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.

 

Views: 74239

Reply to This

Replies to This Discussion

Here's the two autotune attempts.  The first one seems to go ok, but with these pid's it still flies like a dog lol.  The second attempt simply wouldn't start.  Never had that happen before.

https://www.dropbox.com/s/vh3gxbbupd3xi4x/2016-09-22%2018-14-27.bin...

Note this autotune never announced it was done, just stopped.

And this one wouldn't even start:

https://www.dropbox.com/s/uqcasyd8irsl38y/2016-09-22%2018-36-43.bin...

As you can see channel 7 goes high for AT, then goes low and back to high and finally stays low so not AT.  Maybe a radio glitch.

Went out again for another AT, but this time at .05 aggr setting.  Not very good.  hovers ok, but start moving and it get pretty unstable.  During the AT near the end of the pitch step it lost control for a bit but recovered.

Log from today's AT at .05aggr: https://www.dropbox.com/s/e7jyfkxdld1tupt/2016-09-23%2010-05-19.bin...

As expected the pid's at .05 were really low.  PID's from today's tuning followed by those from yesterday PID's.  

Note that you can also get the tune values from the log file from an incomplete autotune.

Note also you can tie up your quad so it can only rotate on one axis and can't hit anything and adjust parameters that way.

You saw the picture of this X8?  Pretty frustrating working with this the last few days, but that made me laugh.  Image that spinning around all angry on a rope lol!



George Roberts said:

Note that you can also get the tune values from the log file from an incomplete autotune.

Note also you can tie up your quad so it can only rotate on one axis and can't hit anything and adjust parameters that way.

Hi Leonard,

Maybe starting with a better manual tune would help.  I saw somewhere (youtube) that you stated by putting D on a rotary know and adjusting until an oscillation.  Then the same for P (I=P) and finally stabilize.  I presume backing off a bit at the oscillation point.  Never done manual tuning like this, but I'm willing to try something new.  

leonardthall said:

Hi Shaun,

Could you do a 4 minute flight with the Log Bit Mask of : 786429

Could you make sure you have a fast SD card in the pixhawk. (like as fast as you can get)

Just a Stabilize for the first minute

Alt Hold for the second minute

Loiter for the third minute

and a minute of flying around.

I am seeing some things that look like noise and I am trying to track it down.

Thanks,

Leonard

Sorry for the delay mate,

I have been very busy but I will get to you. It looks like you have vibration but I need to characterise it and it will take an hour or so of time and that is a little hard to find right now.

This is what's supposed to be attached to the X8...  I think if I saw the X8 freak out in forward flight like it did with this attached I'd have a heat attack!  Again, I really appreciate you looking at this.  The vibration must be some lower frequency?  In Vib3.3, there's barely anything there.  But with these slow motors and big props, I can see there could be issues I would see in a smaller unit. 

Hi Shaun,

The problem I think we have here is you you some very high rate vibration. I think the main problem frequency is 65Hz or 3900rpm. We don't log this (I think we should) so I don't know exactly what normal is.

This is probably your blade passage frequency or the interaction between the upper and lower blades.

How much flex is there in the copters frame?

Without making some serious changes to the frame to stiffen it up I think we can only drop the filter frequencies. This gets a little hairy and I don't like suggesting this and I would be very careful doing it myself. I may try this and quickly decide that I have went too far. You will need to do a manual tune then evaluate the stability and if the manual tune is good enough do an auto tune and see how that goes. This is very much experimental though and I would normally only do this if I was flying and tuning the copter myself.

 INS_GYRO_FILTER = 10

 ATC_RAT_RLL_FILT = 5

 ATC_RAT_PIT_FILT = 5

You have to start the tuning from scratch with these parameters and it is risky. I have sent you my skype details so I can help talk you through the process.

Sorry for the delay getting back to you. This took a few hours of log analysis and some custom analysis code to try to give me enough confidence to suggest this course of action.

I would have thought the frame pretty stiff.  It's a Foxtech D130 with 30mm booms.  I also added an cross stiffener to make it even more rigid.  This frame has been used a lot with DJI A3's and WooKong AP.  But not many flying a pixhawk.  Maybe give me a time when it's convenient for you and I'll skype you.  Cheers!

Hi Shaun,

The other thing I noticed is you are doing all this with a hover throttle of 12.5% and your minimum throttle is set to 25% of the PWM range. What was the reason for setting the Minimum Throttle so high?

The other thing that might be worth looking at is doing a tune with your expected payload weight secured to the frame to give the copter some more low throttle room to tune with.

Can you also provided some photos of the pixhawk mounting?

Folding frames generally have some issues that you need to work around as they are rarely as stiff and as well aligned as we would like.

The DJI controllers tend to be setup very mushy in comparison to what Autotune is designed to achieve (and the pilots don't tend to know any better). We may need to drop our expectations back to DJI levels to get this flying acceptably.

In any case, lets get a communication line going on Skype and we can report back to the community how we go.

Sorry again for keeping you waiting.

I presume the hover throttle being 3.4rc5 would have been a learned parameter.  As for the minimum throttle setting at 25%, I never actually set it.  Just did the usual radio calibration.  Maybe I should reset all the parameters and start again.  I'll get some photos of the pixhawk mounting and setup a payload dummy weight of 3kg (lidar weight).

Reply to Discussion

RSS

Groups

Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service