Advice please: What params to tweak to get fast turns?

Hey Folks,

I today did a “speed” test with my Solo, and found that, somehow, I could not get fast turns, that is to keep on going at speed and yet be able to pull the Solo through a turn. Whenever I tried that, and rolled significantly, the pitch and thus the speed was gone.

This does not feel right at all !!!
(I’m coming from 3D helis and have an expectation for how it should feel then one goes through a turn at full speed)

To me it feels as if the MAX_ANGLE setting (30° here) is somehow a combined angle for pitch and yaw. I went through the params, but it wasn’t totally clear to me what to do. So, my question:

What parameter(s) do I have to tweak to get rid of that behavior?

EDIT: forgot to mention, flew with PosHold, if that matters.

BTW: The Solo has the UC4H mod to the motorpods, and AC3.6-dev loaded. The test’s aim was to see if the UC4H mod works reliably, and as expected I couldn’t observe any issues. However, it seems that the speed test has not really stressed the Solo, so I need to do better (next time MAX_ANGLE = 45°).
Strangely, the currents did not went above hover-level !!! I doubt this can be right, even at full stops after full speed the current stays low. Is there something wrong with the Solo battery and AC3.6?.

Thx, Olli

PS: some pics with data of the flight

I don’t fly multirotors so I don’t know a lot about the settings there. But MAX_ANGLE does apply to both pitch and roll. On helicopters I set it up around 60 and that’s still not enough bank angle for a helicopter for a tight turn, but it kinda works. I modified the angle limiter for heli and PR’d that to 3.6-dev because it didn’t work right

Helicopters are essentially different than multirotors because heli’s don’t require much pitch angle to go 30+ m/s. But they require the roll angle to turn. I modified it to go to 95% of max collective before it “kicks in”, but a multi-rotor can’t stay in the sky using that much “throttle” since it doesn’t have anything left for pitch and roll control. But this COULD be the problem with your multi-rotor too - the angle limiter is kicking in.

1 Like

many thx
yes, ANGE_MAX applies to both pitch and roll, but - as it appears - sadly not independently, i.e., both pitch and roll at ANGLE_MAX is not possible, which produces this wrong flight behavior
yes, increasing ANGLE_MAX beyond any limit, so that one never touches it, would be a workaround, Yet, the ANGLE_MAX mechanism is IMHO just not behaving correctly.
I guess this is what I will do tomorrow, but great is this not.
Hopefully further suggestions appear :slight_smile:

That’s kind of what I do for helicopters, and then use the angle limiting feature to limit roll angle if the collective would happen to max out. The only way I’ve ever gotten it to max out with a heli when testing the mod was to make a max power run at 100 mph and toss it into a tight turn.

On multirotors I’d have to look at the code again, but I think it’s limiting it at a certain throttle percentage that is hard-coded to allow enough “overhead” for attitude control. In your log you should be able see what the throttle out is and see if that corresponds to the hard-coded limit when it starts limiting frame angle. I’m thinking that it is, and that’s what’s causing the behavior you see.

1 Like

this makes sense, I will check the log for that
this would imply that I really can’t do much, wouldn’t it, since the throttle range available to pitch&roll is already exhausted … oh well, the Solo isn’t maybe for fast flying
thx

yes, ANGE_MAX applies to both pitch and roll, but - as it appears - sadly
not independently, i.e., both pitch and roll at ANGLE_MAX is not possible,
which produces this wrong flight behavior

It’s called the “circular limit”, and is quite deliberate. IOW, some
people consider it “correct” flight behaviour.

