apm 2.5 loiter twitchy in winds

[optiplex]

hey guys

my apm has fw 3.1.2 , at optimal conditions of no wind or light wind loiter is pretty stable
but as the wind grows and not even what i would call strong wind, the quad becomes twitchy in loiter mode
anything i can do to tune it ?

thx

[HowDoIfly]

Check the logs for problems between accel/gyro readings and corrective inputs?

[optiplex]

Check the logs for problems between accel/gyro readings and corrective inputs?

thx ... but ...
what exactly am i looking for ? what am i suppose to see that defines as problem ?

[lakis21]

hey guys

my apm has fw 3.1.2 , at optimal conditions of no wind or light wind loiter is pretty stable
but as the wind grows and not even what i would call strong wind, the quad becomes twitchy in loiter mode
anything i can do to tune it ?

thx

Hi there,
do you have foam over your baro?

[HowDoIfly]

thx ... but ...
what exactly am i looking for ? what am i suppose to see that defines as problem ?

Sorry, I don';t have the link to hand, it';s well documented on the arducopter wiki.
Have you followed the setup guides and recommendations (save covering the basics like baro cover etc)?

[Hands0n]

I would suggest running Auto Tune if you haven';t already.

But before that, ensure that you';ve done work to get vibrations down to the acceptable limits posted on the Arducopter wiki here: http://copter.ardupilot.com/wiki/ac_measuringvibration/  That is always a first with APM (and Pixhawk).  Work on finding and eliminating the vibration source before trying the damping measures. I had good motors so only had to balance my propellers and their hubs to get well within the margins.

It might be as well to do the Auto Tune on a gently breezy day - nothing strong, just light breeze.  You will have to repeatedly bring the quad back as the breeze takes it away, but thats okay, Auto Tune will pause and resume as required.

Make a screen copy of your PID before and after the Auto Tune so you can see for yourself what has changed.

[Flyinegypt]

One comment about using Auto Tune. Do not trust it completely. I ran it and wound up with PID settings so high that it became extremely sensitive. I had to back them right off again until I was happy.
Certainly worth checking the Baro sensor is correctly covered with foam or similar. Mine is the genuine cased version APM and I can see the foam pad on mine.

[Hands0n]

One comment about using Auto Tune. Do not trust it completely. I ran it and wound up with PID settings so high that it became extremely sensitive. I had to back them right off again until I was happy.

True, dat. Same for me with my Firefly 450, I';ve chronicled it somewhere on here  ::)  Contrary however, the TBS Disco required higher values than the Firefly';s that it was set with, and Auto Tune got it stable as if it were nailed to the sky   

In my experience it is always worth taking screenshots of before and after so you can review what Auto Tune has done.

[HowDoIfly]

You can trust autotune but maybe not the way you think. If it creates unusual settings, it';s because it had some unusual reactions to deal with. As they say GIGO. 

[optiplex]

thx for the replys guys

i did try autotune but the results are worse, so always have to revert back to the old config
my baro is covered good with the foam inside the apm';s case
i think maybe its not LOITER issue at all, 1st i need to make the stab mode perfect, i started notice its not perfect in wind, when that will be good im sure the loiter will be good as well

[HowDoIfly]

I know it';s not what you want to read, but the APM does require some study-time and attention to detail to get the best flying experience out of it and, imo, learning to read the logs is paramount. Having said that, I';m still largely ignorant of all the data it gives but taking it one step at a time can be very rewarding.

First point of call is the vibrations. I can';t overstate how important this is for super-smooth and stable flying. The less vibes, the less correctional loading is put on the flight controller and there';s only so many cycles per second the processor can achieve.

I say this because you';re obviously not happy with loiter, autotune and you also mention stab-mode could be suspect. The one thing they all have in common is vibrations and the processing power required to overcome them.

If you set logging to include IMU, then go for a gentle hover, 30seconds should be ample.

[Hands0n]

Absolutely concur with that - Make vibration elimination your starting point. 

There is a splendid article on the Arducopter wiki on the how to and what to look for --> http://copter.ardupilot.com/wiki/ac_measuringvibration/

What you';re aiming for is this:


My own vibrations are less than the wiki';s - and mostly it required good prop balancing and very little damping.  I used a single layer of Moongel on each of the four corners of the APM 2.5 case.

NB: Don';t forget to disable IMU logging when you';ve finished, as it does eat up some considerable CPU and log storage.

