Crash during Autotune with my Bell 230 1.8m

ArduCopter Traditional Helicopter
1

@bnsgeyer
Hello Bill,

last week I tried to tune my big Bell 230 with autotune.
The Bell 230 has a rotor diameter of 1.80 meters, weighs about 11 kg and is equipped with an electric Skyfox mechanics from VARIO.
I admit that the Skyfox mechanism is not a modern design and its load capacity is not comparable with a modern mechanism for 3D flying.

However, the load capacity of the gears etc. is perfectly adequate for scale flying.

I have carried out three flights. First I flew the ROLL_FF parameter. That worked.
Then I tried to fly Pitch_FF. That did not work. I always received the message “Leveling failed, tune manually”. I therefore left PITCH_FF at 0.15 and tried to fly RAT_P and RAT_D for the roll.

That didn’t work either. Here, too, I always got the leveling error.

But then the worst came. The helicopter crashed because a pair of gears in the main gearbox broke. Fortunately, the damage wasn’t too bad.

I suspect that the strong movements during FF tuning overloaded the gears. Looking at the flight movements of this large and heavy helicopter during FF tuning, I was afraid for the model. I was always close to canceling the tuning.

Have you ever tuned such large and heavy helis with autotune?
Is there a way to reduce the violent movements of large scale models?

I am hesitant to use autotuning on such large models in the future. I might lose your support in case of problems, but who wants to lose their expensive and labor-intensive helis? Maybe you have a solution oder advice for me.

When does the message “Leveling failed” appear? And what can you do then?

I am attaching the log files here:

https://1drv.ms/u/s!AjSq727ChpJzgw8ZOFqGDNHrqYE0?e=PnGKwV

https://1drv.ms/u/s!AjSq727ChpJzgw5BdTw7EgQSCPrO?e=m6IcY2

https://1drv.ms/u/s!AjSq727ChpJzgw0bOY1N6Q0LdewN?e=g4zUYO

Many thanks for your help

BR

Heri

2

Hi Heri,
I’m sorry that you had a crash during autotune.

I have conducted autotune on a larger aircraft. The biggest that I witnessed was the Velos Rotors heli. It has a 1.9 m rotor diameter and we tuned it at 17kg. I have also used it on my T-Rex 700 which is only 6kg at takeoff.

Not currently. I have been considering adding a way to scale the maneuvers based on the rotor diameter. I hope to get to that this summer. I realize that the aggressiveness of the maneuver is not suitable for really large heli’s.

I totally understand your situation and you won’t lose my support. As for a solution for tuning, you would have to use manual tuning until I can get this scaling feature implemented.

I will take a look at the logs to see if there is anything else that I notice.

2 Likes
3

@bnsgeyer

Hello Bill,

Thank you very much for your help.

It would be very helpful if there is the possibility to adjust the autotune maneuvers to the heli size in the future. I’m already looking forward to the addition.

Maybe you can see in the logs what I can improve and use for my manual tuning.

Can you explain to me what the message “Failing to level, please tune manually” is all about?

Many thanks for the explanation in advance.

BR

Heri

4

Hi @heri remember SkyFox mechanics (mid '80s, basically a rework of Heim mechanics) were designed for 1.4/1.5 m rotor diameters and “old” .60 class engines. The maximum power output you get from those engines is around 1.2kW. Nowadays electric engines are at least 4kW rated in that class. Add the extra load of significantly bigger rotor diameter and maybe a little extra pull on the collective…gears won’t last long. If you look at the Benzin mechanic gearbox, it’s a totally different design (still poor in my opinion) but it copes well enough with 1.8m rotor diameters.
Hope this helps.

5

@Ferrosan
Hi Ferruccio,

Nice to hear from you.

First of all, I didn’t want to criticize the autotune function.
Like all of us, I’m glad that this function exists. And I’m aware of how much work Bill has put into it.

