Another AQ quad - has suddently started twitching

The last Bastion to fine tune your FC matching your airframe & setup

Another AQ quad - has suddently started twitching

Postby phynix » Sat Mar 02, 2013 5:40 pm

Hello, hopefully this is the right Topic for posting this.

I am tryng to set up an AQ on a light quad (Warthox Frame, Suppo 2212, 9''x4.5'' Gemfan, 20 SimonK, 700g w/o battery). This setup has flown perfect with a Nanowii, so I thought it is good for starting with the AQ.

It worked quite OK in manual mode with the AQ, but not altitude hold in PH. So I tried some variations in the Altitude-PI settigs. But no stable result so far.

Today I tried to reduce Vibrations, set all parameters back to the values in the wiki http://autoquad.org/wiki/wiki/flying/tu ... on-tuning/ (except factor throttle, so it hoovers at centered throttle stick (with 0.5)). Then, I did another testflight (in light wind). Flying in manual mode worked good as usual (without trying extreme maneuvers). But it was still not holding hight. it "Jumps" successive upwards or downwards. I have to move the throttle stick quite far to get a reaction with altitude overwrite (sometimes i cant give enough, so i have to switch back in normal mode to safe the craft).

Well, then I tried some more aggressive maneuvers in manual mode. When I was stopping from higher speeds, it imeadiately flipped and tumpled to the ground. Hight was almost 10m I think. there was no increasing whobble or shaking, just a simple flipping and then no more controll at all.

The result are at least slightly bended arms, a damaged Lipo and broken proppellers. In addidion, when i tried another flight with new probs afterwards, it continuously rolled to the right, so that i hade to continuosly increase the roll command to the left to counteract. Is it possible that the sensors got damaged?!? The AQ-bard itself didnt get a scratch so.

20130302_132533.jpg


20130302_132215.jpg


When I removed the AQ-Board from the frame for inspection, I discovered, that the cable from the GPS-antenna was somehow loose in the plug and detached completely when I tried to disconnect the plug. Unfortunately it broke when I was trying to reattach it.

So, I have a youpe of questions so far (a logfile of the flight decriped above can be found here https://depot.uni-konstanz.de/cgi-bin/e ... 5kzmekasga ):

What is the reason for the failure of altitude hold? Noise? Parameters? I know that I have vibrations on this frame that I cant get rid of completely. But it flies good in manual mode with the AQ and flew also with the nanowii. In addition, it is used by other people... I simply dont have the experience to decide what is the most important thing to focus on for solving this issue.

What is the reason for the crash? is this the "divergence issue" descripted in other posts? Is it probable that the sensors got damaged as a result of the crash?

Any ideas what to do with the plug for the GPS-antenna?

Is the Lipo (see picture) still safe to use? Any experiences about it?

Thank you in advance for your efforts and with best reagards, Sören
Last edited by phynix on Sun Jun 30, 2013 6:16 pm, edited 1 time in total.
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: Another AQ quad - support is welcome

Postby phynix » Sun Mar 03, 2013 12:11 pm

Finaly, I managed to upload the video to the descriped above! (the logfile is https://depot.uni-konstanz.de/cgi-bin/exchange.pl?g=5kzmekasga)

http://www.youtube.com/watch?feature=player_embedded&v=3U88G5V5uEA#t=2s

Hopefully that helps to determine the reasons for not holding altitude and for flipping,
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: Another AQ quad - support is welcome

Postby JussiH » Sun Mar 03, 2013 3:37 pm

You have quite sever mismatch between your pitch and roll motors which is a known cause for this kind of problems. I see you are using SimonK ESC´s, and you should calibrate them to a 1000-1950 uS throttle range and recalibrate them regularly.

Mismatch between motors can also be caused by some motors beeing tilted on the horisontal plane.

Notice that Pitch motors are working a lot harder than roll motors:
Motors.jpg


Try and recalibrate the ESC´s and check if the roll and pitch output is matched in a hover - it should also make your hover much more stable. You should be seeing curves of same magnitude and placed over each other.
JussiH
 
Posts: 2103
Joined: Thu Jun 21, 2012 9:25 pm

Re: Another AQ quad - support is welcome

Postby JussiH » Sun Mar 03, 2013 3:38 pm

PS: Be nice and dont post links to secure webspaces! Use google docs or similar...
JussiH
 
Posts: 2103
Joined: Thu Jun 21, 2012 9:25 pm

Re: Another AQ quad - support is welcome

Postby bn999 » Sun Mar 03, 2013 5:23 pm

phynix wrote:Hopefully that helps to determine the reasons for not holding altitude and for flipping,


I had a look at your log. The cause of the crash was saturation of the rate gyros.

rates.jpg


The gyros are only 500 deg/s. You can see that you exceeded this significantly up to the point where the gyro saturated to the positive side. Once this happens, the attitude filter integrates rates that are lower than what is actually happening. This results in an attitude quat which is way out of sync with reality and there is not nearly enough time for the filter to re converge. Once this happens, it is game over.

I guess no one has ever actually stated this in the docs before, but you absolutely cannot allow the machine to rotate any faster (or better yet, no where near) 500 deg/s in any one axis.

