Request for yaw test on r1539

This forum is dedicated to software development related to MultiWii.
It is not the right place to submit a setup problem.
Software download
Post Reply
Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Request for yaw test on r1539

Post by Alexinparis »

Hi,

I've just uploaded a new dev (https://code.google.com/p/multiwii/downloads/list)
I would like to have some feedback about yaw behavior:
- on tricopter
- on quadcopter

Use the yaw rate setting to increase yaw rate which is now proportional.
The yaw rate can be set up to 1.0 to have approximately a 2 turn/s yaw rate ability.

You should be able to do this kind of thing without a mag:
https://vimeo.com/28125726 at 3:00
the angle should be like "memorized"

some hint to setup a tricopter with this:
1) set I setting to 0, and leave P setting to default. D term is not used.
2) yaw stick must be centered at 1500 (no trim)
3) fly and adjust the linkage to remove any yaw drift
4) only at this step, reset to default I setting
If not well adjusted, you could have asymetrical overshoot

-ralf-
Posts: 215
Joined: Mon Dec 03, 2012 7:08 pm

Re: Request for yaw test on r1539

Post by -ralf- »

Alex, I can test it next weekend with my hexacopter.

Will you be so kind to implement the bug-fix discussed here:

http://www.multiwii.com/forum/viewtopic.php?p=38772#p38772

Thanks

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

Tested it. Hold does seem better than 2.2, but yaw hop is, well frightening. I have Yaw rate at 0.50 (1.00 in 2.2), and when I Yaw in either direction there is a small jump up at start of yaw and a big jump up at end. If I move the quad with my hand there is a very big jump upwards.. I can lower the rate, which does lower yaw jump significantly, but I like a quick yaw on my quads. Which of the P I or D would have affect on jump if any?

Thanks, I think this is definitely progress! It wants to hold it's position as the user sets it.

Andrew

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

Most testing tonight, Yaw holds great as long as I is set to greater than 0.00, the default which I think was 0.024 is good on my quad. Played with P, I, D and rate to see if I could get rid of the Yaw hop, if I reduce yaw to very slow it's fine, but significant yaw causes a jump, letting it spin like a top (boy can it ever) and then stopping it results in a 5ft jump upwards.

Andrew

**Edit: I should mention this is a quad, a tricopter wouldn't hop....

User avatar
Hamburger
Posts: 2578
Joined: Tue Mar 01, 2011 2:14 pm
Location: air
Contact:

Re: Request for yaw test on r1539

Post by Hamburger »

I will do TRI and (later) HELI testing asap

wilco1967
Posts: 156
Joined: Thu Aug 18, 2011 6:04 pm
Location: Winterswijk, Netherlands

Re: Request for yaw test on r1539

Post by wilco1967 »

Hi Alex,

Works beautifully on my old tricopter.

Even with mag deactivated (aux switch off, but it is still in the config), the tri will hold its heading much better then in earlier versions with mag (heading hold) active.

When I hold it in my hand, and force it into another heading, it will forcefully try to come back to the memorized heading.
A slight blip on the yaw stick, and the current (forced) heading is taken as a new memorized heading.

I had to take down the Yaw P quite a lot (from almost 7, to about 3, to have similar feel).

I like it ! much more locked in.
(Even with my old beaten up yaw servo construction which has a lot of mechanical play.)

PS. this was using Flyduino mega, with #define MEGA_HW_PWM_SERVOS active (if that makes any difference).
I use yaw P = 3 (was 7), and I = 0.045 (as before). D=0

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Request for yaw test on r1539

Post by Alexinparis »

doppler wrote:Most testing tonight, Yaw holds great as long as I is set to greater than 0.00, the default which I think was 0.024 is good on my quad. Played with P, I, D and rate to see if I could get rid of the Yaw hop, if I reduce yaw to very slow it's fine, but significant yaw causes a jump, letting it spin like a top (boy can it ever) and then stopping it results in a 5ft jump upwards.

