The Tri Rotor Drone: Why Has It Been Overlooked?

A DJI Phantom 3. Zimin.V.G. [CC BY-SA 4.0]
If you are a watcher of the world of drones, or multirotors, you may have a fixed idea of what one of these aircraft looks like in your mind. There will be a central pod containing batteries and avionics, with a set of arms radiating from it, each of which will have a motor and a propeller on its end. You are almost certainly picturing a four-rotor design, such as the extremely popular DJI Phantom series of craft.

Of course, four-rotor designs are just one of many possible configurations of a multirotor. You will commonly see octocopters, but sometimes we’ve brought you craft that really put the “multi” in “multirotor”. If the computer can physically control a given even number of motors, within reason, it can be flown.

There is one type of multirotor you don’t see very often though, the trirotor. Three propellers on a drone is a rare sight, and it’s something we find surprising because it’s a configuration that can have some surprising benefits. To think about why, it’s worth taking a look at some of the characteristics of a three-rotor machine’s flight.

A Chinook helicopter in service. UNC - CFC - USFK [CC BY 2.0].
A Chinook helicopter in service. UNC – CFC – USFK [CC BY 2.0].
If you think for a moment about a typical small helicopter, it will have a tail rotor. The tail rotor is there to provide a sideways force to counteract the natural tendency for the fuselage to spin in reaction to the rotation of the main rotor. With a twin-rotor helicopter such as the famous Chinook military aircraft, the tail rotor is superfluous because the tendency to rotate imparted by the front rotor is counteracted by an equal but opposite force from the contra-rotating rear rotor. If you take the counteracting forces in a Chinook as analogous to those on a multirotor, you will then understand that equal numbers of contra-rotating propellers cancel any tendency for the craft to rotate, and allow the craft to be flown .

Now imagine the same interplay of forces in a trirotor, and you will instantly appreciate that it has an odd number of propellers, and thus it will have an excess of rotational force that is not counteracted by another motor. A simple three-rotor design will therefore naturally spin in flight, and be next-to-useless as an aircraft without some means of counteracting that rotation. In the simplest of flyable trirotor designs one of the rotors is simply mounted at an angle to vector some of the force sideways in an analogous manner to the tail rotor on a small helicopter. This has the desired effect of creating a flyable machine but offers no advantages, and a few disadvantages, over a four-rotor design.

The layout of a trirotor, showing the yaw servo axis in green. Toglefritz [CC BY-NC-SA]
The layout of a trirotor, showing the yaw servo axis in green. Toglefritz [CC BY-NC-SA]
Where three-rotor craft come into their own though is when instead of being mounted at a fixed angle, one propeller is mounted at a variable angle. When the force counteracting the rotation is under variable control, this ability for the rear propeller to yaw gives the craft some of the characteristics a fixed wing aircraft gains from its tailplane. The result is that tri-rotor craft can deliver a more stable and more agile flight than their multi-rotor cousins, at the expense of an extra servo with its control circuitry and software. The additional benefits of the layout are that with fewer drive motors it can be more energy-efficient, and with a larger gap between rotors it can have larger propellers and provide a more unobstructed view for a camera.

So given these advantages, why do we rarely see a trirotor? There are probably several factors behind the proliferation of quadrotors as the most common configuration. The first is simply market inertia: fashion, if you will. The majority of customers now expect a quadrotor, so the manufacturers make the machines they want. Then there is the cost of fitting that yaw servo. It’s a slightly complex mechanism which probably requires a few precision parts, so if you are a factory in China making millions of them you are likely to pick a design that saves you a few cents. If the extra motor on a quadrotor is cheaper than a servo and linkage, then you’ll make quadrotors. And finally, there is a reliability angle. If you crash a quadrotor, as you inevitably will, all its arms are equally strong. The yaw servo is an expensive weak point on a trirotor, so in the event of a crash there is a good chance that it will be first to go.

All is not lost for the trirotor though. If you fancy owning one then you can build one yourself by following this Instructable. We’d suggest becoming proficient at flying a quadrotor first though, otherwise the shop selling yaw servos is going to like you a lot.

