On the topic of dampeners, i have the softest ones available in the head right now. That went a long way twords eleminating jello and unwanted tendencies at LHS when i tuned it before.
Tim
Cool. OK. I had missed where you recommended that. I didnāt realize it would accept that without looking up what the range and increments are.
2017-05-18 19-22-53.bin (2.6 MB)
2017-05-18 19-16-43.bin (1.9 MB)
2017-05-18 19-11-41.bin (1.3 MB)
2017-05-18 19-05-10.bin (2.2 MB)
Okay,
So I flew with the D gain added in, no real excitement there. I dunno if it was high enough to have an effect? I have raised the I gain up to .26 on pitch and roll and P gain is still down there at .082. I started to bring it up a bit from .07-.08 and since it was windy, gusts would cause it to wiggle a little and it looked like after quick jabs in cyclic it would wiggle just the tiniest bit. The biggest issue I have right now is a bounce on the pitch axis. Doesent do it on roll, just pitch. I moved the packs around and got CG all but perfect and that didnāt help. If I jab the cyclic forward it does a good up and down bob with the tail if I just hit it and bring the stick back to center. Very much like a rubber band effect.
Tim
Tim, just for giggles I just made a short vid clip of hovering the Trex 500 in what amounts to a wind turbulence tunnel between a row of pines trees and my shop with wind swirling and 10-15mph. With the settings I posted, except I set the VFF back to .15 on both pitch and roll, which really livens it up.
Just wanted to demonstrate that you can make a FBL helicopter hover perfect in Loiter using more VFF and I-gain and less P. It is as close to flybar stability as Iāve gotten that helicopter to fly. Little bit of GPS drift, which is normal. But Iāve never been happier with the tuning in this heli with the DFC head.
So donāt be afraid to try a little more VFF gain. It wonāt make it oscillate. It just makes it sharper and sharper and gives both you and the FC more control rate response, the more you turn it up. Iām running more VFF with this heli than Iāve ever seen anybody use with a FBL, and it works. Thereās no toilet bowling or anything. Itās just about like itās mounted on a post in probably the worst place you can have to fly in. I know for a fact that if I was trying hover my TriCopter there it would dancing around like a drunk bumble bee
.Chris,
So the VFF is used by the flight controller as well? I guess I was under the impression that it was just for the pilot and bypassed the loop. I know its been explained before but I still am at a loss I guess?
What I will say is the helicopter is definitely not lacking in response, even with the P gain much lower where I started it still never felt sluggish at all, it just was harder to keep in one spot with lower P. The I gain has gone a long way towards making it feel more like iām used to and it does stay in one spot better. Still have that darn drift though. Dunno if re-doing the Level is what I need? or trim? Iām quite confident the swash is level and blade tracking looks awesome in a hover at eye level.
Still so much to learn about this system. Wondering if more D gain is a good idea since it didnāt eat itself with the .0005?
Tim
So I still am not confident about what Iām looking for but, I looked at my last flight with the highest I. P & D gain and if I graph out R & Rdes and P & Pdes the plots seem to follow each other closer than what Iāve seen in other plots. Is this a desired thing? Is that even what Iām looking for in the data?
Tim
Oh and Chris, it looks like that 500 is handling the wind well. That is a feature I am hoping to get out of my UAS at some point. Itās usually windy here in Maine and most locations I need to be at are just naturally windy ergo near mountains or in wide open areas In a valley etc.
Tim, normally once you have it tuned you can look at the ATT message and you should see the desired and actual attitudes staying within 1Ā° of each other.
It has to be. It is definitely used by the flight controller in flybar mode. And I think it is in FBL too. The first time I ever experimented with turning it up on the FBL was about 2-3 weeks ago. It was just part of my experiment to see if I could get the thing to fly like a flybar. So I dunno, but Iām happy with it
On your tail rubber band effect, did you try turning up the headspeed about 50 or so? Just about without fail thatās what causes it is headspeed too low.
Bill, thanks for the explanation. Ill keep tuning and see how it goes i guess. Wonder if i should feed any more D gain in? Bad idea?
Tim
Chris, hmmm. Okay, ill add a little more VFF in and see what happens. Im going to split the pitch and roll P gains and see if i can get one up higher than the other. I would like to get this flying a bit better and add the gimbal in. I think for safetys sake ill leave the Canon 5D and Atomos Ninja off it so itll be a few pounds less than it will be when in service but i think ill get an idea how it will handle the underslung weight? Its mostly going to be flown with the gimbal so I guess it makes sence to start tuning with it sooner than later?
Tim
Bill, so using your methodology for viewing the logs it does look like the graphs are getting a bit more similar on pitch and roll. Yaw on the other hand looks like a mess? Something does not look quite right there?