Andrew

**Edit: I should mention this is a quad, a tricopter wouldn't hop....


Hi,

thank you for testing

I suggest 2 independent things:
1)
add this line in multiwii.cpp:
#ifndef COPTER_WITH_SERVO
PTerm = constrain(PTerm,-GYRO_P_MAX,+GYRO_P_MAX);
PTerm = constrain(PTerm,-100-abs(rcCommand[YAW]),+100+abs(rcCommand[YAW]));
#endif

2) lower P setting

But in any case, I think reducing yaw jump on a quad would result in a less accurate yaw stop or start.

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Request for yaw test on r1539

Post by Alexinparis »

Hi,

I have also very good result on my TRI :)
MEGA_HW_PWM_SERVOS should help a little bit as there is no jitter with HW PWM and more resolution.
But even without it, the result should be interesting.



wilco1967 wrote:Hi Alex,

Works beautifully on my old tricopter.

Even with mag deactivated (aux switch off, but it is still in the config), the tri will hold its heading much better then in earlier versions with mag (heading hold) active.

When I hold it in my hand, and force it into another heading, it will forcefully try to come back to the memorized heading.
A slight blip on the yaw stick, and the current (forced) heading is taken as a new memorized heading.

I had to take down the Yaw P quite a lot (from almost 7, to about 3, to have similar feel).

I like it ! much more locked in.
(Even with my old beaten up yaw servo construction which has a lot of mechanical play.)

PS. this was using Flyduino mega, with #define MEGA_HW_PWM_SERVOS active (if that makes any difference).
I use yaw P = 3 (was 7), and I = 0.045 (as before). D=0

copterrichie
Posts: 2261
Joined: Sat Feb 19, 2011 8:30 pm

Re: Request for yaw test on r1539

Post by copterrichie »

Personally I really don't understand the problem people are having with servo jitter. All of my bicopters are using two servos and Pro-mini Arduinos. In the very very early days, I have a few problems but I resolved that by powering the servos from a separate power source.

User avatar
shikra
Posts: 783
Joined: Wed Mar 30, 2011 7:58 pm

Re: Request for yaw test on r1539

Post by shikra »

Have to say - my Scarab Tri is amazing already with Multiwii.
It will hold position for absoloutely ages. Longer than I can be bothered to keep in position. 30 secs I get board and wanna fly around.

But I do tune as Alex suggests. Trim mechanically with I=0...

http://www.rcgroups.com/forums/showpost ... ostcount=1

scrat
Posts: 925
Joined: Mon Oct 15, 2012 9:47 am
Location: Slovenia

Re: Request for yaw test on r1539

Post by scrat »

Sorry for posting this question in this thread..but it is for r1539 dev:

Is this suitable for MWii version 2.2 too with CRIUS AIO Pro v1.1?

// avoid using 115200 baud because with 16MHz arduino the 115200 baudrate have more than 2% speed error (57600 have 0.8% error)
#define GPS_BAUD 57600

Thanks.

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

Alexinparis wrote:
doppler wrote:Most testing tonight, Yaw holds great as long as I is set to greater than 0.00, the default which I think was 0.024 is good on my quad. Played with P, I, D and rate to see if I could get rid of the Yaw hop, if I reduce yaw to very slow it's fine, but significant yaw causes a jump, letting it spin like a top (boy can it ever) and then stopping it results in a 5ft jump upwards.

Andrew

**Edit: I should mention this is a quad, a tricopter wouldn't hop....


Hi,

thank you for testing

I suggest 2 independent things:
1)
add this line in multiwii.cpp:
#ifndef COPTER_WITH_SERVO
PTerm = constrain(PTerm,-GYRO_P_MAX,+GYRO_P_MAX);
PTerm = constrain(PTerm,-100-abs(rcCommand[YAW]),+100+abs(rcCommand[YAW]));
#endif

2) lower P setting

But in any case, I think reducing yaw jump on a quad would result in a less accurate yaw stop or start.


Thanks Alex, I've added the line but haven't had a chance to test it yet, hopefully today. I will also play with P, you mentioned somewhere that D has no impact? I swore in my testing it did, but perhaps that was just placebo effect?

