Shop OBEX P1 Docs P2 Docs Learn Events
Controlling Tricopter esc's - Page 2 — Parallax Forums

Controlling Tricopter esc's

2»

Comments

  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-05 14:22
    With the motors evenly spaced, a tri naturally has them 120 degrees apart, whereas a quad has them 90 degrees apart. Most sports cams have a field of view of 170 degrees, so you're more likely to have the arms in view on a quad than a tri. If you take a quad and make the positions of the motors "non square" you may have to compensate for it in your flight code. If you don't, it may "cross contaminate" the axis, for lack of a better way of putting it. Applying "roll" might introduce "pitch" and vice versa.

    A number of people are using H - shaped quads so they still get "square" motor placement, but can position the camera at the very front.

    Like this: http://www.rctodayshow.com/forum/viewtopic.php?f=7&t=645
  • Heater.Heater. Posts: 21,230
    edited 2013-06-05 14:41
    OK. That makes sense.

    But what about the fact that a quad has all motors and props the same (give or take push vs pull) , two rotating one way and two rotating the other so there is a balance in the yaw forces. With three props the balance is blown.

    Isn't that harder to design or control?

    The H layout appeals to me more.
  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-05 15:08
    Tri-copters uses a servo on the tail to control yaw. An example : http://rcexplorer.se/projects/2013/01/the-tricopter-v2-6hv-delux/

    You can also do six motors on three arms (three above, three below) and still use torque to control the yaw. (usually called a Y6)

    The tail servo adds a bit of weight, but I doubt it's much different to control. A decent servo would be more responsive than the motors - much more direct control and less inertia, so you probably have better yaw control.
  • Duane DegnDuane Degn Posts: 10,588
    edited 2013-06-05 16:14
    I saw a "tri" copter on YouTube a while back that had four motors. It had two up front like a normal quadcopter but the rear motors were on a single center boom in the back. It looked pretty cool and I was a bit amazed it could fly like that. I suppose it could be controlled a lot like a "+" quad with a bit of tweaking to the code.

    Edit: Cool tri-copter page. Thanks for the link Jason.
  • RambomanRamboman Posts: 101
    edited 2013-06-06 09:35
    To stabilize a Tricopter, you need no more than 4 gyro : one per branch and a fourth one for the yaw servo...
    To control it you need an heli programmable Tx...
    If you want more, you need at least a 4 channel Tx/Rx and a propeller (in our case)
    I am using a Jeti DC-16 with a Rsat2 that gives a PPM stream... one pin to decode instead of four.
    I found nowhere code to decode a PPM stream... any idea ?
  • AnubispodAnubispod Posts: 42
    edited 2013-06-06 10:13
    Ramboman wrote: »
    To stabilize a Tricopter, you need no more than 4 gyro : one per branch and a fourth one for the yaw servo...
    To control it you need an heli programmable Tx...
    If you want more, you need at least a 4 channel Tx/Rx and a propeller (in our case)
    I am using a Jeti DC-16 with a Rsat2 that gives a PPM stream... one pin to decode instead of four.
    I found nowhere code to decode a PPM stream... any idea ?

    4 Gyro ?? Mhh i fly mine with 3 but ok :)
    Its the same as a quad just that instead of the 4th esc a servo is with slightly different mixing table and pid settings.

    And the copter with the 2 motors on the back what are on a 45 degree angle are better in yaw respnse then a quad, but tricopter are the best in yaw and feel more natural to fly,
    specialy with FPV equip. They are not so banky to fly like a quad.
    [video=vimeo;52700855]

    And to come back to the ESC thing , yes of course on some Tri/Quad/... what have a bigger size you get away with lower esc rates.
    But as soon you have a smaller one specially quad you have to deal with way more turbulences, when you for instance descend.

    A tricopter has less of those , but as soon he gets small and you want to have good video from your gopro in HD you will face the situation where higher ESC rates make the big difference on some copters.
    And also for acro flying and in windy gusty weather higher rates improve.