So, the next question is why was the machine rotating that fast? Well, it is a small frame, which makes it physically easy to do, but I think the tuning of your attitude PIDs is to blame. I won't get into PID tuning detail here, but you need to tune them such that they will not allow rotations of such high magnitude. I would first look to constraining the output of the angle P term which might mandate that the I term is also constrained some more. It is also the rate PID's job to resist large changes in rate, so its output might need to be increased and/or its F term decreased to spread the counter force out more over time.
bn999
 
Posts: 1559
Joined: Thu Jun 21, 2012 11:40 pm

Re: Another AQ quad - support is welcome

Postby phynix » Sun Mar 03, 2013 11:24 pm

Thank you very much for you support! I realy appreciate it.

About the webspace, there shouldn't be a problem. At least I hope so! It is a service of our University that can be used by any member. You should not be able to access anything else except the file I provided, please let me know if I'm wrong!

@ Jussi: I calibrated the esc with the RC (Graupner Hott) after I adjusted the servo settings so that I get values in QGC from -700 to 700. Is that OK? I will check the alignments of the motors. Do you think this could be the reason for both, the failing in Altitude hold and the flip? The position in PH was relatively accurate. Its only the altitude that i dont get stable.

@ Bill: That makes totally sense to me. I was not aware of the danger of saturation. I will try to familiarise myself with the two PID loops. In the wiki it is stated, that the rate-loop is running inside the attitude( if I remember it right). Does this mean that the attitude-loop is providing the setpoint for the rate-loop? Understanding the interaction between the two loops might help me finding a good set of parameters. I am confident concerning this matter. Do you have an idea what might be the problem in altitude hold?

Right now I have another issue: I mounted new props and tried whether everything is still working. But the quad is now slowly, but constantly increasing its roll. I have to constantly increase the roll command in the other direction to counteract until the limit of the stick is reached (about 10 sec). Is it possible that the sensors got damaged? The board didn't get a scratch so.. (see picture above). A run for static calibration is on its way...
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: Another AQ quad - support is welcome

Postby phynix » Sun Jun 30, 2013 1:48 pm

Hi, it has been a while now since I posted the last time... everything went well since today.

With a new board, my small quad was flying nicely right from the start. I was impressed by the stability of the system even though it uses simonK esc and Suppo motors.

I combined it with a Alexmos VectorDD gimbal, and the video that I was getting is extremely good for such a relatively cheap system - I am still amazed about it :).

But after many flights without any troubles, today my quad crashed. Camera still OK, the Gimbal is probably destroyed. While hoovering in manual mode, it started to somehow twitch, constantly increasing its pitch and finaly resulting in a crash. One could hear the throttle of the motors somehow peaking at a frequency of about 5 Hz.

I had this sometimes before right after takeoff. Till today I thought it was due to movements while plugging in the battery. It was easy to land quickly, reset the AQ und everything was fine again.

But today it occured in mid flight, after maybe 2 minutes without problems. I checked the logfile, the RC was not the reason, battery voltage was fine. Gyro and Motor outputs seem to match, the peaks are in the motor-outputs, and the gyro shows small spikes that seem to match to the observed movement off the copter. I can post the log here if necessary.

I changed the broken probs, now the quad (without gimbal) is flying perfect again. For testing purpose, I varied all angle PID- parameters to see whether they can somehow create a divergence, but in a first attempt I was not able to reproduce it. It is quite robust to changes in the PID-parameters ... recovers nicely even from extreme flying and overshoting its target angle (with bad settings).

Has anybody observed this issue before? Someone has an idea? I still like the AQ very much, but I need to find out what was the reason and eliminate it before I put it on a more expensive Hex carrying a DSLR ...
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: Another AQ quad - support is welcome

Postby JussiH » Sun Jun 30, 2013 4:56 pm

Pls post your mixer table. If using the old 50/50 mixer table, this can happen. The solution is to use the 100/100 mixer table that is available in recent versions of the GCS.
JussiH
 
Posts: 2103
Joined: Thu Jun 21, 2012 9:25 pm

Re: Another AQ quad - has suddently started twitching

Postby phynix » Sun Jun 30, 2013 5:36 pm

I am using the 100/100 table (default for quad x in the new qgc) - there is no saturation in the logs of mot_pitch / roll in the logfile when the spikes appear. At the end, direct before impact, it reaches the 300 limit. It was not realy an divergence that was building up in both direktions (in +/- pitch), bur rather a continuos shaking in one direktion. Unfortunately, I have only the onboard video - and the VectorDD Gimbal removes the shaking :).

I have to admit that the quad is not perfectly balanced in the pitch axis - but I also made this imbalance artificialy worse with no negative effects observable.

I just have flown four more batteries (without the gimbal) ... I cant reproduce this behaviour. I tried realy to fly aggressive to force some kind of divergence, with no succes yet - as soon as I let go the sticks it always recovers nicely.
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: Another AQ quad - has suddently started twitching

Postby phynix » Tue Jul 02, 2013 1:26 pm

Is it possible that this behavior comes from one of the esc and not from the AQ itself? I tried to apply fast throttle changes on them manually direct from the RC and did not see any problems. BUT one of them sometimes has some troubles to find its kommutation when starting...

I have no clue, but realy want to figure out what causes this. I remember someone reported the observation of a similar behaviour, but I can't find the corresponding thread...

Besides this the AQ flies so reliable so far that I realy like it :).

Does soemone has a good Idea for narrowing down the possible causes for this problem? Thank you in advance ...
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm


Return to PID Tuning

Who is online

Users browsing this forum: No registered users and 5 guests

cron