(Tradheli 4.0.2&4.0.5) do not go to waypoint and somehow lost control and went missing

Dear TradHeli Community,

Our team has been developing a 800 Size UAV Helicopter base on CubeOrange autopilot and 2 Here3 GPS via UAVCAN

We have setup the systems on two of our Helicopters with CubeOrange+V-bar(Flybarless system) They were set up as follow:
Prototype 1.1
FC: Cube Orange+Vbar Neo
GPS1: Here2 via CAN1+CAN distribution board
GPS2: Here2 VIA CAN1+CAN distribution board
RangeFinder: Lightware LW20
Telemetry: RFD900
Firmware: Arducopter TradHeli 4.0.2

Prototype 1.2
FC: Cube Orange+Vbar Neo
GPS1: Here2 via CAN1
GPS2: Here2 VIA CAN2
RangeFinder: Lightware LW20
Telemetry: RFD900
Firmware: Arducopter TradHeli 4.0.5

We skipped Firmware 4.0.3&4.0.4 because of the servo sluggish movement which happen to be fixed on 4.0.5

The tuning take a long time but everything went well and both aircraft become very stable and solid in AltHold and Loiter mode.

We move on into Auto Yet something weird happen. Both aircraft cannot progress into Auto mode once press the switch. The aircrafts will only slowly descend and turn side to side without progressing to the waypoint.

Is there anyone who experience the same issue? Is there any parameter that might affect the Auto mode?

The second issue happened soon after, we experienced high compass variance on our Prototype 1.2(Bar on EKF window reach red level). Therefore, we change our compass priority in the following setup;
FROM:

  1. AK09916 via SPI
  2. HERE3 via UAVCAN
  3. HERE3 via UAVCAN

TO

  1. HERE3 via UAVCAN
  2. HERE3 via UAVCAN
    and we disabled internal magneto from Cube Orange

For a while we try to tweak several Navigation Parameter to try that is there any parameter that prevent the Heli from going Auto.

And in one moment while we trying to land the aircraft and call it the day.

The aircraft start spinning and totally losing control, The VBar Neo still keep it flying by help maintaining the attitude. we raise the altitude to prevent it from hitting anything near by and trying to regain control. I(the pilot) switch to Loiter mode from Althold mode hoping it will help kept the aircraft in place. In Loiter mode, the aircraft stop spinning but start traversing randomly at high speed. Like GPS glitch problem, the aircraft start traveling in big circle and moving far away from the test site. At the same time, the operator keep shouting that there is nothing wrong on EKF status but there is some message keep saying “Glitch clear”. I try to switch back and forth between AltHold Mode and Loiter Mode with no result the aircraft keep spinning in AltHold Mode and traveling in circle in Loiter mode. Then the aircraft move far away enough from the RC controller, it changed to RTL mode, Still it keep traveling in circle. From all the fluctuation the aircraft keep losing altitude and lost connection.

I would love to share the Flight Log if somehow our aircraft come back to us.

We suspect that the UAVCAN might be the main culprit, since the symptoms is like we are losing both GPSs and Magnetos at the same time plus we disable internal magneto(which we might never do that again, ever). Is there anyone have similar issue before?

The last known working perfectly well firmware we have tested was Arduheli 3.9.11 with L1, we are force to change to Arducopter Tradheli 4.0.2 and 4.0.5 despite all the obvious bug(like sluggish servo movement in 4.0.3 and 4.0.4). Sadly, Arduheli we have, support only Cube Black and we kinda miss it.

Ps. Everything happen in the span of 15 seconds and it feels like 15 years of my life has gone down the drain.
Ps2. We drive around the neighborhood for hours to make sure that our missing heli(weight 8 kg) did not hit anyone or anything. Luckily, seems like it is just disappear.

@Eakarat First I am sorry to read of your misfortune. It is always disheartening to crash and not be able to recover the aircraft.

I was unaware of the sluggish servo issue but I am glad that it was fixed.

I am curious about your set up especially that you are using a flybarless controller with ardupilot. Just wanted you to know that the flybarless controller is not necessary and ardupilot firmware can handle stabilization.
Could you post a param file for your setup?

Could you post a link to a log file showing flight in loiter and althold flight modes? I would like to see how it flies with your setup.

please provide a log of this behavior.

I am not smart enough on the interaction of the controller with the compass. I would suggest that you post this on this thread.
I have in most cases, set up the controller so it recognizes the internal magneto but don’t have it use the magneto. The only one that I have set for the controller to use is the external magneto in the GPS puck. I have never had an issue with this but I am not using UAVCAN either. Hopefully the dev team can give a little better assistance in analyzing the problems you experienced with this.

