7 secondsSaturday, May 7, 2016 · 2900 words · approx 14 mins to read
It flew for just 7 seconds, before the wind took one side of it back to the ground, one of the propellors hit the dirt, and the drone tipped over, shedding two props and associated nuts into the grass nearby. I throttled back, cut the motors, and a friend and I started a forlorn search for the now missing nut. I can’t fly it again until I replace it, and there’s nothing I want to do more. Argh! These words mark the end of three days, starting from an impulse buy of a self-assembly kit from RadioC and ending with it flying for those 7 seconds. I want some more keyboard practice, and writing about the experience seems like a good excuse. Plus, my future self will hopefully look back at this as the start of a grand adventure in making new things.
It all started on Monday lunchtime, at a barbecue at a coworker’s place. They have a huge field of empty space adjacent to the house where we all sat, cooked, ate, drank, and talked until early evening. Then another coworker asked if we’d like to fly his quadcopter. Fuck yes I’d like to fly it. It’s one of those reasonably sized quads that’s almost all plastic and has no smarts in how it flies, with everything that happens down to operator skill or error. Armed with 5 or 6 separately charged batteries, I ran most of them flat with a series of 1-2 minute flights. It was easily the best fun I’ve had in a very long time.
I went home that night and dreamed about flying it. I woke up thinking about it. With that May day bank holiday the precursor to a week off on holiday, I knew I’d end up with a quad rotor drone of some kind by the end of it. Tuesday was research day. I started with the cheapest, mostly Chinese ones that you can easily get on Amazon. I had every intention of getting pretty much the exact same one I’d flown the day before, and there it was on Amazon with a bunch of batteries for take-a-punt-and-if-I-break-it-who-cares money, Prime next day delivery. 1-click shopping very nearly happened.
Then I spoke to Tim. Tim is my guy when I need to know about new things. Want to get your head around anything decently technical in a short amount of time? Tim. Doesn’t matter what it is; American military hardware, audio equipment, rockets…you name it, Tim has it squared away in his head and can get you going quickly, usually with links and referencing so you can keep going afterwards. I’m fairly sure he never sleeps, or at least he falls asleep to the Wikipedia audio book (wouldn’t that be awesome?!).
“Talk to me about drones”, I said. When we established that I didn’t mean a Predator or its ilk, he showed me the RadioC website. They’re a racing drone vendor here in the UK, not too far from where the future Mrs. Ryszard lives during term time, while she’s doing her masters. They’re friends of the London Hackspace where Tim spends a bunch of his time, and the Hackspace is home to a bunch of first-person view (FPV) racing drone pilots who regularly shop there.
It didn’t take me long on their website plus a look on YouTube for FPV drone flights to understand that’s what I wanted to do eventually, combined with aerial photography rather than racing, so why not start with something a little more serious, from RadioC, rather than the cheaper but more restrictive models on Amazon? In at the deep end and all that.
So on the way down to see Christine on Wednesday, I decided to pop in to see RadioC in person. They had a couple of ready-to-fly configurations showing as in-stock on their website, so I figured a quick chat with them to decide between the two configurations would be all I need, and I could be flying later that evening with something pre-built, that I could tinker with later. It turns out that the parts for the drones were mostly in stock, but there was a queue for assembly, and even if there wasn’t and one of the kits was sat ready to go, the recommended transmitters and receivers to control it for new flyers weren’t available until next week at the time of writing.
Rather than walk out empty handed, I sat down with one of the RadioC guys and we put together a kit that I could build myself, where I could get a transmitter later on when I was just about finished and about ready to fly. With no idea how long builds usually take, but with nothing better to do all week, I walked out with a box of parts that I thought would be enough to build a fully working drone, minus the Rx/Tx side of things. When I got home later, I sat and read manuals and watched build videos on YouTube. It quickly sunk in that I’d have to learn to solder properly.
Today’s racing drones are usually hand assembled from bare parts, from the frame upwards, and there’s not much standardisation in terms of things like connections between the motors and their speed controllers, or in the layout or pins — if there even are any pins — of the flight controller board. Nothing is plug and play, so to make any electrical connection between any two components in the drone, I’d have to make them myself. I’ve soldered complex things in the distant past, over 20 years ago I think, but I’ve only really soldered wires together ever since, nothing small or fiddly. What better thing to relearn the skill than on something reasonably expensive and error prone?
