LibrePilot Forum
Users => Vehicles - MultiRotors => Topic started by: Muskoka on February 02, 2016, 08:41:28 pm
-
Hi folks, quick question. My buddy and I both have the Totem 250 X quad, and it flips/rolls great with Libre. The only concern is the slight "bounce" when it's completing (coming out of) the 360 degree flip or roll. It's nothing major, and isn't so bad that the quad gets out of shape, it's just something we observed, and wonder if there is some way to tweak it out.
Were observing this in rattitude mode with cruise control, with rate at 700 degrees per second. It does it and all rate settings, just were at 700 now.
I'm also flying a f330 with close to the same settings as the 250 and it doesn't bounce at all. Is it strictly a mass/size issue, the 250 being that much lighter?
Thanks for any thoughts, input.
-
To remove the bounce you should tune PID. See the Easytune feature.
Take a look also to the Attitude "P" factor, maybe reduce a little.
-
We've both used Easytune, and they fly great, solid as a rock in flight. Hate to mess up regular flight just to get the bounce out of flips/rolls. I also used Easytune on the 330 and it flies just as tight, but no bounce coming out of a flip or roll.
The Outer Loop pids are still the default of 2.50 (on my 250 and 330), are those the values you're suggesting we lower for pitch and roll? Again, my 330 is fine, just stating it's on default as well for the Outer loop.
-
Adding a little D gain will help with the bounce. But may dull the response. To high P gain may have the same issue. Adjust P till it barley bounces then add some D to get rid of it
-
Ran Easytune again and here are the numbers for Rattitude mode using Cruise Control. It flies real nice, and flips / rolls great, but still has a bounce at the end of a flip or roll. Not a big bounce, but it's there. It should be crisp, as in snap over, and stop, and it's not. No amount of tweaking the pids seems to get rid of it totally. Again, it does not exist on my 330.
Easytune Numbers:
Roll and Pitch Inner Loop Prop .00713 Int .02496 Der .000096
Roll and Pitch Outer Loop Prop 2.5 Int 0 (Defaults)
These numbers are a long way off from the defaults for a "Generic X Quad". But there what Easytune is giving me.
I've tried Easytune from .00300 to .01000 on Proportional, and accepting the values it spits out for everything else, it makes no difference. It's not a huge deal, it's flying great, but that bounce is annoying.
With these numbers it's right on the edge of oscillating, actually does a bit with quick stick movements. If I go lower than these values the bounce is worse, again not horrible, but more than with the numbers above.
I should add, I've been trying acro+, and it flips real fast, but I don't think there is a bounce it that mode, but everything happens so quick it's hard to see for sure. I'm thinking the bounce is a cruise control issue myself, like it's applying too much trust at the end of the flip, thus causing a "rebound", bounce. No cruise control in acro+.
-
I think it's the transition from rate to acro+ and back to rate. In rate it's using the PID loop to handle the quad when in acro plus I don't think it uses any of the PIDS and goes full manual. That's why it's only used for really fast flips. It's to be able to use low rates and still get fast flips. Try using just RATE mode with like 5-600 degrees a second and like 66-75% expo. Also you may be exceeding your setups ability to stabilize your quad. Some escs just arent up to the tasks of really high rate moves then sudden stops.
-
Thanks for the input guys, but I'm pretty sure were not on the right track yet. If it's a pid issue it's a problem. You can't tune pids for normal flight, getting that locked in feel when flying, and then have to re-tune your pids for flips and rolls in the same "flight mode", because it can't handle a flip properly. That just doesn't make sense. If the pids are good, they're good, and should handle every aspect of flight. We've both been flying for years, everything, so were not knew to setting things up. But we are new to "CC3D" and Libre Pilot in the last few months.
Raining today so I can't test but I'm sure it's the cruise control causing the issue. Like I said, I have my 250 setup almost exactly the same as my 330, same flight modes, same rates, both setup using Easytune, yet the 330 doesn't bounce. And when I say bounce, it's more like a "shake" with a hint of upwards thrust. It's like something needs to be smoothed out with the amount of force/thrust that's being applied (by cruise control) at the end of a flip, because it's really not noticeable in acro+.
Ok, so no pid loop in acro+ (?), but more important I think is there's no cruise control in acro+. Cruise control works fine on my heavier 330, but not so well on either of our 250's. Still say it's a weight issue, and the way cruise control is working.