    And the same goes for the tail servo on the tricopter yes a standard servo will work, depending what your goals are to get in the air and cruse around good, precise acro flight not good, there would be a digital servo better means higher rates



    MfG
    Oliver
  • Duane DegnDuane Degn Posts: 10,588
    edited 2013-06-06 10:34
    To stabilize a Tricopter, you need no more than 4 gyro[/quote]

    I thought there were only three spacial dimensions? I can't imagine why one would ever need more than a three axis gyro?
    Ramboman wrote: »
    I found nowhere code to decode a PPM stream... any idea ?

    There are a couple mentioned (and attached) to this thread.

    Spektrum satellite receivers output a serial data stream that can also be read with just one I/O pin.
  • Heater.Heater. Posts: 21,230
    edited 2013-06-06 11:25
    That's some interesting designs coming out here.

    I have heard it said that people used to say that: According to the laws of physics it is not possible for a Bumble Bee to fly. How true that is I don't know but I'm having the same feeling now about some statements made in this thread.

    1) I still don't see how a 500Hz pulse rate on your ESC can possibly help unless the whole control loop can work at that speed. Are the gyros also read at 500Hz? Can the attitude of the machine possibly respond significantly in that frequency range, even if it is a small copter?

    2) As far as I can see we live in a three dimensional world. A solid object can only rotate around three orthogonal axis. What is the fourth axis that a tricopter needs another gyro for? A Tardis, for example can rotate around the time axis but I'm not sure how a rate gyro handles that and it must be a bit weird to watch:)
  • W9GFOW9GFO Posts: 4,010
    edited 2013-06-06 11:58
    Heater. wrote: »
    That's some interesting designs coming out here.

    I have heard it said that people used to say that: According to the laws of physics it is not possible for a Bumble Bee to fly.

    Right. If you apply the same equations to the Bumble Bee as you would to a fixed wing aircraft then no, it doesn't work. Same with a helicopter. If you calculate the lift of the rotor as if it were an airplane wing then you would have to conclude that helicopters can't fly either.

    Those types of sayings are a minor pet peeve of mine.

    I am skeptical about a tricopter yawing "better" than a quad. I think all things being equal, motors, props, mass etc., a quad should yaw equally well as any other multi rotor - I suspect even slightly quicker since it does not have to wait for the thrust to be vectored. I do believe that that nothing will yaw as quickly as a conventional helicopter with a tail rotor.
  • Duane DegnDuane Degn Posts: 10,588
    edited 2013-06-06 12:55
    Heater. wrote: »
    I have heard it said that people used to say that: According to the laws of physics it is not possible for a Bumble Bee to fly.

    People have said a lot of things. People have said I'm the best looking person on the planet. (Okay only one person has said it, but when my wife gets home, I'll ask her to say it and then "people" would have said it.)

    What the people should have said is "According to someones limited knowledge of the laws of physics, they computed a bumblebee shouldn't be able to fly."

    The actual laws of physics (which I also have a limited knowledge of) don't have any problem with a bumblebee's flight ability.
    W9GFO wrote: »
    Those types of sayings are a minor pet peeve of mine.

    They're more of a domesticated peeve rather than pet peeve of mine.

    Anyone who has flow helicopters with a tail rotor has probably noticed an asymmetric yaw ability. One of the things I like about quadcopters is the symmetrical way it yaws (and flies sideways). I'd imagine a tri-copter would have asymmetric yaw and sideways movement traits. Not that asymmetric traits are big problem in many applications, but I think there are situations where the symmetric traits of a quadcopter would come in handy.
  • Heater.Heater. Posts: 21,230
    edited 2013-06-06 13:15
    It' seems you have to be a bit more tolerant of you pet, or domesticated, peeve about the bumble bee myth.
    Looks like people did actually conclude, from the state of knowlege at the time, that bees should not be able to fly. That story is no myth.
    It's a bit wrong because the problem was not a hole in the physics but a hole in the understanding of how bees and even birds can generate lift. It was the biologists who had not studied the problem hard enough yet. It was not "According to someones limited knowledge of the laws of physics..." but rather "According to someones limited knowledge of biology, i.e. insect flight....".
    Turns out this may not have been understood at all well until as late as the 1970's

