Hi, I am having slight oscillations on my quadcopter, here are the specs:
“DJI F450” frame
10inch props
Matek H743-Slim v3
PWM 30a ESCs
Matek m10Q GPS
Radiomaster rp1 elrs receiver
When in flight and the controls aren’t touched, the drone wobbles slightly, but other than that it’s largely controllable and usable. This is noticed in both althold and loiter mode. I’ve lowered the P value till the drone feels incredibly sluggish and hard to fly, but the wobble only slightly decreased if anything.
Apart from that, I’ve used qgroundcontrol’s basic tuning to make it feel a bit better to fly. My vibration levels are all below 20 on both imus 90% of the time. I have not done any harmonic notch filtering.
I’d greatly appreciate any help on this subject, thank you very much!
I’ve done all of them and no change. The only thing I noticed is that with no wind, the drone flies very smooth, but in wind it becomes quite unstable. In my opinion it’s a mix of imperfect PID tuning and the drone reacting too harshly to drifting from wind. Do you have any other suggestions? Thank you again for the help.
P.s. I can’t upload logs because the anonymizer isn’t working in MP. The initial logs should still be relevant, and I can provide params if needed.
Your Rate PID P&I are unconventional. Set them equal, keep them equal, and tune from there. The Vertical Acceleration Controller gains are too high for the hover thrust you have.
Your rate output seems to be oscillating. And you seem to have an imbalance in the frame judging by your motor outputs. I would balance the frame, one of the motors are working like 200-300 pwm lower than the other. In some instances it could hit the minimum output which would cause issues.
So to adress the oscillation. I think it could have something to do with P/D banalce. I think your D gain needs to be lowered. Well, lower in relation to your P. When you lowered the P gain your would only increase how big D is in relation to P and worsen the issue. I would try lowering the D gain little by little until the outputs stop oscillating.
Once the oscillations are gone, set up filtering and spend some time on it. The reason the D gain could cause this is amplification of noise which can be reduced with filtering. When you have working filters, you can start to tune for better performance.
params.param (34.5 KB)
Here are my parameters after quiktune, notch filtering, and a magfit tune.
As for the imbalance in the frame, I cannot find a major source of imbalance, the only potential imbalanced I see are the GPS mount or the loose power wire.
I have used quiktune since and I’d imagine it’s improved my drone a good bit, but I will still look into your PID recommendations tomorrow because the oscillations persist.
I’ve also flown one of those. An S500 with those plastic arms. They suffered from lots of vibration issues. If vibrations are not too high on this build, one could assume the mechanical vibration dampening is quite good, maybe too good. @Bloxxer, how is the FC mounted. If it’s mounted too softly, it could cause the FC itself to move around which in turn could cause all sorts of issues.
It’s an aliexpress knock off f450, the arms came like that with two different colors. I can’t remember the material but it was something like nylon fiberglass? The aliexpress listing is down now. Also I apologize for not being able to actively diagnose the drone, I am away for a week. Do you think I should try a different frame or arms?
The vibes are acceptable on this drone, usually below 30 and often below 20. I use a 40x40 => 30.5x30.5 mounting adapter with silicone grommets in the fc’s screw holes. I don’t think the fc is moving that much, though, pressing against it only moves ~2cm.
Yes. But putting other arms on those frames is like lipstick on a pig. I had some solid G10 arms at one point which were very rigid but that just chased the compliance back to the frame. Hardware To Avoid
You really mean 2cm? That seems lite lots of play in the mounting. Sure you don’t mean (mm)?
It’s also worth mentioning, if you can’t change the arms, you could still get the thing to fly decent. But you most likely won’t be able to use autotune or any other automated tuning script. Flex in the frame or arms can really throw it off.
I have been there, with flex in the center-hub of the frame. It flew quite great but autotune couldn’t get anything right.
