17 May 2013

My First Quadrotor: Parts Selection and Build

Note: All this happened in mid-January. I've just been too busy to finish this and post until now...

Yupp. I built it! Here's how it happened...

Being a practical man, I want to build a quadrotor for a few concrete reasons: 
  1. I want to write the flight controller from pretty much scratch, as an exercise in my control system knowledge. 
  2. I want to show to the world just how easy and inexpensive it is to choose parts for, order, and build one of these. 
  3. I want to to document my software design process in a way a blog reader can learn from. 
  4. THEY'RE F****ING COOL. I mean, LOOK AT THESE: 
AWWWW ^_^

Shane's 4PCB at MITERS. 


Parts Selection: 
I began the process by asking the experienced quadrotor-ers around MITERS and the IDC about where to start. Shane suggested I find a 330mm-size frame. Searching the Hobbyking Quadcopter frame section for "330" yielded the following beauty: http://www.hobbyking.com/hobbyking/store/__28172__F330_Glass_Fiber_Mini_Quadcopter_Frame_330mm.html. It seemed big enough to make the controller run smoothly while being small enough to comfortably fly indoors (my flying domain of choice, for now). 

So I had a frame. To get from here to working quadrotor I need to spec out: 

  • Frame
  • 4 x Brushless Motors
  • 4 x Electronic Speed Controllers (ESCs for short, one per motor)
  • n x sets of Propellers (Banks Hunter let me know these break often, so I ordered like 5 sets)
  • Battery Pack
    • Battery Charger (I did not purchase one because I already have one)
  • Programmable microcontroller board  (Arduino should do)
  • IMU for angle feedback
  • Other sensors for other kinds of feedback (usually come with the IMU and ucontroller board)
  • XBee pair and USB interface, in order to give the quadrotor commands from my computer program. 
Looking at the F330 frame's listing on the HobbyKing website gave me the proper specifications I needed for the various components. I chose the components necessary simply based on what was least expensive. I picked the SK3 Aerodrive 2826 motors with a Kv of 1130 rpm/V and many spare sets of red 8", 4.5"pitch propellers (These break often, I hear), which are within the specification range recommended by the F330 frame page. I chose 15-amp ESCs, which is more than I need for this size copter, because they were also well-priced. I then chose the lightest 1800mAh 3s (11.1 V nominal) battery I could find.  In order to tie them all together with minimal wiring work, Banks suggested I use a nifty Power Distribution Board to get battery power to the ESCs.

As for the XBees, I chose to get this starter kit from Makershed, which was the least expensive option. It comes with a breadboard-able 5V-3.3V logic level shifter (Many electronics run on 5V, the XBees run on 3.3V) and a USB adapter so you can connect to a serial device (like a quadrotor!) wirelessly. 

I'll worry about the sensors and microcontroller board later, because I have Arduinos and IMU breakout boards readily available nearby. The ESCs each take a 5V servo-style PWM input signal, so an Arduino is be able to command all 4 brushless motors. Apparently the ESCs also generate 5V for the logic power, so no need to use a regular! 

Total cost? About $150.00. NOT BAD! 


Building the Quadrotor:

About a week after placing the order, I received the XBee starter kit from Makershed and the four SK3 2826 motors from the HobbyKing USA Warehouse in the mail. HobbyKing has both an International and USA warehouse, and you can isolate the online selection available from each right on the website. The International warehouse has a wider selection of products, but shipping to the USA is costly and takes forever. On the other hand, the USA HobbyKing warehouse does not have as wide a selection, but shipping to US destinations is both cheaper and more timely. For this build, I ordered everything but the motors from the International warehouse. 


After waiting a few more weeks, a package from China came!




And check out all the new toys! From left to right, I see two frames (I gave the extra one to Shane Colton), the ESCs, battery (above), and a shitton of props. 



Somewhere in the pile of new toys was the Power Distribution Board that Banks highly recommended I purchase. As you can see, it makes hooking everything together really easy. Kinda looks like a floating nervous/circulatory system quadcopter without a backbone...



Speaking of the backbone, let's put the sexy F330 frame together! The kit came with an instruct- I mean INSTRUCITON MANUAL! It was pretty well-written, though the kit was simple enough to figure out myself. 



The flip side of the sheet had the same thing, but written in Chinese and probably with better and more detailed instruct- I mean INSTRUCITONS. 



What really bugs me about the kit is there is only enough hardware (screws, etc) to put the thing together, and no extras should you lose any. Metric hardware isn't as easy to come by for me, so this may prove itself to be a pain. 




The frame consisted of four identical fiberglass arms placed in an equilateral X formation sandwiched by two carbon fiber plates. Lacking the room to hold the battery and control board (which will presumably occupy the very top) and attach the Power Distribution Board (PDB?), I decided to fasten the PDB to the underside of the vehicle. Sure, live power lines are just exposed for now, but I'll insulate the whole thing before trying to fly it. 



The tiny (D'AWW SO CUTE!) SK3 2826 motors each come with the proper mounting hardware (Again, no extras in case the tiny M3 screw wanders before it is secure...). I decided to leave off the top carbon fiber plate for now until I could figure out my microcontroller situation. 



The ESCs and extra wire folded up pretty nicely on each arm of the copter. I used small black zipties to keep everything down and prevent general self-destruction from the propellers from chopping things up.
The ESCs are sensorless, relying instead on current sensing to commute the motors.  For now, I plugged the 3-phase SK3 2826 motors into the ESCs in a random configuration. I will later need to shuffle the leads in order to ensure that each propeller is spinning in the proper direction, producing downward thrust. 



