In the last post I went through the process of designing and realizing my first Carbon Fiber-based H-quad. While the process itself was successful, the design was based on an open section, a fact that made it twist excessively even though it was quite resistant to bending. I thus went on to redesign the central “unibody” frame to be torsion resistant.
Closed (tubular) elements exhibit orders of magnitude better torsion resistance than open, thin sections. The figure below says it all:
An interesting thing to note is that the shape of the section doesn’t play much of a role in torsion resistance. The key factor is whether the section is open or closed. With this in mind, a closed, tubular section would provide much better performance.
With this in mind I fired up Rhino3D once again, and set out to create a new design. The final, revised design is below. You will notice that I have reserved space for the newly arrived NAVIO2 flight controller, along with a Raspberry Pi 2, instead of my old CC3D controller. This gives me infinitely more capabilities in terms of aircraft control, video streaming, even advanced features such as computer vision using OpenCV, or interfaceing with sensors such as the VL53L0X rangefinder. It also requires additional space and provisions for connections, mounting etc. All these have been designed into the frame before production. A stylized render of the frame is below.
The above render also includes custom motor arms, which were supposed to be 3d printed and constructed out of tubular CF cloth. I didn’t get around to building them yet, so I’m sticking to 3k CF tubes and 3D printed motor mounts for now, as you will notice later on.
In order to have a clear print without imprecision due to print height, I reinforced the section with inner flanges, and printed the body in four pieces. I then joined the pieces together with cyanoacrylate glue. In order to achieve better precision in joining, I glued plastic guides at the inside of each section, so that when bringing them together, the last cm or so would slide in precisely. The guides were made out of some pieces of zip-tie.
The CF cloth was just rolled around the mold just like a burrito in this case. Below you can see the prepared mold, mold wax and PVA release agent added, on a holding rail that was temporarily glued to the interior:
I started by adding a layer of epoxy to the mold1. Then, I started to roll the cloth, and as soon as the mold was completely covered, I kept on adding a bit of epoxy between layers of CF, and applied a bit of tension while rolling, to make sure the layers stay tight together2. I did about four layers of CF, which resulted in a quite strong structure.
Once left to cure, I had to remove the interior mold, which practically meant destroying it. The removal was actually easier than I expected, with large pieces of 3D printed PLA coming out in one chunk. Then, I went through the usual procedure of trimming out the edges of the CF cloth, and sanding them to get a nice finish.
The result, together with motor arms, is below:
The frame is really strong and light. I tested it both in bending as well as in torsion, and there is almost no displacement. There’s still a lot of work left, namely fitting of components, and (most importantly) cutting out the necessary openings for both components as well as cable harness to pass through. It is expected that after the openings are cutout the frame will lose a bit in rigidity, but still I expect that it will be quite strong.
Below you may see a snapshot of the frame after just one opening has been cut out (the one for power distribution), and hanging strips are placed on the exterior to help dry-it the components.
As a conslusion, the tubular “unibody” fuselage has been a success. The tubular construction indeed minimizes torsion, while minimally impacting weight. Stay tuned for the next post on how the frame is finalized, the hardware installed, and, of course, the maiden flight!
- For more details on the preparation and lay-up of the CF, see my previous post
- You should not apply too much though, because the cloth will start sliding on the still liquid epoxy bed!