Were having a tough time deciding on which setting in cruise control to "test" with. We're thinking the "MaxThrust" might be the one, but we're just guessing. It say's 80-90 for multis, and we're both still on 90. Maybe lowering that value will apply a little less thrust coming out of the flip or roll and will alleviate the shake/bounce in our lighter quads? How will lowering that value affect regular flight?
If that doesn't work I guess we'll just have to get away from using cruise control. We both can fly in any flight mode, that's not a issue, but being new to this controller we like to try everything it offers. :)
-
Here's a short video from the other day with a Mobius strapped on showing what were talking about. The bounce / shake isn't extreme but you can see it, and hear it.
https://www.youtube.com/watch?v=rSI9WwY90PA (https://www.youtube.com/watch?v=rSI9WwY90PA)
-
First tune in Rate (inner loop), i see your PID numbers are the same in Roll and Pitch a this should never appends because the weight are never the same in Roll /Pitch axis.
Try to reduce the Attitude Proportional OuterLoop, this can give this result i see in your video assuming you are in Rattitude.
Acro+ is rate around center stick and manual at full stick: Manual means motors are "connected" directly to stick demand without any Gyroscope limitation or stabilization.
-
It's a fully symmetrical "X" quad, not a "H", roll and pitch are the same, weight is balanced evenly, it's fine, and the way it should be. It flies great so the inner pids are tuned just fine. I've already adjusted the Outer Loop from the default of 2.5 for pitch and roll to 1.0, didn't change anything with regards to bounce. Just tried with Cruise Control off, that didn't make any difference either.
The only flight mode where it's hardly noticeable is acro+.
Ok, just tried something else. I lowered the rate for roll and pitch in Rattitude from 700 to 550 and it seems a lot better. But of course now the flips and rolls are no where near as tight. Outer loop pids at 1.0 for roll and pitch still.
I've only tried PwmSync and Pwm at 490 as these Emax simon series esc's aren't capable of oneshot without flashing. Really shouldn't make a difference though, it flies great.
We're going to leave it for now. Fly them, and just accept the way it is. Lower the roll rates, little to no bounce, higher rates, accept a bit of bounce. No biggie, thanks guys.
-
Your quad can be H, X or + the weight are never symmetrical.
If your battery is a rectangle (is really a perfect cube ?) and placed in Pitch axis this cause bigger values needed for Pitch axis.
If you still see rebound in Rate or Acro+ you should tune the Rate (inner loop) first.
OneShot give better response and motor braking, most of the time PID are higher.
-
Yes the quad is perfectly balanced symmetrically, if the battery is placed diagonally across the quad, corner to corner, like it is. Symmetrical quads can always been tuned with same pitch and roll pid numbers, they have to be. Laws of geometry say so. ;) Anything other than a symmetrical "X", ok, but it's a non issue with a symmetrical quad.
And like I've stated numerous times, it flies great. If the pids were out it wouldn't fly great. And, my 330 is setup exactly the same way, using the same techniques, and it flies and flips great, it's only the lighter 250 that has a slight bounce in a flip or roll. My 330 escs are Oneshot125.
Like I said, not a problem, we'll live with it.
Thanks again.....
-
Well, seems you know exactly what you are doing. Please consider i try to help you.
Not only "geometry laws" are taken in account but also inertia and more "laws".
A small example without rebound taken from test flight using a almost well tuned quad, most of them take 0.5s so around 700deg/s
https://www.youtube.com/watch?v=eupiMDGgOUA
Please post a picture of your symmetrical quad.
-
I know your trying to help, thank you. And yes, neither one of us are new to flying or setting up anything rc, but we are new to CC3D (last 2 months). I've flown a Vector, Naza Lite, Naza V2, Tarot, and KK2. Were not new to this, or pid tuning. :)
We're both flying these Totem 250's. My 330 is exactly the same, and has no issues. Here's mine upside down with battery attached. The sat is under the front, opposite to battery leads, balances fine, front to back, side to side.
-
Just attach a picture of your quad, no need a frame picture but your quad RTF.
I you can see the difference from your flip video and mine, you are able to tune properly your quad and matches the same behavior while your quad rotate and stop without rebound.
-
Posted a pic of mine above. My buddies is basically the same. I do notice in a pirouette that it will jump a bit when you stop the pirouette, you can hear the motors rev up a bit as well but only for a split second.
