The Picture above is of my FlameWheel 450 with one added propeller motor Unit. It is for testing the feasibility and potential value of providing 4 opposed motor propeller units for providing position control instead of copter pitch and tilt. For testing purposes the horizontal motor / ESC is controlled manually from a separate receiver channel controlled by an analog "dial" on the transmitter.
Oliver Seeler and I are both pursuing the possibility of providing a more stable multicopter platform by separating the stability control of the copter from tilt and pitch based position control. Oliver is using a Hex Flamewheel with a setup very similar to the above at the moment.
This Blog is an outgrowth of a Forum item started by Oliver under "Aircraft platforms" entitled "The Witch Gets a Broom".
First test flights are complete on both of these copters and some significant data has been learned.
On Oliver's Hex considerable thrust was required to move the copter horizontally even at a slow pace. On my copter a very small horizontal thrust (< 1/3 throttle) moved the copter smoothly at a slow walking pace and my Quad's motor thrust is tiny in comparison with Oliver's Hex.
Oliver's Hexacopter's center of gravity is considerably lower than mine due to tall landing gear, a camera mount and GoPro camera but his horizontal motor and prop unit are mounted in the same location as mine. This means that the horizontal props center of thrust is considerably further above Olvier's Hexacopters CG than is true for my Quad.
From this result we have concluded that it is very important to have the horizontal thrust line aligned closely to the copters center of gravity. Failure to do so induces tilt which the stability mode of the APM counters by applying opposing vertical thrust, doing a surprisingly good job of cancelling out the horizontal thrust. My copter as seen above does not have this problem and requires very little power to smoothly produce horizontal flight.
Also, Oliver's Hex has the prop mounted in a pusher mode and on my Quad it is mounted in tractor (puller) mode. It appears either mode can be made to work satisfactorily. I was concerned that mine in tractor mode would have the horizontal prop wash interfering with the copters downward prop wash and might induce instability. At the power levels tried so far, no negative effect is observable and transition to and from horizontal flight is very smooth and without tilt.
My conclusion from the results achieved so far is that using a horizontal propeller for position control actually does make more stable flight possible with little or no pitch or roll of the airframe. I will continue to test in this configuration at higher power settings to insure stability is not significantly degraded.
But I am also beginning work on a setup that will place prop motor units between each of a Quads 4 vertical thrust rotors.
The intention of this configuration 4 x 4 is that the 4 horizontal motors will completely take over horizontal position control.
The existing APM programming and the Vertical thrust motors will be responsible for stabilization, altitude control and the yaw or horizontal orientation of the air frame. This way only balanced symmetrical vertical thrust on the air frame is required to adjust altitude and yaw and the only asymmetrical thrust is in direct response to outside forces trying to pitch or tilt the air frame. Basically it's entire vertical thrust resources are used keeping it level and there is no purposely induced tilt or pitch as is conventionally necessary for position control.
Control for the 4 x 4 unit will necessarily be more involved as it will be desirable at a minimum to produce differential control of all 4 props using a stick. This means that zero thrust of all 4 motors would be at "stick center" position and pushing in any direction would produce differential activation on one or 2 motors in that direction producing a horizontal displacement similar to what is done now with a quad copter using the APM.
This is not achievable from a transmitter directly and requires at least translation of stick center to = zero throttle on all four motors.
My initial solution to this will to be to either use an auxiliary (in addition to the APM) Arduino processor to take normal stick input servo data and translate it to the controls for the 4 motors or I may attempt to simply incorporate the control into the APM if my programming capabilities and available resources prove adequate to the task.
The hardware part is easy, the programming part, less so.
Oliver and I very much welcome and would like to solicit your comments, thoughts and contributions on this project, and if somebody with more APM programming experience than I would like to jump in there, please feel free.
Keep in mind, the silly thing in the picture actually works and there is no perceptible air frame tilt or pitch when you use the horizontal motor to move it.
This is a new kind of aircraft which so far appears to be feasible and also that it might have some really significant advantages and capabilities.
Current Design Above.
Comments
Randy, thank you very much for the informative and useful input.
