So two sets of parameters is a viable option? In theory it sounds good, but as I said there may be something im missing?
Is the process of uploading parameters to the Pixhawk stable? Have there ever been issues with parameters not saving correctly I dont know about?
So the pendulum oscillatory movement that you experienced is a natural long term dynamic mode of the aircraft. To stabilize this dynamic mode the flight control system needs attitude feedback. So the stabilize mode of the pixhawk should remove this dynamic mode of motion if properly tuned.
It’s very doable. You can even pick and choose which parameters that you want to modify. You would do it from the config tuning tab with the full parameter list option.
Bill, have you actually ever tried it? I’m a little long on experience, short on theory?
This is what I use my heli’s for. I have flown everything from MAPIR to towing 12 foot gliders at the RC club. Have tested the FBL extensively to determine its suitability compared to flybar for lifting and flying various loads. It is the nuts and bolts of commercial use of RC helicopters. You do not drop in a different “tune”. You do not do the ICBM Power Launch like the 3D/FBL mentality. You do not just plop it on the ground like the 3D/FBL mentality. You do not have the “locked in feel” like the 3D/FBL mentality. When carrying loads you gain quick respect for the scale pilots that practice velvet smoothness in their flying until they got it picture perfect.
I would normally never fly a loaded helicopter this way. This is just a demo. In gusty wind in the worst place I could find with wind being funneled around two buildings, in a location where I could get some fairly good camera footage.
Oh yes I’ve tried it. With a 46 size flybarred heli. But I certainly pale in comparison to the years of experience you have doing this for a living. There is no doubt that this requires pilot skill. We have a guy that brings in an S-55 to give our students a chance to fly something different. He puts air conditioning units on buildings for a living and other interesting stuff. He demos to the students how to do long line (150 ft) operations with a 50 year old aircraft. The work he does is impressive with an aircraft built in the 1950’s.
My point was that there is a big difference between an external load and internal load and how dynamic modes set up in an aircraft.
Nice video. What was the P gain for pitch and roll? I am assuming it was maybe 0.15. If it was then maybe Tim won’t have issues but I don’t want him to take that chance.
A helicopter should get more stable carrying a load. Not less stable. It is why you don’t see big multi-rotor drones or Osprey’s on the Bakken oil field carrying 20,000lb bundles of pipe to the drill site. They use S-67’s because it’s the only thing that flies in a VTOL platform that can do it.
My personal opinion is that there is a bit much FBL/3D mentality here in this tuning procedure going on? The wrong platform has been selected in the first place. A lightly loaded big rotor is “floaty” and inherently unstable. The stock 700 size blades would be much better and more stable, and turn 'em faster. And it is not a matter of having to have a 800 to lift the load. I hold the current record in the Unlimited Class in the RC club in the Beer Lift with a 700 helicopter. Seven 12 packs (64.7 lbs) air lifted and delivered to the “bar” within the time limit to still be judged “ice cold”. I beat out a 12 motor multi flown by a DJI NAZA that crashed attempting five 12 packs. The multi was a total loss. We recovered the beer from the crash site, except for two cans that broke open.
You have to have some fun too. It’s not all work.
The helicopters with FBL units are TOTALLY unstable trying to do a lift. They are barely controllable at even half their takeoff weight. I don’t know why. But Pixhawk ArduPilot handles it with extreme stability and it somehow compensates for it in a way the FBL units don’t using basic attitude control (Stabilize). With Pixhawk I have found you can place the load off center and it does not make one bit of difference. That will flip a FBL unit-flown heli right over. I think, but am not sure, that is one of the advantages of using EKF State Estimator for attitude control vs just using gyros, and why Pixhawk handles it so well. Pixhawk seems to automatically “trim” the helicopter out for an unbalanced load, and it does it VERY well. It will fly an auto flight with that 2 x 4 strapped in there just like I showed, and it will have no problems. Just like an S-67 carrying a big bundle of pipe though, you keep the speed down when carrying what is a big load for the helicopter. It doesn’t handle as quick, and you don’t want it to handle as quick because you’re moving more mass around.
No, my P gains aren’t even close to .15. I think they’re around .06. I am using VFF about .15 though. I will get the current params out of the little 500 and post 'em so you can take a look. I’m flying custom code in that helicopter to get the DDFP tail to work though. The current release code is broken.
Bill, here is the current params, freshly exported from the heli. A couple notes - I use a V throttle curve. For the load demo I have the top end of the curve at 100% power. That would normally be at 85%.
And the tail settings are for DDFP. If anybody uses these as an example, DO NOT use those settings for a standard tail. It won’t work.
