I am trying to fine tune PID's on my quad and have two questions:

1. In Acro mode does raising the D value make it less likely to overshoot when coming out of forward flight? The wiki seems a little ambiguous on this one. It is entirely possible I don't understand what is written here, but raising the D term seems to make the quad transition out of forward flight much smoother (in ACRO mode)
" A negative D value is used to help the ArduCopter change faster to a level position after forward flight."
 
AND 
 
"Once you are happy with hovering performance, try moving the ArduCopter into forward flight, then stop to a hover. Does the ArduCopter dip too much when transitioning from forward flight to a hover? If it does, the P gain may be overcompensating (too much motor power) when transitioning to a hover. To reduce this, apply a negative D value (try decreasing it by decrements of -5) to reduce this effect."
 
 
It looks like this might be an old out of date webpage that is apparently still live. I thought it was pretty useful however. 
 
2. In Stable mode when transitioning out of moderate or greater forward flight (by simply letting go of the stick) the quad tilts totally the opposite direction and comes to a grinding halt. I would like it to smoothly level itself and continue to glide forward (like a normal real RC helicopter) What should I change? 
 
Thanks!
 
Ted
 
Attached is my current PID config screen

Tags: pid

Views: 1891

Attachments:

Reply to This

Replies to This Discussion

What size motors/props/frame are you running?  I can't take my Rate D over .004 without it going nuts.  Also, I'm curious on how much of the climb is just the increased lift generated by having more copter level?  I know I do ramp up the throttle some to maintain altitude while moving forward.

I have 2836 motors with 12x3.8 props. All up weight is 50 ounces with a 3s pack. I think the rotor to rotor measures 25". Its a homebrew. 

I find that if I am in a hover, slam full forward cyclic and do not touch the throttle, move for a few seconds and then let go it levels and it immediately climbs. Maybe its like a real helicopter and the disc makes lift in forward flight? I'm not really up on the aerodynamics. But when I pitch forward I can hear the motors spin up, so I'm pretty sure this one is in the code. 

My guess on the rate D is vibration. Mine is at .014 without a hind of oscillation. I have isolated the APM from vibration pretty well with thick gyro mount tape (the green goo kind) and also use balsa wood booms which I think dampens vibration. I however have way out of balance props from nicking the driveway a few times. 

I think I will further increase my STAB_P until it oscillates then tune down the sensitivity of the quad with EXPO on my radio. Raising STAB_P made it significantly more sensitive. 

I am really glad we found a solution that seems to help...

Yeah, I'm still going to do a little more logging just because I'm curious, but I would say this is the fix.  What still strikes me as odd as you're the first other person I've seen report this problem.  I open a few threads on this back when I was flying another frame, but never got an answer.  

I'm curious if a lot of people are not flying FPV in stabilize mode or that people are so focused on tuning there loiter they just don't get to this.  Regardless, I updated the wiki with this answer (I know I read it 10 times trying to find the solution to this).

My gut tells me that for the type of flying most users do, this behavior is acceptable and probably not noticed. Us coming from traditional whirly birds we expect different behavior.  And turning up the STAB P makes the quad way more sensitive, so unless you tune it down in your transmitter, most people would avoid flying like this. I say use expo or Dual Rate. 

I am still looking forward to ACRO balance. I would love to only have a gentle force leveling the quad for me, but otherwise be in ACRO mode. And no motor compensation for tilt either. I like to take the automation down a notch or two. 

I'm thinking a hybrid mode would be nice.  Coming from flying helis, I like the rate based control for general flying around because I don't want the system fighting me to go back level, but when I'm just hovering I would like it to stabilize.  Therefore maybe we could get a hybrid system where around 0 stick out to some (adjustable) value (say 10% stick) we are in stabilize mode.  Beyond that it becomes rate (Acro) mode.  It would probably need to blend the transition so we don't get an abrupt change.  Thoughts...

I'm still learning the code, but something the lines of "If stick input > X%" the add in something that cuts back on the rate buildup.  I need to get some cheap small motors build a small balsas frame so I can play around without too much worry.

Hey guys,

So glad I found this thread - I knew there's was something odd about how acro mode stabilizes itself.  I also come from a heli background and after years on the sticks there - watching the quad snap back to level in acro mode just left me baffled.  I've spent the last 3 flying sessions fighting this and although I've gotten it close to where I want it for the type of flying I was expecting - I still fight it on the turns and during stops.

I realized it wasn't me after I saw this video:

http://www.openpilot.org/pid-tuning/

If you skip to 6:00 he does a demo where he locks one of the axis.  He's running a split screen with throttle on the left and ael/ele on the right with his quad tied to two chairs.  At one point he pushes forward then lets go of his stick, and you can clearly see the quad still holding it's forward angle.  That's precisely the behavior I'm used to in helicopters.

