Randy Slavin created a 'Day of Drones' at the Liberty Science Center as part of a weekend of events for his NYC Drone Film Fest. We did some racing outside as kids watched on monitors and goggles in a special viewing room, and there was also a netted off area inside for combat and demos. I brought my Blast for some target practice on plastic cups, and it was a big hit. It was hilarious to watch the disks fly through the netting and (harmlessly) hit spectators while LSC employees scrambled to retrieve them like ball kids at a tennis match.

LSC CEO Paul Hoffman is a huge fan of drones, and the event just made him even more enthusiastic. He asked us to design something perfect for that venue, something that kids could learn to fly on that would be really durable and not get tangled in the netting. My initial thought was a simple H with animal shaped canopies and DAL unbreakable props, but didn't satisfy the 'not get tangled' requirement. There seemed to be no way around prop guards. 

Which of course brings us to Butters, but I wasn't sure Butters could survive long term abuse in the hands of rambunctious children. Besides, I hate to miss an opportunity to design something new. So I sat on this for weeks hoping inspiration would strike, and it finally did when someone mentioned ducted fans. Now watch these two videos and I'll catch you on the other side.

I've been curious about a ducted fan quad for a while now, but I was certain one could never perform as well as a regular quad, so I never bothered to make one. But now that I needed to make a slow stable LOS quad with prop guards, this seemed to be the perfect opportunity. The first thing I did was print a tiny duct to see if TPU (my new best friend) would work. It turned out really well, even at only 1mm thick, so it was time to design.

My idea was a carbon central structure that would hold four ducts using standoffs. To save weight the center would only grip the inside half of the ducts, leaving the outside half to act as flexible bumpers. Of course, as they deformed they would rub the props, but I was hopeful that the motors would be able to power through that and keep the quad aloft. Here's the first sketch:

Next I read this and trusted that it would be correct, 'cause after all it was on the internet. I drew up a 35mm tall duct profile for 4" props with those parameters (ideally the duct should be way taller than 35mm, but Google searches showed lots of ducts that aren't as tall as they should be, so I took another leap of faith there). I printed several tests 'til I found what I felt was the minimum thickness necessary for a duct that would hold its shape, and it was on to the next step.

With the duct shape locked in the rest was just a matter of connecting the dots. The overall size was governed by the minimum distance between the ducts that would leave a strong enough piece of carbon. The only oddity was having to offset the middle mounting point of the ducts so they didn't interfere with the corners of the fc stack. A 45 degree fc stack would've fit in there real nicely, but then there'd be no usb port access.

I'm always excited to build up a prototype, but I was especially excited by this one. I was dying to know if my crude TPU ducts could indeed increase thrust enough to make up for their extra weight. The build was relatively simple, I had to do the reacharound to fit Littlebee 20's on the relatively short arms, but other than that it was straightforward and extra easy due to the lack of fpv gear. The props rubbed at first so I gave the corners of the bullnose tips a little sanding and let the motors do the rest. The gap was nice and small, at around 1mm or less.

Since this frame is shaped like Butters, with lots of mass out wide where normal quads have nothing, I started with roll and pitch P at 7 as a guess. When I fired it up I was pleasantly surprised to find that the 603 gram quad hovered on just 25% throttle – the ducts work! I could pop the throttle and it would recover from rapid descents almost immediately, like a much lighter quad. With a minimum throttle of 1050 it was tough to land, shooting up as soon as it tapped the ground, a sign that P was too high. I eventually brought it down to 4, which is very surprising to me. It seems the motors have enough thrust to have pitch and roll authority with default pids. I dropped min throttle to 1020 and it got less jumpy on landing. The next day I took it outside and dusted off my terrible LOS skills.

I wasn't able to get it to right itself in forward flight, so it seems the flight controller can easily overcome any self leveling tendencies of the ducts. I never expected it to be so nimble, so next, of course, I had to put fpv gear in it.

There were no mounting points on the frame for fpv gear, so I drew up a 35 degree cam mount that wrapped around the duct standoff sleeves, and an antenna mount that snapped into a top plate cutout. I also made a vtx holder for a pigtail vtx, but it burned out before I even got a flight in, so a 90 degree sma vtx sits under the top plate.

FPV testing was slow going, for a couple reasons. One, I'm working on so many prototypes at the moment, and I can only take 2-3 quads on my bike rides to the field. Two, I'm having a hard time tuning it for FPV, for some reason I can't get it to fly fluidly FPV even though it's pretty smooth LOS. But I am improving my LOS skills, which helped my throttle control in FPV quite a bit.

In the meantime I built the two quads for the Liberty Science Center, in nice patriotic colors. It was cool to build two more and rev the props against the ducts as they ground away their own clearance.

I've been real busy working on multiple prototypes, so I haven't been able to name this baby and create graphics. So here's a slow rollout with the entry in the store, and a build log.

But Wait There's More

This baby was a fun science experiment, I never expected to sell more than a handful. What surprised me was number of requests for adaptations of this frame. I spaced out the ducts for a version to hold a Pixhawk for a tech firm, so that it could play Simon autonomously. I also made a 6" version for a falcon trainer to hover and dangle targets from.