Oh and before i forget again, yes it is a one two bounce. Not something that builds and stays.
Angular acceleration comes from the rate gyros. And it shouldn’t matter where the unit is placed relative to the CG for measurement of angular rate or acceleration. Now because the attitude is most likely calculated with linear accelerations then the placement of the pixhawk relative to the CG then becomes important and I believe it is more about distance from the CG than it is that it is forward or aft. I think what Tim is seeing is more of a dynamic mode of the aircraft. Something like the phugoid or long term mode that airplanes inherently have. That’s why we have flight control systems. They attempt to modify the aircrafts natural behavior to make the aircraft easier to fly. Sometimes designers don’t do such a great job and the control system ends up causing the dynamic modes to change. So the arducopter attitude controller along with the rate controller may be causing this. As I showed in my post regarding technique to tune without instabilities, I had a low frequency oscillation that contributed to the toilet bowl effect. For that PID setup, there was a lightly damped dynamic mode of motion. I used increased P gain to ultimately remove or heavily damp that mode.
That is why I’m interested to see what Tim came up with to fix this or minimize it. Since the notch filter will push some instabilities to the knee of the filter, it is tough to see if it is a feedback instability or just a dynamic mode of the aircraft. Since he was having this problem before incorporating the notch, I leaning more to the dynamic mode.
I would estimate about 1300rpm on the headspeed. I have to get off my laurals and hook up the rpm sensor. im basing my figure on the sound of the rotor and the power out figures and rpm data in the castle logs.
I do have to try the tach you speak of for android though.
I will have to look through logs and match a couple up to changes in the notch frequency. As I said before, even a .5hz change up or down has a dramatic effect on whether I get an ossicilation or not. Its certainly interesting.
Reading the posts, I think more and more that you may have to go thru this tuning process again once you hang your camera gear on this thing. I’m wondering if it wouldn’t be a bad idea to hang some weight on the aircraft in the location that you plan to mount the camera gear. Certainly it will not be exactly the same but at least you will be a lot closer to the configuration you intend on flying this at than you are now.
My other thought is that modes can change between hover and forward flight. If you think that you got this settled down then I would say you should do some low speed forward flight. High enough to get the leaky integrator turned off and start looking at it. Maybe do 10 or 15 knots. Fly it in stabilize mode first and then look at the data for any weird stuff going on in the background. If it looks good then try an auto flight or put it in loiter and drive it around as fast as it will go in loiter. I don’t know but I would think the velocity controller would probably be similar to auto mode. You will gain some confidence that the auto modes aren’t going to excite any badness.
Just my thoughts.
it could be. But I can get that “rubber band effect”, if it is what I think it is, with my flybars too, even in Acro mode (or with no Pixhawk at all) if I run the head too slow.
Being fixated on running excessively low headspeed to get 1 minute more flight time is counter-intuitive to having a stable helicopter at the expense of a little higer fuel burn. And it’s not that much. I’m kind of new to testing electric helicopters, as I like piston engine power. But flying my Trex 600 at 10.2lbs takeoff weight with 8300 6S in it I get 16 minutes if I run the head at 1, 800 or 1,900. Doesn’t make one bit of measurable difference except that I get way less blade chatter in the corners at 1,900 rpm. Turn it up to 2,000 rpm and it’s smoother yet at the expense of 45 seconds of flight time. And that’s still far below the typical 3D headspeed of 2,300-2,400 rpm that most would run with a 600.
The ESC governor simply doesn’t work right below 70% throttle setting (1,800 rpm). And I question if Tim’s is working right. In the one vid I saw him fly, with an unloaded helicopter, he pushed in a fair amount of positive collective and I could hear it audible lug the motor when he did that. Mine will only do that if I use a throttle setting lower than 700 in the params. At setting of 700 or above it will hold perfect headspeed at full collective pitch, regardless of battery voltage.
Chris, unfortunately going up tonsay 1500rpm with my 800 has a dramatic effect on flight time. At 1500 rpm with one set of packs im at 5-6min +/- 30 seconds. At 1200-1300 im at 9 min safely. 1300 is a sweet spot for the helicopter where I usually get plenty of stability and acceptable flight times, i get that to 15min with another set of packs.
I should not those times are assuming hovering videography style work. Not getting any effeciences from forward flight. Im quite interested to see what that brings in the way of flight time increases. ?
Sure. It’s a dynamic mode of the helicopter that seems to be sensitive to rotorspeed. Which also makes sense because higher rotor speed means more damping from the flybar and more damping from the rotor itself. Which would damp the dynamic mode.
I agree, most likely adding the gimbal will change things drastically. just the 6’ wood dowels made a large difference in gains before ossicilation. I was over .1 with the dowels on it, the first flight after renoving them it ossicilated badly and back down to .03 I went and only made it up to .04 in the end without the notch.
Bill, I have to disagree on that? The one and only thing I have adjusted for different loading is the headspeed
Running low headspeed the helicopter has no real lifting power, and whatever it can lift it can’t fly with anyway. The amount of lift of the rotor increases with the square of the rotor speed. With my flybars I don’t have to change a single thing to lift a load except ramp them up. I tested this with my little under-powered Trex 500 on 4S. I test lifted a 2’ foot chunk of 2 x 4 with it on a tether tied to one side. The only adjustment I made was to turn up the throttle setting to 100% power. It lifted that piece of 2 x 4 with no problem, hovered with it in Loiter and I flew it around some with that 2 x 4 swinging in the breeze with no stability issues at all. It simply would not do it at the 2,100 rpm headspeed I normally fly it at.
