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
I had to add bit 16 to the log I think. Anyway I got two ATUN events this time. When I first switched to autotune it definitely rolled hard (in a perfect, controlled, safe way) and then did one more 16 seconds later and then never again. The wind was very light - gusting to 5mph. But I'm in a small area - about 60 feet by 60 feet and had to adjust probably too often for autotune to run. I guess I need to find a larger place to fly where there aren't any people.
There were many times where I didn't touch the controls for 10 seconds but maybe autotune had already given up? If I switch out of autotune mode and in again will it start from the begining or continue?
Anyway - here it is - attached. The pitch needs tuning badly as you can see from the log.
69-12-31_19-00-00_1.bin
Ok, there are two issues that will cause the problem.
The first is you don't appear to have done your radio calibration because all the RCX_TRIM numbers appear to be 1500. I looked at this because I could see your Roll request is not zero and this will stop autotune from happening. This means your radio calibration is way out because you have the default, large dead bands set.
The second thing that may be a problem is your copter appears to be wobbling around with up to 10 degrees swings. Was this in windy or gusty conditions. You still may be able to get autotune to run with a proper radio calibration but if you can't you will need to work though a basic manual tune before you can proceed.
Can you tell us a little about your copter and prove some photos?
Ah! When I upgraded to 3.2 it seemed to have remembered all the parameters so I didn't do the radio calibration again. I'll do that before I fly it again! It remembered the gyro calibration. Don't know why it lost the radio calibration. Thanks for that tip!
Well the roll oscillations were all me - hitting the stick too hard maybe. The pitch oscillations are definitely there - that's why I want to tune it. I have a long moment of inertia in pitch now that I have the camera mounted way the hell out front of CG and the battery way the hell out the back. It was fine before I did this so I think that's the problem and I'm hoping I can compensate that with better PID values. Some day.
Since I took this picture I strengthened all 4 arms to be like the 2 thicker arms. But on new props (these ones are beat up). Cut a few inches off that aluminum bar and the battery is also about an inch closer but still way out there. I don't want an acrobatic quad but I don't want it to oscillate in pitch so much either. I've also mounted the APM better and have less vibrations - it's glued to thin neoprene foam which is glued to the yellow plate. I spent quite a bit of time getting the vibrations under "1" whatever the unit is. So the APM gyros are nicely connected for fast no-slop but vibrations are still reduced.
DSC_0591.JPG
Sorry to jump in but a small point is that the the RC calibration has been done at some point. The ground station is responsible for implementing the radio calibration routine and I suspect it's not setting the RCx_TRIM because we don't actually use them (as far as I know anyway).
I agree though that the RC calibration needs to be done because, as Leonard says, there is a very small but persistent desired roll (it's about -0.5 degrees of lean) that is likely the cause of why autotune is not starting (because it thinks you're trying to override the roll).
I just noticed I also get ERR,16,2 quite a bit: intertial nav bad variance according to here:
http://copter.ardupilot.com/wiki/common-diagnosing-problems-using-l...
Ah, there we go Randy. Thanks! :)
George,
The bad variance error is normally caused by a misbehaving compass. It could be anything that causes the compass to give a bad heading including:
1. incorrect orientation
2. compass calibration hasn't been done since more equipment was added to the vehicle which has affected the offsets
3. bad interference from power wires, escs, motors
I take it back. I *did* do the radio calibration step. So maybe I should calibrate the transmitter instead? It seems to be a crappy quality transmitter - turnigy 9x - $40 included transmitter and 8ch receiver. The yaw joystick needs constant fixing (occasionally I need to adjust some offsets inside the thing to keep the yaw around 1460-1500 centered) and each time I turn on the receiver I never know what I'm going to get. That's probably it - 80% of the time the yaw doesn't change but occasionally it does. Often it's bad enough that I can't even arm. There must be loose wire inside the roll potentiometer or something.
So I got it figured out. The main thing seems to be that when my controls on my transmitter are completely centered they tend to put out a signal of around 1466 instead of 1499. So I had to trim it all to be happy for the apm software. I wish the radio calibration would use the values of the controller. Anyway I got autotune to work, thanks guys. And I learned a few other things (like my compass was off by 20 degrees).
George,
To echo Randy's point, the RCx_TRIM is important in this case as it tells the copter when you're not requesting any angle demands of the copter, and the copter won't engage AutoTune if it thinks you're trying to re-position (a lean demand of .5 degrees is still a lean demand)
To further investigate the problem, I suggest the following:
The main goal of the exercise is to make sure the autopilot is not perceiving any "pilot demands" to roll, pitch, yaw, or altitude. If it is, it places the AutoTune mechanism on hold because it believes you are trying to reposition the vehicle. It doesn't know that what it is seeing is unintentional, it just knows the radio is saying "wait, hold on.. need to do something first."
With your Intertial Nav variance, the first thing that comes to mind there is making sure that when you power the vehicle up, it is perfectly still for the first few moments until the flashing lights stabalize. If the vehicle is moving while the LED is rapidly flashing red/blue, the gyro start-up is goofed which is one of the ways to trigger that log event (unless I am mistaken - Randy nor Leonard like bad info getting out so if they disagree with this I'm certain they'll correct me.)
Good luck
-
18
-
19
-
20
-
21
-
22
of 39 Next