Possible radio issues causing uncontrollable oscillations

ArduCopter
1

Hello folks. This is quite long winded, but I’m struggling with resolving an issue.
I have been trying to use Ardupilot 4.6.2 on smaller drones.
My tests have been with small woop frames and even with larger frames like a Tarot 650.
The basic arrangements have been with a Matek H73 mini and the F405 MiniTE and the radio has been the Jumper 15/ Radiomaster TX16S and receiver is the Radiomaster XR running ELRS 2.4GB.
The ESC is based on the frame size, Anything from using the Skywalker 50A per motor, or the Flywoo Goku 55m, or the Holybro Tekko32 65A. Motors are based on the frame size, T-Motor MN3515, to the Axisflying C-157. Again, these are just test frames.

The issue, on all of these airframes, ALL experience oscillations upon launch and in short flight that increase in frequency almost uncontrollably. The smaller drone motors get really hot. These flight cannot last more than 30 seconds or so before I have to abort the flights.

After testing, correcting, checking and rechecking the motor placement, balance, ESC settings I could not isolate the problem.

I thought the issues were with the FC, so, I swapped out the Matek FC with a Orange Cube+ on the larger drones, but had the same problem.

So, I went back to my old workhorse test platform that never had an issue in flight. I swapped out the Cube+ and Herelink and installed the Matek FC and used the Radiomaster XR1, and had the same problem. I then swapped out the Matek FC and out the original Cube+, configured it to use the Radiomaster XR1. The drone started oscillating and swinging radically as soon as I lifted off. The only difference was the XR1 receiver. I put the Herelink back on the Cube+, reset the configuration for the Herelink, and it flies great again.

I’m thinking the ELRS system is causing the issue, I don’t know if I’m missing a configuration setup, or whether it’s noise or signal issues.

The whole point was to try to develop a smaller system that was reliable steady platform.
Could it be the ELRS radios and receivers?, and if so, what could I be doing wrong with radio and receiver configuration?
Thank you.

2

Post a .bin log file, otherwise we can not diagnose it.

But usually the problem with output oscillation is bad or missing PID tuning.

3

I might need to reproduce the conditions to get the a fresh .bin log files. This might be a flight one from the Matek FC. If not, I will need to submit a fresh file.

4

There’s no flight or anything very interesting in that log. However there is some movement on the RC channels, I presume when you are touching the sticks.
It could be shakey hands :slight_smile: or the transmitter could do with a hardware calibration. Be sure to do the RC calibration in each copter after you’ve done the transmitter.

But as per this graph the variation is within the deadzone of the RC inputs, so the copter would be ignoring those tiny inputs. The deadzone is 20PWM except for Throttle which is 30PWM.

image
image2028×1203 31.6 KB

If you’ve used BetaFlight with that transmitter you may have turned off the analog smoothing too, try turning that on again.

5

Thank you Shawn. I don’t think I have jitter fingers, LOL, but. I will rebuild a test vehicle and start fresh with a new program build and get a clean data capture to resubmit. I haven’t used Betaflight and was doing the builds straight from Mission Planner.

1 Like
6

Hello, here is the rebuild and tests.
I tested it in Hover only and a simple out and return.
In hover only, to avoid any additional stick inputs, I only used the throttle to climb above the pad, and used RTL to complete the flight. There was a lot of drift.
Both flights had a lot of oscillation and the motors were very hot at the end of the flights.

Access to the logs

Airframe configuration - Woop
Electronics used -
Mateksys H743Mini – Arducopter V4.6.2
Flywoo Goku G45M
Mateksys M10Q GPS
Axisflying C157

Receiver -
Radiomaster XR1
EdgeTX 3.5.2

Radio -
Radiomaster TX16S
EdgeTX 3.5.6

7

As one would expect with default Rate PID’s and vertical acceleration controller gains on a whoop.

This will get them cooking:

RCout
RCout3410×1049 421 KB

You need to follow a tuning process:
Tuning Process Instructions
Or, if you have limited tuning experiance:
Methodic Configurator

If you want to take a short cut to tame the output oscillation then:
Reduce the ATC_ANG Pitch/Roll Rate PID’s by 50%
PSC_ACCZ_I,0.20 (2X Hover Thrust)
PSC_ACCZ_P,0.10 (Hover thrust)

Did you do this? Setting Motor Ranges

But, you still need to proceed with standard tuning.

2 Likes
8

Thank you Dave.
What could the cause for the PIDs to be so out of whack?

9

They are not “out of whack” they are at default which is totally unsuited for a whoop size craft. Arducopter defaults were initially produced for a ~10" propped craft on 3S power like the venerable 3DR Iris (550/10"/850kV motors). But, there are many variables which make defaults unsuited for most craft.

There may be some pilots that come along and offer up proof that defaults work well on their craft. Some actually do for a narrow range of characteristics but it’s pretty rare.

You have a nice Flight Controller though! My favorite for small craft.

1 Like
10

Thank you Dave. I do follow the motor configurator methodology in the Docs. You explanation for the size makes total sense. I will make the test corrections and try it. I’m new at building the smaller drones and using EdgeTX. This was my test solution for building a mini autonomous drone. I’ve never had these issues with my larger platforms.

11
  • ISSUE SOLVED -
    Again, thank you Dave. I reduced the PID values as you stated by 50% and that corrected the oscillations. The drone is flying smoothly, no motor heating issues. I will be able to attempt auto tuning.
    Again, Thank you.
12

You need to check for output oscillation in the log. Did you also set the PSC parameters suggested?
You should configure the notch filter before running Auto Tune.

2 Likes
13

Hello Dave. Yes, I applied the config adjustment per your instructions. I failed to configure the notch filters before attempting the Auto Tune.. Here is the screen shot of the motor outputs.

image
image1264×411 16.6 KB

14

OK, Auto Tune will probably run with that. But, set INS_RAW_LOG_OPT,9 and make another hover flight to collect data for notch configuration first.

1 Like
15

Thank you Dave. Will do.

1 Like
16

Post a link to that log if you are unfamiliar with how to configure the filter using Web Tools.

17

Thank you. I did have the INS_RAW_LOG_OPT set to 9. I will post a hover only log for you. I have not worked with configuring the notch filters. Should I go ahead and enable INS_HNTC1_ENABLE, and INS_HNTC2_ENABLE before the test flight?

18

No. You can use Web Tools to configure a virtual filter and view the result then set the parameters. And it’s not common to need 2 notches.

1 Like