    Here is a nice summary of the history of that "myth": https://www.wolfson.ox.ac.uk/~ben/zetie1.htm
  • Heater.Heater. Posts: 21,230
    edited 2013-06-06 13:28
    Whilst we are on the subject of Bees. Those little buggers beat their wings about 200 times a second. Which is about the response frequency you would get if your copters control loop could work at a 500Hz rate. As suggested by those who want 500Hz ESC drive. If you have a copter that can do that I would expect to hear it humming like a bee. Apart from the normal motor whine of course. I look forward to hearing this on some YouTube video of copter flight.
  • teganburnsteganburns Posts: 134
    edited 2013-06-06 18:08
    back to the subject of gyros, i had orded this one (https://www.sparkfun.com/datasheets/Sensors/Magneto/Tilt Compensated Compass.pdf) wondering if it would be sufficient for my tri-copter? probably run it at fast mode 400Hz.

    other links : (https://www.sparkfun.com/products/10703)

    and while I'm at it can any one shoot me to a link that can help me understand I2C interface /w the propeller? (looking at the C code in the pdf it should be really simple)
  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-06 18:55
    That's not a gyro - It's a compass, with an accelerometer, so it's made for low-speed applications where you want a heading, not high-speed rotation. Check out:

    ITG3200 : https://www.sparkfun.com/products/9801 (just the gyro)
    MPU6050 : https://www.sparkfun.com/products/11028 (gyro + accelerometer)
    MPU-9150 : https://www.sparkfun.com/products/11486 (gyro + accel + compass)

    I have a pair of 9150's at home waiting for me to figure out. Once I do, I expect to have a version of the QuadX code that'll use it.
  • Duane DegnDuane Degn Posts: 10,588
    edited 2013-06-06 19:17
    I'd strongly suggest using one of sensors Jason is using so you can can adapt his code for your needs.

    I posted a link to an inexpensive ITG3205 board in Jason's thread. From what I've read, it looks like the ITG3205 should work the same way the ITG3200 does.

    The ITG3200(and 3205) use I2C for communication. Jason's code does the I2C communication from within the PASM section of the gyro driver. The program also uses I2C to communicate with the EEPROM using Spin.
  • RambomanRamboman Posts: 101
    edited 2013-06-06 21:52
    To keep it very simple, you have one gyro per branch controlling directly the esc of that branch... the last control directly the servo.
    The 3 esc and the servo are controlled in heli mode.
    You don't need any add'l electronics... for one add'l gyro ;)
  • Heater.Heater. Posts: 21,230
    edited 2013-06-07 03:09
    Ramboman,
    To keep it very simple, you have one gyro per branch controlling directly the esc of that branch... the last control directly the servo.
    This gets interesting. The complex answer might be required...

    From what you say I could surmise:

    1) In the extreme I could have a regular quad copter with four 3 axis gyros, one on each leg.

    2) Each gyro directly controls the ESC on it's leg.

    3) Clearly each gyro can tell when it's leg rotates about the vertical axis, yaw of the craft, and can change it's motors speed to compensate. Motors on opposite leg's being controlled in the opposite sense.

    4) Each gyro can tell if it's leg is trying to "point up or point down". This could be pitch or roll of the craft or a combination of both depending on if it's an + or x configuration machine. Hence each leg's gyro can affect control of it's own motor to stabilize the crafts pitch and or roll.

    5) That leaves one gyro axis unused, pointing along the leg direction. If the leg rotates about that there is nothing the attached motor can do about it.

    In this way we could stabilize the craft with four independent gyro/ESC circuits. Kind of cool I guess.

    At first sight this seems wasteful as we have four times as many 3D gyros as we actually need. Gyros on opposite leg's read the same yaw and "tilt" but in opposite directions, so one is redundant. Then we have one axis of each gyro not used at all.

    But these things are small and cheap so why not?

    How do we then mix actual control inputs into these four circuits?