Vibrations between 20-30 should generally be looked at, depending on what you’re going to use the drone for.
yeah mm sorry. about the same mobility as any m4->m3 silicone grommets give
I’m finding it hard to do more PID tuning because I have no telemetry yet, and no real skill or experience with drones. If you would be willing to manually review each log and tell me what to alter, I would be willing to change as you suggest, but if not it’s totally understandable and I’ll buy a different frame. Again, I don’t know about drones but I assume that fixing a hardware issue with software can’t be optimal. What do you think? I don’t want to occupy your time because of my mistake.
Try a takeoff in Stabilise mode, so you can land quickly and easily if necessary. If it seems stable enough switch to AltHold , hover for a while, try some gentle movements.
A more ridged frame would improve your experience of ardupilot. But you cold try to tune it some first to see if you get any good results. Shawn, (xfacta) is really great at tuning. Try his parameters before jumping the gun on a new frame.
The great thing about trying to tune out a hardware issue is that you’ll probably learn lots about tuning, rather than having a perfect frame, running autotune and all done after that.
Read the documentation regarding tuning. There’s some good youtube videos out three as well and search for posts on the forum regarding tuning issues!
I agree the issue here is frame rigidity. The original DJI “Flamewheel” design was so bad, even DJI discontinued it themselves. What you have is the knockoff which are even more flexible.
Hi Shawn, thank you very much for the parameters. I have test flown the drone and it flies far better now. I have no issues but I flew it in a pretty low-wind location today. I did however observe oscillations when i was just above the ground, but not touching, in loiter mode. That is possibly because of ground effect or the propwash and not really solvable.
Can I change back angle_max, and was there a particular reason you initially changed it? I find 30° to be fine for me.
Again, thank you very much. I would really like to learn how to get PID values just from logs. Do you use a simulator for this, or just look at the graphs and interpret them? Do you have any resources about this method of tuning that I could learn from? Thank you!
Attitude control looks quite good, except yaw could do with a bit of work but it’s not critical.
Overall it’s good for that type of frame. They usually end up looking like a flying ball of cable ties and velcro
Set this for the next flight to gather harmonic notch filter data correctly. INS_RAW_LOG_OPT,9
We’ll get that working then you can probably just do an Autotune on yaw.
Those base tuning values came from an old flamewheel that seemed to be a bit sensitive, so they should work OK on almost any flamewheel.
Hi there, an unfortunate update. I flew again a while back(a day after I made my last post), and there was an issue that developed. I may not have extensively tested the drone before declaring it fixed, or this may be a new issue(though no changes happened except an edgeTX update). When I move the drone from still, level position(only after no input for a second or two) to max degrees on a diagonal axis, it will act like it’s going to turn fully, then stop for a quick tenth of a second, then continue to the angle it should be at. I can’t tell if this is a PID issue or radio problem or something else. It also has other rather bad stability issues with rapid diagonal movement, where sometimes if will completely flip out and take a second or two to stabilize. I can’t see anything in the logs but I don’t really do well at reading them.
Attached are two log files, the first was a flight that just demonstrates the “hesitancy” issue and the second has a crash at the end. I really can’t tell what caused it. It was rather close to the ground so it’s possibly my fault(the drone tends to fly slightly down when it should fly in a straight line) so if it looks like a physical collision happened that would explain it, but from my point of view which was around 7m away from the drone, it didn’t look to have hit the ground until it flipped. It flipped before it was disarmed, and fortunately there was minimal damage, just the props and gps stand snapped. If you could interpret the log I would greatly appreciate it.
I have considered switching my current frame for a Holybro S500 V2 frame as an alternative that should be sturdier than the f450 while still lighter than carbon fiber frames, especially because I have plans to add a gimbal and camera to the drone and I can imagine that the extra weight could cause even more sagging. Do you have an opinion on this?
I greatly appreciate all the help you’ve provided me with this drone so far.
I just realized that Axel said he had bad vibrations on the S500 frame. I might look at 3d printing or constructing my own, as carbon fiber ones are super pricey and weighty.