On your comment about less accurate yaw stop or start, yes perhaps, I would say if you have your quad spinning at 120RPM that an accurate stop less important, just that you stop and don't jump up into the sky.. Other boards can do this (KK, KK2 are one's I've used in the past), I'm not sure how it's done, perhaps there is a ramp down from the speed, perhaps we need a Yaw Rate correction on P where low Yaw rates retain the P setting, and higher rates chip away at it...

Every time I think I understand PID, I learn something new, so I could be speaking out of a place with no mouth....

Thanks again!

Andrew

Noctaro
Posts: 280
Joined: Thu Sep 08, 2011 11:15 am
Contact:

Re: Request for yaw test on r1539

Post by Noctaro »

Hey all!
I tested the new def release on my quad-x and for now, it does exactly what it should.
If i turn it by handforce, it will recover to desired position. (Of course tested in acro mode without active mag)
The only thing i expirienced for now was a bit of yaw overshoot, but i did not do any PID-meditation for now. My standart PIDs(V2.2) at LPF 42khz.
Also YAW seems to be less critical in fast turns. Behaves quite cool!

Thank you Alex! :)

Could we just dampen the return at big yaw changes? As it makes copter unstable in height and also a bit in attitude if reacting that strong.
Maybe its not needed as i start playing around with YAW PIDs.
Btw. i did fly at 0 Yaw rate, wich feels quite perfect for capturing videos, as most cams cant capture good footage if yawing too fast.
Ill keep on testing this today.

Greetz Noc

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

Noctaro;

I'm seeing the altitude gain, quite significantly after some strong yaw movement, if you find PID's that address it, can you please share it?

You might also want to try adding the link that Alexinparis indicated may help.

I still haven't tried it yet, it's loaded and ready to go, just need to get outside.

Andrew

scanman
Posts: 74
Joined: Thu Jun 21, 2012 9:26 am
Location: Durban, South Africa
Contact:

Re: Request for yaw test on r1539

Post by scanman »

yaw servo not working at all on this version. no response at all.

I'm using Crius AIOP (yaw servo is on pin2).
Yaw servo works fine on previous versions.

User avatar
Hamburger
Posts: 2578
Joined: Tue Mar 01, 2011 2:14 pm
Location: air
Contact:

Re: Request for yaw test on r1539

Post by Hamburger »

r1541 @ mega2560 with hw-pwm on a TRI - working good (after some retuning from old r14xx PIDs)
Holds the tail well, could not initiate yaw spiral of death.

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

So I flew today with r1539 with the extra line Alexinparis asked me to try. P for Yaw was anywhere from 4.7 to 9.0. At 4.7 it drifts and then slowly brings itself back to where it should be, it's a little unsettling to watch, feels so loose. Up'd the P to 9.0 holds much better.

Now for the YAW hopping (altitude gain), this is improved with the line, but however, now it overruns when you stop fast and long YAW movements, but then it slowly yaws in reverse back to where I suspect I let go of the stick.... also unsettling to watch, do not like... again hopping is improved, not resolved.

Andrew

User avatar
Hamburger
Posts: 2578
Joined: Tue Mar 01, 2011 2:14 pm
Location: air
Contact:

Re: Request for yaw test on r1539

Post by Hamburger »

I think the current yaw implementation does not take the yaw midpoint into account as the neutral point?

gektor
Posts: 10
Joined: Thu Jul 18, 2013 5:20 pm

Re: Request for yaw test on r1539

Post by gektor »

Will this new Yaw algorithm also work with AlexK PID Implementation?

User avatar
Hamburger
Posts: 2578
Joined: Tue Mar 01, 2011 2:14 pm
Location: air
Contact:

Re: Request for yaw test on r1539

Post by Hamburger »

no. completely separated

scanman
Posts: 74
Joined: Thu Jun 21, 2012 9:26 am
Location: Durban, South Africa
Contact:

Re: Request for yaw test on r1539

Post by scanman »

does anyone have any tips for me to get the yaw servo to work on this version? It is completely dead
Crius AIO Pro v2.

I did all the same settings on the earlier versions that worked.