    Does that sound like the plan you have? I'm not sure why you mentioned it in respect of 3 legged copters, seems to be the same with four to me.

    Perhaps I need some high level diagram of how this is all put together.
  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-07 03:29
    No, he's talking about using off-the-shelf single-axis gyros, commonly used on model helicopters to control the tail. Mounting a single gyro to each arm, such that the gyro is sensing vertical rotation around that arm, is enough to stabilize a tri. You use a transmitter with CCPM mixing. Each gyro feeds directly into an ESC. The 4th gyro is set up "normally" and controls the yaw servo.

    Building it this way means you don't need a control board - no centralized micro controller making decisions. I can't decide if I think that's good or bad. It certainly means it'll never auto-level, but it's still quite flyable.

    J
  • RambomanRamboman Posts: 101
    edited 2013-06-07 03:38
    Correct ;)
    You have a stable base...
    If you want to improve : up to you...
    Example: insert a propeller between Rx and gyros,
    sense altitude with a ping
    modify signals accordingly...
    (I am not so far)

    PS: with a propeller, it's very easy to read Rx signals, sense captors, modify signals, send them to gyros...
    I did it with a wheel robot...
  • Heater.Heater. Posts: 21,230
    edited 2013-06-07 04:38
    Jason,
    Mounting a single gyro to each arm, such that the gyro is sensing vertical rotation around that arm,
    Just to be sure, all these axis give me headache: We have a propeller on a horizontal stick. We have a single axis gyro on the stick. That gyro is oriented so as to detect one end of the stick trying to get higher or lower than the other.

    Boom we have a stick stabilized, trying to remain horizontal. Cool.

    With four such sticks in a quad I would have a fifth gyro responding to yaw, but now it has no motor to control unless I add a fifth "tail rotor". Hence the thing about using this setup on the tri copter.

    I'm fascinated by the idea that the motor + propeller is a gyro. Consider as the end of the leg with motor on it tries to rise or fall with respect the centre of the machine that rotating motor+prop is applying gyroscoic forces to it's mounting. If we could measure that we have the gyro we need.
  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-07 15:02
    While that could potentially be made to work, I'm not sure if it'd be sensitive enough to keep stable flight. The props will never be perfectly balanced, and are affected by things like small pressure variations in the air, the presence of the booms, etc. The force readings you got would have to figure out how to reject all that extraneous noise, which wouldn't be easy. I've never tried it though, so who knows? Perhaps it's possible.
  • teganburnsteganburns Posts: 134
    edited 2013-06-07 19:43
    I could see this being done to reduce manufacturing costs...


    And i ended up getting the (https://www.sparkfun.com/products/11486) MPU-9150

    I looked for the ITG3205 and found it on the international page for hobbyking but that takes for ever to get to me...

    So when i get the chip i will mess with the Main Hoverflyopen.spin and see how that goes.
  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-07 21:54
    I'll try to get my driver up and running for it soonish. I'm anxious to try that one too.
  • teganburnsteganburns Posts: 134
    edited 2013-06-11 19:33
    I had a little trouble with shipping and sparkfun but i am expecting it to arrive tomorrow or Thursday.

    But out of curiosity I was thinking about the zero g detection function on some of the other modules i saw (i would think this would be easily replicated with the prop chip)....?

    So my idea was to hold the tri or quad out with your arm and then drop the craft. Before it hits the ground it initializes and recovers / stabilizes.

    This would be an impressive way to start an autonomous demo or ect.

    Possible? Flaws?
  • JasonDorieJasonDorie Posts: 1,930
    edited 2013-06-12 09:37
    I wouldn't want to be standing beside it. :) Without GPS or optical flow, there's no way to get position hold, so while you can make sure the craft is level, you can't keep it in one place. The drop recovery may be possible though.
  • teganburnsteganburns Posts: 134
    edited 2013-06-12 16:10
    JasonDorie wrote: »
    The drop recovery may be possible though.

    HMMMMM :D

    It came in the mail so I will up up really late tonight...
    1024 x 1365 - 149K
Sign In or Register to comment.