Why does arms has to be perfectly ninety deg?

Arthur P. Meiners December 1, 2010 at 10:58am

If the arms are (accidentally) bent, then the may also be twisted which means your motor thrust will not be pointing straight down. In addition they may be bent not only in the horizontal but also in the vertical plain. The latter wil again result in a thrust vector which is not vertical. A non-verticual thrust vector will cause either roll, tilt, or yaw effects depending in which direction is most affected. If minor the software may be able to compensate.

There are quadrokopters which do not have the arms at 90 degrees to each other. As long as the diagonal axes of the motors go through the centerpoint of the craft, this is possible. In fact having a wider distance between both front motors and both rear motors than from front to rear motors provides more room for a camera to look up between the props. For such layout you can use either an X-type frame with a larger than 90 degree angle between the front arms (and thus smaller than 90 degrees between front and rear arms), or an H frame with a longer horizontal in the H than the two verticals. You will probably have to have different PID values for roll vs tilt axes. In an ideal world you might have a mixer table which allows the different strength of thrust on the roll vs tilt axis to be entered (with the motors further away to the side the same thrust level on two side motors produces a stronger roll efffect than that the nick effect of the same thrust level on the two front motors).

I don't know whether this has been tried with the ArduCopter, but it does work with various other flight controlers. The "H" frame approach has also been used for octocopters to get the motors out of the camera image.

Developer

Jani Hirvinen October 20, 2010 at 3:45am

To have best and most accurate calculations for all thrust vectors, arms should be in 90° angles but it does not mean that they really need to be.

Like Chris said that he has been flying with bend arms so have I, I have even bend my arms for testing purposes and at least ArduCopter can fly still nice even arms are bent. Ok it is most efficient way, no but it flies.

And on Hexa's and other bigger VTOL's arms are not any more 90° but still same thing, they follow certain pattern. ArduCopter and other similar VTOL machines know nothing about aerodynamics, it's pure mathematics that keeps them flying. And these mathematic algorithms are already rather complex so we just try to keep them as simple as possible and also CPU friendly.

Craig Burden October 20, 2010 at 2:12am

mine deviate by probibly 1dec. should be fine, ill tell you what mines like in a few weeks when my APM arrives!!

Ed Stewart October 20, 2010 at 1:49am

I can imagine it flying with possibly reduced performance with different sized propellers, the smartware would adjust to balance the thrust output. And I can imagine it would still fly fairly well if opposing arms are both bent down (or up) at approximatly the same angles (possibly after a hard landing) as the opposing thrust vectors would cancel each other out. I'd really be impressed if it could compensate for a single arm being bent or twisted in any direction beyond a very small diformity. But then I'm easily impressed ;-)

Sebastian Gralla October 19, 2010 at 1:58pm

it should be 90deg but software can correct this thing
you can even fly with different propellers ;)

3D Robotics

Chris Anderson October 19, 2010 at 12:03pm

I've flown with bent arms (bent down, not sideways) with no problem. The software handles it fine.

Ed Stewart October 19, 2010 at 11:42am

Just my 2 cents worth but imagine in your minds eye (or on a sheet of paper), the thrust vectors if the motors were not perfectly aligned with each other. The motor on the bent arm is going to be constantly pushing the Quad around (or over). I envision the electronics trying to level it back out but then it get's this weird thrust vector from the motor pointing in a non perpendicular direction that pushes the quad again. Some nasty osilations would occur that you might be able to hold steady in manual mode by adding in corrective yaw, pitch or roll inputs but the quad will not be level. In some auto mode the electronics arn't going to be smart enough to know that the thrust vector of one or more of it's motors has changed and the fight will be on.

Not having finished my quad yet, someone please correct me if I'm wrong.

This brings up a question that I have though. How well does the Quad hold position in a steady wind? Will it lean into it and maintain a 0 ground speed?

Discussions

Why does arms has to be perfectly ninety deg?

Replies

Activity