SergioC
Posts: 19
Joined: Sat Dec 22, 2012 11:33 pm

Re: Request for yaw test on r1539

Post by SergioC »

the yaw jump problem is caused by not taking in account the MINTHROTTLE motor lower limit. When a sharp yaw correction is required one motor will spin up the other will slow. If the slow motor is to be commanded below MINTHROTTLE it will be instead tied to MINTHROTTLE but the opposite motor will be sill commanded a higher throttle even if the slow is saturated. Hard to explain with my not proper English :)
It is a similar reason because the line n.1424 in Output.cpp exists

Code: Select all

if (maxMotor > MAXTHROTTLE) // this is a way to still have good gyro corrections if at least one motor reaches its max.

But you cannot code something like

Code: Select all

if (minMotor < conf.minthrottle) // bla bla bla

because it will exaggerate the jump

The fix is to limit yaw correction when the slower motor hits conf.minthrottle, but the main loop should knows the mixing table, breaking the code abstraction. I still need to sharpen the picture (more caffeine).

At an extension this could fix also the "turbulence/buffets induced jumps"


Sergio.

trello
Posts: 19
Joined: Wed Apr 10, 2013 1:04 pm

Re: Request for yaw test on r1539

Post by trello »

scanman wrote:does anyone have any tips for me to get the yaw servo to work on this version? It is completely dead
Crius AIO Pro v2.

I did all the same settings on the earlier versions that worked.


I believe the motor pinout changed sometime between release 2.1 and 2.2 although i don't know where this is documented. I'm flying a tri with aio v2.0 and i think i had to use the pins from left to right (2, 3, 5, 6) while most diagrams show pins 9, 10, 11 and 3 for tri.

I can check the correct order when i'm at home.

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

SergioC wrote:the yaw jump problem is caused by not taking in account the MINTHROTTLE motor lower limit. When a sharp yaw correction is required one motor will spin up the other will slow. If the slow motor is to be commanded below MINTHROTTLE it will be instead tied to MINTHROTTLE but the opposite motor will be sill commanded a higher throttle even if the slow is saturated. Hard to explain with my not proper English :)
It is a similar reason because the line n.1424 in Output.cpp exists

Code: Select all

if (maxMotor > MAXTHROTTLE) // this is a way to still have good gyro corrections if at least one motor reaches its max.

But you cannot code something like

Code: Select all

if (minMotor < conf.minthrottle) // bla bla bla

because it will exaggerate the jump

The fix is to limit yaw correction when the slower motor hits conf.minthrottle, but the main loop should knows the mixing table, breaking the code abstraction. I still need to sharpen the picture (more caffeine).

At an extension this could fix also the "turbulence/buffets induced jumps"


Sergio.


Your thoughts on this matter look very promising Sergio, and it makes sense that it could be exactly that given the way the quad reacts, when I'm doing fast Yaw, spinning it like a top, you can clearly see two motors barely spinning, my min-throttle is 1100 which with my ESC's and motors is a very light spin. I would love to test any changes you can suggest that may fix this.

thanks
Andrew

SergioC
Posts: 19
Joined: Sat Dec 22, 2012 11:33 pm

Re: Request for yaw test on r1539

Post by SergioC »

doppler wrote:Your thoughts on this matter look very promising Sergio, and it makes sense that it could be exactly that given the way the quad reacts, when I'm doing fast Yaw, spinning it like a top, you can clearly see two motors barely spinning, my min-throttle is 1100 which with my ESC's and motors is a very light spin. I would love to test any changes you can suggest that may fix this.

thanks
Andrew


Glad to help!
I think mixTable() needs a broad rewrite so I think the issue worth a new thread and extra care, I'm trying to put together all the math required, with a multiple equation system to be solved, seek for caffeine and dig my brain for old, very old and dusty school topics :lol: :lol:
I'll open it in a while and edit this post with the link.

Sergio.

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Request for yaw test on r1539

Post by Alexinparis »

trello wrote:
scanman wrote:does anyone have any tips for me to get the yaw servo to work on this version? It is completely dead
Crius AIO Pro v2.