I think it’s most important in something like a headless mode. If it
wasn’t present in headless mode then yawing the vehicle would lead to
changes in velocity, which would be a little exciting. I don’t fly
headless mode, however…

Where the pilot does care about the orientation of the aircraft… it’s a
bit more interesting. Do you think there’s call for having separate
roll/pitch limits and a further option to have the current circular limit?

well, I’m really not in a position to give recommendations here … this today was really the first time I tried to push a quad like this … otherwise I would have noted that earlier. So, that’s something those with more experience in this matter should comment on.
I know this circular thing, and when I was much in heli’s (a couple of years back, since then I fly very occasionally a heli) I actually tried that, you know, there is this cardboard mod to physically limit the movements of the stick. In more acrobatic maneuvers this was a no-go, but for fast crusing with fast turns it was inconsequential. If this would be the explanation, it here would be extremely consequential. Is this known for quads? My current feeling is that ChrisOlson might be right, that I’m already at the limits of what the Solo can do. I guess tomorrow I will retry with ANGLE_MAX = 45 or so.

That’s called the “cyclic ring” or “swash ring” on a helicopter, but it’s pretty much different than what Peter is talking about. My theory is that what you’re seeing is the angle limiter working. It basically prioritizes altitude over pitch angle, in multi-rotors by looking at the throttle out. If the throttle is getting to the hard-coded limit it means there’s just enough overhead left for attitude control, but she can’t handle any more frame angle. So the angle limiter doesn’t allow any more frame angle at that point, regardless of what you have the ANGLE_MAX set to.

If ANGLE_MAX is set to a lower value than what it can handle for frame angle without losing altitude, then it will limit at ANGLE_MAX.

Only on helicopters would it make any sense to have different limits on pitch vs roll, because most helicopters don’t even use the same PID tuning on pitch as they do on roll. Helicopters, although they can fly in any direction, are more like an airplane in flight, i.e. you pull back on the elevator and a helicopter will climb and if you hold it will do a nice loop (with a little collective management). Or if you bank a helicopter in flight, it will do a coordinated turn just like an airplane.

Multirotors tend to fly in the direction they are tipped. They do not, for the most part, have a nose and tail like a helicopter. Nor do they have cyclic pitch to make them go. So excessive frame angle in ANY direction can cause one to run out of throttle overhead and lose attitude control. Therefore it only makes sense to use one ANGLE_MAX that applies to any angle of the frame, pitch or roll.

ok, I guess I’ll then need to think more about that
since this also gives a correlation between pitch&rolI of the type that you can’t pitch then you roll I was assuming that it’s something like this which is operative here

yes, having heard your argument I think too that this is what it is. Indeed, just checked the log, and the motor outputs quite touch the boarders. So, I guess we actually can conclude that as a given. As I said: I’m obviously already at the limit of what the Solo can do.

I think I can live well with that conclusion :). I had from day one the feeling that this is rather an elephant than a panther (which means, the solo’s color is mimicry LOL).

I nevertheless will try 45°, just for the sake of fun.

[quote=“ChrisOlson, post:8, topic:25727”]
Multirotors tend to fly in the direction they are tipped. They do not, for the most part, have a nose and tail like a helicopter. Nor do they have cyclic pitch to make them go.
[/quote]I’m not sure I want to agree with this. At least I flew it like a heli with nose to the front, also in the turns (or what were supposed to be the turns). Watching racer videos they seem to do this too. So, this “headless mode” as peterbarker called it is not totally un-quad-like, IMHO.

I’m talking aerodynamically. Helicopters, as soon as they start to move faster than about 12 kts start creating lift just like an airplane. With helicopters it is called “translational lift” and with one with wheels on it that won’t otherwise take off vertically because of a load, they can make a takeoff run like an airplane and get airborne with the load.

As a helicopter goes into translation lift it noses up - this is called “blowback” when it starts to fly. In hover they are pretty much like a multi-rotor and just beat the air into submission. In flight they are different than a multi-rotor and more closely related to airplanes - even the control inputs are the same as an airplane, short of collective management.

