Dangerous divergence: FIXED and understood

[afernan]

Thanks to all you for addressing this important issue.

To start with the right "initial conditions" I´d like to know from main developers what is the "deffault frame" for which the "deffault PIDs" are tunned?

Frame: quad,hexa?
ESC´s (I guess ESC32)
Motors, props
size, weight rtf, etc

The output of this "tunning tests" we are doing will be some PID´s recomendations for every type/size of frame, hopefully small wrt deffaults (i´ve sugested to built a database with fully succeded projects)
Angel

[JussiH]

As far as I know, all default parameters are based on Bills "Tank" frame - its a Rusty frame, quad +, AXI 2212-34 and 1047 APC. AUW around 1kg (I think)

I have had this issue on one of my quads earlier. In that case, going to ESC32 solved it, so I am willing to bet that in your case its tied to uneven throttle range of your SimonK ESC´s. But tuning down the P factor and limits of the Navi controller will probably help too. I have only seen it on X-quad layouts, so it maybe has something to do with the 50/50 mixer table, I would try and use a 100/100 mixer table for that instead. Both my Xquads are grunded at the moment, so I cant test.

[afernan]

Jussi:

could you please post the mixtable 100/100 for quad-X you propose?

[Max]

[phynix]

Hello, I was also goint to use the warthox quad with SimonK-esc for tests of the AQ ... maybe I schould enclose the electronics waterproof and do the testings over the lake .

The thing is, that this setup flies perfect with a Nanowii, so in my mind, the AQ should be able to handle it without problems, too. If not, something is suboptimal with it . This is no criticism on the AQ itself, I rather think that there is a reason for this behaviour that can be determined and eliminated. The AQ should simply outperform a multiwii in terms of stability and safety - but I'm confidend that it will do so soon.

I cant wait to get a working one, mine got shipped to Jussi for examination... it is so hard not to become unpatient .

[kinderkram]

phynix,
you should be good with your setup. During the proto phase the warthox or similar frames were used a lot - some are still in use.

[skew]

The best settings that work for my f450
all stock setting except
CTRL_FACT_ROLL .4
CTRL_FACT_PITCH .4
CTRL_TLT_ANG_P 55
CTRL_TLT_ANG_D 900
CTRL_TLT_RTE_D 4500

I played with just about all of the PIDs for 2 months nothing seemed to keep mine from freaking out if you tried to fly it aggressive and crashing until I lowered CTRL_FACT_ROLL .4, CTRL_FACT_PITCH .4 . After that I quit crashing and could trust it flying missions. Its doing really good right now with these settings probably 95% tuned perfect unless there are strong gust of winds that will cause problems if its at the right speed and blows just right.

[Max]

So this could be a "radio calibration" problem? I set my radio up by adjusting the throws (end points) so that the values shown in QGC match what's described here: http://autoquad.org/wiki/wiki/autoquad- ... ppm-notes/ (~ +/-700 on roll and pitch). I only adjust factor_throttle as needed.

I've been wondering if some kind of radio calibration function in QGC would be useful. There's already some code in there we may be able to leverage.

-Max

[skew]

My radio is set for +/- 700 also. I didn't have to give full throw to cause a crash just going in 1 direction fast and giving opposite control 1/4-1/2 stick and it would over correct then over correct the opposite way and it does this until it flipped upside down and drove itself into the ground. Before lowering CTRL_FACT I would have to center the stick and let it coast to stop then I could move it in another direction. After I lowered it I can command it opposite full stick it will stop and switch directions.

[LPR]

I know how hard it is to find a PID solution.

I found if there is vibration causing you to reduce the position P in order to have a non oscillating hover that you will not be able to fly as aggressively as the default PIDs will push your multi.

Lowering the P value made my multies not able to stop right on top of the home position with RTH. When the copter would go beyond the home position it would occasionally over correct and if I did not switch to manual it would end in a crash.

When I fixed the vibration issue I was able to hover in default PID and even lower the Tilt rate D.

Lowering Tilt rate D will cause oscillation if lowered too far.

If your flying well in default PIDs and you see the copter over shoot the home position even by just a foot, lower Tilt rate D.

This is what I found using the AQ with ESC32 in both a 9lbs octo with 12" x 3.8 props and a 3lbs quad with 15" x 5.5 props.

Larry.