After the last experiment, I went on to design a frame that would have reduced weight but still be stiff enough. This time I planned to use the center plates of the F450 frame, because of their stiff connections to the arms. I started by looking at the weight breakdown of the F450 frame:
- Arms: 50 grams or 1.76 oz (x4 = 200 grams or 7 oz)
- Lower Plate: 60 grams or 2.11 oz
- Upper Plate: 15 grams or 0.52 oz
- Hardware: around 20 grams (or 0.70 oz) in total
It is clear that there is considerable potential in saving weight by redesigning the arms. There are many very good arm designs for this frame, but almost all of them aim either at increasing stiffness, or extending the arms themselves. While all this is good, my main goal was to reduce weight. I found just one ready-made arm design made of carbon fiber (CF) that manages to do so, but it is no longer produced, apparently.
I went on looking at arms using CF tubes. The thing with most of these arms is that, while the tube itself is lightweight, the weight of the additional hardware needed to fix the plates and motors usually counteracts the weight savings. Especially the motor mounts usually inherit a very inefficient design featuring two aluminum clamp mounts, on top of which is an additional CF plate, on top of which the motor is mounted. Phew! What a waste of material!
Since I have access to a 3d printer, I decided that I could do a bit better and design a 3d printed mount, both for the plates, as well as for the motors. Initially I found some designs online, such as this one on Thingiverse.
However, after printing it turned out too heavy for my intended use, in fact heavier than the original DJI arms. So instead, I went on to create my own clamp design. The finalized design is shown in the photos below. I was printed on an Ultimaker 2 Extended, using white PLA. The tubes used were 14mm (or 0.55 inch) dia. 3k CF tubes.
The initial design for the plate mount came from this design found on Thingiverse. However, the motor mount on that design turned out to be very heavy, and so I redesigned it completely. It is interesting to note that the main difference with the original designs is that I used zip-ties for securing the tubes to the mounts, rather than the screw collar of the original design. For many it may seem an insecure way of fixing critical structural components, but on the contrary the zip-ties turned out to be very reliable. The only time a zip-tie snapped was on a quite rough landing (i.e. crash 🙂 ), where the snap actually helped to reduce stress to the motor mount itself, thus saving it from breaking.
Another advantage of this design was that it allows for the cabling to be neatly routed through the tube. This is not possible if you have separate ESCs on each arm, but in my case I was using the QBrain 4in1 esc, so the electronics were centralized, and only the power cables needed to go through the tube.
Overall I’d consider this a relatively clean build, with a reasonable flight time of around 15min using a 3S 3300mAh battery.
The STL files for the mounts, as well as a newer version of the motor mount are available in Thingiverse.