I knew I had a good, basic design with the Tri^2, but the arm vibration was killing any aerial imaging efforts. So I decided on a further refined design with a much stiffer set of arms and body. I knew it would be heavier, but I was willing to trade off some of the phenomenal flight time of the Tri^2 for the ability to take clear pictures.
The first change was using very stiff 21 mm diameter tubular carbon fiber arms. These required special Delrin clamps to connect the round arms to the flat surfaces of the body and motor mounts. The arms and clamps all came from AGL Hobbies. Heavy, but wow they are sturdy.
For the body I decided to increase the thickness of the G10 fiberglass plate to 1/16 of an inch. I also increased the size of the body for the sake of stiffness. The arms ended up being big enough so that the top and bottom plate separation was large enough to stick the ESCs in between them. That was neat and tidy.
One of the very big improvements I made was how the vibration isolation is handled. Tim Nilson, designer of the QAV series of quadcopters, was one of the early developers of the concept of “clean” and “dirty” plates. In this type of construction the dirty plates connect all the vibrating bits and the clean plates hold the camera and other sensitive equipment. The plates are separated, but connected by silicone vibration isolators.
I used this concept with the Tri^3. The tricopter arms are held in place by a sandwich of two G10 plates forming the main body and dirty plate. Connected to this basic body by four silicone vibration isolators are two upper plates containing the FPV camera, Pixhawk flight controller and other goodies. The fairly heavy battery is suspended beneath the tricopter but connected to the upper clean plate with four aluminum standoffs. These four rods pass through the dirty plate to the upper, clean plate, without contacting the dirty plate. By “mass loading” the clean plate it makes it much less susceptible to vibration. And….it works.
Another enhancement I added to this design was the use of an EzUHF RC receiver. I had been getting familiar with them on the fixed wing aircraft I was experimenting with and there were some potential benefits to using them on a tricopter. In a straight line this can easily provide control of the aircraft out 10 to 15 miles. Operation at that range for a tricopter is pretty nonsensical, but it provides a much more robust RC connection closer in. It’s possible to fly behind some objects without losing aircraft control.
Finally, I wanted to have improved interchangeability for the various cameras I planned flying on this tricopter. To that end I created a flat nose with attach points. This allowed me to build four different camera mounts, all of which quickly attach to the nose of the tricopter with four screws. Two of the mounts are for my Canon SX230 GPS tagging camera, both looking straight down and oblique (tilted forward 15 degrees), and two mounts are for my Mobius HD video camera. One of the Mobius mounts is fixed (and thus very light) and the other is a brushless gimbal with an AlexMos controller board.
I retained the basic yaw mechanism design I used on the Tri^2 so no big changes there. Likewise with the G10 landing struts, but I did add lightening holes. And since I was dealing with round arms instead of square ones, I used nylon conduit clamps with nylon screws to make the attachment of the landing struts to the carbon fiber arms.
Like the Tri^2, this iteration retains the ability to fold for transport (sort of the point of this exercise anyway!). The folding arrangement is a little more robust than the Tri^2, where in this one screws hold the folded arms in position and prevent them from unlocking. The use of large carbon fiber props generally requires the use of socket head screws for attachment, so as they say, some assembly required. But it doesn’t take much time to get up and going.
The Tri^3 is still in a testing phase but it has already accomplished some notable flights. I flew it on a flight out to 2.3 miles (3.7 kilometers), then safely back. It could have flown further but that was pushing the limit of its 5.8 GHz video link. I’ve done a few local OC flights, reaching to the top of Loma Ridge, 1,100′ above the launch point and about 1.75 miles out. Nice views from there, not to mention flying over all that Irvine Company land where trespassing is prohibited. Very satisfying feeling.
Flight times vary quite a bit depending on the camera and battery load. With my heavy Canon SX230 I can get perhaps 15 minutes, which pleases me. That’s a lot of weight. If I fly my lightest payload, my Mobius HD video camera in a fixed mount, I can fly for at least 17 minutes.
At this point there’s not much further room for improvement. I could possibly use slightly thinner G10 stock for the body and maybe make the arms a bit shorter. With a lot of care I might be able to save 40 or 50 more grams, which would equate to another minute or so of flight time. That’s not a heck of an improvement and probably not worth doing a rebuild. So for at least this moment, I seem to have my ultimate tricopter.
- Motors: T-Motors MN3110-26 470KV
- Props: Hobbyking 17×5.5 carbon fiber
- ESCs: Castle Creations 25 Amp Multirotor
- RC receiver: ImmersionRC EzUHF 4 channel with VAS 433 MHz dipole antenna
- Flight controller: 3D Robotics Pixhawk
- Onscreen display: MinimOSD-Extra
- FPV cam: SC2000 CMQ1993X
- Video transmitter: ImmersionRC 600 mW 5.8 GHz with Circular Wireless Skew Planar antenna
- All up weight with 3,300 mAh 4S LiPo battery: 1,700 to 1,900 grams, depending on camera load
- Distance from tricopter centroid to motor (i.e., “radius”): 400 mm