So I dunno, man. If it was me I would not retune ANYTHING once you get it flying good. Just put your load on it, turn it up until it hovers on 5 degrees collective and it should not make one bit of difference in the tuning. It will handle different and not be near as snappy as minimum takeoff weight. But big (full sized) helicopters are like that too.
You forget that Tim doesn’t have a flybar. So the stability that you gain from the flybar with just increasing rotorspeed with not be the same since most of his stability comes from the pixhawk. Plus I as Tim has said his tuning did change by taking off the wood dowel. Hopefully he can quickly bring up the gains to the settings he has at the current weight but I wouldn’t assume anything at this point.
Bill, please note my 500 is not a flybar. And that is specifically why I tested it. I wanted to know, if I ever went to a FBL helicopter, if it would be capable of lifting and flying with various payloads. A helicopter tuned for one payload like a heavy DLSR, then take the payload off and replace it with a lightweight MAPIR for another job, is totally useless because it is now unflyable.
Sorry. You mentioned flybar in there somewhere and I guess I didn’t catch that you were talking about the 500. Yes I agree that the setup needs to be tolerant of weight but at this point since we are suppressing instabilities I would start in a safe place and then move quickly to the settings he was using at the lower weight. So hopefully we will find out that the PID setup will work across the board with a notch filter in place
What about the prospect of capturing 2 different parameter sets? One for gimbal work/heavy lift and another for lightweight higher speed mapping missions? It should be as easy and uploading parameters correct?
Or am I missing something?
If you cannot load the disc heavier on your helicopter without inciting instability then there is a serious problem with your basic tune.
The helicopter has a minimum takeoff weight 7.2lbs. The green treat chunk of board weighs 3.82 lbs - 53% of the helicopter’s own weight. Dangling and swinging back and forth on a 10 foot rope. Flying manually in Stabilize flight mode. Stock empty weight tune - nothing changed except ramped up the throttle curve (I’m using RSC_MODE 3 with a V throttle curve in this heli). Tether tied on about 1" ahead of the mainshaft centerline.
Absolutely zero problems with stability of the helicopter even with the load swinging side to side jerking the helicopter back and forth while I’m fighting the swing of the load. Strapped to the landing gear on the helicopter so it’s not swinging around on a rope, it’s like the board isn’t even there.
Im going to go out on a limb here and say that adding mass to the helicopter would have a positive effect on stability? Through my testing with pihawk, adding roughly one pound of weight in the form of the 6’ wooden dowels and likely the fact the weight was further out, I found that P gains could be 50-75% higher before issues popped up. Im assuming the weight of the dowels and with the mass further from COG acted like a form of mechanical filter? Am I way off base here? What im getting at is within the control system there should be a way to filter out the instability, kind of like digital wooden dowels if you will?
In any case, if I can get some time tomorrow, I am going to strap the gimbal on and see how it handles it the way it sits now. Ill add some lead to the gimbal plate to simulate the camera/lens weight and another peice of lead where the SSD recorder sits so I dont have to risknso much gear but can still get a very close approximation as to how it will fly.
I think its I’m pretty good shape for FFF now, so I will see if adding the weight and changing the CG to a lower point will require two different tunes?
Unless you do something really radical like hang it all on the tail boom right by the tail rotor, yes. Adding disc loading has a positive effect on stability. But it can have a negative effect if you run out of lift.
Hovering helicopter with a weight dangling 10 feet below it is much different than weight that is strapped to the aircraft and rigidly mounted. This would not likely excite feedback instabilities in the aircraft and thus probably wouldn’t require a change to the tuning of the PID’s. Weight more rigidly attached to the aircraft will change its weight but more importantly will change its moment of inertia characteristics which do play an important part of the aircraft flight characteristics. Again I agree that it would be nice if the tuning was the same and Tim makes a good point that it is easy enough to just load a new param file if you change the configuration of the aircraft
Chris, in this instance the weight hangs under the mainshaft which is where the frame is CG’d as well. I can pick the heli up, turn it on its side and hold the mainshaft/headblock. It will just sit there, no bias to the front or rear.
Now when i strap the gimbal on, the gimbal is balanced as well and its CG point matches up with the helicopters. The only weird thing about the load is that its quite heavy at the bottom, ergo where 3-5lbs of camera and lens hang along with a 700tvl Sony sensor for lining up camera shots. I have the Ninja SSD recorder mounted right under the jello plate up on top of the gimbal as well as the video downlink, lipo etc. I tried to keep as much weight up towards the heli as I could.
I do know it changes quite a bit in the way of handling with the helicopter from prior use. I had to work quite a bit to tune out a very slow pendulum like motion when hovering before. I am interested to see how Pixhawk will handle the gimbal given the difficulty of tuning prior, albeit with a standard FBL. I have access to far more parameters with Pixhawk though so here goes nothing…