Ok good to know, thanks Bill, will give that a try and see if it improves.
I flew earlier this afternoon, with the CG adjusted, that helped a little, maybe re-tuning with Chris’s method as you described will resolve the remaining pitch issues, will let you know.
@223Wylde The Trex helicopters seem to have a tendancy to be bouncy on the tail. And I’m not 100% sure why. I think, but not sure, that the tail on those is too light and doesn’t provide adquate damping of the pitch axis. That’s only a theory based on every flybar Trex model I’ve had does it until I increased the headspeed enough to make the flybar more effective.
My 500DFC used to do it until I put a DDFP tail mod on it. The extra weight of the 3508 motor out on the end of the tail boom totally solved it.
A 500DFC I set up with a APM2.8 a couple weeks ago did it.
It has also been asked about literally dozens of times on HeliFreak and the common advice is, “increase your headspeed”. But I’m wondering if there isn’t a couple design flaws in those:
1.) The tail is too light
2.) The location of the FBL unit or flight controller is far enough from center of rotation of the pitch axis that it senses excessive acceleration in pitch and over-compensates for it.
That could be, but I managed to get one flight in late today, was hoping the wind would go down, but it was still pretty windy, the pitch oscillation is gone after re-zeroing the PIDs, I did one short flight to log some pitch & roll data with the VFF set to .15 for both, but I had trouble maintaining height with the wind gusts, it wants to rapidly drop when the wind backs off and I bounced the skids kind of hard a couple times, but I could definitely see the VFF rates need to increase for both. Will try to finish the VFF rates tomorrow, then will carefully tune the PIDs using your method, I have a feeling its going to work this time, but will post the results back here when I’m done. Thanks for all the help, I really appreciate it!
Yes, it’s definitely the PID loop that makes it worse. But why it does it to a greater extent with the Trex helicopters is still somewhat confusing to me. And it’s not unique to ArduPilot. The FBL units do it on the Trex models too. Since most people aren’t setting them up for UAV use the common recommendation is always increase headspeed to get rid of it.
Align went thru (I think) three frame revisions on most of the models. The lastest ones have the FBL unit mounted up front and internal frame battery tray to get the weight of the battery down lower. The earlier ones had the battery mounted on a nose tray at an angle. I’m pretty sure the way the helicopter is balanced has something to do with it too.
I think you’re definitely on the right track about the pitch oscillation cause, I built this 500 about 3 or 4yrs ago, the battery is mounted above the ESC pretty far forward at an angle, so its difficult to balance CG. I later upgraded it to a DFC head but everything else is stock. I think I’ll end up using a very low P pitch rate once I get through all this. I’m just using this 500 to get the new tuning methods figured out and get everything functional, then I’m planning on building a 700 which I plan to use for aerial survey work, I’m a Certified Photogrammetrist, so I typically only do vertical photography not video, and I prefer to carry DSLR cameras directly below the CG on a fixed mount, I don’t use brushless gimbals (waste of payload in my case).
I’ve built two large quad X multirotors which work great, but I believe I can much better flight time with a 700 carrying the same payload setup the right way, and a TRAD heli will be a nice option that will have certain capabilities much closer to my fixed wings. I’ve just been waiting for the firmware to get caught up, its definitely vastly improved from when I originally built this 500, I had a lot of difficulty just getting a stable hover 3 or 4yrs ago!
Yep. I think those frame revisions they did was to make it quicker handling and better balance. They had the v1 E models (which are still my favorite with a few mods), the v2 L models, and now the v3 X models. Very few frame parts are interchangeable between those models, the most notable difference was moving the battery under the mainshaft in the tray and the new electronics deck they got in the front. And, of course, wider frame to accomodate the in-frame battery tray.
I agree a gimbal is wasted weight on a helicopter for still imagery. They are way ahead of multi’s in stability. This is an example of what kind of rolling shutter-free video I can get from a well-tuned Synergy 626, no gimbal, solid-mounted GoPro under the nose. If you can get this kind of video you can get super-sharp still imagery at the shutter speeds common in decent lighting.