Thanks [Jared] for the impetus for this piece.

Header image: Toglefritz [CC BY-NC-SA]

77 thoughts on “The Tri Rotor Drone: Why Has It Been Overlooked?

  1. so what is the benefit again?

    If you are going to go to all the effort of adding yaw to an arm, you can do it just as well with any number of arms, going to three doesn’t gain anything.

    mandatory extra complexity doesn’t seem like much of a beneifit, if you really need drastically more yawing force than a traditional quad gives you, adding a small horizontal motor seems much simpler (and probably not a significant weight penalty, especially when comparing it to a pivoting main engine and the servo to drive it)

        1. the only thing I see claimed is greater yaw maneuverability, and that’s a matter of the twisty boom, which will work just fine on a 4+ rotor craft, and would be heavier and weaker than just putting out a tiny motor aimed sideways like a traditional copter tail rotor on a more normal craft (assuming that the traditional craft are actually suffering from insufficient yaw force, which I question)

        2. I’m guessing the physics of rotorcraft might be like the physics of rockets where even if you hang from your thrust it doesn’t stabilize you because thrust is actually out, not up. I’m guessing things with more than 2 points of lift might be more stable and turn over less? Not that that seems to matter for the IR controlled coaxial helicopters. But I’m guessing the heights in a linkage with the slope of the rotors has something to do with why those are stable. In other words maybe you can’t have fewer than 3 rotors without tilt or a much more complicated control system.

          1. So there’s this old thing people keep saying about rockets, that if you could put the engines at the top of the rocket and pull then the bottom of the rocket would form a pendulum and naturally stabilize. The people who actually knew the math would say it doesn’t work that way. And once people got ahold of Kerbal Space Program (where rockets aren’t damaged by their own exhaust yet) lo and behold it isn’t any more stable than any other unguided rocket. The flaw in the thinking is thinking that the thrust is pulling up. In actuality the thrust is pulling in the exact direction the vehicle is pointing at this moment in time, which is not exactly up. As a result errors in direction don’t get corrected, and they may even get reinforced by the thrust of the craft.
            Helicopters are a little different because the pitch of the propellers isn’t fixed, instead being controlled by a complicated linkage system that varies the pitch over the course of the arc. In a manned or full-size model helicopter the pilot has control over this and uses it to maneuver the aircraft. In the smaller “3” or “3.5” axis IR controlled helicopters with coaxial rotors (one axle is a tube running outside the other in the opposite direction) each rotor has a linkage to a free-flying pair of weights on a crossbar below it controlling the pitch and if this plastic linkage pops off the helicopter is wildly unstable.

          2. It doesn’t guaranteed stability, but hanging from the thrust will be more stable than standing on the thrust because the force of gravity on load continually tries to pull it down and thus provides a re-orienting force towards vertical. Standing on the thrust is the opposite, where gravity continually tries to pull the load away from vertical.

          3. Hanging from the thrust doesn’t help a helicopter/multicopter or a rocket and it’s common misconception, as noted by the OP. Look up the Rocket Pendulum Fallacy or similar, comes up everytime someone talks about manned multicopters.

            The gravity works equally on the motors and the rest of your copter airframe, it pulls them down. The thrust vector is independent of gravity and is attached to the body frame, and also acts on all of the airframe unless differential thrust is being applied. In that situation neither of the forces stabilizes or destabilizes the copter. If your airframe has rolled by 30 degrees, the thrust is also tilted by 30 degrees, it doesn’t work against gravity. If it’s upside down, the same applies. The copter only sees the sum of the two forces, and the sum of accelerations, until a speed where drag comes into play.

            On the other hand there’s some claimed aerodynamic advantage to have the propeller below the boom that the motor is mounted to, over above the boom, but makes it harder to design good landing gear.

        3. Mainly efficiency, but also stability and a better layout. And the complexity of adding a servo is much lower than having another engine and engine controller.

