High pucker-factor oscillations on climb out - gain help needed!

SJBrit

Member
Today I was dialing on the gains on my Tarot 650 build (details here:http://www.multirotorforums.com/showthread.php?22168-Dual-Mode-(Hopefully)-Tarot-Ironman-650-Build). I felt like I was getting the gains where I wanted them but I did have an issue which caused some alarm:

When doing a full throttle climb out, the quad started off very nicely - good response, stable climb. But then a slow roll oscillation began which rapidly got very large indeed (maybe +/-45 degrees although it's hard to say exactly). When I centered the throttle it came back to a level hover very quickly, but I really need it to be stable in a high speed climb.

My thoughts were that this kind of oscillation was due to gains being too high, but I lowered and lowered the vertical gain right to the minimum of 50% and it wouldn't go away. The pitch and roll gains seemed to be stable - nice attitude hold and one small overshoot when centering the sticks from about half throw.

Any idea what's causing this roll oscillation when climbing? It really looked like it was about to flip if I didn't let go and let it settle down.
 

kloner

Aerial DP
most likely a motor or more that's not happy.... pull the props and power it up, feel arms for any vibes as you try varrying throttle, from your explanation, the higher range of throttle,,,, use manual mode to do that with and feel each arm
 

SJBrit

Member
I just tried that - the motors seem fine. They all spun up and changed speed with throttle position as expected. No vibrations in the arms - these are brand new KDEs so they are nicely balanced. Checked the props too and they are perfect.
 



Vertical gains have nothing to do with roll oscilations (and i guess that You are expiriencing those). Check Your atti gains, thay may be too high.
 

SJBrit

Member
Vertical gains have nothing to do with roll oscilations (and i guess that You are expiriencing those). Check Your atti gains, thay may be too high.

I've been thinking my ATTI gains are too high based on some other issues too so I'll see if I can get some flying time in later to make that adjustment.
 

SJBrit

Member
Well, I finally got to mess with this again. All the motors checked out for vibrations under load, so I got back in the air with my gains on two dials on my Futaba and started playing. I found that I had the ATTI gains too high which was causing the oscillation. I actually noticed that this was quite sensitive, with gains set at one point everything was smooth and stable, but just a bit higher and the oscillations kicked in rather significantly.

Thanks for the help!
 

min0nim

Member
I know that each setup has different gains, but can you give an idea of what range of gains you were testing and where you ended up? I am hunting down a similar issue, but related to stability in wind.

I've made a lot of progress with MUCH lower basic gains than I would have expected (still some way to go though until it's perfect), but I haven't really touched the atti gains yet. I'm really not sure how they interact with the basic gains at such a low setting.
 


SJBrit

Member
Here are my gains - this is for the configuration with the 4S batteries and the 17" props (see the link to my build thread in the first post - 380kV motors):

attachment.php


Since I knew from the vertical gain exercise that the settings were going to be low, I believe I centered my gains at 100%, which gave me 50% to 200% on the dial.

The first thing I did was to get the vertical gain set. The manual says to lower that until it can't hold altitude anymore and then turn it back up so it hovers. Well, mine hovers just fine at the minimum setting so I locked it down there. Then I put P/R/Y Basic gains on a dial on my Futaba and the Atti gains on another dial and went out to fly in GPS mode which is the one that will exercise the FC the most.

First off with the Atti gains set low I hovered about 30 feet up where I had a little wriggle-room if things got sticky. This was in a wide open field so I had plenty of space. It was also fairly windy and gusty. I dialed the Basic gains lower and lower until it was having trouble staying in one spot - it was drifting quite a bit as the gusts took it. It always corrected back to the right spot but with the gains low you can expect to see it wander far and take a while to settle back. As I cranked up the basic gain it got better and better at keeping a tighter and tighter radius until it got really quite twitchy. I never got it to oscillate - I think the wind was giving the FC enough work to do to mask that, but I could certainly see when the gains were too high from the way it started reacting very strongly to gusts, but in the ned there was no real improvement in the station-keeping performance.

The advantage of having the gains on a dial is that you can vary them up and down and really get a feel for the performance. I turned my gains up until I wasn't seeing any appreciable improvement in its station keeping, keeping well below the point where it was getting twitchy and "angry" (the only way I can think to describe it!).

Next I worked with the Atti gains by starting with low gains, putting a full forward command on the stick and then snapping it back to center once I'd built up a good head of speed. With the Atti gains low it takes a while to accelerate and then when you center the stick it takes a while to stop but it does it very gently. As you crank up the gain it becomes more responsive and at a certain point makes a very violent maneuver to arrest its forward speed when you snap the stick back to center. This can get into an overshoot and correction, and in the extreme you will lose altitude as the motors are maxed out just controlling the attitude.

In the end there was quite a band of Atti gains that gave me acceptable performance in terms of balancing acceleration/deceleration with the vehemence of the correction when you make step changes, especially back to center (which is where you really see the difference). However, when it came to the problem that started this thread, I found that roll oscillations in a full throttle climb started to happen at rather a sharp cut-off with the gains. So, with the gains in the middle of the band that seemed acceptable from my pitch tests, i went into a series of full throttle climbs, moving the gains up until I saw a gentle roll oscillation starting, and then back down again till it just disappeared. As I said, there was a very distinct edge that got hte oscillations going so this setting ended up being the most precise for me.

And that's it! It flies really nicely now and I have to repeat the procedure on my 6S, 14" config to get that dialed in too. Hope that helps! Take the time to get the gains on a dial or slider on your Tx if you have the capability - I really don't know how you would do this without the ability to vary the gains up and down as you are flying and get a good feeling for the sweet spot.
 

Attachments

  • Screen Shot 2014-10-06 at 6.38.25 PM.jpg
    Screen Shot 2014-10-06 at 6.38.25 PM.jpg
    13.7 KB · Views: 245

Top