I had play as above, and found that the balls on the swash plate driving the blades, was only 4.75mm dia and the links were 5mm. sockets. I had to update the swash plate balls to 5mm dia. That took all the play out, now have no slop at all.
Ok, just got back from the field and got through the tuning all the way, I really like this new method. I did manage to get a small amount of D term in this time, that definitely helped. The pitch oscillation is pretty much completely gone, its barely perceptible now, I tried reducing the P pitch a little after the final values, but it didn’t matter, so I think the only thing that would eliminate it is adding a little weight on the tail as Chris mentioned. It flies beautifully now, and I left it in position hold for about 5min with winds at 15mph gusting to 20, and it hardly moves, now its rock steady and extremely stable. I still need to tune the yaw, still running mostly default values, but its flying extremely well.
Here’s my final values for reference, TREX 500E DFC, Align stock servos, linkages:
ATC_ACCEL_P_MAX 85000
ATC_ACCEL_R_MAX 100000
ATC_ACCEL_Y_MAX 27000
ATC_ANG_LIM_TC 1
ATC_ANG_PIT_P 4.5
ATC_ANG_RLL_P 4.5
ATC_ANG_YAW_P 4.5
ATC_ANGLE_BOOST 1
ATC_HOVR_ROL_TRM 300
ATC_PIRO_COMP 0
ATC_RAT_PIT_D 0.0004
ATC_RAT_PIT_FILT 10
ATC_RAT_PIT_I 0.2
ATC_RAT_PIT_ILMI 0
ATC_RAT_PIT_IMAX 1
ATC_RAT_PIT_P 0.04
ATC_RAT_PIT_VFF 0.2
ATC_RAT_RLL_D 0.0004
ATC_RAT_RLL_FILT 12
ATC_RAT_RLL_I 0.17
ATC_RAT_RLL_ILMI 0
ATC_RAT_RLL_IMAX 1
ATC_RAT_RLL_P 0.04
ATC_RAT_RLL_VFF 0.17
ATC_RAT_YAW_D 0.003
ATC_RAT_YAW_FILT 20
ATC_RAT_YAW_I 0.12
ATC_RAT_YAW_ILMI 0
ATC_RAT_YAW_IMAX 1
ATC_RAT_YAW_P 0.18
ATC_RAT_YAW_VFF 0.024
ATC_RATE_FF_ENAB 1
ATC_SLEW_YAW 6000
Ok yes, that will definitely help, loose ball links would be really bad
Awesome Chris, I agree Trad helis fly significantly more stable than multis, I think due mostly to the gyroscopic effects a massive main rotor generates, and even large quads(mine have 17" rotors), still don’t even have close to the disk area a 700 or 800 has. And I do plan to mount the battery packs low on the 700, I’m not going to use the battery tray(my packs won’t fit anyway), I’m going to mount all my electronics on the top side of the bottom plate, battery packs will go on the sides of the frame down a bit lower.
These params here can be adjusted for the limits of your I-gain. I think, unless they changed some scaling again, 1 is a lot of buildup for your I-term. Typically around .44 would be used on pitch and roll and maybe .3 or .33 on yaw. This is the percentage of your I-gain that can build up in the event the helicopter doesn’t change attitude right away.
The I Leak Minimum param sets the minimum value for your I-gain - the value that it is allowed to leak down to on the ground and in hover. The I-gain does not become fully active until the helicopter is moving at 5m/s for 2 seconds. Otherwise the rate integrator has a “leak” that leaks it off so it doesn’t tend to cause your helicopter to tip over on the ground due to I-term buildup. For instance, if you set the ILMI to .03 that is the minimum amount of I-gain it would carry both on the ground and in hover, and it provides a sort of “auto-trim” for hovering carrying a little bit of I-gain.
I typically don’t use it myself because I don’t do any hovering work. But if you do hovering work you might want to consider putting in a little minimum I-gain. You have to experiment with it. Putting in too much can cause your heli to tip on the ground if it’s sitting on a slanted surface.
Ok good to know, I’ll adjust those, I don’t think I changed them from the defaults. I typically only fly survey missions, so the only hovering will be just after takeoff and after it does an RTL, very little hovering at all.