I did all the same settings on the earlier versions that worked.


I believe the motor pinout changed sometime between release 2.1 and 2.2 although i don't know where this is documented. I'm flying a tri with aio v2.0 and i think i had to use the pins from left to right (2, 3, 5, 6) while most diagrams show pins 9, 10, 11 and 3 for tri.

motor pinout never changed, it is documented here:
http://www.multiwii.com/connecting-elements

what changed is some servo outputs on mega.
by default for a tricopter on mega, it is no more 2 but 11 instead.

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Request for yaw test on r1539

Post by Alexinparis »

doppler wrote:Thanks Alex, I've added the line but haven't had a chance to test it yet, hopefully today. I will also play with P, you mentioned somewhere that D has no impact? I swore in my testing it did, but perhaps that was just placebo effect?

yes, yaw D has absolutely no impact. it is just not used.

On your comment about less accurate yaw stop or start, yes perhaps, I would say if you have your quad spinning at 120RPM that an accurate stop less important, just that you stop and don't jump up into the sky.. Other boards can do this (KK, KK2 are one's I've used in the past), I'm not sure how it's done, perhaps there is a ramp down from the speed, perhaps we need a Yaw Rate correction on P where low Yaw rates retain the P setting, and higher rates chip away at it...

On other boards you tested, would you say you can achieve an equivalent yaw rate without the "yaw jump" effect when you stop the yaw ?

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Request for yaw test on r1539

Post by Alexinparis »

SergioC wrote:the yaw jump problem is caused by not taking in account the MINTHROTTLE motor lower limit. When a sharp yaw correction is required one motor will spin up the other will slow. If the slow motor is to be commanded below MINTHROTTLE it will be instead tied to MINTHROTTLE but the opposite motor will be sill commanded a higher throttle even if the slow is saturated. Hard to explain with my not proper English :)

I'm not sure it is the cause of this effect.
When a motor is at MINTHROTTLE, its trust is nearly 0.

My interpretation: When a sharp yaw correction is required one motor will spin up the other will slow. <- here we are agree
But due to inertia, the motors that should slow down will reach the target speed after a longer delay than those that slow spin up. => the result is a sudden over trust
Someone noticed that with special ESCs than can dynamically brake, this effect is much reduced. Probably because the motors that should slow down reach the target speed much sooner.

not a simple problem to solve

SergioC
Posts: 19
Joined: Sat Dec 22, 2012 11:33 pm

Re: Request for yaw test on r1539

Post by SergioC »

Alexinparis wrote:
SergioC wrote:the yaw jump problem is caused by not taking in account the MINTHROTTLE motor lower limit. When a sharp yaw correction is required one motor will spin up the other will slow. If the slow motor is to be commanded below MINTHROTTLE it will be instead tied to MINTHROTTLE but the opposite motor will be sill commanded a higher throttle even if the slow is saturated. Hard to explain with my not proper English :)

I'm not sure it is the cause of this effect.
When a motor is at MINTHROTTLE, its trust is nearly 0.

My interpretation: When a sharp yaw correction is required one motor will spin up the other will slow. <- here we are agree
But due to inertia, the motors that should slow down will reach the target speed after a longer delay than those that slow spin up. => the result is a sudden over trust

I agree Alex, and this is more true as more overpowered the copter is. That was my first interpretation also (actually I red somewhere and agreed).

Anyway I think it worth to check how much a motor is driven below minthrottle during these sharp yaw corrections, something like find the minimum motor value and debug it. My copter is currently undergoing... ehm.... periodic servicing :oops: otherwise I would love to try myself.

Someone noticed that with special ESCs than can dynamically brake, this effect is much reduced. Probably because the motors that should slow down reach the target speed much sooner.

not a simple problem to solve

I see!
Maybe the Z acceleration could help, even more difficult. 8-)

Sergio

User avatar
Hamburger
Posts: 2578
Joined: Tue Mar 01, 2011 2:14 pm
Location: air
Contact:

Re: Request for yaw test on r1539

Post by Hamburger »

flew r1541 on a heli.120. driving yaw servo with hw.pwm @ 200Hz, yaw pid= 2.0 , 0.001 , 0 , yaw.rate =0.40 with gyro.smoothing=70, 70, 60.
I get some tail shakes (from overcompensation?). It is not a rock solid yaw yet. Higher yaw.P or yaw.I will result in pendulum oscillation of tail in stationary flight.
In all, it works ok.