Trex500DFC-PX4-AC3.4.6-June 3 2017.param (15.5 KB)
I really appreciate the experience being brought to bear here, the objective opinions and genuine good advice. What I am most perplexed by at this stage is how far behind I am in progress vs when I set the machine up two years ago and used a regular old BeastX FBL. Maybe 15 flights tweaking a few settings and it just flew. Later on I changed to a Spirit Pro that had built in stabilization based on accelerometers and flew with that up until I tore the heli apart to start this Pixhawk build this winter.
I was using the helicopter with the gimbal to good effect already, no wobble, no pitch bobs and lower headspeed than I am currently running.
Now I understand the difference between an FBL designed for 3D flight vs the Pixhawk which is why I am where I sit now. What is eating me up is the difficulty I am having getting it to fly remotely as good as I had it before? If I could take a controller that was not designed for that type of work and get great results from it, then I most certainly should be able to get a workable stable platform with the Pixhawk. There are infinitely more variables to tune witg the pixhawk vs say the Spirit I was using, so it should be possible. But maybe that is the problem as well? Far more variables to dial in.
Hands down the helicopter will fly at the headspeed it is at and lower cpparrying a gimbal and heavy camera gear as that is exactly what it was doing up until December of 2016, I know its more than possible as it was doing it, just not with a Pixhawk.
Well then why did I evem change people might ask? Well, a $300 3D FBL unit was not designed for professional work where thousands of dollars of camera gear is at stake. There is no built in redundancy like there is with Pixhawk. I needed to move forward with GPS positioning to just plain make my life easier in the field as manually piloting a big heavy helicopter in windy conditions gets more than demanding, I feel Pixhawk is just what the doctor ordered. I was getting to the point where if I had a job to fly I would watch the weather and dread anytime there was more than a 5mph wind. Why? Because although I can pilot the aircraft perfectly well, the more I did it the more stressful it became dealing with the customer, the gimbal/camera operator, the weather and the rigors of piloting a helicopter all at once all the time. Just because you can do everything at once doesent mean you should? Add to that the need for waypoint navigation with mapping… Enter Pixhawk, and here I am… I dont know if its filtering I need or just better tuning , but I know 100% the frame is capable as its already done more with less.
I just know for certain there is no way I am going to settle for good enough with this unit as if I can get a great flying helicopter platform from a peice of flashy consumer hardware than I better be able to do twice as good with a product like Pixhawk.
I know im not using a professional industrial platform. But I have made enough changes with bearings and increased tolerances in gear mesh and linkages to have made it a viable platform already. If things go my way I will be using a custom frame next year. Just alot of CAD and machining to be done and there are only so many hours in a day.
Unfortunately the wife got ahold of me this weekend and stole some more of that precious time to finish the ceramic tile in the master bath… So im further behind with the project now.
In any case soon enough I will get some more stick time to hopefully get closer to the finish line.
Just have to figure out what makes the Pixhawk tick and get this puppy flying good.
Bill, Chris, thanks again for all of the help! It has been immensely helpful, id probably be staring at the helicopter while it collected dust without it!
Lets just say your approach is somewhat unconventional in putting together a camera ship
I don’t even know, at this point, where this idea came from that putting a load on the helicopter causes oscillations? Who thought this up? I certainly hope my demo of what two different types of loads on a little 500 does, unequivocally shows that assumption to be false. Now, if we were dealing with a multi? Yeah. Totally different outfit. Does the helicopter handle different with a load? Yes - the same as any other aircraft. It is no longer any more “locked in” than your car is carrying half its own weight in cement blocks in the back seat.
I’m hoping to see the results of an auto flight with this thing to see how stable you actually have it. A really handy tool, besides the logs, is to hang a GoPro on it and shoot in-flight video. Analyzing the twitches and moves it makes in flight is just as useful as the logs. And the GoPro only weighs a few grams that doesn’t even affect a little 500.
Hah, I think I spit my beer out reading your post. I guess my methods are a bit unconventional… Facing the facts almost every helicopter I have seen carrying a DSLR or other large camera uses a front mount gimbal and has extra lipos in the back to balance out the CG. Me on the other hand, I went my own road and hang a large gimbal under the helicopter. 2 reasons why I chose that after my first build some years ago with a front mount. 1, i have full 360° rotation of the camera independent of the helicopters orientation l. 2, I have a large octa that also uses the same gimbal and with 4 straps and some thumb screws the gimbals are interchangable between frames depending on the need.
So far it has worked, albeit likely by the seat of my pants… Hopefully this time it comes together with a bit more purpose.