          1. how is diverting some of your lift to the side more efficient than having it all pointing up?

            speed controllers come in packs of 4, as do engines, adding another one is trivial (plug them in) and they attach on a solid boom

            how is that more complicated than adding a servo (which needs different software to run it) and making one of the booms tiltable? Especially since the majority of the controllers assume 4 motors?

            as far as stability goes, software handles that now (to the point that you can attach wildly different props to the different motors and it ‘just works’

            why is the 3-rotor layout “better” than a 4-rotor layout?

    1. The advantage of 4 rotors over 2 is that you can have independent control of all three rotation axes. With a 2-rotor design, yawing by increasing torque on one rotor relative to the other also introduces a pitching motion. With 4 rotors you always have opposing pairs you can use.

    2. Right. A quad rotor doesn’t need any kind of mechanical movement other than the change in RPM of the rotors to balance itself. This is inherently easier to manufacture, since mechanical errors are cancelled out by the motor controller.

      The Tricopter requires a very precise and very strong mechanical hinge for the tail rotor, which something cheap Chinese manufactures aren’t particularly good at. Which is also why you never see any mass-produced tricopters.

      I’ve built a tricopter myself, and frankly it’s really, really hard to balance one. They’re incredibly sensitive to vibrations from the tail rotor, so if your prop isn’t perfectly balanced, you’ll end up overheating the tail servo because it’s trying to correct for oscillations generated by the tail rotor vibrations.

      A bicopter will need two swash plates to achieve full maneuverability, and swash plates are mechanically complex and require precision to maintain stability. It’s not impossible, but expensive, and I’m not sure of any control algorithms exist for this configuration yet with any of the mainstream control boards ( like KK board or Naze32).

  2. I’ve flown both 3 rotor and 4 rotor designs a lot. I wouldn’t say the tricopter is more stable than a quadcopter, not by a long shot. 5 years ago tricopters were more agile – the yaw effect was much more responsive than a quad of the time. But today with modern ESCs and flight controllers my little 210 racing quad is both far more stable than my tricopter, and far more responsive in all axises.

    Part of that is better controllers and sensors, but one other important thing which is much harder to control: servo jitter. The servo controller the angle of the yaw motor on a tricopter isn’t perfect and any positional noise it sees will manifest itself in the stability of the aircraft.

    The larger gaps between rotors is hardly a benefit anymore. You can put a camera on a quadcopter without props in the shot. My 210 has about 2mm distance between the tips of its props and I never get props in my shot.

    Larger gaps mean that larger props can fit, sure, but you’re still constrained by your motor and ESC. If your motor can’t swing the larger prop it’ll burn up. That’s if the ESC doesn’t burn up first. It needs to be rated to spin higher torque motors as well.

    I think tricopters could be improved to become comparable to quadcopters of today, but a lot of that requires better software. Someone needs to be motivated enough to account for servo noise and improve PID routines aimed specifically at tricopters. Given the added hardware complexity and cost, why bother?

      1. I do, actually. The tricopter I have in mind is 240mm, motor axle to motor axle and uses the same props that my 210 uses.

        I’d still consider it a comparison of apples to oranges though. That was something I neglected to mention in my previous comment. Even at similar sizes they’re totally different in the air.

  3. I’m still not seeing the advantage, and that’s speaking as someone who owns both 3- and 4-rotor models. It might be surprising to the hacking community, but among the radio-control enthusiast community, the 3-rotor models came FIRST and were fully supplanted by the 4-rotor models within 2 years, as soon as the software was sorted out.

    Yes, the 3-rotor can provide a wider space to mount a camera. It also plays nicely with simpler gyros and with the control mixing already built into most mid-range RC radio gear, which is why they came out first. That’s about it for real-world advantages.

    The article seems to claim some unsubstantiated stuff about how it’s more efficient, but provides no evidence for that. I don’t believe the larger motors are any more efficient, but the larger propellers certainly are, and the extra space between them helps reduce turbulence as well. Of course, you could also put huge, slow props on long arms on a quad. And a 3-rotor has one HUGE disadvantage, which it shares with conventional helicopters and is indeed the whole reason a Chinook exists in the first place – spending your energy budget to counteract rotation by directing thrust sideways is very wasteful when you could just add another rotor and use torque to counter rotation instead. You’re producing the torque regardless, so why not split it evenly and make everything cancel out?