With my post I wanted to ask Bill for help, if he has a solution for me. And I also gathered from Bill’s old posts that Bill no longer likes to deal with problems caused by manual tuning. Which I can understand.

I agree with you 100% that the Skyfox is not a modern mechanic and flies at its performance limits today. However, it is still used in many scale helicopters. This is because scale helicopters live much longer than 3D helicopters. In addition, you can still buy the mechanics new from Vario and they are also recommended up to 1.8 m. In the latest issue of the German ROTOR magazine there is a construction report on a VARIO Airwolf with Skyfox mechanics.

In Germany, the motherland of RC helis, many “old” mechanics are still flying, often self-developed designs. I just bought a Bell 47G from SSM from the 90s for Christmas. Also with 1.8 m.

The 2.5m Jetranger from Vario is on my wish list. The new XL-Vario mechanics are certainly very stable. However, it is installed almost self-supporting in the dome of the fuselage. The forces are transferred to the fuselage via frames. This is not as stable as a full carbon trainer in e.g. 800 size. I think you understand what I mean.

I think I can continue to tune the roll and yaw of large helicopters using autotune. I will tune pitch manually. The movements on pitch seem the most violent to me when I watch them.

And Bill has assured me that he will help me with manual tuning if I have any questions.

Greetings from Germany to Italy? :wink:

Heri

6

I know all of them very well :wink:. I started with a semirigid head from the great Dieter Schluter without even a gyro in the tail rotor (it was an Hely Baby if I recall correctly), then moved to Kobold, Skyfox, Benzin and on. I also have an XL which I was converting to Sph5 turbine power, but I moved away to another (much more reliable) Eu turbine engine manufacturer.

1 Like
7

You’re my buddy.

I started at the end of the 70s. With a Mini Boy from SchlĂĽter. Even without a rear gyro! What a dance. :smile:

Then I switched to Heim Mechanik with tuning parts from VARIO. Back then I was sitting at Uli Streich’s kitchen table. He fetched a few plastic bags of parts from his cellar and a few sheets of paper with drawings. That was all. Wonderful times.

3 Likes
8

Hi, Bill, in the other thread Trex 450 Tail Rotor Tuning you stated the ANGLE_MAX value with default of 30deg (3000 centi degree)

Could it be possible to reduce this values just during autotune to reduce the violent movements as @heri asked for?

9

Hi, Heribert, oh my god. The good old Helis. My starting point was also late 70th with Graupner Bell 47G without gyro.

1 Like
10

No. The autotune was designed for the FF tuning to go from -20 to 20 deg attitude to determine the gain. The sweeps are designed to not cause more than 5 deg but how much the aircraft actually pitches or rolls is dependent on the aircraft. In @heri case, the pitch axis was pretty lightly damped and the roll frequency sweep excited the pitch axis causing a divergent oscillation. The code limits the pitch and roll attitudes to 30 deg. If during any autotune maneuver, the attitude exceeds this angle, the maneuver will stop and reset and try again. So I don’t think the message was correct. I need to change it so that it is clearer as to what the problem is.

@heri we may do something a little different to tune the pitch axis but we will want to tune the rate P and D first before doing any autotune in roll or yaw. So when you are ready to tune that pitch axis, I will provide directions. I think we may try to use autotune to tune the pitch axis but we willl do it first. As long as the aircraft behaves with no large pitch motions, then the autotune will work. If we get divergent oscillations again then we will use the system ID mode to get the data we need. I tuned a 125kg tandem that had a similar issue with the pitch axis and couldn’t use autotune. So I used the system ID mode and was able to get the data I needed to tune it.

Standing by when you are ready.

11

@bnsgeyer

Hello Bill,

thanks for your support.

However, I think I didn’t understand everything from your last post.

In addition, I don’t understand anything about SYSTEM ID. I’ve read the documentation on this, but it’s a closed book for me.

I’m in the process of repairing the broken gears and other minor damage. When the repairs are finished, I’ll do a few hovers to check if everything is OK again.