I will definitely look into the Simon K ESCs, I understand they are fast response in any case. Fast reverse on a brushed DC motor is easy, just reverse the polarity and apply as much current as you can without exceeding the motors or ESCs ratings. On brushless it is more complicated and even sensing speed it ispossible to apply current in a way that reinforces it continuing to go in the same direction (can't tell the difference in actual motor direction). Generally on a brushless setup you need to sense direction in addition to speed for instant reverse and I am not sure that the ESC's / Motor hall devices on our RC stuff are adequate for that. It is also a complex algorithm. Still you could have forced braking, sense stop and then apply maximum acceleration in the other direction and that would be a lot faster than what I have seen so far in cars and boats.
I have already decided to go straight to APM control of the 4 motors, since it is really the only hope of achieving integrated control and I was thinking that APM output channels 5 - 8 would be the logical ones to use since they are the ones that would be used by an Octcopter and so should be compatible with PWM.
I greatly appreciate and will definitely use the information in your last paragraph which zeros in on the functions and modifications that I need to address. nav_roll and nav_pitch seem like exactly where I need to start, this info truly saves me hours of wandering around in the code.
One of the toughest things in dealing with the APM is it's enormous sprawling software and correctly ferreting out where the actual critical thing is that needs to be changed in order to accomplish a given result.
I have been updating the Arducopter Wiki periodically and one thing I would like to do when I am more familiar with the APM software is to produce a comprehensive flow chart for it. (Right now way too daunting).
Thank you again, hope you don't mind if I you ask a question or 2 as I try to get this to work, I will try to keep it to a minimum.
Gary,
Curiously, the wiki about contra-rotating propellors mentions improved efficiency and increased noise. In practice the sound is different but the noise level is certainly not excessive, and while I'm pretty sure that the bottom props operating in 'second-hand' air do not help efficiency, there seems to be no problems at all due to interference or turbulence.
As to the decoupling, you will also need it if you want to take advantage of the additional 2 DOF. It should not be too difficult, certainly not compared to the code that performs the actual stabilization. Randy provides some pointers above. I intend to start with stabilize mode (case ROLL_PITCH_STABLE) first, though. The part of control_in below a threshold will be sent to two virtual output channels, the part above to roll/pitch, starting with a low threshold to avoid unforseen weirdness.
For my setup, I've derived a class from AP_MotorsMatrix that takes two more virtual rc channels as input and mixes them into the 8 motors.
Gary,
I think the simon k firmware has a car mode that allows you to reverse the thrust. I haven't actually tried it though and it may suffere from the same issues as you're seeing with other ESCs. Certainly an instantly reversing ESC is important for when we get to handling a motor failure on a hexacopter.
Getting arducopter to use the additional motors for position control is a bit tricky. It's possible of course though...what you need to do is divert the nav_roll and nav_pitch variables and make them output to channels 5 ~ 8 which you would attach to your extra motors.
The exact point where the nav_roll and nav_pitch are sent to the roll/pitch stabilize controllers is just below "case ROLL_PITCH_AUTO:" in the update_rollpitch_mode function here. You could write a function which takes that nav_roll and nav_pitch (which are numbers between -4500 and 4500) and translate it into a pwm and send it to the motors using something like this:
APM_RC.OutputCh(CH_5, 1500);
If anybody does know of a lightweight (<20 grams) reversible brushless ESC of at least 10 amps please post it here. There are boat and car/truck ones, but the ones I have found are all too large and incorporate heavy heat sinks.
Another issue that I have seen even with the car and boat ones is that they do not instant reverse, rather they wait till the motor comes to a stop in the initial direction then there is a short latency then they start up in the opposite direction. For copter use, instant reverse would be an important feature. For brushless motors, instant reverse can certainly be accomplished, but the ESC's / Motors we use in our hobby may not have enough sensors or smarts to actually determine motor direction and might therefore be incompatible with an instant reverse feature.
One thing mentioned in Jake's video is that this design is being used with a HobbyKing board that only incorporates gyros for keeping the platform level but does not have accelerometers which could act to counter the applied external thrust.
This is certainly an issue with my APM based system but at the slow acceleration rates I have been using so far it does not appear to be a problem.
At faster or higher performance levels it may be necessary to interact with the APM's programming to counter its accelerometer based stabilization response.
