Strange behavior in loiter mode under version 4.3.6

ArduCopter Traditional Helicopter
1

Hello to all,

I have been searching my log files for a whole day now to find the cause of a problem that occurred in loiter mode during my last two flights with my scale helicopter (800s with 10kg).

I am now quite good at log analysis regarding vibrations. However, I am far from being a log analysis expert.

Therefore I need your help. Many thanks in advance.

During my last two flights the following phenomenon occurred:
I was flying in loiter mode (which I do most of the time) when the helicopter suddenly rocked violently back and forth over “pitch” in forward flight.

After the first moment of surprise, I switched back to Althold mode. Hereby the back and forth wobbling disappeared.
I flew the heli to me, hovered briefly and then switched back to loiter mode. The rest of the flight was normal again.

With a second flight on this day it happened after short time again that the heli in the Loiter mode (and in the forward flight) over Pitsch again back and forth wobbled. I switched back to Althold mode, where the effect stopped again, and landed.

On this day it was almost windless.

On the first flight, I was able to remember the approximate flight time at which the phenomenon had occurred by looking at the flight time on the remote. It had been about 3 minutes since the rotor started up.

I looked at the log file of the two flights. In doing so, I noticed that I had the following error message in both flights:
“Thrust_loss_check-1”. This error occurred about the same time that my flight clock on the remote control said the wobble also occurred.

I downloaded all the log files stored on the FC from the FC to my PC and looked in all the past log files to see what “Error” messages were stored.

I noticed that in none of these past flights does this error message appear nor does the wobbling itself appear.

What I also noticed: Until these two erroneous flights I had version 4.3.0 on my FC.

After I updated my FC to version 4.3.6, I made these two flights and both have this error and the wiggling back and forth in loiter mode.

Hence my guess that it is due to version 4.3.6?
Or did I make a mistake that I updated from version 4.3.0 directly to 4.3.6? Should I have done all the intermediate updates?

Here are the links to the two bin files.

Thanks for your help

BR

Heri

2

But maybe I’m totally barking up the wrong tree and the fault lies in something else entirely.

Since you are possibly not so blind, I would be glad for any help.

BR

Heri

3

Hi @heri , looks like you are costantly flying very close to the higher collective limit:

image
image1495Ă—938 9.34 KB

Also, considering you have set 13,5 degrees of maximum collective pitch you might be running very low headspeed or very “heavy” setup?
I normally aim to have a hover collective around 50 to 60%.
See what @bnsgeyer Bill thinks

4

@Ferrosan

Thank you for your quick reply.

I have not used the parameter CTUN_ThrO consciously yet.

What unit of measurement is used for this parameter? What does e.g. the value 0.92 mean ?

Does the parameter show what percentage of the cyclic path is used by the FC?

The heli is hovering at about 6 degrees. Actually it should have enough reserve to the 13.5 degrees maximum cyclic motion. The speed also does not drop audibly when the helicopter climbs. However, I have no telemetry values for this.

I will test my rotor speed again exactly.

What does the error code “Thrust loss” actually mean exactly?
Does the FC assume that the model does not climb fast enough, therefore, the “requirements” of the pilot, or in Althold/Loiter mode the “requirements” of the FC can not follow fast enough? And what is the effect of this error? What does the FC do with this error information? How does it react to it? Can this lead to “wobbling” on pitch?

Since I have not changed anything on the helicopter, I wonder why this error suddenly occurs. Is the version 4.3.6 more critical than the version 4.3.0?

As you can see, I take your objection very seriously and have many more questions.

Thank you for your help.

BR

Heri

5

@heri the big question is whether you did your collective setup properly in ardupilot. In 4.3, I believe I added the H_COL_ANG_MAX and MIN in that version which helps either the determination of the PWM for zero thrust and collective land min. I will have to take some time to review your setup. I’m not sure how that would have changed things. I would like to see a log prior to 4.3.6 to see if you were encountering the same high collective but it just wasn’t reporting it.

6

Hi Bill @bnsgeyer

here is a Log-Flight with 4.3.0

Thanks for your help

I do the collective setup properly. I promise :slight_smile:

BR

Heri

1 Like
7

@bnsgeyer

I have made some flights with version 4.3.0. All were without problems.
I can upload more log files with version 4.3.0 if you want.

With version 4.3.6 I have made four flights so far. In all four the THRUST LOSS error is displayed. However, I only noticed the wobbling in the last two flights. Perhaps in the other two flights also smaller wobbles were to be seen, which I did not notice however or attributed to other influences (e.g. wind).

Do you want more log files?