I ordered a soldering iron, solder and some helping hands — a little contraption with a couple of crocodile clips on movable arms and a magnifying glass, for holding something while you do more intricate work. What else could I need? I read the build manual for the frame and realised that frame assembly and motor attachments were all M3 bolts, so I grabbed a box of nylon M3 washers, nuts and spacers, just in case. The RadioC images of the final assembled drone looked neatly packaged and obviously well done, but because I’d upgraded parts from their standard kit that uses the same frame, I didn’t expect to get to that level on my first try. I asked Tim to help, he said yes, so the next day would be build it, test it and fly it day. He arrived early in the morning and we got started. I had build instructions and links to YouTube build vides on my iPad, Tim had his laptop to run the flight controller programming software, Cleanflight, so we were ready to rock.
First thing to put together was the frame, and we immediately hit problems. Counting out the screws, posts and locking nuts we needed to effectively sandwich the power delivery board between the two main carbon fibre frame plates, we realised we didn’t have enough metal screws or M3 locknuts. And because those parts are the major load bearing parts of the drone’s structure, nylon screws and nuts were no good, especially nuts that couldn’t lock. That proved to be a good call, because on subsequent test flights it’s sheared the head off of a nylon screw I used to quickly fix the top plate in the rush to fly it again.
We drove around all of the usual hardware store suspects here in the UK, to find the right M3 screws and locknuts, but to no avail. The locknuts were key. We could use nylon screws here and there during the first test flights and order metal ones to arrive later, but we definitely needed the right lock nuts or for the particular frame I have, for fixing the motor arms, or it just wouldn’t survive any kind of a crash without coming to pieces. We couldn’t find them at Wickes, Screwfix or B&Q, so we pondered a trip back to RadioC to get them to top the kit up with the missing bits. Realising we wouldn’t be able to get there in time before they close, we put that off until the day after and decided to keep going with the build and do what we could with the parts we had.
Then we realised that the motors I’d picked as upgrades didn’t fit on the arms of the frame because of their hole pattern on the bottom. The pattern was rhomboid rather than the square. We’d need to either drill new holes in the carbon fibre — the dust of which does not sit well in the human body — or take it back to RadioC. Along with the missing frame fastening parts, we stopped for the day and decided to head to RadioC on Friday, another mutual friend, Mat, along for the laughs.
We arrived and talked to the RadioC guys (after eating cheese on an industrial estate — no, really) and explained what was up with the missing parts and the holes for the arms. While they drilled new holes, we talked parts as we went through what I was missing, and we walked out with an upgraded flight controller and some extra batteries, on top of freshly drilled arms and the frame assembled using the fastening bits missing from the original kit. It was easier to drill out the new holes on the arms that way. Finally, everything we needed to complete the build into the waiting frame! I was so excited to get home, get it done with Tim and Mat, and fly it before sun disappeared later in the evening.
I got the motors onto the arms with a little muscle-powered persuasion to get the screws through the arms. The holes RadioC had drilled for the rhomboid hole pattern were ever so slightly too small or out of alignment, depending on the arm, so I either filed the holes to be slightly larger using one of the M3 bolts left over from the kit, or screwed very carefully, effectively threading the carbon fibre of the arm between the screw and the motor. Then it was time to put the flight controller in place to get a handle on how our wiring would be laid out. We’d picked some right-angled PCB headers to connect to the bit of the flight controller that connects to the controllers that govern the speed of the motors, so that we could connect the speed controller wires horizontally line with the frame, rather than come down on top of it.
The speed controllers I picked as upgrades are the type that don’t have a +5V line coming out of them, which you can use to power the flight controller. So we needed +5V from somewhere else. Thus began the wiring adventure. Missing from the original kit I walked away with, RadioC gave us a voltage regulator that we’d supply with voltage from the battery, via the power distribution board, adjusted until it was outputting +5V. The power distribution board for the frame I bought starts with an XT60 battery connector, then routes that to well-placed pads at various spots near key bits of the frame. It also has support for a camera, for flying FPV, which starts at a (not supplied) capacitor to help smooth out the power. That’s because it suffers from feedback from the motors and can disrupt the camera signal before it’s transmitted back to whatever receiver you have on the ground, for your screen or FPV goggles.
