Unsteady descent resulting in near crash

Hi, I am pretty sure this has nothing to do with v.4, but I didn’t know where to put my topic.

After a small break from the hobby, I decided to reflash my old F550 and do a new configuration, as I have never really been satisfied with my configuration.

My F550 experienced had a near death experience, after a successful autotune. I did an autotune with all settings as described as the documentation suggest, and the aggressiveness of the autotune set to default. Everything went great and I was pretty happy with the tuning. It flew very well. It was a calm evening.

But when I decided to do a fast descent which I did notice was a bit unstable, it almost went wrong. When I raised the throttle to maintain the altitude the vehicle started banking from side to side (90* bank angle, to 90* bank angle opposite) for a looooong time (seemed like forever). I went into panic and accidentally hit landing gear switch, tried to change mode, but nothing helped. The autopilot kept things in “control” and suddenly it went into a stable hover…

I have attached my log file and parameters, if anyone could give it a look, and give suggestions for what to change, it would be appreciated. Do I need a more/less aggressive autotune? I know I could change the decent rate but it wasn’t even going that fast so I would prefer to fix the problem, rather than hiding the problem.

DJI F550 frame
Tmotors (MN3110-15, 780kv)
10" master airscrew props
Cube Black (with all imus running (mask set to 7)
Mauch PSU
Here GPS
SimonK ESCs (working fine, never had desync)
5000mah 5s graphene lipo battery.

Any help is appreciated. Thanks…

Log and parameters:

Just a wild guess…
Are you sure you were not in auto tune mode?

Seems like you have some serious vibration issues. Your x axis has a couple hundred clips and your z axis have over 30k clips. You should ideally not be getting any clipping in flight. I would recommend sorting the vibration issues out first and then run auto tune again.

I just checked my radio… And it is in the high position for the autotune switch… DOH! But isn’t the autotune disabled when it receives pilot input, which it almost did the entire time? Especially during descend… Does the log indicate it has been in autotune mode? Thanks for your comment.

Well, this frame is prone to have vibration issues but it still flies well, and no matter what I am not able to lower the clipping. It has been commented before. Motors and props are in perfect balance, Cube is hard mounted as recommended. I don’t see what else there is to do. Are you sure the uncontrolled flight isn’t raising the clipping count? I am not very good at reading the logs, sorry :frowning:

Yes, vibrations will definitely contribute to uncontrolled flight. I’ve seen noticeable improvements in copter handling when I’ve reduced vibrations and clipping from tens/hundreds down to zero. The clipping on your z axis starts 10 seconds into the logs and is well into the thousands a minute into the log, indicating there’s another issue and not only caused by the uncontrolled portion of the flight.
Your pitch axis seems pretty well locked in, however your roll axis has quite a few oscillations and some major over shoots at time. May be worth re running auto tune for the roll axis to see if it produces better results.

Also, no you weren’t in autotune mode. Logs show you switched between alt hold, loiter and RTL a few times only.

EDIT: just had a look through your parameters, your pitch and roll PIDs are quite different, normally there’s not as much of a difference between them. Try ticking the lock pitch and roll gains options in mission planner and giving that a go before you autotune the roll axis.

So, a few problems I see. What actions would you recommend taking in order to minimise the vibrations/clipping? My Cube is hard mounted which is by far the most recommended way of mounting the cube. My motors and props are in absolute balance. Nothing there. And there isn’t much more to it. Its a frame, with a Cube, motors, ESCs and a receiver. Its down to the bare minimum right now.

When you say that my pitch is pretty well locked in, and my roll is not, is this regarding to the PID tune, or my vibration problem?

Glad to hear I wasn’t in autotune mode, even though it would have explained the problem.

Can you explain the lock pitch and roll gains options for me? I am not sure what you mean about that.

Thank you very much for your time so far…

1 Like

BTW I uploaded another log file to my drive folder from the flight before, which was the autotune flight… Just if this should show a more realistic picture of the vibrations.

Motors can cause very powerful and noticeable vibrations.

I had a case where I had 400kv on 16mm arms, all seemed perfect but clipping like crazy. the simplest solution was to steal one of my wife’s silicon baking Tray, cut it to the shape of a cross for each motor, little holes for screws and most importantly loc-tight the screws because you want that flexibility

I don’t have a wife… How can I get one of those?

Lets assume I reduce the clipping a bit, will there be other things to do in order to reduce the amount of instability during descent? Just to get back on topic… I guess I should do a re-autotuned when I dampen my motors…


Something similar to what I did
[₪ 4.15 | Silicone Material Brushless Motors Soft Mount Vibration Dampener Drone Motor Guard Landing Pad for 2205 2207 Motor https://a.aliexpress.com/_dZeEjUR](₪ 4.15 | Silicone Material Brushless Motors Soft Mount Vibration Dampener Drone Motor Guard Landing Pad for 2205 2207 Motor https://a.aliexpress.com/_dZeEjUR)

And of course you would have to re tune you vehicle

Look at mounting the cube on a vibration dampener. I’ve had success with this, however I’ve done this on large drones (18" quad and 29" octo) which are known to cause issues with the Cube’s internal damping due to the large props inducing low frequencies (<120Hz) which the Cube doesn’t do a great job of damping. You have 10" props so it may not help, but its worth a try. Are all your cables tied down, nothing free to flap around?

Regarding the pitch and roll, I am referring to the PID tune. You can see in the following that the actual pitch follows relatively closely with the desired pitch whereas the roll has more variance. This is a zoomed in portion but it seems to be the case for the majority of the log.

There is an option in mission planner to lock roll and pitch gains, it basically copies the pitch gains over to the roll gains. Its a good starting point, and seeing as your pitch seems better than your roll it may help you to lock the 2. When you run an autotune again, it will unlock them and tune them individually.
I would recommend rerunning the autotune but do one axis at a time to prevent battery failsafes.