8

@heri no. Let me look at what you gave me. I have flown 4.3.5 without issue but I haven’t done any considerable forward flight in loiter. I will try to look at it this week.

9

@heri I had a few minutes and looked at your setup because i was curious.

image
image1920Ă—1080 283 KB

so here is your collective set up. You said earlier that your collective in a hover was 6 deg. I’m not sure how you estimated that but if your collective set up numbers are accurate, you are hovering at approximately 10 deg. If you look at the H_COL_HOVER, that is the fraction between H_COL_ANG_MIN and H_COL_ANG_MAX that the collective is at in a hover. So in your case both flights before and after the switch to 4.3.6, the value was around 0.8.

So my first question is whether your H_COL_ANG_MIN/MAX are the collective blade angles you get with H_COL_MIN/MAX in PWM, as per the setup instructions. If they are, then your estimate of hover collective blade angle is wrong. I would agree with @Ferrosan that you are using a very low headspeed for the aircraft weight and my thought is that you are seeing some sort of blade stall event, possibly. But why you only saw that after the update, i’m not sure.

1 Like
10

It is adimensional, range 0-1.

it means the current collective position is at 92% of the available collective travel.

It is a check perfomed by the FC over the altitude control authority, shortly.
It is triggered whenever the following conditions are met for 1 second:

  1. throttle over 90% (CTUN.tho in the logs)
  2. desired attitude smaller than 15deg
  3. angle error (i.e. desired attitude angle vs measured attitude angle) smaller than 30deg
  4. vehicle not climbing
    In your specific log case, these conditions were all met (look at the below highlighted time span):
    image
    image1920Ă—1080 110 KB

    Also in your previous FW logs the hover collective was quite high, so I do not think the FW upgrade played any role in this, it was already a bit borderline.
    The only difference I see, but it is just an idea(so take it for what it is…) is temperature. Between the two flights there are little more than 10degC in the measured OAT, which means higher density altitude in the most recent flight. If we mix this aspect with the fact that you might be having a borderline setup(in terms of rotorhead speed/ take-off weight) it may have been enough to trigger the condition for potential thrust loss.

it is a notification that you have reached the limit of your machine…the consequences are not related to the autopilot nor there is any “recovery” logic as far as I know.

1 Like
11

@bnsgeyer
@Ferrosan

Thank you both for your initial advice.

I will take the heli once again to my workbench and measure everything again.

I had already reduced the rotor head speed a long time ago, because my belt drive slipped through to the tail. This was almost a year ago. In this scale heli, the belt runs freely in the fuselage and not in a tube, so it is not optimal to tension. I bought the heli used, then you have to compromise. By reducing the speed, the torque to the tail was lower. Throughout the flights so far this went well. :smirk:

Thanks for the lessons, which I received from you both to read log files. I appreciate any teaching on this subject. If only I knew more about it!

BR

Heri

12

@heri The other give away that you were hovering in the upper half of the collective range was the fact that your stabilize collective setup had IM_STB_COL_2 and IM_STB_COL_3 were 80 and 90 respectively. This means that with your collective stick at mid point, the output collective to the head was 0.85 or roughly 11 deg collective.

13

@bnsgeyer
@Ferrosan

You are both right.

When I lowered the RPMs, the hover point was no longer correct when switching from Stabilise to AltHold. So I adjusted the IM_STB_COL_XX values until the switch point from Stab mode to AltH mode was right again. After that I forgot to measure the hover value in degrees again on the workbench. The helicopter flew fine.

I looked at the old log files and found that I was also always at around 0.90. It is then enough when the air is a little thinner and I need a little more degrees. Or I flew a little more dynamic and cyclic and collective have added up.
And already I come to higher values e.g. 0.95

I have to retune the heli.

The warning message I now understand and is understandable to me that it appears. What I still don’t understand is how this wobble then occurs in the loiter and switching back to AltHold stops the wobble. Whereby I probably at this moment also involuntarily ended the forward drive and the roll movement.
In both flights, I was flying a large circle at the time of the wobble.

What do you think?

BR

Heri

14

Hi @heri
As Bill said, aerodynamics might come into play in that wobble. Remember that cyclic pitch deflection is added on top of the collective travel, so the actual geometric blade pitch angle can be greater then 13.5 deg.
I wouldn’t say it’s a stall yet, but it might be unstable/ severe turbulence caused by the high angle of attack setting.
Also the fact you centered the cyclic while switching to AltHold, immediately stops the wobble, leads to the idea above.

15

@bnsgeyer
@Ferrosan

I again learned a lot from you. Thank you for your help!

BR

Heri

1 Like