Thank You Jake, Great Video
Response to Flying Monkey I must say I am aware of the feathering and even reversible variable pitch props for indoor models and the motor sizes I am using are probably compatible with them, but I am not really looking for a fast platform nearly so much as a stable video platform and at low speeds, using a motor that has enough intrinsic drag to prevent the prop from turning when not powered should keep drag low enough to not be a significant problem. A very interesting idea though and the reversing feature could even provide a 2 horizontal motor solution (you'd have to add 2 servos of course to control the props and trying to figure out how to mount them in line with the copters CG could also be a problem, and they do not have much of a reputation for reliability. Even with those problems though it is worth considering).
Hi Jesse, In fact I was concerned about induced torque roll in the frame, but actual experience has so far demonstrated that is not a problem at all. In fact the stabilization mode of the APM completely counters and cancels any torque roll and the compass based directional maintenance cancels any induced yaw.
You bring up an interesting point about loss of lift with the copters vertical blades moving sideways through the air stream, of course the APM in alt hold compensates automatically, but the fact is that at low speeds the effect is pretty negligible. In a stable hover using my horizontal motor to move the Quad sideways does not affect altitude at all. At higher speeds I am sure this would be a more noticeable effect.
As for the far less weight issue, I am planning on all motors, props, secondary subframe and ESC's weighing less than 8 ounces. (These are tiny in comparison with the primary drive system). And the problem remains, no matter how good your camera platform stabilization it cant counter side to side or vertical excursions. With a bit of refinement this system can.
I very much like the idea of being able to reverse the motors and if practical reversible ESCs were available for aerial use (the car ones are not) I would seriously look into using them. If you know of any appropriate lightweight reversible ESC's please let me know. Although not commonly available a symmetrical prop that works equally well in either direction can be made that gives up very little in efficiency.
Actually one of my first thoughts was a motor on a pivot using a servo to turn it to the appropriate direction. The problem is latency. If you are responding to variable gusting winds you need near instantaneous response in any direction to effectively counter them. Just cant get there fast enough. Also, I have discovered the need for aligning the horizontal prop thrust center line with the copter's CG, not easy with that solution.
No problem. It's not my video. But I thought it might help you.
Hi All, First let me thank you for all the great input.
Second, let me profusely apologize to Jake Wells and ask him to please re-install his excellent video of his pusher Quad.
I completely accidentally deleted it. I didn't even know I could, sorry. There will be no more deletions (by me anyway).
That is a really great video and certainly illustrates both that side thrust can work and also seems to illustrate that tilt is induced if the horizontal prop thrust center line isn't exactly aligned with the CG of the copter.
Next Patrick, wow what an interesting approach to the same dilemma. A few questions, do the non-aligned coaxial prop wash's tend to produce interference with each other possibly turbulence or noise at some speeds. It seems to me that although this is mechanically straight forward it is going to be a really serious undertaking in software to effectively do the decoupling you refer to. Still I do see that the always loaded and dynamic nature of your solution could permit faster and better integrated response to keeping the copter level than the 4 horizontal prop solution I have described which has a potential dead or slow response spot in the middle. Mine could be made fully dynamic also by running all 4 props at a low speed which would burn extra power of course, but on yours, extra power is consumed because of the intrinsic losses relating to the coaxial misalignment as well. I totally agree that your design merits a Blog of it's own. Please keep us posted on your progress.
I had the same idea some time ago, and ultimately decided on an X8 coaxial configuration where the motors are all tilted 15 degs outwards. Another member here (Aerhead) has tried something similar before using a dual quad config.
It already flies extremely stable with a stock X8 ArduCopter build. Efficiency obviously takes a due to the coax setup, tilt angle and 3 blade props, requiring 350W for hover with a 2800g AUW, but efficiency has never been a goal, experimentation is.
I'm currently working out the integration in ArduCopter. The goal is to add two linear degrees of freedom (x and y) to the current 4 degrees of freedom (rx, ry, rz and z) and decouple them so that the stabilization acts on the rotationals while positioning uses the linears in low speed conditions with a smooth transition to the classical control in the higher speed range.
Unfortunately a high professional workload and bad weather are conspiring against quick progress.
This project merits a separate blog post as there are some innovations in the design of the frame. Someday.
-
1
-
2
-
3
-
4
of 4 Next