[optiplex]

ok, will check IMU logs for the amount of vibs

thx

[optiplex]

thats my graph, according to this vibrations are not the issue

the problem occures only when the wind starts, harder for it to handle even medium wind
i should add that APM and quads are not new to me, i had a reptile with apm, that was stolen
so i know what to expect from a quad in wind, but for some reason this new one just cant handle a wind as good as the stolen one
cant really compare them hardware wise ...but still, seems like when little wind starts it really affect it, like the apm dont know what to do with some wind :)

[Hands0n]

Those are good vibe levels ... thats a tick in that box then

Have you tried setting the PID in-flight using a spare channel and a rotary dial on the TX?  It may help to work on turning the Loiter P up higher or lower than they are now, but do it in flight while in Loiter.  One way will make Loiter unresponsive to the wind, too far the other will make it overreact (twitchy).  You may need to intially set the endpoints in MP quite wide and then narrow them down successively to fine tune.

My TBS Disco was like that initially - the Auto Tune did an amazing job such that it is as stable as anything can be in a breeze or light wind. So I left well enough alone.  I guess this time I was lucky with Auto Tune. It wasn';t such a good story on the Firefly 450.  I had to manually adjust afterwards to get that stable in the various modes.

[HowDoIfly]

thats my graph, according to this vibrations are not the issue

According to that you have no vibrations at all. I';ve never seen such smooth lines.  But why do they slowly increase and decrease like that?

[optiplex]

According to that you have no vibrations at all. I';ve never seen such smooth lines.  But why do they slowly increase and decrease like that?

the apm mounted on some sort of foam that blocks the vibrations i think
no idea why they go up down like that but they are well withing the acceptable limits
its like the PID I thas suppose to handle outside forces like wind is not working 

[HowDoIfly]

the apm mounted on some sort of foam that blocks the vibrations i think
no idea why they go up down like that but they are well withing the acceptable limits
its like the PID I thas suppose to handle outside forces like wind is not working 

I';m still disbelieving those are vibration logs from a simple hover. Could it be the APM is overdamped? It really doesn';t look like real-world examples I';ve seen before. Do a google image search for apm vibration and see what I mean.

http://api.ning.com/files/YMrfRKkQteQWUu-bGAVjjzoVZ4qYmOCu7q1Ux7NDrt4E-0mjAm7vGjV4Q7vbZl1*2RN31y87KsbrI9JZflhSig__/BadVibrations.png

[optiplex]

I';m still disbelieving those are vibration logs from a simple hover. Could it be the APM is overdamped? It really doesn';t look like real-world examples I';ve seen before. Do a google image search for apm vibration and see what I mean.

http://api.ning.com/files/YMrfRKkQteQWUu-bGAVjjzoVZ4qYmOCu7q1Ux7NDrt4E-0mjAm7vGjV4Q7vbZl1*2RN31y87KsbrI9JZflhSig__/BadVibrations.png

i know, its strange to me as well  the former apm had a different graph also within the acceptable limits
but i did exactly as it said in the tutorial, and it was simple stabilize hover, in zero wind (yes we have that in israel)
those are the IMU graphs and data, what can i say .... :huh
ill try again

[HowDoIfly]

ill try again

I think you plotted gyro data by mistake. This is how a smooth vibe graph would look

[optiplex]

nope, those are the imu logs
heres another capture of the same logs, u can see its IMU and the values
the graphs are just zoomed in

[HowDoIfly]

the graphs are just zoomed in

I did consider that but thought "nahhh, that';s too zoomed in" 

[Hands0n]

I am really not trustful of those IMU logs of the X Y and Z accelerometers.  They';re definitely not right.

Not sure if I';ve asked you this before - but do you get anything like the random HUD movement in Randy';s video here that explains the cause is a blown 3.3v regulator ..

http://youtu.be/1q6jjha5rgU

[jferfre]

Hi all.
My name is Pepe, and I';m new to the forum (not to the quads). I';d like firstly say hello to everyone and thanks the forum admin for their job.

Secondly, I';d like to ask the post owner if he finally could find a solution to the problem. I have exactly the same. If wind is zero, bot loiter and stabilize, are pretty good, but with some wind, the behaviour is worst than a Copter CC3D that I have had before, for example.

Today I';m going to purchase better props to see if it helps (my vibs are pretty good, I guess, too) and I must say that my frame is not good in terms of stiffness (7€...Q380 from banggod).


Thanks
^_Pepe_^

[Hands0n]

Hello Pepe,

Welcome to the forum.  This is an old thread but the information is still relevant. 

If the quad is overly twitchy in the winds at Loiter it is going to be that the gains (PID) are too high. 

Since this thread started a lot of time has gone by and the documentation on the Arducopter Wiki has improved remarkably.  I would suggest having a read of this page and work through the parameters carefully

http://ardupilot.org/copter/docs/loiter-mode.html

Pay attention to Alt Hold tuning also ...

The set up order is important, get Stabilize working first, then Alt Hold, and the Loiter before any of the other advanced assist modes.