This is a discussion re the ArduCopter-3.1 AutoTune feature. A place for people to report successes and failures, ask questions and get answers.
The AutoTune wiki page is here.
You need to be a member of diydrones to add comments!
This reply was deleted.
Replies
Leonard, problem ESCs were Exceed-RC Proton 30A, stock (unflashed), props were 11x4.7, but also 12x6.
I might add that these ESCs were old (bought several years ago, flew them first on APM 1 …) yet no problem running them at full throttle on set-up with no overheat. Just happened I had a bunch of them, handy for testing … Did not have a current sensor when running AT but tested them many times running several minutes at max, no overheating at 20-25A where the motors maxed out … They are still sold and reviews appear to be good so have no clue why I’ve been experiencing the problem. Of note, also: I was *never* able to arm them without first powering the APM board via usb. Not a board or set-up problem as this happened on many boards, and does not with other ESCs. Nor was I ever able to calibrate them “all at once” via APM, always had to do that manually via the receiver . Suspect some minimum pwm end-point requirement, or related initialization checks those particular ESCs do. May or may not be related to the initial problems I’ve had with AT…
Hi,
here is my log of a successful AutoTune. I was tuning with the gimbal attached but tighened the gimbal so it can not jiggle.
Parameters before were Stab P 3.5, Rate P 0.1, Rate I 0.1 and Rate D 0.003 but not tuned very well.
neue pids.PNG
2013-12-01 12-52 4.log
After some setbacks due to mechanical/electrical problem that had nothing to do with AT, ran autotune on rc7 with stock pids on a 2.5kg quad, completed without a hitch, in a bit over five minutes. (Already had had one successful AT test on a smaller one ). Copter flys and loiters rock solid.
Lesson learned (for me): Any of my copters that will not pass the autotune test is not air-worthy. In addition to the tuning, it's actually a good test. Don't want to find out in an auto mission or high up that a few fast rollls or pitches could bring the copter down ...
Autotune pid results:
Stab Pitch and Roll: 7.8750
Rate Roll P=.11, I=.11, D=.006
Rate Pitch P=.135, I=.135, D=.0085
Love the fact Roll and Pitch PIDs are different: AT "figured out" and took into account the fact that I ws flying with a single rectangular battery that made the copter sligthly weight assymetrical!
Bottom line: Autotune shakes, rattles and rolls! Thanks Leonard and the devs. Pardon the exhuberance but ... AC, ... you've come a long way, baby. ...
I can't seem to engage the auto tune at all. I have my channel 8 set to auto tune and when I flip it to high nothing happens. I thought it might be my trims like they said in the last post so I did some did some auto trim to get rid of the TX trim and still nothing. How long after switching on should it start?
2013-11-29 17-18 21.log
2013-11-30 08-12 23.log
Ok, I used auto tune today for a "normal" quad and it worked great! I have awesome position hold now. But, I then tried it on my X8 and it didn't work. I got a couple twitches but then it refused to do anymore. I think the problem is that I am flying with some added trim and it sees the trim as stick input, which disengages the routine. So I am thinking I need to save trim and try it again.
Does this make sense?
Got the courage to try auto tune today. Took off vertically with stock parameters, had some wobble effect after roll/pitch input. Flipped into alt hold, and autotune.The roll portion seemed to go ok - i did have to move it around a bit, but danced left and right nicely. However, once it started the pitch part, did a few back and forwards, then dropped its rear right arm, and headed towards the ground. I guess it thought the grass needed to be cut. I tried a few more times with the same result. I am still novice at reading logs - but i do see that the GPS location got wildly inaccurate at one point. And you can see the massive "Roll" in ATT. unfortunately, i didnt have any special logs on, since i already knew the copter flew fine in 3.0.1
This is on a f330 frame, 8045 props, 920kv 4s, arducopter on top of dubro foam.
2013-11-28 16-02 2.log
2013-11-28 16-02 2.kmz
More fun testing and investigating as "Kamikaze" quad is working hard to live up to its name (:
(It's ok, that's what it's for ... Just one prop this time).
Enabled IMU and motors this time as I wanted to check vibration in autotune test + see what the heck the motors were doing. Started with reduced rate P and R from stock. AT went through four rolls before copter went down. Here are a couple screenshots ....
This one clearly shows the AT rolls before the 40deg+ that sent the copter down. Can't find fault in excessive vibration from IMU raw accels (?) ...
A closeup of what (the heck ?) the motors are doing ...
Each right/left pair seem to be doing their thing as expected, until the fatal roll where right before 12.7 motor 1 appears to be shoot up too far and then react too slowly?
I am more and more convinced that this strange issue may have nothing to do with auto-tune, at least not directly ...
Replace motor 1? Oh, and throwing this out there just in case ... Could it be that my channel 7 is acting up and auto-tune quit right in the middle of the roll, back to alt-hold or stab where sticks were such that the copter could not correct fast enough? In any case beautiful thanksgiving day here, time to ...
at test 3 2013-11-28 12-15 1.log
Hi,
I give a try to ArduCopter3.1rc7 AutoTune today and I am not very pleased with the tuning result.
When I do some aggressive flights (full throttle), it feels like I hit the redline off a combustion engine and engage the rev limiter.
Which value should I decrease ?
Thanks
Pierre
PID.jpg
Leonard, thanks for your precisions on the auto-tune "algorithm". Here is the log where I was trying to "test" manually my AT failure by manually doing fast roll "twiches", until I went too far ... Not making sense of roll_in and roll graphs, btw, roll seems to happen without roll_in ... Also frankly this was so imprecise that I don't know it will say much ...
I've tried AT again since, I've had virtually the same thing happen as in the first test. A couple or three twiches, then a roll past the point of no return. Log attached. This time I just had one prop break, checked all motors while still on the ground and all working fine, and after a quick replacement, I flew the copter again and it went perfect ... So an interesting enigma if it's electrical or motor failure ... Have been flying several lipos since, not one problem, alt hold and loiter very solid ... I've also reduced stab P to 3.5 and rate P to .9ish, may try again with these and see if that makes a difference ...
PS: Not 100% sure the first file is the right one. Is there a fast and easy way to correlate a tlog with a dataflash log given different time stamps (I downloaded the data flash logs much later ...) I tried to convert gps time in dataflash log but was not successful ...
Roll manual twiches test .log
AT Test 2.log
Randy , which version firmware do you suggest I use . I have installed
-
50
-
51
-
52
-
53
-
54
of 55 Next