I will do my best to look at the basic set up to ensure it is correct and your issue with switching into auto mode. I do NOT use a flybarless controller with ardupilot so I will do my best to look at your setup.

Regards,
Bill Geyer
Tradheli Maintainer

Nice Machine, sorry to hear that.
Why didn’t you fold back to stabilized ?
Did you find the heli and downloaded the recorded data ?
As bill said, there is no need for the Vbar, it makes integration harder and has no benefit.

@bnsgeyer Bill, thank you for your detail attention to our problem. It is kind of heartfelt feeling having good community support our development journey
@ZvikaF The main reasons for integration of Flybarless module is to have the flybarless systems act as a attitude stabilizer in any situation(it’s what it does best) and only have Arducopter system work as a pilot that easily control the movement of highly stabilized vehicle. We also use it for so many functions like governor, collective compensation, etc.

This is the parameter we saved right before we loosed our aircraft.
Versar Cube Orange Here3 21-11-2020.param (18.9 KB)

I will try to provide you with the flight log as soon as possible with AltHold, Loiter and Auto mode behavior.

@ZvikaF Sadly, we have yet to recover it, but the best thing is there is no additional damage to people or building. We can always assemble the new aircraft.

Regarding the firmware version, do you have recommendation for the version that work quite well without known issue?

Thank you again,

Eakarat

@Eakarat, bill has better knowledge on the systems.
My two cents on integrating pixhawk and rc helicopters, I did start with a flybarless controller thinking it will shorten learning time and make the system more resilient, found out that the addition only complicate things, checked behavior without the FBL controller, and liked it. collective handling is done within the autopilot and governor function is done,quite good, within the ESC. I kept the ability to change assigned RPM with the pass trough function.
I plan to have a large heli flying autonomously but decided to do the learning with the 450 type helis.
I did use quit extensively version 4.0.4 with no significant problems. most of the problems were traced to my lack of understanding. moved fluently to 4.0.5 and starting to expend the flight envelope.
My best advice is to start with a lean low cost heli platform, and advance to the larger platform after you have gathered enough knowledge and confidence on the basic autopilot and integration.
I do not have enough knowledge to comment on your parameters data, but might do so on the logged data.

@Eakarat Looking over your perimeter file I see some things that concern me with the attitude controller even though you’re using a flybarless controller to stabilize the aircraft. I would really like to see a log file in loiter and althold flight modes to determine whether this was causing problems.

We are also looking over the log file of the other aircraft with similar setup, we saw some weird behavior in target velocity and actual velocity.
this flight log is from the another prototype which is yet to be tuned well.
https://drive.google.com/file/d/15J8cJkT8ybwcUkQTvzgX0aYL2y2NGWKU/view?usp=sharing

Looking forward to your analysis

Thank you so much,

Eakarat

In order for me to diagnose your problem I need to see a log of an aircraft you consider well tuned. I would agree that this aircraft was poorly tuned
Also it looks like you were able to put the aircraft into auto mode which you said that you were having problems.

Lastly please provide a wiring diagram of the connections from the RC receiver to the cube and Vbar and to the servos. I need to understand the control path between the vbar and cube. I am guessing that the RC receiver makes inputs to the cube and the cube makes inputs to the vbar and then the vbar outputs to the servos?

What mode is the vbar in? Are you in a 2D mode or 3D mode. Does it provide attitude hold in the 3D mode? There are a whole host of issues that can happen with using the vbar to stabilize the vehicle. I noticed that you had non zero PIDs for the rate controller (ATC_RAT_XXX_X) which is bad if you are using the Vbar to stabilize the vehicle. Also in noticed that your attitude controller P gain (ATC_ANG_XXX_P) was very small which is also bad because it doesn’t drive the actual aircraft attitude to the desired attitude for the cube.

I have assembled 30 CubeOrange (version 4.0.5), and the flight test did not happen. The only difference is that I did not install V bar

I just want to confirm that you did assemble 30 cube orange with Tradheli 4.0.5 and still working fine in all mode? If so we would focus more on finding the root in our setting

Thanks everyone for pouring in

@Eakarat
Looked at your log, indeed this is a badly tuned heli.
are you using radio transmitter with the neutralized mode collective stick ?
looks like you are advancing to auto modes without having good tuning.
This may lead to an unpredictable issues.
also the reading of the pitch is noisy, might be due to installation or bad balance.
RPM of 975 ? do you have good altitude authority ?
Do you use filters ? you’re having main vibration lobe on 8 Hz…