Motor oscillation 50% throttle Octacopter

ArduCopter Copter 4.1
21

@dkemxr It’s not two problems, only one. The oscillation of the engines with the throttle at 50% on ground.

I commented in the last message about the vibration in the landing gear, because shawn made a suggestion to reduce the values ATC_RAT_YAW_P and ATC_RAT_YAW_I and when it was done the vibrations in the landing gear were noticed, I didn’t notice differences in the flight, only in the landing gear that transmits very well vibration due to being long. But with the values set by Auto Tune I don’t have those vibrations.

As I mentioned in the message above to Shawn the values of MOT_SPIN_MIN and MOT_SPIN_ARM are low and I didn’t make any additional increments. But I will revisit and do new tests to identify if I can reduce even more.

22

That’s good if you’ve already set MOT_SPIN_ARM using the correct method. You would start by finding the best MOT_SPIN_ARM then adding a bit to get MOT_SPIN_MIN.

I’ve had this same yaw oscillation/landing gear issue myself, and without changing anything physical we needed to carefully select filter settings and other tuning parameters. With BLHELI ESCs we were also able to lower the RAMPUP power which helped.
There’s an amount of tuning that can be done to minimise or maybe even eliminate this yaw oscillation effect, however it’s unlikely that just changing one parameter will fix it. There will be a combination of params and a lot of testing.
Really it’s caused by a physical deficiency. The desire is to have long/high landing gear to allow for large or low-hanging payloads. This means flex and opens a pandoras box of resonant frequencies and other problems.
As a temporary test you could try adding guy-wires or some stiffening bar - such as thin carbon fibre rods used for stiffening in foam plane wings - and join them from the ends of the landing gear back to the main base plate. Anything to dampen or reduce the flex in the landing gear.
As a permanent solution you may need a thicker/stronger base plate and two landing-gear struts on each side instead of 1 each side.
It may even be that adding a payload will put enough weight on the landing gear to reduce the problem.

What size props do you have?

Are you saying that reducing ATC_RAT_YAW P and I made the yaw oscillations worse?
If so then put them back to
ATC_RAT_YAW_P,0.8
ATC_RAT_YAW_I,0.08
and add
ATC_RAT_YAW_D,0.005
also try increasing ATC_ANG_YAW_P slightly.
Another parameter to try is MOT_SPOOL_TIME, say 1.0 → allowing 1 second for motors to spin up, even though they don’t take that long, it may allow RPM to stabilise before the flight controller is trying to actively control attitude and getting caught in an oscillation.

Definitely put in the HNOTCH settings too. They will help with tuning and may even help reduce the yaw oscillation. There is no down-side to adding them.

I mentioned the pitch and roll parameters because of the attitude control I see in your logs. Those adjusted values along with the Harmonic Notch filter will give a better result when you next run Autotune.
I believe after this yaw oscillation is solved, or at least reduced, you should run Autotune again.

23

Yes Shawn, the MOT_SPIN_ARM and MOT_SPIN_MIN values were set as per the support documentation.

Unfortunately it is not possible to configure the RAMPUP power of my escs, I am using T-Motor alpha. I also believe that several parameters should be changed to minimize this problem, I’m just afraid of changing something that creates a serious problem.

I’m already using dual landing gear and a 2mm carbon fiber base.

X8 (2)
X8 (2)1920×1297 186 KB

I’m using 21 inch props.

Yes! Lowering ATC_RAT_YAW P generated small fluctuations during flight, but reduced the oscillations initially.

I’m going to do a new test making increments in the MOT_SPOOL_TIME parameter and I will also adjust the values of HNOTCH.

24

@xfacta e @dkemxr I made the adjustments suggested by you and the oscillation continued, but I felt an improvement in the stability of the equipment by increasing the Pitch, Roll and harmonic filter gains.

I am sharing with you the two logs of the flights I made today, the difference between the first and the second is in the INS_HNTCH_FREQ I reduced from 50 to 48 according to the FFT frequency graph.

I made the increment ATC_RAT_YAW_D = 0.005 but when arming the drone I didn’t feel safe to take off due to the vibrations that started, so I reset the value again.

I also reduce the speed of the motors when arming and the oscillations are maintained.

https://drive.google.com/drive/folders/1ogtZoBAYVrvHWxLBO6JHhHrLwUvbJyEd?usp=sharing

25

That second log definitely looks better.
You have this very low, what happens if you raise it to 15 or 20 ?
INS_ACCEL_FILTER,10

You could also try these if you get time, you said previously lowering them seemed to make the oscillations worse
ATC_RAT_YAW_P,2.0
ATC_RAT_YAW_I,0.2

With those T-Motor Alpha ESCs normally you’d set these too:
MOT_THST_EXPO,0.2
MOT_PWM_MIN,1100
MOT_PWM_MAX,1940
You can experiment with the MOT_THST_EXPO - you currently have 0.75, maybe try 0.4 and see what difference it makes.

  • set too high you can see instability at low throttle
  • set too low you can see instability at high throttle

It could be that your MOT_THST_EXPO is fine how it is.

26

I spent the whole week doing tests and with the changes I got a good result.

Initially I increased the INS_ACCEL_FILTER filter to 20 and I noticed the motors involuntary revs up in seconds, so I reduced it to 10 again.

I increased ATC_RAT_YAW_P to 2 but the oscillations were maintained, so I started to reduce the value and did a flight to check if I would have stabilization problems.

These were the values I came up with:

ATC_RAT_YAW_I = 0,04
ATC_RAT_YAW_P = 0,4
ATC_RAT_PIT_D = 0,004
ATC_RAT_RLL_D = 0,004
ATC_ANG_YAW_P = 7
ATC_ANG_PIT_P = 8,54
ATC_ANG_RLL_P = 11
INS_HNTCH_FREQ = 46

Reducing ATC_RAT_YAW_I and ATC_RAT_YAW_P reduced the oscillations, but the motors momentarily increased the revs as if they had high INS_ACCEL_FILTER, as I mentioned earlier, so I had to reduce ATC_RAT_PIT_D and ATC_RAT_RLL_D too.

To make it clearer what I’m talking about, below is an image of the log.

image
image916×425 15.1 KB

This situation was occurring very often before reducing values ATC_RAT_PIT_D and ATC_RAT_RLL_D.

To compensate for the reductions that were made in ATC_RAT, I made increments in ATC_ANG_YAW_P, ATC_ANG_PIT_P and ATC_ANG_RLL_P. I noticed improvements, especially in Yaw which was very soft in the answers.

After making all the adjustments I worked again on the INS_HNTCH_FREQ.

Analyzing the LOG, the oscillation is seen, but with a much lower value that is not perceptible to the naked eye.

This is the log of the last flight I took today.
https://drive.google.com/file/d/1b303TzCdricKO9BxLRX0acqKM8ty6ZKj/view?usp=sharing

27

In that log you have INS_ACCEL_FILTER,30 so 20 should be OK to use.

You need to set INS_HNTCH_REF,0.3 to 0.16 or the harmonic notch filter wont be working at hover and low throttle levels. This could make quite a difference. I cant check the other HNOTCH values though because the batch logging is off. That’s OK if you’re sure the frequency is correct.

28

Ok! I will make the adjustment, thank you for the support.