Afterwards I will report back here and link the new logs from the hover flights

BR

Heri

12

@bnsgeyer

Hi Bill,

I was able to repair my Bell. In a windy winter, we had two days with very little wind. It wasn’t completely calm, but it was better than nothing.
During a first test flight, I noticed that everything was working again. Now I didn’t want to let the windless days go by without making more flights.

I used this to manually tune the pitch. If you look at the helicopter from the outside, the tuning works quite well. I don’t expect a scale heli to have the snappy engagement of a 3D heli. It looks like an old-school paddle heli. I could live with that. The soft-damped rotor head is probably to blame. When I look at the ATT log values, it still oscillates above the zero level in the opposite direction for a few degrees. Loiter works quite well. This is my preferred mode.

Then I wanted to tune YAW with AUTOTUNE. That should work, I thought. Unfortunately not quite.

First I wanted to tune RAT_YAW_D and _P. Here, too, the message “failing to level, pleas tune manually” appeared frequently. Was it too windy after all? Is that why AUTOTUNE had problems with leveling?
However, AUTOTUNE completed this step successfully.

Now I wanted to tune YAW_ANGP. That did not work!
I got the message “Exceeded frequency range” and “Failed”.
Are the minimum or maximum frequencies too low or too high?

First I attach the logs from the manual tuning of the ATC_RAT_PIT_D and the _P values.
https://1drv.ms/u/s!AjSq727ChpJzgxAYb_pDcNXkVe3b?e=vLqagZ
https://1drv.ms/u/s!AjSq727ChpJzgxGNNdAUnRtBqWwA?e=dH5Opg

And here the logs from the AUTOTUNE of ATC_RAT_YAW_D and _P and the log of ATC_YAW_ANGP.
https://1drv.ms/u/s!AjSq727ChpJzgxMovWrPs1FnCUos?e=wO9V8a
https://1drv.ms/u/s!AjSq727ChpJzgxIMKHB6-S9Rv6Co?e=aYGi37

At the moment I am not unhappy about the behavior of the heli, but I think it could be better.

Thank you for your support.

Heri

13

The fact that the traces of desired and actual are offset from each other leads me to believe that you don’t have enough I gain in there. Currently your ILMI for pitch and roll is 0.05. I would suggest making that 0.1. However it could be that your swashplate is not mechanically trimmed. If it was a near windless day, then the I terms for pitch and roll should be within +/- 0.02. Yours are at least 0.05 or greater.

image
image1920Ă—1080 417 KB

Looking at the pitch axis, it looks like the controller is holding a constant 3-4 deg/s down of target pitch rate. You can see a similar thing in the roll axis. It is commanding left roll rate. Based on my calculations, I recommend you change your servo trims to the values provided below
SERVO1_TRIM 1521
SERVO2_TRIM 1516
SERVO3_TRIM 1515

Conduct a short hover on a calm wind day and look at your Pitch and roll I terms. That would be PIDP.I for the pitch and PIDR.I for roll. They should be closer to zero, hopefully less than 0.02.

As far as the tuning, I will ask you to do some frequency sweeps using the SysID mode. You can leave the aircraft tune the way it is. I will write more once I see that your I terms are centered better.

1 Like
14

@bnsgeyer
Hi Bill,

Thank you for your work analyzing my logs.

To be honest, it was not a windless day. It was just much calmer than the many days before. The days before were more stormy than windy :wink: The wind speed was between 7 and 10 km/h. And the wind also changed direction. I always tried to keep the helicopter in the wind direction when tuning. So I always had to keep the nose of the helicopter down, to keep it standing at one point (or near one point ;-)).

The wind flow to the blades also leads to a rolling movement when the wind comes from the front. I will change the I-values and wait for a calm day. I will then send you this log.

To be honest, I’m really looking forward to tuning with SYSTEM.ID. It will greatly increase my knowledge in this area

Could you tell me why the YAW ANGLP Tuning showed errors?

