So..... I guess you could consider me a bit of a nerd gone corporate sell out. I'm all for open source, home grown, new idea, etc..... but in the business world, it almost always means a loss of high availability and redundancy. So my brain is just wired that way!
What does mean for my ArduCopter? Motor redundancy!!!
I'm not looking for much wieght, don't need to add much power..... the camera and gear I'm lifting isn't very heavy at all. Its a Canon 5D Mark II and some really wide open lenses. Total weight is actually pretty stable on my quad with 4 880 motors and 12" props.
Here's the issue.... between the quad, camera, lens and other small gear.... that's like $7,000 floating around and a single motor failure sends some pretty nice gear into peices.......
So that's the question: What's the best way to build a stable quad that can sustain at least 1, if not 2 motor failures during flight and still be able to get back and land without flipping and hitting the deck hard??
The Y6 looks promising, but I wonder about loosing load capacity. I've heard rumors about an "X8" with the same style inverted motors at the Y6, and of course there's the normal Hexa and Octo configurations. Anyone like one over the other? or have an advice?
Replies
If you want redundancy, Y6/Hexa is not recommended. If you lose a motor you will lose considerable lift. I have built quads, coax quads, hexas and y6s. The only configuration that still actually flies very well in a motor-out scenario is the coax quad. I am sure an octo would also fly well, but I haven't built one (yet). If you are a 3D heli pilot then you can handle a motor out on a hex as it will be spinning. I really like the y6 for orientation purposes and for aesthetics, but if you insist on lifting gear with it, you will have to build it to fly, not build it to fail.
As an aside, if you are lifting a 5D with a quad, you are very likely underpowered. You should be able to pick up your payload at just over 1/2 throttle and I doubt that is the case. You are most likely at 3/4 throttle lifting a 5D, so you do not have enough reserve power. The 5D is a heavy camera, even with primes.
For those in doubt, let's update this topic a bit.
I had a flat hexa and spent lot of time for tuning PID for failsafe redundancy (disconnecting one motor while running for tests). I could never make it as stable and reactive as I wished but I thought it was enough.
The problem highlighted on the bench was that stable PIDs were too slow for recoverying a sudden motor failure so that when the motor was disconnected, the copter started flipping and slowly recovering (depending on how good PIDs were).
Some weeks ago it just flipped for unknown reasons and died forever, don't know if it was a motor failure or what but it's gone.
Now I'm building a new one and I'm thinking about Y6 (which is better for filming)... BUT...
I made some simple static physics calculations and turns out that Y6, hexa and coaxial quad (8 motors) are the same as for pitch/roll momentum imposed by the remaining motors (octa is obviously better).
Coaxial builds (Y6 or coaxial quad) suffer from yaw compensation, though. I'll explain better... in any case, if a motor fails, the copter will fall on one side and start spinning on the yaw axis as it's missing a prop. The APM will compensate both but here comes the difference:
- In a flat solution, compensating pitch/roll and yaw act together in incrementing power to the damaged side.
- In a coax solution compensating pitch/roll will increase the damaged side while depowering the opposite motors while compensating yaw will reduce the damaged side (the coaxial motor to the dead one) acting opposite to pitch/roll compensation.
This behaviour might slow down the recovery and give enough time for the copter to flip.
Here's a picture. Red means CCW prop, green means CW prop, yellow means two props (one CW and one CCW).
The good point for coaxial is that losing a prop in such a setup will probably not split the lift by half because of the lower efficiency of coaxial props. It would be let's say losing 0.9 props :P not sure about this, though. Does anyone knows or can do tests on lift of a single prop and coaxial setup with same motors and propellers? I wish to do it but I'm waiting for new parts...
...So, if you're looking for redundancy and safety, use flat setups and you might get some help by increasing the yaw PIDs also (might speed up the recovery without making pitch/roll PIDs too high).
Now... I'm stupid and I really want to build a Y6 'cos it will be smaller and with more space for the camera and wider clear angle so, my ideas should be right but I don't know how big of a problem it is. Does anyone knows if yaw compensation is strong enough to slow down pitch/roll compensation in case of motor failure? I know it's a matter of PID tuning but with usual PIDs is it such a problem?
Also I noticed on the bench that when my hexa started flipping it will never stop but just gain more and more speed. I can live with a flip or two in case of motor failure but it would be nice if the copter would stabilize after a while instead of flipping faster and faster.
I hope I'm helping someone.
Y6 seems to work. I found this video of a motor failure
http://www.rcgroups.com/forums/showthread.php?t=1400855
and even a better example a few days ago but I can't find it for the moment.
Is the X8 allready available through the code?
The problem here is the high disk loading (weight of the craft per area of propeller) so that when one thrust unit fails, it is so far out geometrically from the center of mass that creates quite a torque moment in the pitch or roll directions. Using longer props and turning them more slowly would help this. It would of course be a compromise in maneuverability, but if a stable camera platform is your goal, it's worth pursuing.