Unusual swashplate behavior using AC 3.4.5 in ACRO with full cyclic input

I’m looking for some assurance that this behavior is considered normal or the result of a poorly designed test for trad heli community. I loaded AC 3.4.5 from mission planner to my helicopter with pixhawk. I set up the helicopter and even have flown it with this version of firmware. I had ATC_FF_ENABLED=1 and the pitch and roll accel max values at 162000. This test was conducted with the helicopter on the bench with motor ESC disconnected. I armed the pixhawk and raised the collective to half stick. Then I put in full right stick and kept it there. The swashplate initially tilts to right but after a second it flips and tilts to the left. The longer you leave in the stick input, it may tilt right and left a few more times. This seems like an unexpected behavior.

I’ve tried the same test with AC 3.3.3 and this behavior is not present. The swashplate tilts in the direction of the stick and stays there.

I have a theory but need Randy or Rob to confirm it. So in AC 3.3.3, there is a piece of code that limits the overshoot of the target angle to the actual aircraft attitude which prevents the angle from getting beyond 180 deg in roll. I can’t find a similar piece of code in the AC 3.4.5 firmware. So I suspect the issue has something to do with the target angle exceeding 180 deg.

The big question is whether we should be concerned with this? My guess is no since both the multicopters and helicopter firmware builds use the same attitude controller.
Here is a log file of my test.
2017-03-11 19-57-22.bin (1.2 MB)

AC3.4.5, Trex 500 flybar. Duplicated your bench test, unplugged motor, armed. Doesn’t do it here. In Acro whatever direction I tip the swash, it stays there as long as I hold the stick. And I get full swash range both cyclic and collective. And the tail lock gyro feature works when I yaw the heli.

@rmackay9
I tried AC 3.4.5 heli and AC 3.3.3 heli and get the same behaviour as Bill using left aileron. Elevator snaps back only once but stays ‘blocked’ with tiny oscillations, yaw is fine.

In addition to that unusual behaviour of the swash plate I have a problem with the safety switch in 3.4.5. I can press it in 3.4.5 and 3.3.3 to get the solid red LED as expected, but pressing it again in 3.4.5 I get no reaction (LED stays solid) in 3.4.5. In 3.3.3 the reaction is as expected (LED starts blinking). Any suggestions ?

This phenomenon I also found, Acro mode, the swashplate will slowly crooked to the side, even if the helicopter did not change any attitude!

Like this?

This definitely does not happen with my flybar heli with H_FLYBAR_MODE set to 1.

Pitt,
Yes the response if the swashplate when you made a lateral cyclic input is what I’m referring to. I’m hoping that one of the developers, @rmackay9 or @tridge, could explain it. I know they moved to quaternions in 3.4 which makes keeping track of the aircraft orientation easier but it appears the target roll angle keeps on incrementing based on requested roll rate. In 3.3 this error between target and actual was limited. Thanks for the video!

Chris,
Thanks for your input. However the attitude controller handles acro mode differently for flybarred heli’s (H_FLYBAR_MODE=1). That is why you aren’t seeing this behavior.

I had long understood that ArduPilot is no replacement for a FBL controller for Acro, and has been potentially broken at various points in various releases of the firmware for aerobatics. So what folks are seeing for flybarless is not really that surprising.

For flybar it works great. I just test flew my Trex 500 about an hour ago in Acro, specifically to make sure it works because it’s my favorite flight mode for takeoff/landing. It pretty much acts like the controls are just straight thru in Acro mode with flybar. I tried a couple loops, two high speed figure 8’s, one roll, and several takeoffs and landings. Despite my test heli being a heavy underpowered pig on 4S and sort of falling out of the top of the loops and loosing a lot of altitude in a roll when it goes inverted, no problems with full-stick Acro with a flybar in 3.4.5.

Very interesting about Arducopter version and FBL consistency. I read this after removing my MicroBeast from my FBL Heli and just installed a new Pixracer form AUAV. This is my first Arducopter Heli setup and been struggling with 3.5-rc1. I’ve been using it on my hex without issue but it sounds like there are very few heli beta testers out here if you guys just realized this issue with 3.4.5.