Thank you very much.

Heri

15

Heri,
I understand your comments about the winds and I would agree to not change the servo trim values but increase the ILMI for pitch and roll to 0.1

For the system ID, I would like you to use this mode in each axis (pitch, roll and yaw). The general settings are the following.

SID_F_START_HZ, 0.25
SID_F_STOP_HZ, 10
SID_T_FADE_IN, 10
SID_T_FADE_OUT, 2
SID_T_REC, 90
In the LOG_BITMASK parameter, uncheck the FAST ATTITUDE setting. This will cause the system ID mode to collect data at 100 hz rather than 400 hz.

You would set SID_AXIS to perform the sweep in each axis
For the roll axis sweep set
SID_AXIS, 1
SID_MAGNITUDE, 3

for the pitch axis sweep set
SID_AXIS, 2
SID_MAGNITUDE, 3

for the yaw axis sweep set
SID_AXIS, 3
SID_MAGNITUDE, 5

The SID_MAGNITUDE sets the attitude maximum to 3 deg for pitch and roll sweeps. For the Yaw axis this sets the max yaw rate to 5 deg/s.

You never want to takeoff or land in System ID mode. I would suggest taking off in stabilize mode. Once you are in a stable hover, switch to system ID mode. This mode requires you to control the aircraft and keep it from hitting things. It is not designed to hold position over the ground. It is basically stabilize mode with the frequency sweep inputs being added to your inputs. Once 90 seconds has passed, switch into stabilize mode and land.

Hopefully you watched the videos in the autotune wiki about flying the aircraft while the frequency sweep is being performed. You will do a similar thing here except the sweep will take 90 seconds to complete. Initially the aircraft will drift quite a bit so ensure that you are flying in an open space. It may drift up to 10-15 meters initially during the low frequency portion of the sweep. As long as it isn’t constantly moving away from you, let it oscillate back and forth over the ground. If it does start to constantly move away while it oscillates, then apply a small constant stick inputs to stop it from moving away from you and keep it oscillating about one point over the ground. Once the frequency gets high enough the aircraft will stop drifting back and forth and will stay pretty stationary as it continues to oscillate in the axis being swept.

Be sure to do this on a fairly calm day (< 5-10 kph winds). The lower the winds, the better. If the winds are coming from a specific direction then have the axis being swept pointed perpendicular to the wind. for example, if the winds are coming from the north and you are sweeping the pitch axis, then point the nose of the aircraft to the east or west for the sweep.

Hopefully this all makes sense

1 Like
16

@bnsgeyer
Hi Bill,

Thank you for the instructions on SysID.
I think I have understood everything.
I’ll wait for a calm day and use Sys.ID
starting with Roll.

I will then link the logs here.
Many thanks for your help.

Heri

17

No problem. Hopefully you could get all three done in the same day.

18

@bnsgeyer
Hi Bill,

I have not yet found an opportunity to perform SYS.ID flights. Too windy.

One question in preparation: If I switch from ALTHOLD to SYS.ID mode, will the altitude still be maintained automatically?
Or is the SYS.ID mode based on the STABILICE mode?

Thank you for your answer.

Heri

19

@heri
Hi Heri,
The Sysid mode uses Stabilize mode as the underlying mode. So, no, it will not maintain altitude once you switch into sysID mode. That is why I would recommend going into the Sysid mode from stabilize.

20

@bnsgeyer
Hi Bill,

I made the three test flights for SYS.ID Roll Pitch and Yaw. There was little wind. It went quite well. The oscillations were low. I hope that the tests were successful.

https://1drv.ms/f/s!AjSq727ChpJzgxfrc5klgKkkob8w?e=n2wQo8

https://1drv.ms/u/s!AjSq727ChpJzgxWvj9BNGR0SaYlN?e=ypgHEY

https://1drv.ms/u/s!AjSq727ChpJzgxSJdGnqlKc4X6mY?e=M0olX8

BR

Heri