      1. Both quad and tri rotors aircraft designs have been around for almost 100 years. It’s hard to say which came first.

        While I don’t have a link to a single definitive source I can confirm that tricopters have been around longer, at least in the RC hobbyist world. We didn’t have the economies of scale 10+ years ago (which is when I first recall seeing people trying them). No 4 packs of motors and ESCs, no bulk bags of props for a $10. Our gyros came from gutted Wii controllers. It was a lot cheaper to build a tricopter.

        Tricopters were also easier to fly compared to quads with the software of the day. We didn’t have the software we have today, so you almost had to fly things manually. The kinematics of the tricopter were easier to understand (at least for me). I was able to write my own logic for one without having to think too hard about PIDs and sensory input.

        But as quads started seeing more adoption, and as things got cheaper, and as sensors and software started focusing on quads, tricopters started making less sense.

        Some hobbyists still prefer tricopters. David Windestål of http://rcexplorer.se for example, loves them and sells some fantastic tripcopter kits. It’s one of the only “modern” tricopters that I think compares with modern racing quads.

  4. Servos fast and precise enough to provide tight yaw control tend to be much more expensive than typical quad outrunner motors which are nonsensically cheap due to volume.

    1. Also, if one motor/prop dies in a quad, you still have some hope of bringing it down in some way other than brick-out-window fashion. If one drops on a tri? You’re falling out of the sky.

  5. I have not flown a tricopter, but my understanding is that the flight characteristics are more “swoopy”. They don’t change direction as abruptly as quads, so they can be nicer as Aerial Video platforms. Other than that, Michael Buffington has it right: quads have come a long way and are just as stable and more aerobatic than their three-bladed brethren. For those wanting to try them out, rcexplorer.se has three models: a full-size, a mini, and a baby, each with very different characteristics. (I have no affiliation with that website/store, other than being a fan of what the crazy Swede does)

      1. They’re more maneuverable than a similarly sized quad most of the time, but they don’t have the “jumpiness” that a lot of the quads have. (This is a common problem for me in aerial video work and is why I used a tricopter for a long time.) This was especially an important factor back when 2-axis gimbals were standard rather than 3, as the “jerky” yaw of a quad makes otherwise beautifully smooth video feel weirdly and almost artificially jumpy. You have to fly a tricopter to understand why some people prefer them.

  6. “. . . it’s something we find surprising because it’s a configuration that can have some surprising benefits.”

    “This has the desired effect of creating a flyable machine but offers no advantages, and a few disadvantages, over a four-rotor design.”

    whuuuuuut?

    1. The second comments deals directly with the original design of tri-copters where one rotor is mounted at a slight angle to counter-act the vehicle’s tendency to spin. The “better” design that they claim is one where one rotor is on an remote controlled variable angle arm.

  7. Link to apples to apples comparison?

    Would love to see a ‘traditional’ helicopter with tail rotor compared to quadcopters/tricopters/counter-rotating bi-copters/etc.

  8. I started learning to fly a multirotor via a tricopter that I built myself; that was a learning experience in many ways, but the main lesson was — like the article closes by mentioning — “every time you crash or possibly even just land roughly, the yaw servo is going to get hammered and probably need to be replaced.”

  9. You have a processor in the thing so go ahead and let it rotate, it can handle it with a compass chip. Skip the yaw servo and just fix one boom at a tilt. It’s just a coding problem. Can be tested with the current model and code changes holding the tilt boom steady. Once the thing is 40 ft away can’t tell which way it’s pointed anyways. Will require a compass chip in the transmitter as well, but for early test flights just keep it North of you, mod the transmitter later.