There’s a pair of solder pads you can jumper to enable some extra filtered pads behind the capacitor. Since I don’t have the cap or the camera while I’m learning to fly it, instead we used the front camera supply pads as the input to the voltage regulator. It’s ahead of the flight controller mount in the middle of the frame, so it all lined up nicely and leant itself to cleaner wiring — not anything that happened in practice, but it will be nice when I rewire it soon!
With +5V coming out of the regulator, what we did next was to get that fed to the flight controller, followed by getting the speed controllers connected to they could be signalled properly. I’d mistakenly asked Tim to solder some pin headers to the flight controller, for power and speed controller connections, in the wrong way, which meant we didn’t have much length on the ground plane pins. Creative chopping of plastic to get the pin as short as possible to sit on the pin as much as possible ensued. The flight controller will control up to 8 motors, but we only have four of course, so 4 channels sit empty. +5V comes in on a motor/servo pin, so we hooked just one of those up.
Then it was time for the final bit of the build: getting the flight controller programmed and connected to the receiver, with the basic functionality tested, so we could attempt a first flight! Cleanflight is really two parts: a nice UI to tweak the settings, and the firmware to which the settings apply. You connect your controller via USB, Cleanflight can see it and reads the sensor data, then you can tweak a huge number of settings from what inputs your receiver gets from your transmitter, to ask the controller to move your craft, to the minute details of how the motors respond to your inputs and stabilise themselves in the face of that data coming in over the air.
I setup the flight controller, a Seriously Pro F3 Acro (short for acrobatic!), as best I could in Cleanflight, flashed it with the resulting firmware, then we connected a battery, paired the transmitter and receiver, and stood back to do the first test arming of the motors. I’d set it up so that turning one of the transmitters potentiometers to a certain range of values would arm the motors and ask them to spin up. Doing so would only spin up opposite motor pairs, without more throttle applied; I’d arm the motors and just two would spin up, then stop and the opposite two would spin instead. A quick adjustment of the minimum throttle values was all it needed. We had successful arming, so we went outside into the back garden for the first flight!
I put the quad down on the decking, connected the battery (a big 4-cell Li-Po battery that weighs almost as much as the rest of the quad put together!) and armed the motors. The props spun as they’d done in the kitchen…and then Maisie, my sausage dog, walked towards it for a sniff! My gut instinct was the right one and I disarmed the motors without thinking about it. We needed more space, so Mat drove and I carried the quad, transmitter and battery, and off to the park we went.
There were a few people hanging around, but my local park is pretty huge, so we found a quiet patch of grass with nobody within 50 metres or so. I did preflight, Mat started filming, and I throttled up slowly. It took off! It took off! Well, kind of. I panicked of course, cutting back the throttle enough for it to bounce a bit and tip towards the ground. And that was that; 7 seconds after takeoff (it’s actually 5-6 now that I’ve watched the video a bunch of times!) one of the propellors hit the ground, stopped, and then torqued against the nut holding it on to the motor. Those supplied nuts for the motor are dreadful, and it flew off into the grass — swiftly followed by the propellor I might add — never to be seen again.
Fin! Without spare prop nuts or tools, my little 3-day drone adventure was over, but it was capped off by success. The thing I’d spend all that time researching, buying, assembling, programming and testing with my friends had made it off the ground and into the air at the first time of asking. I haven’t felt elation like that in a long time, so even though we just got in the car and went home again after that one flight (and a fruitless search for the missing prop nut), I couldn’t have been happier. In my adult life, the most making I’ve ever done is Lego. Now that I’ve built the quad, including soldering and the electronics side of things, I’ve got the confidence to finally build some other things as well. An Intel Galileo sits in front of me as I type, waiting for me to solder sensors to it.
More on the quad/drone as I go! Here are the two videos we shot that day. The motor prime and the first flight. Yay!