Anyway - the bird's flyable, just not what I was expecting for acro mode.  What I can say though is that 2.8's running very smooth for me generally speaking, moreso than the previous release, so the improvements definitely helped.  

My setup:
HT FPV

NTM 2826 1200s

8x4 CF props

3300 3S batteries

AUW: 1350g

My latest PIDs

STAB

P: 3.5

I: .010
D: 001

ACRO

P: .063

I: .020

D: .002  <--- (FYI .008 nearly flipped keep this number low!)

AXIS_ENABLED = 1

I tried lowered Rate P to mitigate the recovery snap, all it ended up doing was making all the controls sluggish.  Guess we'll see where this goes.

Hey Lito,


You were seeing this problems in acro mode?  We were having trouble in stabilize mode.  Just wanted to clarify that first.

Yes in both modes.  Flying aggressively in stabilize mode is like flying through a field of flak.  This "locked in" feeling people are getting continues to elude me.

I get the autoleveling and why it's there, but for me - acro's practically no different than stabilize other than it's a little "zippier".  

I did in fact try to disable axis_enabled and let's just say I'm glad I flew helicopters - because nothing else could have prepared me for what the quad threw my way.  At this point - I'm just not sure what I *should be seeing in acro mode.

Cheers

That's interesting.  My understanding of turning off axis_enabled was that it didn't slowly move back towards the center.  I'll give a shot and see how it goes.

You might try the video tuning guide, basically just stet you stabilize P, then wire it up on a jig and do all the tuning inside.  I started that way and I'm really happy with how my acro is flying.


Also your rate P seems kind of low for your motor and prop size, does it feel sluggish, or did you do that to help with the aforementioned problem?

Turning off axis_enabled turned it into a wobbling mess, but I suspect that had more to do with the low rate settings as you mentioned above than anything else. I probably could have started raising all the values, but I'll be honest - I was so fed up at that point I just slapped it into stabilize just so I could get some time behind the goggles.

My tinker:fly ratio sucks right now - it's like close to 4 hours of tinkering for every minute of flight.

Seeing both of your settings helps me establish at least a baseline of where I'm at, so I may yet revisit this, but only after I get a little more experience behind the goggles.  I do in fact think axis_enabled=0 was what I was after, but I quickly realized that it was more than I could handle this early in the game.  I'll certainly revisit it after I'm more comfortable with the controls, but by all means - please share your experiences if you happen to test it out.

Got some vacation time coming up so if the weather's a bust - I'll see if I can rig up a tuning jig.

Thanks again!

<ETA>

I forgot to mention - when I was flying in stabilize mode - everything you guys mentioned tracked with my experience.  I can't "yank and bank" it's more like yank -> stop -> jitter to level -> yaw -> then pitch forward.  Forward flight - it fights so it's really hard to get a smooth run without constantly pitching up or down.  I've gotten a little better at it just by modulating throttle, but watching the flight videos is like a dashcam on a car driving through a mountain road.

On turns - it wants to maintain it's heading even if I roll right and pitch up and even yaw right (typical coordinated turn in a plane right?).  It's not so bad if I slow down, but when I crank up the speed it just wants to keep heading in the direction it's going.

On turns - it wants to maintain it's heading even if I roll right and pitch up and even yaw right (typical coordinated turn in a plane right?).  It's not so bad if I slow down, but when I crank up the speed it just wants to keep heading in the direction it's going.

I may not totally understand what you just wrote here, but its not like an airplane where with aileron-only input the heading will begin to change as well, you have to basically "rudder" through your turns. When I fly FPV I hardly bank at all, I just yaw to where I want to go and push forward, maybe only a little roll to stop sideways movement. 

When I fly without the goggles I bank a lot in my turns but also use a TON of rudder to make the ship point the way I want...this makes it looks more like a normal airplane or helicopter turn. Essentially, these things don't fly like normal R/C planes or helicopters. I haven't flown my Raptor 30 in a long time, but as I recall I could just use aileron (cyclic) and the turn would not look pretty, but because it has a tail the nose will track though the turn and heading will change. Ardupilot will not do that, and even fight it. 

Yep you nailed it.  I discovered early on that the flight characteristics are just "different" not worse or better - just different and it would be unreasonable for me to think I could achieve that type of feel with an aircraft that wasn't built to do that.

Where I got confused is that Tricopters do fly like that, and as I watch my buddies fly - I realize that's more of the "plane" like feeling I thought I might get with the quad.  I like them both - for different things and as I get more flight time, I'm able to suss out the particular nuances of each platform.

So more rudder on coordinated turns - got it.

RSS

© 2014   Created by Chris Anderson.

Badges  |  Report an Issue  |  Terms of Service