The propellers are not all identical. Each set of propellers comes with two pairs (4 total): one pair of clockwise and one pair of counterclockwise props. They are installed diagonally opposite each other on the quad, which is shown in the frame kit instructions. The SK3 2826 motors come with all the mounting hardware necessary to hold the props down. Yet again, no extra metric hardware included should you lose anything. I'm surprised the prop and motor are rotationally coupled by only by the friction force produced by the preloaded washer. I sure hope it doesn't go flying at my face. Or into my arm, for that matter...




Now, for the controller I was going to use an Arduino of some kind, an XBee attached to it, and a spare Sparkfun MPU6050 breakout board I picked up. Shane gave me something that was pretty much all of the above combined! The MultiWii 328P is a board that houses an atmega328 microcontroller (the same that is found on many an Arduino), an FTDI chip for USB-to-Serial communication and programming, and an array of sensors like an IMU, magnetometer, and barometer for stability and navigation feedback. When Shane used it, he wanted to communicate with a receiver via XBee, so he broke out the necessary UART pins on the MultiWii328P to connect to the XBee. He gave me this great looking all-in-one board in exchange for giving him one of the F330 quad frames (Apparently these are never in stock, and I was lucky to be able to find one). 



The ESC leads all plugged into the servo out pins of the MultiWii328P board. It's almost like they designed it to be easy to put together! After fastening the board onto the top carbon fiber plate and attaching the plate to the frame, I had a quadrotor! 


And there you have it! This was really freaking easy to choose parts for and put together. Next up: Write the software and get it flying! 

03 May 2013

I'm Back!!! Here's What I've Been Up To:

Yup. It's been a while guys. 


3 months and 16 days to be exact. It's nice to be back. A lot has happened in that time. 



I didn't win MASLAB! But I learned a ton, and I may just blog about our journey and what we learned. I'm really proud of its ability to drive around to waypoints and come back to its start location using encoder geometry.




 I'm also proud of having designed and built the damn thing, because it had a solid mechanical design. Also, I learned how to weld! Will blog on all that jazz later. 



I build a quadrotor, and kinda got it flying! I still need to rework the control software and get it actually flying rather than kinda flying, but I'm close. Will blog about both the build and writing the control software really soon. 




I got an internship for this summer! My first internship ever!!! Time to experience this thing people call "the real world...". I will be really far away from where I'm used to being, at Aurora Flight Sciences in sunny Cambridge, MA. Really far. Like right behind the Kendall T stop, a few blocks away from MIT. Looks like I'm never leaving beantown =]. 
The company is a defense and NASA research and development contractor, with strong ties with and a style reminiscent of MIT. 



Here's their front desk of the Cambridge office! Yup, that's a jet engine converted to a desk. Looks kinda like Tony Stark's estate: fancy! I am hired officially as a Multi-Vehicle Controls Engineer Intern, and may be working on the above Skate project (Which has been in use in Afghanistan for a few months now) as well as other cool projects (Make a Blackhawk Helicopter safely land autonomously? YES PLZ!). I'm not sure whether or not this stuff is classified information, so this may be the last time I mention my work at Aurora 0_0. 



I took a lot of measurements inside an elevator for 2.671! In particular I used a barometer and accelerometer (Funny Note: Google's auto correct doesn't think "accelerometer" is a real word. Silly Google...) to characterize the performance of the elevator in the Green Building, the tallest building at MIT. Using this characterization I will be able to compare this elevator's performance to that of others. 
This class is equal parts amazing and frustrating. It's a lot of work, a lot of busywork, and a lot of writing, but on the other hand it's giving me a taste for the standards to which I will be held should I continue my pursuit of a career in research. Which is important to me. So I'll definitely be posting about my experiences here! 

What else... 



Oh! I've gotten MelonChopper (Which is currently the most popular project on my blog... which now has almost 12,000 views. Jeez...) up to mechanical snuff! LOOK AT THAT CUSTOM SPROCKET! LOOK AT IT, IT'S HUGE!! Now I just need to put together the battery pack, do some wiring, and convince Bayley to finish the motor controller or find (read: buy or beg for) my own. I'll get it done, folks. And I'll blog about finishing the mechanical aspect of this baby. Hell, I may just put an underpowered motor controller on it and take it out this weekend with the many other fledgling EV enthusiasts taking the 2.007 EV section!



This thing is supposed to be the prototype for my Bachelor's thesis! It's supposed to be a robotic hand that emulates the following robotic manipulator: 

Because I got super super hosed with 2.671 and 2.005 and a tough HASS class, and I was a lead in this semester's MIT Musical Theater Guild musical, The Wedding Singer,  I haven't done anything since making this prototype and going to the special guest lecture given by one of the senior designers of the Willow Garage gripper. Which is why I'm continuing the thesis through next semester. WILL BLOG ABOUT THIS, TOO! 

Am I done? I think I'm done. For now. MELONCHOPPER AND QUADROTOR UPDATES COMING UP NEXT! 

OH! I almost forgot!!!!!!!!!! 
MITERS GOT A CNC MILL TWO DAYS AGO!!!!!!!!!!!!!!

It was donated to us by my 2.671 professor, Ian Hunter. He's a pretty cool guy. 
But the mill computer doesn't know what's going on with its drivers. I'm determined to get this thing moving by the end of the weekend, so we'll see how that turns out...

17 January 2013

MASLAB Teaser

Here's a quick teaser of what I've been up to all IAP:









Moar coming soon. Also, why the hell does HobbyKing take so freaking long to ship my quadrotor parts :,(