VIBE Clipping + Rapid Unexpected Altitude Gain

Hi all,

While performing a simple straight-line AUTO mission (TAKEOFF, WAYPOINT, RTL), after a smooth takeoff and initial forward flight my quad unexpectedly rapidly gained altitude to more than twice the desired altitude of 6 m. Checking the logs, all VIBEs were under 30 m/s/s, but there was a lot of clipping. The first errors were, in order, potential thrust loss and gps error. Can being a little under powered or gps error cause clipping?

I will provide the BIN log file when I get the chance.

Thank you.

edit: here is zip-compressed BIN log file: https://paste.c-net.org/SubduralTracking

I will guess that after the GPS error you had a Vibration Failsafe activating Vibration Compensation and away it went.

You never want to see a Thrust Loss error… GPS error doesn’t cause clipping, vibration acceleration causes clipping.

Thanks for your input. I will upload the BIN log file shortly.

Are you saying the GPS error and vibration failsafe are unrelated? I am a bit puzzled as to why the vibration failsafe was triggered, since all values were under the recommended limit of 30 m/s/s.

What do you mean by ‘vibration acceleration’? edit: nvm you mean the vibration amplitude

No, I am saying they are related. Vibration FS will be activated from bad positional data. Read the Wiki:
Vibe Comp

1 Like

Ah ok thanks. I have read that page, but I didnt know what EKF was. Looks like it is position and velocity estimation. Seems like this should trigger EKF failsafe, but whatever.

But that is interesting. Maybe I just got bad luck with the gps working and ill just try flying it again to see if i get the error again.

1st update to latest Stable V4.1.1. Then configure the Initial Tuning Parameters, you are at default. Use Mission Planners Alt>A plug-in. Then configure the Dynamic notch filter. Then run Auto Tune.

I wouldn’t attempt another Auto Mission until those steps are complete. The Initial Tuning Parameters are for Pre-maiden flight…

Note: On that climb out one motor was commanded to max triggering the thrust loss error. Most likley it’s underpowered/overweight. At a hover it looks like the average commanded output is ~1650µs. It should be close to 1500.

Actually your battery sagged to ~8.2V. A plethora of problems but perhaps all interrelated.

Thanks for looking at the logs.

I’ve done on the order of 10 auto missions with all sorts of crazy modifications that shouldnt be stable, and the autopilot seems to handle it pretty well at default settings (I also dont fly too long each trial)… so I will keep it default for now until I learn enough to know what to change, or if you have specific recommendations (I dont have experience tuning or configuring, this is my first drone built from scratch).

I am fine with being a little overweight as I dont intend for aggressive maneuvers, just gentle flight (though enough margin to handle wind gusts etc). Of course it would be ideal to increase thrust margin but that is more time and money towards a build i would prefer not to spend if not necessary for my target performance.

On flights where the vibe failsafe was not triggered, batt voltage does not drop that much. If the unintended max thrust was not triggered the batt would never sag to low levels.

So against your well-thought advice I will try and fly again and see if I still get the error again. I will update with the results, which hopefully is not a crash report :slight_smile:

Yes I made them. Use the Mission Planner Alt>A tuning plug-in.

A little overweigh can be fine but if you have thrust loss errors the craft cannot properly stabilize. Many crashes have resulted from that.

Ok I am not familiar with that plug-in. sounds like i have to know how to use it, or does it auto tune or something?

so the goal is to find out what caused the unplanned full-throttle climb. if that can be avoided, then I wont have the stabilization problems from no thrust margin.

We already know what caused that, Vibration Compensation. Did you read this?
The vehicle will not change mode but its altitude hold will be less accurate than normal. The vehicle may overshoot it’s altitude targets and/or respond more slowly to pilot input.
What you experienced is common with this failsafe.

Tuning plug-in. Connect to Mission planner. Hit Alt>A, input values as prompted, make the changes.

Oh so the common safe figure of 30 m/s/s is not actually a common safe figure? e.g. I have to tune it to 30 m / s / s?

We are not only talking about levels of vibration, we are talking about the Vibration failsafe, why it happens and it’s results. Vibration is only one component. But if you have clipping events, and you had thousands, then the vibration levels are too high.

hmmm… so from the logs you can see this order of errors:

thrust_loss_check → gps-2 err → failsafe_vibe

the thrust loss check happens first, and happens due to unplanned max-throttle climb. how do you deduce this is vibration problem, esp. when vibe amplitudes are less than 30 m/s/s?

edit: so looks like clipping starts happening even before above errors, as shown in this pic (note that red is gps alt):

ok so i think i agree with you on vibration being problem now. i was paying too much attention to errors, when the clipping graph vs time is what i really needed to look at. errors dont show when the clipping start ramping up, which is where the real problem starts happening (as far as i know).

is software tuning the right solution though? Based on docs I think i should try to improve physical damping somehow.

Yes agree that would be the place to start. The Dynamic notch filter is not a band-aid for high vibes. In fact it is only active on the gyros. Clipping definitely has to be eliminated or greatly reduced. There may be some events during high demand maneuvers, high wind etc. But I would be worried about the lack of thrust authority also. Look where the clipping events start to accumulate. When the motor outputs are commanded to max and there is a likely loss of stability:


Clipping axis scale on right.

In fact look what happens to Roll Stability when it can’t Stabilize due to a pegged output:

The tune is not that good to start but it gets wonky.

1 Like

The 30m/s/s vibration level suggested as the maximum or grey area is itself a very grey area. For example 20 to 30 is probably in the “this can go either way and I’m just waiting to surprise you” zone. Down below 15 is much safer. Under 10 is good.
Clipping indicates the accels exceeded limits and the vibrations recorded may not be accurate.

2 Likes

alright thanks guys, based on your answers i have more confidence it is indeed the vibes.

now onto the art of using foam to reduce vibes…

Foam might do it if it’s hard mounted now. These small ball mounts work well for smaller/lighter FC’s. It’s a PixRacer sitting on it.

1 Like

Where did you get that case for the PixRacer?

It’s an aluminum one that comes with the Chinese Clones. AFAIK you can’t buy them separately, I asked an eBay vendor once. The nice thing about them is it adds a bit of mass to an otherwise light weight FC which works well with the ball mounts.

The 1st Pixracer I bought from the original AUAV and it’s still flying a 210 quad, my 1st Arducopter “mini quad”. I bought a couple more Clones, one is in a Rover and the other is in a corn field somewhere waiting for a combine to catch it…

1 Like