First, are you sure you have your facts straight? Not many helicopters can external 20,000 lb. I’m not sure that you are talking about the right helicopter, I think you are referring to the sky crane. The H-53 is about the only other helicopter that can lift that kind of load. Second I’m just not sure about this heavier equals more stability. I’m still looking into the physics on this. More resistant to wind effects, I might buy that. In a general sense, systems where you add more weight will have lower frequencies and less damping. And external loads basically change the system completely with much lower frequencies depending on the length of the line.
I don’t appreciate comments like this. I’ve tried to respectfully respond and consider your expertise as a long time operator in this field. I have 20 years in helicopter flight test and my share of experience with fly-by-wire flight control systems. When you take a flybarred heli and add a pixhawk, you are essentially dealing with a helicopter with a flight director. Not using the rate controller removes many of the feedback instabilities. When you strap on a pixhawk to a FBL helicopter, you are now in the fly-by-wire flight control system world with all of the challenges that full scale helicopters manufacturers wrestle with. The FBL 3 axis gyro designers understand these challenges and from what Tim has said, they work very hard to make these systems tolerant to helicopter configuration. From what I have seen with the arducopter code not much has been done to tackle this. There is no offense intended here. This is an immensely complex field. Helicopters are a vibrating mess and it would be very challenging to design a fly-by-wire system that was capable of tuning any helicopter. Full scale manufacturers spend tons of time and money testing and regression testing their aircraft to get it right. So the time it takes to test model helicopters and get a gain set that produces good flying qualities does not surprise me.
Your previous flights with your 500 prove nothing to me because your gains are not close to what Tim is running and he is dealing with feedback instabilities and you aren’t. More weight and inertia could push his instabilities outside of the notch which could result in bad oscillations. I come from a flight test world and I know all too well that configuration changes can kill you. Certainly we are talking about a $1500 helicopter here but that is $1500 Tim does not want to throw down the drain.
Chris, I appreciate the experience that you bring to the table and I work very hard to understand the claims that you make. It is very cool when you tell me about something that you see and I can trace it back to the underlying physics. Please understand where I’m coming from.
BTW, the competition carrying 12 packs of beer was hilarious. Too funny. Glad to know your helis are being put to good use
Typo. Excuse me. I meant S-64.
While it may not mean anything to you, Bill, I hope at least some folks learned that it’s a myth that you have to retune your helicopter just to hang a camera on it. I digress. This is really not going anywhere. I do not believe in attempting to fix a mechanically flawed design in software.
Let me share my opinion.
With my limited experience, adding mass rigidly helps resist movements and instability which allows for higher gain. Which means without retuning, there won’t be any instability.
It may be retuned for more gains.
I don’t know what happens when a mass is removed from a well tuned heli.
There are plenty of examples of FBL helicopters lifting heavy loads out there, and in the tests they usually started at X weight and increased until max lift was determined. The examples cover both long line style independent loads and loads that are more rigidly attached to the vehicle.
Unfortunately most people are not as thorough as one would like, so its left up to speculation which FBL unit they used and what settings they employed. But they worked nonethless and some quite well with very heavy loads.
What I found with adding a rigidly attached reasonably heavy load with the mass distrubuted much further from the rotor was the tendency to swing like a pendulum. It was brought to my attention long before I built the gimbal and attached it to the helicopter that this would happen, and it did. What I was able to do to remedy the problem differed between the BeastX and Spirit Pro I ran on the helicopter. With the BeastX I had to turn the head gain down as there were no filters in the firmware to employ. That unit was designed to be simple and user friendly with the intention of just flying good enough out of the box for most people. Turning the gain down worked, but I always felt like it was a bit sloppy after doing so.
With the Spirit, I was able to run considerably higher gain on the elevator and aileron axis and employed the filtering to eliminate the slow ossicilation that would set in hovering or when I stopped the helicopter quickly. I guess it only made sence it would do that, swing like a pendulum, as there was a large mass hanging down much lower than the rotor and basic physics principles explain it already.
Using a filter I guess I compensated for this instability in the software.
I understand the allure Chris has with flybar as it compensates mechanically on a consistent infinite basis. But what I have been dealt is an FBL helicopter that requires the firmware and sensors to do the job his flybar does for him already. So I think before I move too much further ahead, I am going to see how the Pixhawk responds to the same load I have already been flying successfully with different hardware.
If there is an issue, I will attempt to try and tune it out and if need be, I will have a couple sets of parameters to swap out on an as needed basis.
Secondly, I am going to try a small auto flight next week, I am very interested in seeing how the auto controller likes the settings I am using currently as I feel it is flying the best it has since starting this project.
