Recently I built a hexacopter and have flown it a total of three times now. Yesterday was the first time I tried the yaw while in flight. When changing the yaw it dramatically changes the elevation, like 30+ feet of increased elevation for a 90 degree change in yaw. This is very scary to me. Where should I look to resolve this so a change in yaw doesn't affect the elevation?
I'm using the Turnigy Talon Hexacopter frame, NTM 35-36a 910 Kv motors, and APC 12x3.8 SF props. The Arducopter firmware is version is 2.7.1 with everything at the defaults except for the roll 'P' parameter for stabilize mode which is tuned to a lower value from the default (0.08) as the roll was very aggressive with the default. Thanks for your help!
I would strongly suggest you install 2.7.3 instead of 2.7.1
There are many changes and fixes in the new release.
I'll do that, should be able to try it out in the next few days.
What is your throttle level when you hover? It's odd since this version had a piece of code called a yaw differential which was explicitly there to prevent this. We found it had other negative side effects to stability so it was removed and a different approach was used.
Try and lower your Yaw_rate_P.
The hexacopter will hover with around 40% throttle. It seems to be rather touchy and likes to go either up or down rather quickly, but not hover consistently very well. I'll check out Yaw_rate_P. Thank you!
I have a turnigy talon framed cuad, when i started playing with it I had the same problem, I did varios things to help it,
My first advice is check that you are not moving the throttle when you yaw, its easy to do,
second, make sure you have covered the barometer so that wind doesnt affect it,
third , fly it inside in a big space and yaw when hovering, mine yaws without altitude drops when inside , it dropped maybe 10cm, when outside that can increase up to 50cm
I have the turnigy 9x, and i think the pots on the gimbals are bad and dont return to center all the time, which is causing problems, that and the talon airframe being hard to set up well,
are all your motors level?
Luke, are you SURE that you aren't moving the throttle stick when you do this? It's very hard to avoid moving it.
I made that yaw differential code, but I sort of took a stab at a number that worked for my frame. Other frames might need more or less, ideally it would become a parameter. I created it because the during yaw testing, I found that I could not stop the quad from rising during a long yaw (I was spinning the quad continuously). I kept lowering and lowering the throttle, until eventually it actually shut down the motors. However as Jason stated, we pulled it out for 2.7.3 because it caused some other problems.
Hopefully in the future we can put it back in.
Something else you might try, is to lower your Throttle Max. From your description, it sounds like your hexa is way overpowered. Lowering throttle max might make it more docile.
I'm pretty sure the throttle was not changing but I'm off to fly again today and will look at the sticks when I do the yaw to be certain. I will also try out the throttle max, thank you!
Foiled, city wide soccer league practice at all of the parks and schools in the area this afternoon. I'll get back to you in another day or two when there's a safe location to fly.
I have not messed with the throttle max yet but just flew with 2.7.3 and the height gain from yawing is manageable now, about a foot for 90 degrees. I was careful not to adjust the throttle while doing this so I'm not sure if it was something with 2.7.1 or if it was me moving the throttle last time. If this is useful information I can try 2.7.1 again while being careful not to move the throttle.
Well, given the code that was supposed to fix this was removed for 2.7.3, then I'd guess you moved the throttle. Feel free to test again with 2.7.1 if you're up to it.
Out of curiousity, were you yawing left or right? If you look at your thumb you'll see you have a strong tendency to increase throttle when you yaw left, and decrease when yawing right. Just the mechanics of your hand.
Is there some way to log the input signals from the Rx?
That would help find issues like this, or at least point out user error.
I was wondering how this effected the stability?
Thanks for your time.