ArduPlane 3.0.1 Beta - AutoTune Logs

Posting dataflash logs from Autotune flights here for developer use.

This one was done with a non-tuned Multiplex FunCub.
Only pre-flight calibration and the mandatory setups have been made.
It contains about 4 minutes worth of autotune flight.

No problems or unusual behaviour detected during flight.
Intentionally made with about 50% roll input, as this was a first attempt and only meant to test how it felt and how it worked.
Autotune level at default 5.

Edit: removed log due to missing data.

This one is made with a Multiplex Twinstar II
No other tuning performed, only pre-flight and mandatory calibrations.
Full stick deflections as often as I could get away with safely.
No problems or unusual behaviour detected during flight.

Autotune level at default 5.

It was a long flight and only the last ~5 minutes of Autotune made it to the log.

Edit: removed log due to missing data.

Had a few more Autotune flights with the same two planes.
Tested up to Autotune Level 7 which seemed close to my liking. :slight_smile:

With the FunCub, I used full roll input this time.

As promised, it’s a very relaxed process.
I could not detect any unusual behavior in flight.

I reduced the amount of logging for most of these flights, and used Default + Ctun and Current.

I have parameter files pre- and post-flight that can be emailed to a dev on request.

Edit: removed log due to missing data.

I couldn’t get into Autotune, it gave me RTL instead.

Thanks for taking the time to get these logs, but unfortunately I have some bad news! The log files don’t contain any AUTOTUNE debug information.
The AUTOTUNE was working, but the problem is the CLI log dumping process used to get .log files off the board doesn’t produce the ATRP log messages that show the detailed autotune actions. If you download the log files using the “download log files via MAVLink” button, which produces .bin files, then it will contain the information.
I’m afraid I overlooked this when I added the logging code to AUTOTUNE. It uses “dynamic logging” where new types of log messages are added by a library (in this case the autotune library). The old CLI based log downloader doesn’t understand that new log types can be added on the fly, and so doesn’t dump them. Darn!
I’ll fix this when I get a chance, but meanwhile if you do gather any more logs can you download the logs using the “log files over MAVLink” button?
Sorry for the wasted effort!
Cheers, Tridge

Hehe. Oops. :stuck_out_tongue:
No worries, and no wasted effort. I had fun doing it.
This weekend should provide some excellent Autotune logs.

I had a successful autotune session in strong winds today (8-10m/s). It went well, but I couldn’t get any useful .bin logs. They are all too short to contain useful data I think. I tried to download them using telemetry, which took so long, I think they were over written in the process. My first time downloading logs over Mavlink.
Oh well, I’ll try again.
It seemed to work pretty well. Level 5 was a bit soft, level 7 felt very good.
I’m not real clear on when MP updates the new PIDs for me to see. I tried to refresh and they didn’t change, but when I disconnected and reconnected telemetry I saw the new PIDs.

This is very exciting stuff Tridge. Thank you.

Download logs over Mavlink works really good over USB. It feels like it’s faster than the “old” way.

For the Dev’s:
Same aircraft as before. .bin logs that now contain the ATRP message.
Started out with default PID:s, no other tuning made.

Three flight per aircraft, starting at Autotune level 5.
Level 7 ended up feeling more or less right on both.

MultiPlex FunCub logs first. (Only five per post are allowed.)

And the logs from the Multiplex TwinStar2

(Yes, the TwinStar did end up in the ditch on the level 7 run.
And, yes, it was waterfilled. Single reason: Pilot stupidity. :blush: )

I’d also like to help.
I loaded the latest version this morning.

Hi All,
Due to the discovery of an important I2C bug recently I plan on going a new plane release this week. I’ve just marked it as a 3.0.2beta now.
It would be great if we could get some test flights!
I think we should probably create a “Beta Testers” topic on the forum as well.
Cheers, Tridge

Yes, a Beta Testers topic. You can describe the code changes and specific areas/conditions you need tested.
Is there anything specific you want tested in 3.02? I don’t have anything plugged into i2c, so that revision probably won’t affect me.

I posted a reply on drones-discuss, but that got deleted for unknown reasons.
EDIT: My posts on drones-disuss were deleted due to my mistakes.

Anyway, I’d be happy and honored to do any APM:Plane Beta testing.
Weather will keep me grounded the next few days, but ground testing can always be done.

A dedicated forum, and a lead tester are both very good ideas.

As for where to keep the discussion, I have no real opinion.
I’ll just jump in wherever is best for the project.


Hi Andrew,

Below is atatche the log file from today flight. I was using Autotune level 5, there are 3 flights with Autotune and Auto missions.
I would be glad to hear your comments.

Andrew has left as default “AutoTune Level 5”, imho is a lot conservative, after many tests I’ve done I believe that “7” is the best start value to get PID slightly higher and great for a fast stabilization, tested with “X8 Skywalker, Phantom FX61 and RV-Jet”.
So I recommend you try to redo AutoTune with this value, this can help a lot in windy conditions.
Of course also depends on control surfaces of the plane and its flight characteristics.

I would like to add that from my testing, auto-tune does an inadequate job if no airspeed sensor is installed.
Has anyone else tested without an airspeed sensor that can confirm this?
This was using a Bixler. Level 7. Level 8 caused oscillations. Pitch P autotuned to .7, should be about 1.1.

Hi Marco,
Thanks for reply.
Indeed, in Wiki Autotune is saying to try using a higher level than 5. Me I did try 5 and 6 (as you can see below), but Level 6 gave me some less stable flying, even I did try to continue longer until it stabilize itself. For that reason I did go back to level 5. And I was not confortable to try Level 7, but maybe next time I will do.
I did have airspeed (pitot) sensor installed and activated, also calibrated with preflight check.
Thanks again for your suggestion.

Below is the model I was flying

What dataflash log bit mask would you like us to use for our beta test flights?