Hi All,

I've got a very powerful quad build going on here and have finally managed to get it flyable in Stabilize. After AutoTune failed to tune it to a point where any significant roll or pitch commands would prevent it from "flipping out", I have lowered the PIDs and now she is flying like a dream... in Stabilize Mode.

What is happening now is that Altitude Hold and Loiter are unable to maintain a consistent altitude. When I keep the Quad hovering in place (either by manually doing so or using loiter), I get an altitude change of +/-1-2Meters. This is a bit of an issue but even more so is that when I fly it laterally, it looses altitude quickly and all of a sudden pitches/rolls violently then shoots up into the sky where I have to quickly change over to Stabilize to correct as some of the pitching is close to 90 degrees!

Here are the current PIDs that I have which allow me to fly nicely in Stabilize:

Here are the Specs on the quad:

Motors: T-Motor 3520-11 400KV

ESCs: Phoenix Edge Lite 50AMP

Props: 14x4.7 APC SlowFly

APM Version: 2.6 with Firmware Version 3.1

 

Here is what I am seeing in my log with regard to the sudden, not expected and not requested, altitude changes while flying in Loiter and Altitude Hold:

The logs are attached to this post but as you can see above, right in the middle of the flight before I switch to Stabilize, the altitude starts climbing QUICKLY! Also, at this point, the copter is pitching and rolling out of control hence the switch to Stabilize. In the next Altitude Hold attempt, you can see the copter starts climbing on its own again and cannot maintain altitude whenever in a mode that requests altitude hold!

 

The APM is sitting on Alpha Gel to reduce Vibration and is also covered in a dark soft foam to keep the light off the barometer and to protect the APM.

 

Please, any help would be greatly appreciated! We've been trying to get this thing flying properly for some time now and though it flies fine in Stabilize, we want to fly without this erratic behavior in Altitude Hold and Loiter. I foresee this issue being present in AUTO mode too because it seems to be linked to the Barometer and don't want to lose this heavy and powerful quad.

 

Thanks in advance to anyone who can help!

 

Kind regards,

Oliver Volkmann

Micro Aerial Projects L.L.C.

Views: 17837

Attachments:

Replies are closed for this discussion.

Replies to This Discussion

Hi,

Do you use sonar? If yes it might be the cause of erratic altitude control. 

Another verification is to be sure your barometer does not get airflow from the props; it can indeed disturb pressure measurements and thus altitude.

As for PID tuning, you P rate is quite low, meaning you are way overpowered; why don't you try 13x4 or even 12x5 props ? Verify you have tuned your mid throttle in APM's parameters list to the value your quad hovers (for ex set it to 300 if your quad hovers at 30%)

Also I would set I to same value as P, not higher as you have it now.

Hope this helps

My experience with an overpowered quad: the first param I adjusted was Rate D, and I ended up with 0.016 instead of 0.004. It was buzzing a bit with high-freq oscillation, so I lowered Rate I a bit, then started raising Rate P.

Had almost 5 to 1 thrust/weight ratio on that one.

Also make sure that your THR_MID parameter is set appropriately.

Hi Hugues,

Thank you for getting back to me here. I do not use sonar. The APM is also covered in foam to prevent any "blasts" of air on it. Here is a picture:

I am wanting to use these large props for more lift, hence the choice of motors which are large to say the least. We hope to carry some weight for more than a few minutes using this setup.

Could you tell me why you would put the I value the same as P? To be honest, I don't really know what the PIDs really do even though I have read all about them but for some reason the language doesn't make much sense to me.

Admittedly, the quad mid throttle is too high, meaning I maintain a hover around 40% and I have not changed this so thank you, I will start there.

Yesterday I flew the quad again but this time used a lighter foam underneath the APM suspecting that the Alpha Gel I was using was not "soft" enough and though I tried to flip the copter by flying fast then abruptly letting go of the sticks, shaking the quad left and right and violently moving the stick around, I did not have any of the weird reactions that I had yesterday. I suspect that vibrations were causing a lot of the trouble on this hardware store copter. I looked at the IMU data but it seems I still exceed the recommended vibration levels though... perhaps you can take a glance and let me know?

One thing that I did notice during the last flight yesterday was a pendulum effect while in Loiter. As soon as the sticks were released, the copter would rock back and fourth for a few cycles before coming to a rest. As soon as there was wind while it was hovering, this would start again. When flying into the wind and letting go of the sticks, the "pendulum" becomes quite fast and the copter angles quite a lot. What setting would I have to change to calm it down a bit in loiter and remove the pendulum effect? I don't have toilet bowling... thank goodness.

Log files from yesterdays flight are attached.

Thanks again Hugues for your help!

Thanks Daniel Sisson, I have not done that which might be one of the problems. Will let you know if that solves the altitude changes during flight (where throttle stick is centered and flight mode is alt-hold or loiter).

The log file is larger than 7mb... Uploading elsewhere right now

here it is zipped.

Attachments:

Do you have a log with vibration logging turned on?

If your vibes are too high althold can go nuts, as it can't accurately measure whether you're moving up or down in amongst all the high frequency noise.

I have had similar problems where the copter suddenly embarks on a power ascend prompting a quick return to Stabilize and a manual descent. Thus far I can correlate the behavior to unusual spikes in zacc. Other sensors and parameters seem sane, but VFR_HUD.climb gives negative values while the craft is actually ascending rapidly (barometer alt and gps alt agree).
Attachments:

to test the vibrations, you need to turn on the imu option in the logs.

http://copter.ardupilot.com/wiki/common-diagnosing-problems-using-l...

there are many documented attempts at reducing vibrations. personally, i think adding some mass to the APM/case and mounting it on a soft medium does best. i found a 0.035" aluminum plate, cut it to the dimmensions of the case, epoxied it to the bottom of the case, then mounted to dubro foam from the hobby store

Do yourself a huge favor.  http://diydrones.com/forum/topics/arducopter-tuning-guide

Follow Dave C's tuning instructions.  That will fix all the overshoots, wobbling, pendulum, etc.  It will do wonders for you. The default PID settings will never work for you, and blindly changing them will just make it worse. Been there, done that!  My first 2 weeks of flying were on stock settings, or randomly adjusted values.  Last weekend I followed Dave's procedure and it is like a whole new world.  Then on the next calm wind day, I ran the auto-tune mode to see what it came up with, compared to what I came up with.  The auto-tune backed my PID settings down just a hair, but it was all very very close.  The auto-tune settings were definitely an improvement, mine were indeed just a hair too high.  But I'm not sure how well the auto-tune would work if you started out completely screwed up.  Never tried that.

Secondly, as others have said, the vibrations will ruin you. Wedging the APM into foam isn't necessarily helping.  I don't see any vibration dampening gel pads in that photo. Just the APM wedged into some foam.  This will not help, and may even amplify it.  The APM should be firmly attached to 4 square gel pads.  Those 4 gel pads should be firmly secured to a solid surface that doesn't wobble or move around.  I used Kyosho zel jel pads, and some double-stick tape on each side of the gel.

RSS

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service