      1. No more worry about which way it’s pointing… just move joystick way you want it to go. Skip twist on one boom to compensate for rotation, let ‘er spin! No more control reversal coming towards you vs going away. Still flip and twist stunts, nothing lost. Got easier in fact! Just one compass chip added to aircraft, one in transmitter.

    1. i really like this idea and it’s exactly what i thought myself after reading the article. go ahead and let the 3 motors spin in the same direction, and use a gyro and/or compass to maintain a virtual heading lock as the device itself spins freely. kind of impractical but definitely would be a cool demo, i wish i could find it on youtube. :)

      1. Cox made a “flying saucer” that was a free-flying .049-powered object.

        I could see mounting a camera gimbal below that on a powered platform that used a compass to keep it pointing in one direction, and steerable vanes at to vector the thrust.

  10. This may sound obvious but… why not make the servo controlled arm without cross supports and let it tilt via a wing warping style movement, not bending the structure but allowing it’s parts to pivot.
    That way the structural pressure is still within the frame, not concentrated on the servo.

    1. That’s essentially what the better designs use – the servo doesn’t directly drive the tilting of the motor, it drives linkages.

      That said, you can’t avoid the fact that you have motor spinning at some pretty high RPMs with along with a gyroscopic effect. Anything you design is, at best, only going to distribute the loads. You can’t make the loads go away.

      You could direct drive that motor using a servo that has enough torque and precision to have it be as close to perfect as possible, but then you’ll be spending more money than just adding another motor (making it a quad). That servo would also probably weigh as much or more than the additional motor.

      So many people in the comments of this post are suggesting things that could be done, design-wise, that are in theory possible, but don’t hit the “sweet spot” of meeting the right balance of weight, cost, simplicity, durability, and flyability.

      10-12 years ago, we could make quadcopters, but we couldn’t make them cheaply and they weren’t that easy to fly. Because of those constraints we settled on tricopters. Maybe 10-12 years from today similar conditions that allowed us to shift to quads will motivate us to use all the ideal but too expensive tech we’re seeing today.

  11. In the “good old days” it was a big advantage to safe the 4th BL motor/esc because of the high price. Furthermore we had no flightcontroller, just a bunch of gyros/mixer boards…
    It worked somehow (like “acro” today), but was pisa to tune. You had to retune nearly every 2nd flight because auf temperature drift of gyros etc…
    Then came the Shred/MultiWii… then the chinese stuff.

  12. I would prefer variable pitch quadrotor to a fixed trirotor setup. Get those instant flips and inverted flight profiles. (See Cutler’s MIT paper on this + utube).

    1. While that’s a theoretical possibility, I’m not aware of any common flight controller software that allows for that. Certainly when one prop came off the Phantom I was flying, it started tumbling. Fortunately, it fell into a tree, and the branches slowed it down such that it didn’t smack too hard into the ground.

        1. a 4-rotor can’t fly on 3 motors7 (unless you can reverse one motor, but they are generally not setup to do that)

          but a 5+ rotor craft may still be able to fly if it looses one motor. the more motors, the more balanced the remaining lift is going to be. with 3 of 4 motors, you have no lift on one side of the craft.

  13. David Windestål (rcexplorer.se) was the culprit that i like trirotors/tricopters. The difference with a quad is in the feel/handle while you fly it. I think is not a question of which one is better, more of which one you prefer to fly for the day.

  14. 4, 6, 8, blade drones having a “front side” just is not logical to me. Logically there is no “front” so we artificially define one edge to be “front” to make it human controllable. It is a holdover from the early days of development, back in the dark ages, before reasonable compass chips were available and one had to use flux-gate magnetometers of some significant weight, so made very good sense for those pioneering days when a compass on board was not practical.

    Stunt drones would still have a “front”, but they all fly close to you as honestly doing a cool double-flip half a mile away just isn’t gonna be much fun. For a wanderer you want more autonomy as you likely have no line of sight to it and a spinning vehicle would have a gyroscopic stability with those motor weights out at the perimeter just like you’ll actually need. Google already has proven spinning video cam devices and processing giving up to 360 degree video so that’s a hurdle already crossed, and alternately a rotating camera platform is completely possible today. The baro chips are accurate to about 10″ altitude and better than that for sensing small changes in current altitude. The three rotor drone is perfect for this application. Don’t think I even have to mention gps…

    Wanna improve it? Skip props on booms, make it a disk with ducted fans for a large increase in propellor efficiency and eliminate broken props, reduced battery consumption.

