Pixhawk 2.4.8 F330 quadcopter log/build review

ArduCopter Copter 4.6
1

Hello, so this is not exactly my very first flight but it is certainly the first attempt at an extended flight while trying to AutoTune. Before this I had done a shorter flight test in AltHold only and done filter analysis but didn’t enable the filter before autotune (which I think I should have). Anyway, after some time during autotune, the drone unexpectedly started yawing to the right and I instinctively tried to roll the copter to reposition it correctly but since it was yawing that doesn’t work and I crashed into a nearby structure.

→ So I hope you guys might be able to give me some valuable insights from the data collected during this ordeal.

Log: https://drive.google.com/file/d/1_r7stzXjxTbIBjhDbcr_TZHcLJCqdgHh/view?usp=drive_link

Sorry for the GDrive link but a message here shows I can’t upload sizes exceeding 8 MB for som reason even when that .bin file is around 5MB only.

Param file: DSHot.param (16.7 KB)

→ It is worth stating that I had Baro and GPS ALT inconsistent by about 5800m+ so i made ALTERR_MAX to 6000 to pass it.

→ Any idea why the sudden yawing? Was it part of autotune? I don’t think it would have started yaw so soon in such a short time when my pitch and roll have considerable oscillation.

Hardware:

image
image1040×780 117 KB

Pixhawk 2.4.8

F330 frame

A2212 1000 kv

(:backhand_index_pointing_up: yeah, yeah, hardware to avoid :sweat_smile: )

3S Molicel P45B battery pack (self-soldered :slight_smile:)

Emax Bullet 15A ESC

9045 props (very good quality, it seems to me)

Neo 6M GPS module (Just for AutoTune)

The FC is mounted using earplugs I cut with thing gel tape on either side plus some superglue like stuff on the bottom plate (I guess the earplugs are longer that required). Do you suggest I rather place the pixhawk on another plate that will be supported by the earplugs which get inserted through holes at the corners sort of like in the wiki page on anti-vibration?

Here is my filter review with a single harmonic notch applied and using ESC telemetry (fundamental freq: 85.779 Hz, attenuation: 40 dB, bandwidth: 40 Hz, loop rate: 400 Hz):

image
image1053×413 29.7 KB
only showing one axis due to clutter. X and Y are almost identical.

→ Any idea why there are peaks at 7.9 Hz and 39 Hz? Are these bad? Could it be because of the longer than necessary earplugs?

2

Get rid of the soft mount entirely and simply mount the autopilot with some 3M mounting tape or screws. Those soft mounts can actually induce oscillation modes that screw with the tune more than they reduce vibration. There are few cases where they are actually warranted.

1 Like
3

3M mounting tape OR screws? You mean hard-mount with screws? Also since I don’t readily have 3M tape available, would normal double sided tape or a few layers of this thin gel tape (photo attached) or a layer of some sort of sponge do the job?

image
image1080×388 48.6 KB

4

Yep. Hard mount (or thin gel tape). Your mounting solution should transmit frame motion as directly as possible. Resort to those damping solutions when they prove necessary as a last resort, not as a rule.

2 Likes
5

I saw this comment by @mike_tk in a similar recent thread Review this log please - #189 by amilcarlucas about additional peaks being due to off-center CoG. Could this be try in my case?

6

It’s dangerous to do autotune before you setup adequate filtering. I doubt you need 8khz imu rate on such big rig, all motor noise will be in 100-300hz range, so try to disable fast sampling - this will give you better frequency resolution to analyze.

INS_LOG_BAT_OPT,4
INS_FAST_SAMPLE,0

I see you have bdshot working, so RPM filter is a way to go, but you get errors in communication, try switching to dshot300 or 600 (bdshot should not work with dshot150 you configured now)

image
image491×349 28 KB

after errors are gone try filtering like this

image
image478×593 15.8 KB

Also looks like the PIDs are messed up, try with default PIDs, they are good for 9”

1 Like
7

I was using DShot300. I didn’t see the error logs. Will check that and find whether the errors are on the two forward motors whose cables I had to extend using jumpers.

Can you confirm whether Motors 8,9,10,11 correspond to ports 9,10,11 and 12 respectively?

As for the PIDs, the aren’t the defaults higher? With these values I still have oscillations in Rate out for pitch and roll

8

ESC number starts from 0, so yes, usually ESC 8 and 10 means output 9 and 11, they have 11% error rate. Try twisting the cables, maybe need to shield it with some foil and keep far away from power. You can lower PIDs from default by 50% for test.

9

5 is DShot300.

Yeah 9 and 11 are my front motors. The jumpers aren’t twisted. I’ll twist them.

Also as an aside, one of my front motors are now vibrating. Earlier with a spoon shaped plastic 8” prop, I had it vibrating like this but that was because the prop was badly out of balance. The crash did scratch the props a bit but would that be an issue? Or is it just the collet being shot

10

Yes, you are right dshot300 is 5 , I accidentally used rover docs. You will see if this an issue or not in log, but any vibration is not good.

11

Do you think that my MOT_THST_HOVER of about 0.17 is a bit too low? I mostly flew in AltHold because I can’t reasonably hover in Stabilize given the fact that I also need to pitch and roll to keep the copter in position. Its that hard for me lol

12

No. That’s a good thrust to weight ratio. You just need practice.

13

So there couldn’t be a possibility of an error in this parameter since I flew in AltHold instead of Stabilize? The docs say to fly in Stabilize, that’s why

14

You need to fly in an altitude keeping mode for the hover throttle parameter to be learned.

The docs recommend stabilize for the first flight to minimize chances of fly aways.

Here is your throttle output while hovering in altitude hold. ~0.17 looks perfect.

image
image845×507 7.58 KB

Also, any log file contains the parameters, as well. There’s no need to share that file separately.

15

Thanks I didn’t know that

16

For what it’s worth, I find stabilize mode to be fairly useless in day-to-day flying. It’s good for first flights, learning throttle control, and as an “emergency” mode if the more assisted modes get erroneous input that causes the throttle output to get funky. Otherwise I’m either flying in more assisted modes (alt hold, loiter, and auto) or acro.

2 Likes
17

Should I apply the tape on the four corners or paste one large strip down the middle? The width of the tape is 1 inch

18

I’d do two side-by-side strips about the length of the autopilot to give the most security.

1 Like
19

Okay cool. I’m going to try this but why does everyone suggest soft-mounting, using 3M foams on 4 corners of that vibration isolator mount with the rubber gummies?

20

You’re looking at decades of forum responses, some of which are overcome by events by now.

Those techniques aren’t necessarily invalid, but they should only be employed when necessary rather than as absolutes.

1 Like