So it sounds like 3.3.3 is my best course of action. Does anyone know where I can find it other than GitHub? It was never uploaded.

Use mission planner. It allows you to pick a previous released version.

Good point but I’m now thinking this is the wrong course for me as the older versions will not allow for remapping the ESC away from Channel 8. Thanks for your help but I think my only option is to become a 3.5-rc2 Heli Beta tester. I will mention this issue.

Bob, I think a lot of the folks flying Acro with FBL are using an external FBL unit with 3.4.5. There’s some folks here that know how to set it up. I don’t.

Well I loaded 3.5-RC5 and it doesn’t exhibit the behavior described here. Unsure what other obstacles I may encounter. Now if I can just get the darn motor interlock to clear. I didn’t want to derail this thread so I started an new one in the 3.5 section about that.

I’m not sure if related or not, but a bug was found (and fixed) on loading Heli parameters (https://github.com/ArduPilot/ardupilot/pull/58661). The fix will go out with with 3.4.6 and 3.5-RC3 shortly.

No, this is an incorrect and unfair statement. I’ve been flying Acro on Heli for years, never with an external FBL. I have never installed an external FBL or tail gyro on an Ardupilot powered system, ever. I’ve actually got a few that were given to me and/or came in kits that have never been installed. 1 SK720 and 3 Brain2’s. (Available for sale if anybody wants them)

IMO, using an external FBL controller slaved underneath Ardupilot is a band-aid solution for an improperly mounted Pixhawk. And it really doesn’t even solve the problem with the Pixhawk. You still risk crashing when the EKF blows up. The only thing it buys you is a fully “manual” bailout to rescue it… if and only if you have the piloting skills to fly the heli in rate only mode, from an unknown attitude, and possibly at great distance from the pilot.

As for the original topic… yeah… I don’t know what’s going on here. I will try to look into it. Eventually I will have to be flying on 3.4. I’m afraid that what that will look like, however, is a fork with a lot of cruft removed.

I think what’s going on here, is that, in the past, we used to have a 10 degree limit on the build-up of an target angle error. It appears that is gone. So as you hold the stick over, the target angle keeps rolling around right through 360°.

I have flown AC3.4 in Acro, and I think this doesn’t manifest itself in the air… at least not that I’ve found yet. But it’s still not right.

One issue, is that if your heli does have an attitude control problem, it could get a MASSIVE cyclic output, which could exceed the structural ability of the copter and/or cut the tail boom off.

Have you ever been able to get the damping (P gain) to the level of a flybarred heli or that of what a FBL unit could provide? Just curious.

P is Proportional Gain. D is Damping or Derivative (or Danger, according to some. :slight_smile: )

I don’t know what the damping on an FBL unit would be, or how I would would even measure it. I just know that my helicopters fly very well. Unfortunately I don’t have any good videos of me flying Acro, as I can’t find a cameraman who can keep up.

Best I can do is this:

120 km/h in auto mode. And I have logs with 140 km/h in Acro, pulling 6G loops. with a little Goblin 380. You don’t do this unless it’s tuned in right.

Rob,
Thanks for the input. From what I’ve read and the tutorial on tuning, I thought the issue of limited tuning of P gain due to instabilities was pervasive throughout the trad heli community. Certainly acceptable flight characteristics of an aircraft are pilot dependent. What I might consider easy to fly may not be what you consider easy to fly.

I agree that P gain is the proportional gain and acts on the error of the signal and the D gain works on the rate of change of error which could be perceived as damping because dampers oppose rates. But from the perspective of the aircraft and what the PID is controlling, the P gain is actually affecting the angular rate error and effectively doing the same job as the flybar which is to damp angular rates. The D gain is angular acceleration feedback. I just wanted to clarify my interpretation of impact of the P gain on damping.

Thanks for the feedback on the handling qualities that can be achieved with a properly mounted and sound setup of the pixhawk. I hope to improve my setup with the next version of my X-3. Looking to do some performance testing this summer to look at the benefits of helicopters vice multicopters.