You can force a multi-rotor to have a nose and tail. But aerodynamically they don’t. Nor do they turn the same way as a heli because they don’t actually fly aerodynamically. You can shut down the power in flight with a heli and it is perfectly happy to continue flying. A multi-rotor isn’t. And that basic difference between the two defines some rather significant differences in how they handle. And is why the angle limiting with multi-rotors is quite important. Without differential thrust overhead they can’t remain airborne. A helicopter can because it has an actual rotating wing that gives it VTOL capability and it can fly vertically, upside down, whatever you want, simply by changing blade pitch.

I had this discussion with Leonard some time back and he would like to insist that multi’s are the same and can use the same code. But they can’t - that’s why we got a bunch of if HELI_FRAME statements in the code :grinning:

ah, that’s of course true, I misunderstood this
yeah, at high speed one doesn’t really need the yaw, one can pull through a turn with just roll&pitch … I actually liked that, maybe I should get my helis again LOL

btw, I’ve marked your above post as solution, since I think this is what it is, THX a lot

ok, so I did this speed test with ANGLE_MAX = 45° … and I have to take back quite some of my statements/conclusions
I think the whole story demonstrates my total inexperience with flying quads faster, and that quads fly quite differently to helis

your comment, @ChrisOlson, on the aerodynamics was probably the most precious to me. THX!

but step by step

first, this was now a much better stress test to the Solo, amps up to above 40 A, speed up to above 80 km/h, from that perspective this was much more like what I wanted to achieve. I must admit I’m surprised that with 30° and 57 km/h the current was still nearly at hover level, while going to 45° and 82 km/h made it pushing the electrics. I take this as one point reflecting my total inexperience in this matter. Good lesson.

second, the aerodynamics comment made me realize that the quad just doesn’t lean into a turn like a heli does, the heli is more like a bike in that matter, while with the quad one really has to pull it around with a lot of yaw stick command, it just wouldn’t do such a nice leaned turn. I take this as another point reflecting my total inexperience in this matter. Good lesson.

third, I take back my comment that the Solo is an elephant. It’s a quad. And I was expecting it to fly like a heli. With sufficient ANGLE_MAX and better realizing how to turn it, it can be pushed a bit.

finally, as regards the cyclic, it may well be that the cardboard thing is something different to the cyclic implemented in ArduPilot (I know though that the cardboard thing was coded into heli FBL controllers), but it somehow feels very similar. I could now very clearly feel that then I transitioned from straight into a turn, that the Solo just doesn’t have enough pitch&roll authority while in this transition. This could be this “circular” thing or because it needs additional power to do the yaw and thus limits other things, or such things, but this was not how it felt, it felt like it just takes back pitch when there is a roll. I think it would be interesting to see how it would behave without such a limit, if there is such a coded limit. Anyway, I guess I’m perfectly satisfied now, and this is just a aside. I could push the Solo now to some extend, which was the whole goal of the effort :).

In conclusion, quads are not helis, and this is what fooled me.

As a corollary, I guess I’m never going to like how quads fly at high speed (and they can’t really do acro too) LOL.

Thx so much, this was a good lesson.
Cheers, Olli

Olli, you should be able to leave your ANGLE_MAX at 45, or whatever you are comfortable with, and it SHOULD limit itself to where it still has adequate throttle to maintain stability. I think most quads that hover around 50% throttle are going to be limited to probably about 40 mph or so on speed before they run out of throttle. Of course, a race quad, with much higher power to weight ratio will be able to go faster.

But yeah, they don’t really fly around a corner like a heli. They sort of pull themselves around the corner with power :grinning:

well, the point kind of is, that in straight flight I’m fully satisfied with 30°, I really don’t have to go faster than 50km/h, but for the turns I definitely want at least 45° or maybe even a bit more.
It’s this disparity which makes it flying so strangely. And to me it seems clear that this is not a copter limitation, as suspected before, it obviously can do it, right, as the 2nd test demonstrated. So it’s only and only a control thing, i.e., a thing of whatever code is doing this. Which implies it could be overcome by code :). But that’s another story, I guess.

So, yes, I’ll stick to 45°, not because of the top speed, but because I want to be able to turn within some reasonable time. :smiley: