A Tensegrity Micro Quadcopter, Part 3

This is the lightest tensegrity micro quadcopter so far, with a flying weight of just 46g. That is exactly the same weight as the HubSan X4 that the flight controller, motors and battery are taken from. It doesn't fly like a HubSan, though, as the mass is distributed very differently. In the video above, you can see that it's a bit lively, but the tensegrity cage works very well when I smash it against the walls. I filmed the second part in slow motion as I wanted to see if the motors were twisting on their rods or vibrating. They look like they're held on quite solidly in the video, so that doesn't look like a problem.

The version here has replaced the carbon rods of the last version with bamboo skewers which were bought online. I like the idea of bamboo as it's a more ecologically friendly material, but the carbon rods were slightly lighter. The carbon rods are 1.2g, while the bamboo ones are 1.3g, so it's not a huge difference. Where I save a lot of weight is that the cotton thread is much lighter than the string that was essential for the carbon frame, plus I don't need the twelve 3D printed rods ends for attaching the string to the carbon rods. The 3D printed motor holders attach the four motors directly to the bamboo rod frame, where the last version used a 3D printed quadcopter frame that sat inside the tensegrity frame. The design files for everything are published in the OpenSCAD file here: https://github.com/drones4good/Tensegrity

The build for this tensegrity frame is rather involved. I swapped to the bamboo skewers as it allows me to use the very light cotton thread. The bamboo is essential as the thread is held in place by being pushed into a slit cut in the end of the bamboo with a scalpel. It is essential that the thread can be easily attached to the end of the rod in a way that prevents it from coming out.

Just to re-iterate, this is not easy to build and anybody attempting this should expect to have a few goes at it before getting it right. It requires very skillful threading of the cotton onto the rods and then an assembly involving lots of fiddly cotton.

Here is the build sequence:

First, you need to make six of these.


The 3mm diameter bamboo rod is cut down to 20cm long from a 30cm skewer, bought in a pack of 150 online and designed for barbecue kebabs. In order to cut a 15mm slit in the end with a scalpel, I clamped it in a vice vertically with 15mm protruding from the top of the vice. Then I could, very carefully, press the scalpel down to make a cut precisely 15mm deep before hitting the vice jaws. Make sure the slit at each end is the same orientation, otherwise the frame will twist. It's not necessary to be extremely accurate, but I marked the orientation with a pen first.

In order to get the right sized triangle of cotton, use a template which you can place between the rod and the cotton, as you can see in the picture above. I used a piece of balsa wood with black tape on each end to stop the thread from piercing the wood grain. I cut this to 5.7cm in height, which worked for the quad that you see in the video. It was a little tight, though, so you could try 5.9cm to make it easier to assemble? The idea is that the cotton fits tightly over the template when it is wedged at 90 degrees in the middle of the rod, making a triangle of the cotton with height 5.7cm (the template) and base 20 cm (the bamboo, although that's not accounting for the 15mm each end where it's tied).

Next comes threading the cotton, which is really fiddly. In the picture above, that is one single piece of cotton. Measure a piece of cotton which is four lengths of the rod i.e. 80cm. That's more than you need, but it's an easy way to measure and you need a bit extra. Fit some cotton into the slit in the end of the rod (the left one in the picture above). Tie it off securely, then wrap it around the rod tightly two and a half times, then go through the same end of the rod again, along the rod and then into the slit at the other end (the right one in the picture). Use the 5.7cm template to make the triangle, then wrap the thread two and a half times around the rod to lock it tightly in place and go through the slit again. The two and a half turns is so that you come out one side of the slit, wrap the thread around the rod to prevent it slipping, then enter the opposite side of the slit and come out the same side as you did before. Otherwise you will end up twisting the thread on the frame.

Now do the same for the bottom triangle. Put the thread through the opposite end of the rod (the left side in the picture). Use the template to get the correct length to make a triangle, wrap it two and a half times around, then tie it off securely so it can't slip. It's worth pointing out that the tie at the beginning and the tie are the end are both at the left end of the rod in the picture. You should now have something that looks like the one in the picture. Assuming you're got to this stage successfully without throwing the rod at the wall in frustration, you only have to repeat this five more times to get six rods. I needed to practice this quite a few times before I got the methodology right.

Now you have six threaded rods, it's time to assemble them. This is not easy.


Lay out five of the rods in a cross as demonstrated in the picture above. It's worth pointing out that, in the final tensegrity assembly, each rod end has two threads connected to it. That means that the central, horizontal rod, is already fully connected to its four immediate neighbours. The sixth rod, in the bottom right, is going to be connected above the central rod.


Now it starts to get complicated. The sixth rod is placed above the central one and its two ends are connected to the extreme left and right free ends of the triangles from the last picture.

Now it gets very complicated. The top and bottom free ends are connected to the sixth rod's threads that you added in the last step.


The two inner horizontal rods in this picture form the top and bottom of the structure. The outer horizontal rods, top and bottom, are the verticals, which are going to be pulled up in the next step.



And there we have it. Once the remaining threads are connected to their rod ends, it pulls the whole structure together and we have tensegrity. It's a bit loose, though, as you can see by some of the dangling threads. You can use the tensioners from the last post to cure that if needed. The 5.7cm measurement that I gave earlier will give you a much tighter structure. This one used 6.5cm as a test.

I made a single page diagram of how the assembly is accomplished in case it helps:



It's included in the GitHub repository: https://github.com/drones4good/Tensegrity

All you have to do now is attach the 3D printed motor holders, and wire it all up. In order to have somewhere to put the flight controller and battery roughly in the centre, I added a piece of foamboard sellotaped between the two horizontal rods. This also helped to stabilise the rods holding the motors and prevent them from twisting.

That's all there is to it. Flying was a bit interesting, though. If you yaw too quickly you get really bad adverse roll. It's hard to fly because you don't know where to look. You need to be looking at the motor plane in the centre of the frame, but it's obvious from the video that the frame and the motors are out of line with each other. It feels like it's flying at a slight angle, caused by the imperfect alignment of the frame and the motors.

Moving forward, I'm going to try using a flight controller running BetaFlight so that I can tune the parameters to account for the different inertial moments.

It's been an interesting project, occasionally very frustrating, but I've proved that I can build a tensegrity frame to a similar weight to an equivalent 3D printed micro quad frame. This gives it very good resistance to crashes, but the question of whether it reduces the motor vibration being transmitted to the flight controller is still unproven. It's just waiting for me to add some instrumentation and measure it, so that's my next task. I might give tensegrity a bit of a rest for a few weeks, though, as it's been a bit of a nightmare, but an excellent end result.

Comments

Popular Posts