I’m trying to set up a large agro drone - 57 inches.
X13 45 kv, 18s, 57 inch props motors. And it flies more or less well up to a certain load - it flies normally with 40-60 kg, and further increase immediately leads to a very unstable flight (poorly holds the horizon, sometimes almost falls).
I see from the logs that vibrations increase with increasing speed. What could be the reason for inadequate behavior under heavy load? Strong vibrations? Maybe it’s in the fltt, fltd filtering? We are currently produce vibration isolation - we will try to reduce vibrations to the required < 20. But what scares me most is that the drone simply at some point starts to tilt strongly somewhere and almost falls with a large load.
Logs number 30 - 80 kg load
Logs number 38 - 55 kg load
Yes - in 30 log - vibe very strong.. Next days we trying make vibro isolation for FC. But i’m think maybe problem in MOT_THST_EXPO for X13 motors 45kv.
i’m attach logs - you can see all params what i’m using. I’m little tune pids. but with strong vibrations i’m stop increasing pids.
and very strange that with 50-70kg craft flying normal but from 85+ - very unstable. and some time want to crash :).
maybe someone can help me with the logs and tell me where to look first, besides vibrations.
Don’t worry, we fly where there are no people and we don’t go higher than 3-5m. Paid is good - and free is even better I hope there is someone here on the forum who can help.
Thanks guys Maybe someone will say more after looking at the logs.
From ATT logs i’m see that with 85kg (30 log) ATT very very strange - desired pitch/roll far from pitch/ roll. I’m think this is because very strong vibrations and wrong notch settings.. I’m think need reduce vibrations and tune notch. From 38 log 55kg - ATT more better.
The engine manufacturer replied that MOT_THST_EXPO should be 0.9, I don’t know how true this is. -
"Hi dear sir, The value should be about 0.9, for detailed information, please see the picture below or the link attached below.
Next, we need to set up filtration and remove acillations (RATE.out - more than 0.15).
I also noticed that they were not set correctly SERVO_MIN and MAX (need 1050 and 1950)
I know 100% that there are knowledgeable people on the forum who can direct me where to dig - why others (for example - hoverit x13) were helped but not me - that’s what the forum is for - some kind of bias. I don’t need to set it up from scratch - I need advice on what to pay attention to and what actions are needed for this.
Controller Pixhavk 6c with custom vibration isolation - i’m disable second imu because i’m afraid that in flight ardupilot switch to second imu and did crash
I can send a photo only after the concept is released into mass production. - not now.
Methodic Config - I’ll try to do everything step by step and come back with the results. But I don’t think it will help me set up filtering for this configuration.
yes we have x7 pro cuav, cube orange , cuav v5, pixhavk 6c. Cube orange not testing but all other - yes. and can say that cuav have very bad internal dumping system - dumping nothing at all and for all frequencies. And all fc with cube have upper center mass that need producing proper external dumping system but pixhavk 6c have nothing and simple to using with custom vibration isolation and have two imu and chipper than other.
Drone have weight 106kg with battery’s and without any payload. Payload 40 or 80kg (one or two tanks)
X13 + 5720 props have 15-55kg thrust. we need max 25kg for one motor.
While you do have vibe issues to sort out, you should carefully review the notch filter configuration documentation. For batch logging, enable pre/post, not just pre. INS_LOG_BAT_OPT = 4
You are using ESC Telemetry for tracking, but your settings are limiting how low in frequency they will track, and missing the main harmonic.
This is a graph with some adjustments and the settings I used to get them. NOTE This is an EXAMPLE and should be used to inform how you make changes as you reduce vibes. These changes are rough estimates right now, and you should check the documentation to make sure you understand their impacts.
In the new plot, I’ve lowered the notch filter limit to 23hz and enabled multi-source tracking, which sets up a notch filter for every motor as opposed to taking the average. For this, it is required to reduce the bandwidth and attenuation settings, as now there are more notches and the results are additive for each notch. I’ve also un-selected “triple notch” in case that adds too much load on your flight controller. I also lowered your gyro filter to 10hz. The general rule on the forums is to get your noise below -50dB, and with your current level of vibes, you’ll probably find that you need to keep the gyro filter lower than you had it (20hz).
I must make clear that making such adjustments incorrectly can result in the loss of your craft. Particularly, changing the attenuation and bandwidth due to the selection of “multi-source” can result in over-filtering and undesired response from the craft. Same goes with the lower gyro filter.
Thanks for the answer.
We have reduced vibrations - I posted a new log number 108.
Yes, we need to work with the filters. I did one test with multi source and 4 harmonics turned on -
logs - https://drive.google.com/file/d/1xEo8eEFH29JjWflbQKoqxD39zgHurmQy/view?usp=sharing
it filters well, apparently, but ATT very unstable - i’m think don’t need to turn on 3-4 harmonics - but lower the gyroscope cutoff filter as you wrote, and here are the settings below, apparently, they should give a good result - noise is less than 50 everywhere (the first harmonic is on the edge).
And there is one very unpleasant thing - very often and randomly some motors seem to enter into resonance and the speed on them drops - and the drone almost falls - but then after a jerk it recovers - this can be seen in this last log in 5.34.846 timestamp - because of this jerking the vibrations increase sharply. Perhaps this is due to an incorrect gyroscope cutoff filter
This is a specrogram of Z vibes, it seems like there is a lot of noise coming from motor 8. This is a stab in the dark, but I’d be curious to know if you find anything loose on that motor or arm.