This is my build log post on the quadcopter. Today is the introduction of almost flying the craft, but held-up with just a small problem: Stabilization.
If your just joining in, make sure to catch-up on the previous two parts of the Build Log:
Ok, continuing on, if you haven’t caught the name of the quadcopter yet, it’s name is: Black Dragon. Due to how evil the stabilization tests have been, and for the fact that the color of the craft is black.
If you haven’t caught the video, here’s me trying to fly the quadcopter without any stabilization. This is why I need to build a test platform:
The above photo is the first attempt on figuring out how to test the quadcopter, without it hurting itself while trying to stabilize. The thought was simple. Test each axis individually, and then together, it should work.
Well, after building it, a invisible problem arose: vibrations. Yes, there are visual vibrations, but there are also hidden high frequency vibrations that we can’t see. These vibrations started to distort the values read by the accelerometer and gyro [ Part 1 ] . The code was technically correct though. If there is a tilt at more than 15 degrees in either direction, we need to adjust the motors so the craft can stabilize in the air. After doing some testing and wondering how the motors were turning on semi-randomly, I connected leds to show whenever the angle was 15 degrees or greater. Nevertheless, I got a good light show from the leds, highlighting the invisible vibrations I couldn’t see or notice.
During each test, I have been learning how to perform the “start-up” procedure before the quadcopter takes off on each flight. One of these things to check about every 5-10 flights is the bearings on the motors (bought some cheap Chinese brushless motors). I learned this lesson unfortunately after the damage was done. One of the motors came loose from its bearing, and flies off the craft. The ripping off of the motor was bad, and tore the vital wires to power the motor. Although it is possible to fix the motor by rewinding the entire device, this is a long draw out process by hand, and not something I’m wanting to attempt at the moment. At least until this project is done. I decided to go over to the local hobby shop and get another ‘similar’ motor. This ends up making another problem with the stabilization.
With the problems mounting with the first platform, and the inefficiency of only testing one axis, when we know the code is mostly correct, this new platform shown above is the new version. Same idea as the first version, just we test both axis at the same time.
Again with the testing, I keep the arduino off to the side, and being careful, hold the wire from being cut by the propellers, and check the vibrations as the motors are revved up. The first results were disappointing. The error amounted to almost 600% from where it should be. This did explain why earlier the motors were seemingly turning on and off randomly. With a lot of research, I found that I needed some sort of material to separate the accelerometer and gyro from the vibrations. The first test was with some packing foam that was lying around my home. That lowered the error down to 300%. Making a trip to Lowe’s, I got some ear plugs. The error was further reduced to 150%. After taking a day of thought about it, I rearranged the ear plugs on the quadcopter and got the error to an average of 75%.
I then realized that I was going to have to actually order and wait for some special vibration damping material to come in. The go-to method for quadcopters I found: [ Moon Gel ]. Made for damping drums, this material was the best found so far for getting rid of vibrations.
Right before I get this material, I test the craft again, and one of the cracked propellers comes loose and breaks. It slices into my leg. Thankfully it wasn’t bad enough to need stitches (the propellers move at about 12,000 rpm).
Got the moon gel in today. Testing with a missing prop, I knew I couldn’t finish the tests today until I get new propellers later this week. I decided to test with the three that are good at the moment. With a combination of moon gel and earplugs, the error was reduced to [ 3.35% for the y-axis and 14% for the x-axis ] ! The y-axis is now in the range it needs to be in! Why the x-axis isn’t in the 5% range, I haven’t figured out yet. Some more testing to follow, and I should get it down in the next day or two.
Well that’s it for now! It was probably really boring, but I had to experiment. And we’re still not done. As noted, this vibration thing is annoying. Still experimenting with the moon gel at the moment. Hopefully there will be flight in the next week or two! 😀
As a small thing on some cleaning, if you haven’t noticed, links are highlighted in RED and are in brackets .