Well, yeah. But too many of the defaults are for multi’s and not helicopters. I just checked and they didn’t change the scaling again recently, so 0-1 is the adjustment limit for IMAX, and I’d recommend like .44 for pitch and roll and try .30 on yaw. That’s adjustable in increments of .01, and it corresponds to percentage. With 1 being 100% buildup, which would be rather dangerous with a heli than can snap a 180 degree roll faster than you can blink. Multi’s don’t respond that fast, so I suppose they have it set to 100% for those.
The ILMI params, since they don’t apply to multi’s, aren’t even listed in the wiki Full Parameter List.
The state of the current firmware development and documentation, as far as watching out for Trad Heli users, is horrible. That’s why I fly my own custom firmware.
Ok makes sense, documentation is almost non-existent, and yes I’m also wondering what other parameters need to be adjusted for safe auto missions, are there any default failsafes or other items I should turn off?
Not sure what speeds I’ll use yet, but guessing it will be in the 10-15m/s range, I always have to make sure I’m not out-flying the camera, so also depends on focal length and flying height, but I think that will be the typical speed range.
Turn off the battery failsafe for sure. It doesn’t work. Well, it does. But it has a serious flaw for heli and you’ll lose all control of your helicopter if it kicks in in Stabilize flight mode. I found several other serious problems with 3.5 when I was testing it. I think only one of them that I know of actually got fixed.
There is no heli maintainer anymore. So 3.5 isn’t really even beta quality for helicopters. As I mentioned before, the servo check when you boot the flight controller doesn’t even work right. So just make sure you test every function you intend to use under controlled conditions before you actually fly it for work.
For battery low charge use your ESC’s low battery soft power cutoff instead.
Ok thanks, very good to know. There are some aspects like that in the fixed wing and multirotor firmwares as well, I always thoroughly test everything before I put a camera on board, and I also run Li-ION battery packs, so voltage characteristics are vastly different from all the typical settings anyway. A good preflight checklist is a must, so I’m used to not relying on the internal startup checks anyway, especially those related to control surfaces.
I’m well aware of the history on the Trad heli side, and I know there’s a much smaller user base too, so much more difficult to catch problems in the firmware or settings, that’s why I really appreciate the help here, this is huge!
Thanks again!
Jeff
I think you already have Alt Hold and Loiter working. So those should not be an issue. The default WPNAV Accels are a little low for a helicopter on survey flights. You’ll find the heli slows too much in the turns and won’t accelerate back to target speed on the next pass with the defaults. So you’ll have to turn those up - especially the horizontal accel. I usually fly 13m/s on 5 second image sequences with a MAPPIR camera at 75 feet over terrain. I think I’m running about 400 on horizontal, default is 100. You have to kind of tune that so the helicopter does its accelerating in the turn and comes out on the start of the pass right on target cruise speed. My new one will cruise at 13m/s at only 4 degrees down pitch. I got channel 6 set up to change the WPNAV speed with 5/10/15 m/s presets on a three-position switch, and that switch is mixed with a knob so I can infinitely adjust the cruise speed so it doesn’t over or under-fly the camera.
Ok great, yes I ran the calcs, and 13m/s be the maximum survey speed for me, so will probably fly at 12m/s. My large quads pitch about 8-10 degrees at 8m/s, so 4deg pitch is outstanding at that speed!
I was wondering how to adjust the speed in the turns, so that will really help. What do you use for overshoot on the waypoints at that speed?
Best way is to illustrate with a screenshot of an actual flight plan
The helicopter hits the turn entry waypoint at cruise speed. It banks over at max allowable frame angle at the apex of the turn and only bleeds off ~1-1.5 m/s. Gradually bleeds off bank angle leaving the apex and pours the cyclic to 'er to accelerate back to target speed. It hits the turn exit waypoint flying level and true, right on target speed for the next pass.
That’s probably about 180 feet of overshoot shown there for 75 feet above terrain passes, because the passes are spaced at 200 feet.
Chris,
You use spline waypoints to achieve that smooth transition from leg to leg?
