[muchzill4]

Hi,

I'm a happy flyer, but there's one thing bothering me. My quad oscillates when I apply throttle at certain forward angle when recovering from a drop. It flies really well apart from that.

Problem visible here:


It's much worse when I fly without a camera (i.e. quad is lighter)

PIDs attached in uav.

And here:

Code: [Select]

            <field name="RollRatePID" values="0.00267000007,0.0101399999,3.60000013e-05,0.300000012"/>
            <field name="PitchRatePID" values="0.00350999995,0.0133300005,4.70000014e-05,0.300000012"/>
            <field name="YawRatePID" values="0.00710000005,0.0134899998,5.99999985e-05,0.300000012"/>


[NicholasDavid]

Looks like prop wash. Applying throttle to "catch" the quad and finish the split S

*disclaimer* I am new to tuning. I like to "think" I know what I'm talking about.

I think the D term is supposed to help tame prop wash oscillation.
If not using TPA. Maybe use it to increase D at the low throttle to mid range. If you are already using it maybe just play with the percentages.

I do believe it is the hardest "behavior" to tune out. For all flight controllers across all tuning software. Cleanflight/beta flight, race flight ect.

I have a similar issue with my quick 180 turns. I do the mini,quick split s turn to change direction more often than not. And I get it sometimes and sometimes i don't.

I'm only now diving into tuning my own copters. If you don't mind I have a question for you, your quad seems to fly smoothly. So I'm flying a 220 H quad on 4s. And my I gains are 4 times my P across all axis. Even more than 4x on yaw. Has that been your expirence? I'm looking for some general numbers people are ending up with. But I feel like I'm making solid progress tuning so far.
I believe it was liftbag that told me he uses I gains that are 4x his P on all his quads. So far I have tuned my 3s X quad and this current 4s H and I am finding this to be the norm. Also now that I think of it it makes since that tuning an H quad could require a little more I gain (on yaw)than the other axis.

Basically in forward flight or cut of the throttle it yaws. With I at 4.5x the P on yaw it's much better. Am I doing it right?  Lol.

[muchzill4]

I originally used OPTune (now EasyTune) to tune my quads, so never adjusted each of the values on their own. Meaning I'm probably in no position to make any statements about PID values.

If it comes to magic 4, as a multiplier of P to get the I, it seems that it's more or less true. If you look at PIDs I've posted, they do look like more or less I = ~4 * P. Not the case for yaw though.

The quad I'm flying is 220 too. https://www.instagram.com/p/BF_uthVw6Kd
As you can see on the picture, props are unfortunately really close to each other on the pitch axis, which may (or may not) affect the oscillation effect I'm having.

I'm jumping between dRonin and LibrePilot recently. I've managed to get rid of these oscillations using autotune with a bit lower motor io curve fitting (their equivalent of TPA, as suggested by one of devs).

Autotune is next step from easytune - I like the fact that I get very flyable and responsive quad after the process which takes ~1min, no fuss afterwards. I know there's people in both camps though, but I much prefer less config more flying approach if it's achievable.

[NicholasDavid]

I'm kind of trying it all out and figuring out how the LP PID controller works. I seem to be going in the right direction. I want to learn to tune this cc3d/LP combo effectively.

Well as I have continued to tune this copter more. I have raised my P on yaw higher than roll and pitch. Which seems to account for my higher yaw I gains, sticking with the "about 4x P value idea". I'll keep going and report further progress.

[TheOtherCliff]

Running without a camera changes the weight and thus the PIDs.  In a perfect world we would retune for that.  It sounds like the PIDs are too high.  You may want to reduce PIDs or better yet, put reduced PIDs in a different bank and configure to use the other bank (Flight Mode Switch setup) when adding or removing the camera.

[Mateusz]

Hi @muchzill4,

I just saw your post with idea of "two camps". To me instead it seems most of people here are very enthusiastic and excited about Autotune. It really improved over time, has solid mathematical background and if all works without disturbances it should give you awesome PIDs very quickly. I believe every modern flight software should have this option as it is such an invaluable aid in getting perfect tune!

Now putting how great it is aside, I hope everyone agrees with me, it should not ever be treated as a complete blackbox use blindly replacement for an informed user doing PID tuning! User should always inspect PIDs estimated by algorithm, not only because you might bumped into something, had a wind, or flown close to the ground, but simply because it may just happen to fail for case no one predicted yet.

Two of builds I tested, one was flying great first time, second was consistently getting micro-oscillations and motors were getting hot, that is now fixed by limiting yaw factor. Other person I heard had problems with Pitch to Roll values higher incorrectly for the type of frame he is using (X or H don't remember). I can imagine user may want to use Easytune in those cases if something similar happens.

It's tempting to use AT right away, especially because it's convenient and does not require investing time or even worst case learning about tunning, so there is a risk someone might abuse it. Different story, a guy on forum who had red alarms and was warned at least three times, that this is very wrong and not suitable or safe for flying, he sturbonrly kept setting "always armed" to bypass safety checks of the firmware. He even argued with devs, and some time after came back complain that he crashed in front of university with lots of people watching his quad burning with smoke...

It's great that pilots like you use AT, as it gives more confidence and trust in it that it really works, and useful feedback which benefits community. Other side of a coin is that people like the guy above I mentioned likely will abuse this module...

As for oscillations, I would try to tune for nice not oscillating flight in hover and then use TPA.