Fancied the idea of building a flying replica of a drone from a game or film for ages (the ones in Detroit: Become Human were pretty rad but at the time I didn't have the resources to tackle one); after getting pretty into Arknights (a tower defence gacha mobile game) I realised one character, Silence, who uses a drone to heal teammates could be a fun project. Whilst the drone in her standard art would be a real challenge to make fly, her alternate outfit has one that looked quite doable.

With that decided, I hit up Solidworks. For probably about 6 weeks, on and off. Taking the design from a single stylised image to a working flying machine was more work than I'd anticipated - not least because I wanted to make use of the mechanical aspects of the design, and opted to turn the little linkages above the arms into a functional landing gear system.

To get the approximate sizes, I dropped the image from the game into Illustrator, scaled it so the propeller disc was about 6" diameter, and then measured everything else relative to that (not the easiest task due to the liberal use of perspective in the art!)

The final design was workable, if a very difficult build. Essentially, ignoring the plastic top and bottom aesthetic plates, it's a carbon sandwich, with the flight controller in the middle and the ESC under the bottom plate. However, since the motor wires come through the bottom plate, once the ESC is soldered up, access to the FC is impossible. Luckily this isn't likely to see much flying period, so I can live with that.

When it came to actually making the parts, I opted to work around what materials I had on hand, leading to it being somewhat overengineered from CNC routed carbon fibre in 1.5, 2, 2.5, and 5mm sheet. The legs were made from 2mm delrin, and the top and bottom plastic parts were made from 2mm clear perspex, both again on the CNC. 3d printed parts were made from mostly PETG in Vertigo and copper colours, and yellow PLA.

To make the functional landing gear, I came up with a system where an elastic band would pull the leg down, and a servo with a pulley wheel connected to the leg with a linkage would pull it up. Initially I used small elastic hair ties, but they proved fragile and not elastic enough, with most permanently stretching in short order. As a result, I ordered some proper rubber o-rings (it took 3 attempts to find a size that worked well).

Whilst initially I thought I'd have loads of room to work with, as per usual I was completely overestimating the size of the build when viewing the CAD on a screen, and in the end I needed submicro servos - fortunately I was able to find some good quality metal gear digital 5.7g options which work nicely. I used the included round servo horns, and sandwiched carbon discs on the top and bottom to turn them into short throw winches. 1.5mm dyneema cord was used to connect each winch to the carbon linkages, and nylon washers were used on all moving parts to reduce friction and keep it all moving smoothly.

As for other components chosen, I picked an Omnibus F4V6 as it's one of my favourite FCs, and I figured it would be easy to get it to work with the servos. I wasn't convinced the onboard BEC could run the servos and Rx safely, so added an external regulator which I wired back to the servo rail on the Omnibus (which isn't connected unless you solder the 5v or rx jumper, fortunately). I also made an external servo pdb to keep things neater, using stripboard.

All in all, the build was tricky but not impossible - no worse than some of my race builds. Thankfully the tolerances were spot on for all the interlocking carbon parts and they slotted together first time. One of the trickiest parts was the top and bottom plastic parts - to get the logo, I cut the shape out of vinyl on my Cameo 3, but the lettering was so tiny that getting a good result was a mission. Weeding and application was also a joy!

Once it was all set up, I tried a test hop and was pleasantly surprised that it all worked first time - though it was quite a handful, both being a very heavy quad and with my forgetting to set stabilisation on a switch and to change rates. Stock betaflight acro rates are not ideal for a build like this inside!

So yeah, that's how I spent my lockdown - putting hundreds of hours into a LOS only quad that I'll probably not fly again. Worth it!