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

Alexinparis wrote:
doppler wrote:Thanks Alex, I've added the line but haven't had a chance to test it yet, hopefully today. I will also play with P, you mentioned somewhere that D has no impact? I swore in my testing it did, but perhaps that was just placebo effect?

yes, yaw D has absolutely no impact. it is just not used.

On your comment about less accurate yaw stop or start, yes perhaps, I would say if you have your quad spinning at 120RPM that an accurate stop less important, just that you stop and don't jump up into the sky.. Other boards can do this (KK, KK2 are one's I've used in the past), I'm not sure how it's done, perhaps there is a ramp down from the speed, perhaps we need a Yaw Rate correction on P where low Yaw rates retain the P setting, and higher rates chip away at it...

On other boards you tested, would you say you can achieve an equivalent yaw rate without the "yaw jump" effect when you stop the yaw ?


It's been a while since I've used my KK2 boards so my memory isn't very fresh, but I certainly don't remember a Yaw Hop when Yawing. Did I get the rate we have on MWI right now, not sure. But I also don't think we need a quadcopter to spin like a top at 120 RPM either, fun to watch, not entirely practical for anything.

doppler
Posts: 64
Joined: Wed Sep 26, 2012 1:35 pm

Re: Request for yaw test on r1539

Post by doppler »

Alexinparis wrote:
SergioC wrote:the yaw jump problem is caused by not taking in account the MINTHROTTLE motor lower limit. When a sharp yaw correction is required one motor will spin up the other will slow. If the slow motor is to be commanded below MINTHROTTLE it will be instead tied to MINTHROTTLE but the opposite motor will be sill commanded a higher throttle even if the slow is saturated. Hard to explain with my not proper English :)

I'm not sure it is the cause of this effect.
When a motor is at MINTHROTTLE, its trust is nearly 0.

My interpretation: When a sharp yaw correction is required one motor will spin up the other will slow. <- here we are agree
But due to inertia, the motors that should slow down will reach the target speed after a longer delay than those that slow spin up. => the result is a sudden over trust
Someone noticed that with special ESCs than can dynamically brake, this effect is much reduced. Probably because the motors that should slow down reach the target speed much sooner.

not a simple problem to solve


I have noted in another thread that using SimonK recompiled with COMP_PWM turned on which actively brakes the motors to the desired rate improves this problem significantly. I have a couple of quads right now, the one I've been testing this code uses the bheli flashed Plush 30 ESC's with lightly damped setting. Heavy damping made them emit smoke. I've just replaced the ESC's with SimonK COMP_PWM flashed HobbyKing F30A ESC's. I haven't flown it yet. This will most likely resolve the problem for me, but flashing ESC's isn't always possible, it's hard for the basic user, and will limit the adoption of MWI. My "fun" quad is still running r1380 with a patched mixer table for yaw hold improvements, also with SimonK 12Amp ESC's.

I agree this is not a simple problem to solve, you almost need to know the speed of all the motors to and/or the ramp up and down curves... not sure how other boards software handles this...

Andrew
Andrew

wilco1967
Posts: 156
Joined: Thu Aug 18, 2011 6:04 pm
Location: Winterswijk, Netherlands

Re: Request for yaw test on r1539

Post by wilco1967 »

Not sure if this has been discussed, but perhaps the jump on yaw effect for a quad might be caused by non-linearity of the RPM to thrust characteristics ?

perhaps I'm totally wrong (most likely), but I was wondering if for example a 10% increase in speed (or better, 10% increase in demand signal), will give the same additional thrust, as a 10% speed decrease would cause the thrust to drop on the other motor ?

If this is not the case, a mixing table for yaw might help (might be there already, that I just did not realize it) :roll: :oops: .

Post Reply