Build, fly/test, adjust and repeat…
Just a quick comment, Tim. Pitt is correct. Adding mass to the helicopter damps undesirable (fast) charactertics because the mass resists movement. It may cause a slower movement, but a gimbal is easily able to deal with that. That slower pendulum movement was likely caused by you with stick inputs, and the FBL unit stabilizing your stick input. With Loiter in ArduPilot I do not think that will be an issue because that is what Loiter is designed for. You can make a too-quick stick input to reposition and the autopilot has control and you are just making a request to the autopilot to move it. It is way smoother than any FBL unit can even hope to achieve. Shooting with a FBL unit would be like trying to do it in Stabilize with ArduPilot, which would be a disaster for quality shots or video.
If you tune with a load don’t ever remove it and attempt to fly the helicopter. It will not be controllable.
My thoughts on proceeding with tuning with a heavier load. In my opinion, we need to understand where the instability frequencies lie with the new configuration so that you can more accurately pinpoint the notch frequency. Hopefully you end up with the same freq from before which will help in more quickly tuning this configuration. Just repeat the procedure I gave you but you could probably start with P gains at 0.05 and be sure to use some VFF probably 0.05 so you have sufficient control power. I think this is a safer and more experimental approach than just strapping it on and seeing how it goes. That way you are not guessing which way to move the notch if the aircraft has some undesirable oscillation or poor handling.
This is your aircraft, so it is your decision. I’ll be interested to hear the results either way.
I am going to try a short auto mission coming up snd see how it responds keeping the m/s down, then ramp it up a bit and check logs and watch the helicopter. If it bahaves remotely well, I will save the parameters and come back to it.
Then I am going to add the gimbal into the equation and start all over again with tuning. When I built the gimbal and mount, I designed it to be easily removable for both transport and different modes of flight.
Ideally, as I have said, I will have a parameter set for heavy lift with the gimbal and another once I get into mapping. I dont believe a small gopro sized camera is going to affect flight enough to worry about tuning with it on? Btw I looked into the Mapir Survey2 camera a bit. Is that a good first choice for mapping? Thinking about pulling the trigger on one to get my feet wet, I admittedly have not researched as much as I should regarding the gear and software I need, but it has been in the back of my mind for awhile now. I have seen them in the classifieds here and there and keep hovering over the buy button.
I just really need to get a good block of time to get flight times solidified regarding mapping style missions. To date, all of the testing and usage has been hovering or slow cable shots assuming no forward flight effeciency. I am more than interested in seeing the data. If needed I will drop voltage, increase capacity and move to a bettery with higher energy density per kg. It would be nice not to have to invest in different packs. Its already just shy of $600 for the 4 packs I use currently when shooting for endurance. But as 4 packs really means just one flight pack, for a typical job I hsve enough packs for an hour of flight without charging in the field. So now were at $2400 in packs. Im happy with the packs I use, but there are lighter, higher capacity options for exponentially more money. I really dont want to tell the wife I have to drop $5,000 on packs. She likely is still in shock from last years pack order. Really makes fuel powered helis sound palatable doesent it? But since I am set up for electric, have the support gear and packs it makes sence to stay with what I have…
Definatly going to err on the side of caution with the gimbal. I will take it low and slow and start fresh as you have suggested. A bit of P and VFF sounds like great idea to start. From experience I know it can get hairy if your wandering around alot close to the ground and a gimbal leg catches the ground.
I spent the entire day trying to tune the tail bobble out of my T500 yesterday, but it seemed no matter what I tried, it either had no impact or made it worse from my baseline settings, I found that in FFF in would get a disturbance from a gust of wind, and it would set of an oscillation, and it would sometimes recover and sometimes I took over because it started to look like it could get nasty,
I then reverted to my settings from the beginning of the day, as I’d made lots of changes and now found myself in a state where I didn’t know what was working and what wasn’t… the wind dropped, and I sent it on the same mission I’d been doing all day, and it was beautiful… it’s not the FFF upsetting it, it’s an external gust WHILE in FFF that stuff it up…
any ideas on tuning that? I still have my filters all on 20, my ILMI’s all 0 and IMAX’s all 1 as per default, should I start playing with these? and if so how/why? theses parameters are a bit beyond me…
When you say tail bobble, is that in the pitch axis or the yaw axis. I’m assuming pitch but wanted to be sure.
Hi Bill, yeh Pitch, it started to get an oscillation in pitch, but only when it got hit by a gust in FFF, when the air was calm, it didn’t happen, however in Loiter there was a bit of a twitchy tail in both pitch and yaw, which I assumed I could tune out of it, but nothing I changed seemed to help, so I ended up starting again, I’ll start again by lowering IMAX and the filters, why are there so many params that need reducing from default settings, can the defaults be reviewed?