I've been getting more into aerial photography lately. Niche AP rigs like the Ichabod and Squirt are real fun frame design challenges. I've also been working as Randy Slavin's camera operator while he flies his Alta 8, and that's given me a new appreciation for dual operator AP rigs and LOS flying. Then I was invited by Charles Zabian to be on an AP panel for Canon, where I got to meet some industry leaders in drone design as well as content creation. Even though I was a panelist at the event I ended up gaining way more knowledge than I dispensed.
While at the forum I was challenged to make the smallest possible dual operator AP rig. As far as I know the DJI Inspire is the smallest dual op rig out there, so it'd have to be smaller than that. And given how good DJI stuff is these days, this rig needs an angle for its raison d'etre.
Two things sprung to mind immediately. First of all, it'd be ducted. Ducts generate more thrust, so you can use smaller props for a given payload. They also double as prop guards so you can safely fly close to people. Secondly, this will be a rig for us FPV'ers, not the general public, so it can be piloted with way more precision than a DJI.
The first thing I had to do was revisit duct design. I've made several ducted quads before, but I've always reduced the duct size to make them work with the overall frame. This time I wanted to start with the ideal shape (which I stole from here) and gradually pare it down to see how big it really needed to be. I drew up the ideal 4" duct and a carbon arm to support it, and made a thrust stand for it. I tested a 4 blade prop and a 3 blade prop, and the tri was much better. Then I trimmed off about 15mm from the bottom of the duct and tested it again. Then I trimmed off the outer half of the lip and retested. Finally I removed the duct completely. The numbers are here, under day 2 (ignore day 1 I had an esc fire!). I used the same two batteries for each configuration, and started logging the thrust when it reached its peak, then took 5 measurements at 10 frame intervals.
Based on those tests, I determined that it was worthwhile to keep the full height of the duct, but not the outer half of the lip. Then I realized that I screwed up and made the walls slope out at 10 degrees instead of 5, so I redrew the duct and designed a full frame around it. As you can see, the duct is braced on four sides to better maintain its roundness. This is critical for keeping the gap between the prop and the duct as small as possible.
I wanted it to be as modular as possible, so I added battery strap slots above and below. As with the other ducted frames, the ducts are slightly smaller than 4" around so that you have to grind down the props for that tight tolerance.
When the carbon arrived I hurriedly bolted an arm to a duct and had a nerdgasm over the way the two meshed perfectly.
Next I mounted it to the thrust stand to see if the 5 degree walls perform better than my 10 degree mistake, and was very happy to see that it did (those numbers are under day 3). Now I had a full 250g more thrust per motor, or about 5kg total thrust for a drone that should weigh around 1kg. Time to build.
I cobbled it together with some old Brother Hobby 2206 2600's, a Fortini F4, and Aikon 4 in 1. The unusual thing about this build is the way the wires have to go through the duct. I wired the esc to Racewires, then mounted the duct, then mounted the motors and soldered them to the Racewires. It'd be a PITA to repair, but it's super clean. You could also route the wires through and solder direct to the esc, or use bullet plugs.
Then I drew up a super simple fpv cam mount that bridges the mounting points on the two front ducts.
The rest of the build was pretty straightforward and I was hovering it in my apartment at 4am. At 650g, it was surprisingly powerful and spry.
The next morning I strapped a 3700 to the bottom just to weigh it down, bringing the auw up to 1075g. As you can see it hovers at 50%.
I fpv'd it for a quick lap around the apartment and it flew quite nicely, so it was time to strap a gimbal to it. I'd bought a Tarot ZYX T-3D 'cause it had the option for sbus control. The manual's a bit...opaque, so the first thing I did was to just hook it up to power and let it work passively. The Tarot has 60mm wide rail mounts, so I bought some 10mm carbon rods and drew up rod holders. The holders can be fairly floppy on their own, as they stiffen up once the rods are inserted.
We usually fly the Alta with the gimbal below, so that's how I tried it here first. There's no landing gear here, just a big ol' gimbal sticking out the bottom, so I set up a pair of C-Stands as takeoff/landing rails. The takeoff was really shaky, so I caught it by the battery instead of trying to land it on the rails. As you can see, the catch was a bit hairy, but the quad worked! The ducts were creeping into the edge of the video, so maybe the rod holders need to be extended for more clearance.
I also tried it mounted to the top. The gimbal's meant to be bottom mounted, so to make it work as a top mount you have to mount the GoPro upside down and backwards, which makes it unbalanced on pitch. Fortunately the motor seemed to handle the imbalance just fine. The result was about the same – ducts in view, needs taller rod holders.
Then it occurred to me that I could move the gimbal way forward and get unobstructed video, I'd just have to extend the rods fore and aft so the battery can hang out way back as a counterweight. So I cut two longer lengths of the rod and drew up a battery mount. I'd also made taller rod holders which I repurposed here as landing gear to spare the ducts the impacts of landing.
And here's a quick test, no ducts in view, but there is a Corgi.
Now I had something pretty similar to my old Gimbal'd 540, except much smaller, lighter (1050v 1660g), safer, and with a 3 axis gimbal (in passive mode it damps yaw movements). To be fair, the big one is probably faster and better in the wind. I dug up its broken down carcass for a size comparison.
I took it out to the Bronx and fpv'd it. It flew great on stock pids, the only downside is that it catches so much wind you definitely know which way the wind is blowing. I got 4 minutes easily on 1300's, with 5 possible if I kept it slow and smooth. Everything seemed perfect 'til I got home and reviewed the footage.
So, it turns out the Tarot is a pretty sweet gimbal, in that its IMU is nicely integrated into the gimbal itself, as opposed to the old gimbals I used which had a separate IMU that you'd tape to the GoPro. The downside of this nice integration is that there's wires running through the arms, so they can't rotate the full 360 degrees lest the wires get too twisted and eventually severed. This is fine if you use the gimbal in its intended orientation, since the dead spots are pointed straight back. I had the gimbal mounted upside down, so I had only about 20 degrees of uptilt before it hit a hard stop.
In the video you can see that as soon as I sped up or dove the camera would hit the stop and eventually freak out, yielding a hilarious view of the duct.
Other than that I thought the test was great. If I was really filming something I'd have to get way smoother on the yaw stick, but you can see how this thing could really shoot some nice video.
So, this configuration can only be a slow flyer, which is not ideal. The logical thing then is to find a way to mount the gimbal properly, preferably in a way that doesn't require hand catches.
So now things are getting kinda crazy. Pushing the camera out front gives a nice unobstructed view, and lets me hang the gimbal in the right orientation. But it spreads the weight way out, forcing the battery far back to counterbalance the gimbal. And with the weight so spread out, a wider stanced landing gear is needed. This will make pitch and yaw super sluggish, but in this application that may not be the worst thing in the world.
I originally wanted to draw new struts that angle out wider side to side, as well as fore and aft, but this seems like a simpler solution. The rods don't have to be that long, but I didn't want to cut them down since this length works well for top mount applications.
I tested this configuration on my kid on a windy day, and it exhibited some weird flight characteristics, freaking out on yaw. Between the wind and the tight quarters I didn't fly so well, but I did get a few snippets.