Is there some non-symmetry on this craft causing the Pitch and Roll PID’s to be so different in range? How/where the battery is mounted perhaps? Also, before you do another autotune you may want to drop the INS_GYRO_FILTER back to 20. The suggestion from the tuning guide may be too high for this craft. It was for one of mine. Also before another autotune ATC_THR_MIX_MAN back to .1 Your ATC_ACCEL_P/R_MAX parameters look low probably from the last Autotune. Set them back to guide recommended values.

Yes, the vibe clipping is bad, not sure what to suggest that has already been mentioned. You should try the Harmonic Notch. You will have to use throttle reference but that worked OK for me before ESC telemetry RPM or now FFT was available.

You don’t want one… you think this hobby is expensive?? :joy:

You could try mounting the flight controller on antivibrstion gel pads

Okay, I just ordered vibration pads for my motors, so I will try mount these in the weekend and see if they will produce better results. I will keep this post updated.

Thank you so far to everyone in her…

1 Like

I may be way off base here, but I think your problems are being caused by the power system, and in particular the battery.

Those motors are not rated for 5S. They are rated for 3S and 4S.

FWIW, as a rule of thumb, for larger aircraft, once you drop the KV down below 900, you move out of 3S territory and cross over into 4S territory, and once you drop below 600Kv you cross over into 6S territory.

By the same token, 3S is generally 10" props, 4S is 13" to maybe 15" depending on the motor, and 6S is 15 and above.

Like I stated earlier, my hex is swinging 13" props. I have a 700+mm quad that has 325kv motors swinging 17" props on 6S, and I have an H4 Alien 680mm with 530Kv motors swinging 15" props on 6S.

Another thing I noticed that points a finger at the power system is Hover Throttle.

My 680 is slightly under powered and that is a direct result of using the Tarot motors. General consensus is they are crap, and I agree.

On the other hand your aircraft is severely OVER POWERED, and you can prove this for yourself by looking CTUN > ThI (Throttle In) which averages about 0.11 and then look at CTUN > ThH (learned Hover Throttle) which starts out at .029 and then drops to .016.

This is telling me the props are hitting hover thrust well below mid-stick. Hover Throttle Learning is good for putting hover throttle at mid-stick so you have a muscle memory anchor point, but it masks the loss of resolution between full low throttle and mid-throttle, and for the FC, IMHO this is a bad thing.

The issue here is when the aircraft is in any flight mode that uses Altitude Hold, the FC in an aircraft that is over powered has to work much harder to maintain altitude than an FC in an aircraft with a closer weight to power balance does, and personally speaking, I believe this is the major source of your high vibration levels.

Having said that, what other gear do you have installed and how much does the aircraft weigh WITHOUT the battery?

1 Like

Interesting observation!

First of all, this is my “learning drone/platform”. It has served me well for many years and many motors have been sitting on this frame. Last motors weren’t able to keep this thing in the air if one motor failed, which was something I prioritized, why I wanted to go for overkill (and it is overkill! Easily flying 30m/s in alt hold while climbing and pulling 124a of the battery on a 550mm frame)

I think your observation is interesting and I never considered that the FC will work harder to keep the the vehicle level with much less resolution on each motor, especially with old ESCs that doesn’t support high resolution data rate and only something near PWM.

At the moment it is only caring a raspberry pi, a 5.8ghz alfa radio, retractable landing gear and a few navigation LEDs, beside the FC, motor, battery etc.

AUW is around 1.2kg I think, and with my gopro gimbal it is rocking around 1.8kg why I want the somewhat overpowered motors. Whenever I have requested help in here before I was confronted that my propulsion system was underpowered, why I said enough was enough and bought a proper set of motors this time.

I have ordered motor anti vibration mounts for this, which I will try to mount, do a small flight and see if this has improved, and if not, I might try a mount for my Cube…

Thanks for your comment!

Speaking of resolution, do you think oneshot125 would help the situation? I am pretty sure ESCs are set to use default PWM…

Going with BlHeli ESCs will not help because the throttle resolution issue is between the RC transmitter and Pixhawk.

Another thing you need to take into consideration is the limitations of the F550 frame.

With a long diagonal motor center to motor center distance of 550mm, this frame has a side (adjacent motor center to adjacent motor) distance of 275mm. 275mm = 10.8268 inches. Therefore, allowing for sufficient propeller tip clearance you are limited to 10" props. This pretty much puts your motor choices in the 900-1200Kv range, and battery choices to 3S.

A lot of people are going to push back on this, but I’ve tried running 10" props on a 2212-920KV motors and the difference in flight times was not worth the effort for the simple fact that for the same current capacity a 4S battery weighs more than a 3S battery, and because the motors and props are woefully inefficient with typical grams/watt efficiencies between 7 and 8, flight time gains are insignificant.

So, the bottom line here is, if you want to run 4S and achieve optimum control response and acceptable flight times you need to use longer props, but your frame prohibits that option.

Now about this throttle resolution and how it affects Pixhawk. The easiest way to see what is happening is to review the log you posted and plot CTUN > ThH and then consider that this parameter starts with a default that should be very close to the throttle stick being at mid-stick. In an UNDER POWERED air craft ToH will Increase from this point, and in a OVER POWERED aircraft ToH will decrease.

Now take this relationship and consider how much throttle range Pixhawk has to use to control altitude and how rapidly Pixhawk must adjust throttle. For an overpowered aircraft with small props the changes are going to be very small and very rapid, similar to a “high” frequency vibration.

Now then, having said all of that, building and flying these aircraft is all about making compromises so we can have a stable aircraft that has a respectable flight time FOR ITS FRAME CLASS.

1 Like