Tim
Tim,
It does appear that the aircraft is starting to track the desired attitude better with the additional I gain however when I look at the I contribution to the command signal it doesnāt look like it is that big, maybe 1/10 the P gain contribution. I think it is ok to add more D gain. What this really does is help the aircraft reject gusts because it anticipates the rates due to gusts by responding to the change in rates. As for VFF, this is contained in the PID rate controller but just takes the commanded rates and multiplies them by the VFF gain. Remember that the commanded rates are not your stick inputs but the rate required to have the aircraft achieve the desired commanded attitude(which is based on your stick). Ok hopefully that makes sense. Itās late.
One last thought, you might try making ILMI = 0.1 for the yaw axis and see if it closes the gap between the desired and actual yaw attitude.
Bill,
Thanks for the reply. I will bump up the D gain a bit and check it out and add in a bit more I gain as well. Finally ill try and bump up the P gain a bit, but ill likely seperate pitch and roll in hopes i get a little further with it before bad things happen. I will likely add in a bit of VFF as well.
I will likely do this in a methodical manner and add D gain, then more I then a bit of P followed by a little VFF and hopefully it becomes more like what im used to with a helicopter.
Ill try the change for the Yaw. I will say, now that im really looking at it it does hunt a bit and doesent seem as rock solid as im used to. I havent done any big collective punches yet to check holding power further as it does just seem a bit weak. Still a ways to go. The numbers Chris gave me did in fact give me a good tail for the initial flights though!. Im thinking maybe some more P gain might help as well?
Tim
Disregard my last comment on raising the ILMI for yaw. Keep that at zero. Raise the IMAX term for yaw to 0.3 or 0.35. It does not have enough authority to null the heading errors.
Bill, okay i will adjust accordingly. I need to do some more research regarding ILMI and IMAX, in the end i would like to have a firm grasp on the proper use and effects of the parameters within Copter.
Tim
Tim, that would be likely. I was running .35 with mine. I had suggested to drop it to .25 for yours because you are running a longer servo throw. I was sure that would prevent having a piroutte on takeoff. But it probably isnāt enough P gain to really get it to hold. And I run quite a bit of D too on the tail, and I think you had turned that down. I put in the D gain until the tail stopped wagging side to side all the time.
This is just me. But I like to set the tail really aggressive for Loiter and Auto so it can fight the wind. But itās too aggressive (for my liking) for manual flying. So I use some yaw expo in my radio and set my RC4 deadzone to 40 so I donāt accidentally bump the rudder when Iām working the collective. But thatās getting deep into personal preference territory.
Chris,
I usually have what most would probably consider a very agressive tail, i usually tune it to the edge for wag/chatter vs holding power for fast backwards flight, funnels/hurricanes and other moves like pogos piros and tic tocs and then back it off a bit. I run 15ish percent expo and 100-115 on my rates for the tail and i have the rates in my 3d fbl at roughly 85% of max degrees/s.
Now thats all out the window as that is specifically for my 3d helis and really holds no bearing on how i want this.
What I would like to see with this UAS build when im done is simply a tail that holds well in a hover, doesent bounce on stops and has a good smooth linear feel when yawing.
The rates dont feel bad at all when yawing as it sits. It just has a sloppy feel to it if that makes sence.
No part of the heli really feels too fast at this point, tail included. It is a bit too snappy on pitch i guess, which is weird as both pitch and roll have similar settings and i thought pitch was usually dampened because of the increased mass off axis fore and aft? I mean, in the end ill probably add in some expo and slow the rates in the software down a bit if thats possible? With this UAS, i think having more stick throw for a given movement will lend itself to videography etc more?
What parameter is equatable to something like degrees/s in sayba traditional fbl? Or is there a rate to adjust? I just feel so inept at this point with Pixhawk.
Tim
The pitch sensitivity is likely a headspeed issue. The rest of the moves you talk about I have no experience with as I donāt do that. I only do scale type flight and some limited sport aerobatics, so I wouldnāt know much about the settings for the 3D stuff. I only know that I donāt like the feel of a helicopter set up for that type of flying, and thatās the way most of them come off the shelf. When I get a helicopter I remove or modify anything on it that has to do with 3D, including the mechanical rates built into it, so it flies more like a real one.
Chris, I appreciate the loads of experience and expertise that you bring to the table. Thatās why Iām always interested in the comments that you make regarding the various flight characteristics and how youāve dealt with them. Certainly head speed plays a big part in sensitivity but head speed affects a lot of other things too like performance and handling characteristics in other axes. But there are other ways to manage the undesirable characteristics in one axis without having to change something that affect so many things. The code offers some options to adjust sensitivity in each axis. The parameters ATC_ANG_RLL_P and ATC_ANG_PIT_P can be used to adjust sensitivity between the two axes. Also you can back off on VFF or P gain, depending on how much damping you have from adding P gain.