more powerful microprocessor, with more memory than the Parallax Basic Stamp II CPU that Cliff Cady used. He mentions several other possible CPUs.

This type of system adds new possibilities for a pitch trim runaway, so I would probably want to come up with a separate monitor circuit that interrupted power to the pitch trim if some set of conditions was tripped. The simplest condition would be to cut pitch trim power if there was a trim input that lasted longer than X seconds. This would also provide protection against other failures in the main pitch trim system, outside the pitch axis autopilot. The monitor would reset if the trim command went away for more than Y seconds, so that if the pilot triggered the monitor with a too long trim command, he could release the trim switch for short period, and the trim would become available again.

Anyway, I won't put any real work into a pitch axis autopilot for now. I'll wait until the get the plane flying, then I'll start messing around to see what I can come up with. If I come up with anything workable I'll post plans.