Hey guys I built a 450 sized quad that weighs right at 1200g. It is built around a multiwii 328p board. I have been tuning it indoors in a big multi-purpose building at my church. Indoors it flies awesome. It feels super locked-in and stable and takes great video. But when I take it outside even on a very mild day it gets very twitchy and feels mushy and erratic. Any ideas/tips for tuning it for good stable flight outdoors when the air is not perfectly calm?
Help tuning quad for windy conditions...
Try dropping your gains by about 10-20%, and if you haven't done anything to reduce vibe, do so.
I've found the happy gain levels drop when in wind, fast forward flight and generally higher throttle. My theory is the higher throttle and wind generate more vibe and attack the gyros. Drop the gains and reduce/dampen the vibe and the vibe's feedback drops to liveable levels.
I've found the happy gain levels drop when in wind, fast forward flight and generally higher throttle. My theory is the higher throttle and wind generate more vibe and attack the gyros. Drop the gains and reduce/dampen the vibe and the vibe's feedback drops to liveable levels.
Thanks Dan, I will try decreasing gains. I have been in a battle with vibes for a few weeks now. It was flying great for just messing around, but then I put a camera on it. Lots of jello. So I separated the camera from the main frame along with the battery (for mass) and it really helped the video but then I had vibes going to the FC. grrrr. So now I have the flight controller on moongel and it flies great and video is excellent indoors, but outdoors is a different story.
My take from my observations and the building of about 12 multirotor systems now IMHO.
These "toys" we build are not as simple as they seem. They are VERY complex systems. Each part of the whole has a resonate frequency/s and each part added to the whole adds / subtracts / changes the resonate frequencies of the system. Any little change in ANY variable whether materials / structural changes / electrical / magnetic changes or flight conditions...... will cause a change in the resonance of the system and thus change the flight characteristics, sometimes in a minor manner and sometimes in a major manner. That is why each craft performs so differently from any other home built craft even if the same kit or design.
Thurmond
These "toys" we build are not as simple as they seem. They are VERY complex systems. Each part of the whole has a resonate frequency/s and each part added to the whole adds / subtracts / changes the resonate frequencies of the system. Any little change in ANY variable whether materials / structural changes / electrical / magnetic changes or flight conditions...... will cause a change in the resonance of the system and thus change the flight characteristics, sometimes in a minor manner and sometimes in a major manner. That is why each craft performs so differently from any other home built craft even if the same kit or design.
Thurmond
Tigert,
For a PID controller (or PI in this case) the "I" portion of the control loop becomes a moving average, with the gain being the "depth". While the the signal processing concepts are quite different between "I" loop and running average, in implementation it results in the same algorithm.
. . . but by all means, if you've got the tool chain all set and the Atmel ASM skills, do tinker with it. There's only a handful of developers dinking with the KK2/2.1 roms, and if you can get your bearings in the code and ideas for improvements, I think they'd welcome the branch.
For a PID controller (or PI in this case) the "I" portion of the control loop becomes a moving average, with the gain being the "depth". While the the signal processing concepts are quite different between "I" loop and running average, in implementation it results in the same algorithm.
. . . but by all means, if you've got the tool chain all set and the Atmel ASM skills, do tinker with it. There's only a handful of developers dinking with the KK2/2.1 roms, and if you can get your bearings in the code and ideas for improvements, I think they'd welcome the branch.
Dan,
I am actually on a Multiwii board, not a kk2. In the multiwii software there is a section for gyro filtering and one of the options is moving average. That is what I am trying...here is a pretty good video showing someone using both low pass filter and moving average gyro filter: http://vimeo.com/79349443
I am actually on a Multiwii board, not a kk2. In the multiwii software there is a section for gyro filtering and one of the options is moving average. That is what I am trying...here is a pretty good video showing someone using both low pass filter and moving average gyro filter: http://vimeo.com/79349443
sorry tigert, getting the problems mixed up.
interesting. I'm just scratching the surface of multiwii, and that's, well, different. From a signal processing aspect, turning on a moving average should have a similar effect of turning up the I gain, but I suppose order of operations might be applying here -- do a portion of gyro I, lowpass then do it again. have to ponder that, but please report back what you've see . . . you've sparked my curiosity
interesting. I'm just scratching the surface of multiwii, and that's, well, different. From a signal processing aspect, turning on a moving average should have a similar effect of turning up the I gain, but I suppose order of operations might be applying here -- do a portion of gyro I, lowpass then do it again. have to ponder that, but please report back what you've see . . . you've sparked my curiosity
Fixed it! The problem: the way the flight controller was mounted to the frame. Inside it was fine, really good in fact, but outside, the wind was "wiggling" the FC and causing the twitchy, wobbly effects. I completely redesigned the way that the flight controller attaches and it is 100% better. BTW I messed with the moving average gyro filter and was not pleased with the results. I am still using the low pass filter (in my case 42hz works best) and I am very pleased with how solid it is. So, if you are experiencing twitching, look first to vibrations getting to the FC and it could simply be the way it is mounted.
So the quad is flying so much better now that I have the board mounted properly and the vibrations are under control, but I am still baffled by some of the videos where guys are flying and just seemed to be locked in...in pretty windy conditions. What is the key to getting a stable quad in breezy conditions?
Practice and skill
Are you running simonk on your ESC's? That made a huge difference for me, almost as big as going for original wii sensors to the '5060 sensor. I still get pushed around by the wind...but don't really notice any extra wobble. Hardest part is just anticipating what the wind is going to do next
Thanks for the comment. I am using Turnigy Plush esc's so I can't do SimonK but I did flash them with BLHeli and that did make a HUGE difference. I am pretty happy with some of the little improvements here and there...practice is probably the biggest key now...I think what I am having is A.S.A.V.E. (Awesomely Smooth Aerial Video Envy) lol
You are totally right about lowering P gains...for me, that helped a bunch, escpecially on the roll axis. I think it is because the quad is a "spider" type frame with the battery in back and the camera in front so it would require more power to pitch that mass than to roll it. Should I just assume that any of the ultra smooth multirotor videos that I see are shot with a gimbal? Or are they just shooting on perfectly still days?
Last edited:
