Understanding of motor mixing table

[Ecky]

Hi Tilman,

you descibed perfectly my experience!
One thing make me wonder, you speak from increase/decrease Tilt Rate I. Following the WIKI:
"The rate angle controller only operates with a D term because the controller set point (command) is set at zero and the system changes are quiet fast."

What does changing Tilt Rate I effect?

BTW talking about "default PIDs". What are the default PIDs
We have at least two (three) options:
1. WIKI Pids
2. Aquad defaults after resetting the firmware
(3. Pids from the instruction videos shown as default Pids)

Greatings
Christian

[r0sewhite]

Oops, I'm sorry. Of course I meant Angle, not Rate.

Default PIDs are the values you get when you reset CONFIG_VERSION to 0. Afaik there are differences in default values between different software versions so I recommend to not rely on some possibly old screenshots.

[Max]

Hi Tillman,

Just curious, how did you determine that the current default PIDs are the best for your craft(s)?

Regarding the X8, was it overpowered by any chance? Meaning motors at <~45% in hover. Any other outstanding characteristics? I've seen pics of your multiple AQ builds, so what other configs have you tried and what differences have you found, if any, regarding tuning? Do you consistently find that changing the mixing table first is better than starting with PID adjustments?

I only have 3 builds to compare, 2 of them Y6 configs, so my practical experience is limited. I've had good luck with the current default control PIDs, only needing to tune them if there are too many vibrations in the frame, or to compensate for poor calibration (or try to). A lot of the Nav settings are more of a personal preference, so I'm discounting those.

The main basis for changing mixing table seems to be to keep the default PIDs. To me it makes more sense to adjust the mixing only due to physical requirements, like being over powered, slow props, unbalanced load, and the like (as you said). The flight code should already take care of distributing the power correctly if given the proper inputs. I think it's important to understand why exactly one needs to reduce available power on any given axis. It could be a characteristic of the flight code of course, though personally I have not observed that.

Have you seen my "slipping prop" video? I don't think there's any way it would have recovered from that if it didn't have 100% power on all axes/motors. Though I'm not willing to repeat the experiment

Also I'm wondering if you've seen Angel's paper on this subject? There's a post somewhere here and the PDF is in the d-loads section of AQ site.

Thanks,
-Max

[teramax]

Max i understand what 100% means to you in this specific propslipping case. Sure 100% aviable power on every motor is what every one needs but i have also seen a Hex + with 100% on every motor which was nearly thorn appart in midair cause of overpowering and fighting against it selve.
I had to lower the PID´s very much to get a nearly controlable rig but the first windgust blowed it away and it got uncontrolable again.
Lowering specific motors power made a good PID tuning possible. Now outside influences are no more a problem.
What i whant to say is that playing with the mixing is not an universal remedy but it can help to find the best balance and not to go to far away from the approved and rely good working default PID´s.

[Max]

Right, I understand what you and others are saying about this. What I'm asking is why does that particular hexa need the motor outputs lowered? Or is that true of all hexa+ configs across the board? If it's that particular multi, it might suggest something sub-optimal in the setup (or not), while the latter would suggest that the code/mixing strategy could be improved.

-Max

[teramax]

i had only 3 Hexas to try till now but dont know if you remember the first Hexa mixing theory a time ago was 100/70/100 and i´d never got it to work. Now the mixing like you find it in the QCG/defaults is working flawless for me and it is 100/50/100.

Same example if you try to build a Quad X with 100% on all outputs. It does not work what i´ve seen so far but proof me wrong if i´m wrong.
Rosewhites X8 is somthing like 2 Quads one above the other and so why should be a simple Quad X mix not the best solution.

But please, i am also still learning and share my thoughts

I will try somethin other, i´ll take my DJI Naza were i also can customize mixingtable and try if it behaves similar to AQ, just to prove my theory.

[r0sewhite]

Hi Max,

by now I flew the AQ on following copters:
QuadX, 45cm span, Hacker A20-26M, 9x4,7" props, 3S, AUW 1kg
X8, 70cm span, Smartdrones 550, 15.7" props, 4S, AUW 3.2kg
X8, 70cm span, Tiger MT3506, 13x5" props, 4S, AUW 3.3kg
Flat Octo, 100cm span, Tiger MT3506, 13x5" props, 4S, AUW 3.8kg
Flat Octo, 88cm span, Plettenberg Orbit 10, 12x5" props, 4S, AUW 4.3kg

There were some well powered copters but also underpowered copters. As you probably know a 3.8kg Octo with MT3506 is no rocket at all.

If the mixer values are too high, you need to lower PIDs and thus your reasoning of 100% power for emergency cases is obsolete. High mixer values with low PIDs give the same maximum output as low mixer values with high PIDs. Keep in mind that mot_pitch and mot_roll do not allow higher values than 300 as long as you don't change it. At +/-300 the output clips and this is something we don't want.

Instead our goal is to use the maximum dynamic of mot_pitch and mot_roll to be on the safe side under any circumstances. Your high mixer values didn't help to recover your copter more than lower mixer values with higher PIDs would do because in both cases pitch and roll is limited to +/-300.

I have read Angel's paper and it confirms my experience. If your mixer values are too low, you will lose dynamic in controlling pitch and roll. If they're too high, you will have too much dynamic and you have to reduce it by lowering PIDs. At the right point (maximum mixer values but no PID tuning necessary) you reach the point of maximum dynamic.

[Max]

[Max]

[chschmid]

Gents
I can relay to all sayings below. My conclusion on all the multis seen here with the behavior described and my own findings is this:

AQ default PID's are perfectly tuned for a system with average motor power and average torque where the motor outputs are between +-150 -200 and the pitch-roll outputs in a range of +- 150. I guess these PID's where found on a system with 100% output "per axis" (QuadX 50/50, Quad+ 100/100).

When using Motors with a high torque (and power) @ high mix rate, the system becomes "to dynamic" as described by Mäx and found by myself.

description: But what makes me more nervous is that a gust of wind makes the motors "go wild for a 1/10 sec". It looks like overshooting when fighting the gust. A vibe goes thru the hexa. I can even hear the mots screech.

I am positive that a new set of ideal PID's could be found for a high torque setup. I was trying to adjust the PID's on my high torque hexa using the new standard mixing from QGC but I did not achieve as good results as AQ could. My next try will be standard PID's and lowering motor mix again.

Conclusion:
Underpowered systems could loose stability control when outputs are clipped.
Overpowered systems with high torque are too dynamic and overcorrect in some conditions with standard PID's.
A dampening can be achieved with lowering the motor mixing.
The best target I found so far is motor outputs between +-150 -200 and the pitch-roll outputs in a range of +- 150.

These are just my 2 cents.

It would be very interesting to compare some "very good flying" log data sets regarding p-r outputs, mot outputs and mixing. Maybe there is a ideal relation.

Cheers
Christof