And here's my 330. Used Easytune with it and it has no issues. Perhaps Easytune is hit and miss, and sometimes spits out pids that are no good. I find that hard to believe though, because the 250 flies great in all flight modes, if the pids were off, it wouldn't. And the fact it's both our 250's that are flying slightly different pids suggests that it's more than pids causing the issue.
What flight mode are you flying in your video?
-
If your battery is a rectangular, your PID can't be the same on two axis even if frame is perfectly symmetrical.
As an example, take this dog:
(http://www.atara.com/photo/chiens/2008_08/chiens-Basset-Hound-1d187972-0b5c-76a4-5509-bebe7d31371c.jpg)
Try to rotate around the roll axis and next rotate around the pitch axis (located where is perfectly balanced)
Which axis is more difficult:
- to rotate when start from 0 velocity ?
- to stop when full velocity ?
-
Our quads are symmetrical, in shape and balance. That dog is far from symmetrical, lol. He/She looks more like a H quad. ;)
And even if there not to the even to the gram, the controller should be more than capable of handling a slight difference.
The fact that my 330 is setup exactly the same as our 250's seems to keep getting overlooked. It's exactly the same as the 250, only a bit bigger, and it does not exhibit the bounce behavior that our 2 250's do. And the pids on the 330 are the exact same for pitch and roll in every flight mode. Symmetry is not the issue, nor is having the same pids for pitch and roll, the 330 is enough proof.
If it works for a 330, it should work for 2 different 250's, I can't stress that "fact" enough.
-
Your battery is symmetrical ?
Can you ignore the battery weight and how is the volume ?
What the ratio battery weight / Quad weight ?
Play with your battery between two fingers, roll axis and pitch axis...
This dog equal your battery, attached to a perfect symmetrical frame.
-
Honestly, your argument is moot. The 330 is just a larger 250, setup exactly the same, shaped exactly the same, flown exactly the same (flight modes) and it doesn't have the same issues. That disproves your theory right there that symmetrical quads can't have the same pids for pitch and roll. If that were the case then the 330 would bounce as well, it doesn't.
Regardless, it's bouncing in both directions, pitch and roll, so are you suggesting that Easytune spit out "bad" pids for both axis. There's nothing wrong with the pids, as I've said way too many times. If the pids were off, it wouldn't fly great, end of story.
We've adjusted the Easytune pid values up and down with our pots on our transmitters on the fly only doing flips (one axis) and there isn't anywhere from .002 to .01000 for pitch proportional that they don't both bounce. Yet I did te same on the 330 and it was pretty much fine from the beginning.
Thanks again, but I really don't think we're going to solve anything here. Back to flying.
-
You're just lucky with 350 quad and the 250 quad need to be tuned.
The two quad are not tuned well, first because the PID are bellow the limits (no rebound) and second because too high (rebound)
Nothing more, see and compare videos.
Another approach is the Optune, old method. this consist in finding the UOV on roll and next for Pitch.
Take a look here:
https://www.youtube.com/watch?v=tT4_XwAzhYo
Follow web link from video for calc sheet.
Article here:
https://250drones.wordpress.com/cc3d-tuning-guide-using-optune/
-
The 250's were tuned, using Easytune, just like the 330. Guess we won't rely on Easytune anymore. ;) Oh wait, the 330 "is" fine.
I'm done here, this is going no where. We'll just tune everything (pids) manually from here on.
-
It's a PID issue for sure. The bounce. It's a tell. To high P, not enough D. Something. Let's say we have 2 250 quads and 2 330 quads all running the same motors esc and battery's cameras everything. 2 identical 250's and 2 identical 330's. 2 are just bigger. The two that are small and the two that are big will have different PIDs...period. All four will have different PIDs. Guaranteed. Fly in RATE mode, use high rates to get the speedy flips, expo to back off the crazy... And tune it manually from the start. its really not that hard from a template anyway.
In clean flight the same thing happens in their version of acro+. The reason your quad bounces is your flying in rattitude not just stright rate or acro+. And if that's not good enough. 2 quads, 4 quads no matter how many quads with exact hardware, frames, software will never have the same PID and be perfectly tuned. They will all be different.
now. Fly in rate.... Turn D down a touch. Adjust P where you think it's good. And adjust out the bounce by adding in more D a little at a time.... Maybe yours is just broken... Yeah that's probably it