V-Tail / A-Tail configuration for MultiWii

[joshuabardwell]

I'm thinking about converting my spider quad to a V-Tail or A-Tail to try to address some of the yaw problems that I'm having.

At least for MultiWii, it seems like A-tail should be preferable to V-tail, because with MultiWii's default motor directions, A-tail results in the motor torque working with you instead of against you. In other words on a flat quad, to yaw CW, you would spin up the back-left motor, which is spinning CCW, and you would spin down the back-right motor, which is spinning CW. Therefore, the back-left motor on a V or A-tail should point to the right, so that it pushes the quad CW when it spins up, both through thrust vector and through torque.

If the motors are tilted 20 degrees, that means that they are putting 22% of their thrust to the side, and 78% downward. Therefore, should the "pitch" parameter for the rear motors be increased by 22% to compensate?

I have seen people suggest that the roll parameter for the rear motors should be set to zero, because those motors are so close to the center-line. Then only the front motors participate in roll. Any thoughts on that?

Should the front motors have any yaw mixed into them, or is it better to let the angled (rear) motors do all of the yawing?

 

[Craftydan]

Um . . . you need to do your vector math again. Vector components do not add arithmetically to 100%.

With a 20 degree tilt, the horizontal component for a given thrust "T" is T*SIN(20) = T*0.34, and the vertical component is T*COS(20) = T*0.94.


The rear motors are intentionally close together so the roll impact in a strong yaw is liveable, but expect a rear boom to dip in strong yaw --having a yaw mixed in the front two motors can help this. If the rear motors are set to generate yaw and the left spins down and right spins up, roll is coupled in this. If the roll correction is mixed in the rear motors as well, then the roll loop will add thrust *BACK INTO* the left motor to keep the platform level, speeding it back up and yields an anemic yaw . . .

 

[joshuabardwell]

Um . . . you need to do your vector math again. Vector components do not add arithmetically to 100%.

Dang it. I knew I needed to do trig, but I figured 45 degrees is 50/50, so I let myself get lazy.

 

[joshuabardwell]

The rear motors are intentionally close together so the roll impact in a strong yaw is liveable, but expect a rear boom to dip in strong yaw --having a yaw mixed in the front two motors can help this. If the rear motors are set to generate yaw and the left spins down and right spins up, roll is coupled in this. If the roll correction is mixed in the rear motors as well, then the roll loop will add thrust *BACK INTO* the left motor to keep the platform level, speeding it back up and yields an anemic yaw . . .

It sounds like what you're saying is that you want yaw and pitch mixed into the rear motors, and yaw and roll mixed into the front motors. Is that right?

 

[joshuabardwell]

I took my first flight this afternoon, and I have to say I'm a convert. With 20 degrees of tilt, there isn't a huge amount of extra yaw authority, but there is a LOT less coupling of yaw with pitch and roll. Currently I have my mixes set up as follows:

Pitch - 100% in all four motors
Roll - 100% in front motors, 0% in rear
Yaw - 100% in rear motors, 25% in front

There is a little bit of coupling of pitch and yaw. The copter tends to tilt forwards a little bit when I yaw. Frankly, this is not too bad, since it means that the copter tends to "turn into" the yaw, which is usually want I want anyway.

I like the fact that, with the A-tail, you don't need to do anything fancy with the motor directions and mixing relative to yaw, since the same motor spins up as would if the motors were flat.

 

[Craftydan]

Yeah, Ironically I've been more impressed with the KK2's yaw authority for the A-tail than the Naze. Multiwii was fairly similar, but after dinking with the gains I was able to get quite a bit of the authority back. I haven't had that much luck with the Naze yet, but I haven't tuned far enough into it that I can say whether it'll come back or not. in every other respect the Naze has outperformed, except this.

The Pitch/Yaw coupling is about what you expect -- somewhere between a quad and a tri -- but so long as it's coordinated, it's a good balance between stable and organic flight, IMO.

 

[joshuabardwell]

Regarding yaw authority, my hope is that the reduced coupling will allow me to turn the yaw rate back up again. Right now, the yaw gain is at zero. I had it as high as 0.2 on my flat quad, but it was just too unstable and I had to turn it back down again.

BTW, does the philosophy of my motor mixes seem right to you?

 

[Craftydan]

Seems right enough to be a good place to start . . . and your experiment appears to show some promise.

I take it increasing yaw correction in the front booms increases your coupling?

You could play with adding a touch of roll back into the rear, but everything seems to point to the roll control loop fighting the yaw on the rear. Perhaps not what you want if you haven't yet tuned in more than enough the yaw authority to waste.

Perhaps weakening the roll gains might help, but then it won't snap to the roll angle as well -- my unproven theory to the kk2's better yaw performance is a weaker PI roll correction.

The downside of coupling -- an improvement in one can negatively affects the other.

 

[joshuabardwell]

I take it increasing yaw correction in the front booms increases your coupling?

Correct. With the front motors at 100% yaw, the performance is little better than with the flat configuration. Strong yaw causes the quad to twitch about. With the front motors at 0%, on the other hand, there is a "surge" from one of the motors following strong yaw (not sure which motor). I'm guessing this is the gyro compensating for a dip in one of the booms as a result of only one motor spinning up. Best results seem to be around 25%-35% (for this copter, in this configuration).

You could play with adding a touch of roll back into the rear, but everything seems to point to the roll control loop fighting the yaw on the rear. Perhaps not what you want if you haven't yet tuned in more than enough the yaw authority to waste.

Roll authority seems acceptable with only two motors, at least for the kind of flying I'm doing right now. If I was trying to be super-acro with loops and rolls, it'd be a different story. But with 10" props, this isn't exactly a super acro quad. I don't like the idea of wasting power spinning two motors in the opposite direction.