    Currently available Ardupilot boards should be able to handle it. Create your own ufo scare!

  15. Shedding props? Are they still putting the motor shaft UP so the prop lifts the drone in the same direction that you pull the prop off the shaft? Hello! Simple solutions are the best!

  16. > The additional benefits of the layout are that with fewer drive motors it can be more energy-efficient,…

    Nope. Energy efficiency of rotor-type flying things (helicopters, tricopters, quadcopters, etc) depends mostly on the surface area swept by the rotors. The simple math is that a quadrotor is most easily 33% more efficient than a tricopter. (this is assuming you keep the props the same size.

    1. given the same swept area to mass ratio less motors will always be more efficient, you are multiplying inefficiencies, it it also one of the major arguments against hexa and octorotors, despite their inherent redundancy.

      xcoptercalc has some actual numbers that support this and their sims have been accurate enough for any design i have used it for.

  17. Tricopter overlooked? Tricopters is what I started with, when reading about ‘multi’ rotors. Someone has already mentioned rcexplorer.se great resourceful website.
    I think one is not -better- than the other, just a different setup for your personal liking.

  18. Afaik tricopters are less relevant these days, since the usual quadcopter can be made for loose change from widely available parts, and a tricopter still requires some work.

    Oh and a quadcopter can break a prop and still fly, a tricopter with 1 broken prop can not recover, at best it can try smack itself into the nearest wall rather then drop straight down, but a quadcopter with a broken prop can still be somewhat controlled (for a safe landing)

    Why this ability is relevant:

    1. more to the point, the more motors, the larger the lifting capacity.

      The higher the mass of the draft, the more force it takes to move it, which makes it more stable than a lightweight craft

      1. It’s got more to do with axis stability in winds and when moving. A quad can carry a full frame or 3ccd too. I see people do production quality aerials with a Mavic Pro and Gimble all the time.

  19. I have a few tricopters, a big gimbal-equipped one for video and a small racing-style one, and have had more than one iteration of each. Have flown them in the snow, desert, into clouds, crashed mid-air against other multicopters, etc.

    Their biggest advantage is definitely how easily they can be made foldable and occupy little space. Everything else people talk about is a relatively small difference against quads. The folding frame capability is the only thing I ever mention when people ask about the pros and cons.

    Yep, the camera position is nice but it’s even nicer on an H-type quad.

    The efficiency, I’m not sure… I’ve been playing around with stretching the flight time of my plywood 11″ prop tricopter and have reached 54 minutes of flight with fancy LiIon batteries, in hover, but logically this shouldn’t be related to it being a tricopter.

    Stability, yeah, actually quads tend to fly more stable. It’s definitely a software thing — a few of the mainstream FC firmwares have deficiencies in the tricopter implementations, but it’s understandable as there are more variables to live-tune. There are people claiming better yaw handling, and yes, my big tri handles very nice for filming but it’s just a software thing. The fact that there is even a difference is proof that the software support is imperfect.

    Cost.. it’s just one motor/esc, not really a factor. I use cheap $2 servos and rarely damage them in crashes, the plywood frame breaks much more often (but is also trivial to put back together — with 1h CA glue curing time I may be back in the air before I go home)

    By the way, if you go to an RC model club field you’re bound to see a few, I’d hardly say they’re overlooked, everyone has seen one if they fly regularly. There’s even documentation on rcgroups on building 80mm-size brushed toy-quadcopter-size tricopters based on the $10 Eachine H8 mini quads electronics.

    Want to see great tricopter hacks from other people? check out this guy’s boomcopter videos: https://www.youtube.com/channel/UCGFA15rupNZahitWkAWCpzA, his other projects are well worth an article too.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s