TREX 800 Tuning - Severe Roll Oscillation

I just started tuning a TREX 800, and I’m getting extremely severe oscillations on the roll access when just starting the tuning process(using the new method). When trying to adjust the ATC_RAT_XXX_D term, it happens at about 0.004. When the roll oscillation happens, it gets progressively worse, and each time it becomes so severe the control links on the swash plate break and I usually get blade/tail damage before I can get the skids back on the ground and shut the motor down, its always un-recoverable.

I’ve tuned two 700s using the new tuning method and never experienced anything like this, didn’t have any trouble with either of them being when you start to notice roll oscillations, they’re fairly small and they don’t get progressively worse.

Here’s a link to the data flash log, any insight as to the cause would be greatly appreciated.

@223Wylde Definitely exciting a rotor mode but not sure why it is unstable compared to your previous heli’s. Bigger rotor system and heli means the natural frequency of the rotor modes will decrease. I’m not sure why it is becoming unstable. this could be due to the head or blades you are using. I’ve seen these RC heli rotor systems are less damped than the full scale ones and easier to destabilize.
You will have to be extra cautious during your tuning. I would suggest putting the tuning parameter on a knob on your transmitter using the tuning feature of ardupilot. you can set TUNE, TUNE_LOW and TUNE_HIGH to use this feature which looks for the Channel 6 from your transmitter and then scales the input for the parameter you are tuning based on tune_low and tune_high. I’m sure you can search the wiki on how this feature works. That way if you get into trouble you can immediately turn down the value to keep the aircraft from hurting itself.
I would set ATC_RAT_RLL_D to 0.002 and go out and make some sharp stick inputs to ensure you are not exciting anything. Then move to the next parameter.

Ok thanks Bill, yes I think I need to switch to the tuning knob method so I can quickly turn down the rate before it departs from rotor mode.

I’m running the same Align 3-blade rotor head I use on my 700s, it just seems that with the larger blades and a head speed around 200 RPM lower, its far more rate sensitive and the oscillations are much more severe.

I would be suspicious of a dynamic imbalance in the rotor system with a three-blade rigid head. The flight control, of course, will respond to an imbalance that causes a rotor system to go unstable, and may even make it worse.

Rigid head in manned helicopters with three blades are very rare (I think non-existent) because of this phenomenon with three-blade rotors. While the blades may be statically balanced, identifying a dynamic imbalance in a three-blade rotor system is very difficult without aid of a tool like a RPX DynaVibe. You could try a different headspeed. Dynamic imbalances normally change or sometimes go away at different rotational speed.

@ChrisOlson,
the oscillation is not a 1/rev. it is down around 2 hz. Very low frequency. So are you saying the imbalance is causing the oscillation or exacerbating it creating a more unstable system?

@223Wylde I have never seen this head. So there are no flapping hinges (i.e. the grips are rigidly mounted to the head)?

Not to sound like a dick, but would like to ask the obvious: how is the gear slop in the servos? Also, how about all the ball links? I’ve had and seen Align ball links go way too loose. Could it be possible that there is something mechanical that is causing the issue?

I’m not sure. I looked at the log and see it got worse when throttle hold went on as rpm dropped from 1030 rpm to ground idle. Which indicates a rotor system dynamic balance issue, to me. On the DynaVibe we will see the fundamental and harmonics of a dynamic balance problem with the rotor running at flight speed. Whether or not the logs show the fundamental is hard to say. I rather doubt it. All the logs show is an attitude anomaly, or resultant effect on the frame. The sensors that could possibly measure true main rotor vibrations are damped from those vibrations.

I didn’t look at the RPM or the fact that he switch throttle hold on. I assume he hit the throttle hold to get out of the oscillations and get on the ground. The oscillations seen initially are typical of feedback instability that is seen whenever the P or D gains are raised.

Yeah, I saw all the rate roll PID’s were zero. Unless it was adjusted and wasn’t logged. It looked to me like it was just flying on the VFF at 0.15. But I read the first post and it is possible it was to .004 before that runup.

Having it continue after going to throttle hold and shutdown as rotor speed drops on the ground is not normal. But I read it broke either the swash or pitch links too, which may have caused a severe imbalance.

Doesn’t sound like that head is stout enough to handle 800mm blades.

Three-blade rotors can always be a challenge unless the builder can figure out a way to make the head fully-articulated. And even then they are prone to dynamic imbalance issues from something as simple as bumping the landing gear on the ground hard. Unless the rotor system has all design features of a full-size head, including proper coning hinges and hydraulic or elastomeric lead/lag dampers. I am not a fan of three-blade RC heads (at least any I have seen).

He had the D gain for the roll axis at 0.004 for the last flight. He had been incrementally raising them at 0.001 increments for each of the flights.

Ah ok, sorry. I thought this happened just on VFF, failed to get all the information. The broken control link probably caused what I was looking at.

I’ve been running the same 3-blade head on a 700 and its been awesome, I flew many surveys with it last year, logged 35 hours of flight time with absolutely no problems.

I’m using the same head setup exactly the same way with the 800, its new so there’s no slop in the control links, bearings or anything else.

This severe oscillation has happened twice, both times it was while tuning the D term, I think the first time it occurred at .003, and second time at .004, so I’m just going to have to set the D at .001 and call it good.

I’ll switch to using the CH6 knob as Bill suggested for tuning the P term, that way if a severe oscillation starts, I can lower it down immediately.

Once I got the skids on the ground, I killed the motor but it was too late, it threw all 3 control links for the blades and even bent a servo arm, only one blade was damaged so pretty lucky this time.

Will tune the rest of the parameters once I get some replacement parts and let you know how it goes.

Thanks,
Jeff

Jeff,
about 2 years ago I tuned a Jet ranger (TR700 inside) with a 3 blade Align head.
I had 2 cameras going and put small clips together. If you look at that video I think you see similarities what you have seen with your 800 the other day…
At the beginning of the video you see the untuned Heli flying with 0 PID gain with the VFF gain only.
I did cut corners and put a bit to much D in roll ATC_RAT_RLL_D and the tuning was over for that day.
The new 3 blade Align head ripped apart.

Very short video on YouTube from me.

Fred, yes that’s pretty much exactly what happened with my 800 both times, it took a little longer to develop though (maybe 3-4 seconds total), but when it happens the longer it goes the worse it gets, both times just after I got the skids down on the ground and shutdown the motor, the links blew away.

I do think this problem is related to both the D term and the head speed, I think the 3-blade heads have a very hard limit for the D term, with my 800 its .003 to .004, once you find that threshold its unrecoverable unless you can quickly back it down to zero to avoid damage.

Jeff I agree with that assumption.
The early advice for the PID tuning after you settled VFF was that you start with ATC_RAT_RLL_D at .003 which I did. That was to much for my Heli at that time. Now that advice is no more.
And doing it with the trim knob is much safer. But in my case I should had never used those scale blades for the start. That was stupid. There are so flexible.