Today was the first flight test of my upgraded Easy Star with the ArduPilot installed. I encountered a few problems that will take some more tests to diagnose. Would appreciate any comments.Here is the setup:-Easy Star with the aileron conversion as documented in this RCgroups posting.-FMA Co-Pilot-Towerhobbies 72Mhz 8 channel RX and Tower Hobbies 6XM TX.-EM-406A GPS-ArduPilot with the latest code (22 Feb I think) set to RTL.Previously I tested the Easy Star with ailerons and the FMA Co-pilot and found the performance to be rock solid. The only addition to today's flight was the ArduPilot and GPS. Take off was uneventful and I waited until reaching about 200 feet with the plane flying directly away from me to switch on the Ardupilot. I encountered two problems that I believe are likely unrelated.Problem 1: Shortly after switching on the ArduPilot, the plane started flying erratically, the throttle and all control surfaces moving in an uncoordinated and erratic fashion. I switched the ArduPilot off and was able to recover the plane. A little more testing showed that he plane behaved similarly, even with the ArduPilot MUX switched off. The plane essentially behaved as if someone was on the same channel and/or general radio interference. There was only one other flier at the field (that I could see, its a big park) and they were on 2.4ghz.Possible Solution: Move the Ardupilot and GPS as far from the RX and ESC as possible. If that doesn't work, try another RX, possibly a Berg.Problem 2: During the brief intervals where I was not experiencing this interference, I switched on the ArduPilot. The throttle behaved as expected and maintained altitude. It seemed like the plane was turning towards the launch point, but so slowly that I had to switch off the ardupilot and bring it back manually or it would have flow out of visual range. Again I didn't get to test this much as the interference problem was popping up frequently.Possible Solution: Even though I have enlarged the rudder on the Easy Star, when the FMA Co-Pilot is on and I try to turn the plane with rudder only, it takes almost full rudder deflection to make a reasonable turn. I believe there is a setting in the ArduPilot code to turn up the gain on the rudder which may solve this problem.Would appreciate any comments if it looks like I missed something.
My understanding is that it's due to the increased drag of the down aileron on the underside of the wing, where the air pressure is high, compared to the drag on up aileron where the pressure is low. It tends to yaw the plane in the opposite direction of the turn, unless you have asymmetrical aileron throws. Shows up most in gliders, which have big control surfaces and move slowly.
"Inverse Yaw" is a secondary effect of aileron or "roll"
When the ailerons are deflected to the left (left stick) the nose of the aircraft (as seen from a pilots view) will move to the right
If the stick is moved to the right the nose will move to the left.
To counteract this add rudder, so if the stick is moved to the left add left rudder and vice versa, for short wingspans using aileron and then rudder to ballance the turn will work, for long wingspans ie gliders try and introduce rudder befor aileron.
Diagnosed my problems today. Looks like there were two issues. First, the crystal in my TX may have not been seated completely which caused the poor range which looked like interference. Secondly, one of the four sensors in the Co-Pilot's CPD4 unit failed. As a side note, it doesn't fly half bad with just three! Now I am just waiting on my replacement CPD4 to arrive and I will resume testing.
Chris: Regarding the aerodynamics, I didn't mention it but I did perform the test you mentioned and found the easy star turned well with the rudder only and the co-pilot did a good job at keeping the wings level. I did make the ailerons fairly large to try and compensate for some of the characteriscs you mentioned.
Did some more testing today, and definitely ruled out any problems from Ardupilot. The tx/rx failed a range check even with ardupilot removed. On top of that, the Co-Pilot seems to be malfunctioning. I am going to call FMA and talk that over with them tomorrow. I'll report back when I diagnose some of these gremlins.
Jason: We've only had bad experiences with EasyStar's with ailerons, precisely because of the inverse yaw problem. The plane is too short coupled, and unless you have very dramatic asymmetrical throws in your ailerons (up is 3x down), it's a beast to fly. We will be releasing an EasyStar version of ArduPilot in the next week or two that has built-in stabilization, but it only uses the rudder--no ailerons.
That said, you can test the aerodynamics of your setup very easily, simulating ArduPilot. Just fly it with the FMA gains set so that it wing-levels well, and then with your RC transmitter, try manually flying patterns with the rudder only. That's just what ArduPilot is doing, and if you find that the EasyStar flies weird in manual mode, you'll then know what the problem is in autonomous mode.
@huge panic - I assume you have turned the gain up on the FMA copilot to commentate for the plane's natural tendency to bank in a turn. At least on the Easy Star, the copilot does a good job of keeping the wings level in a turn with the rudder if the gain is high on the co-pilot. Good luck!
By thee way: I am still testing without any aurduino/ardupilot. The airframe (plane itself + FMA-copilot) needs good flight characteristics!! This is more a fma-copilot issue than a UAV problem.
If i apply rudder, i want a constant yaw rate. Any climbing or banking is not usefull.
Comments
When the ailerons are deflected to the left (left stick) the nose of the aircraft (as seen from a pilots view) will move to the right
If the stick is moved to the right the nose will move to the left.
To counteract this add rudder, so if the stick is moved to the left add left rudder and vice versa, for short wingspans using aileron and then rudder to ballance the turn will work, for long wingspans ie gliders try and introduce rudder befor aileron.
Hope this helps
Ben
http://en.wikipedia.org/wiki/Dutch_roll
Did some more testing today, and definitely ruled out any problems from Ardupilot. The tx/rx failed a range check even with ardupilot removed. On top of that, the Co-Pilot seems to be malfunctioning. I am going to call FMA and talk that over with them tomorrow. I'll report back when I diagnose some of these gremlins.
That said, you can test the aerodynamics of your setup very easily, simulating ArduPilot. Just fly it with the FMA gains set so that it wing-levels well, and then with your RC transmitter, try manually flying patterns with the rudder only. That's just what ArduPilot is doing, and if you find that the EasyStar flies weird in manual mode, you'll then know what the problem is in autonomous mode.
If i apply rudder, i want a constant yaw rate. Any climbing or banking is not usefull.
-
1
-
2
of 2 Next