They can't all be winners
I was inspired to try a fully 3d printed frame by this post in the Markforged site. Aside from brushing up my CAD skills, I saw several potential advantages to a 3d printed frame:
Thin aero arms for max thrust
Aero fuselage for minimal drag
Motor leads inside the arms, no more chopped wires
Elevated motor mounts to raise the CG, so you don't have to slam the components to get perfect Z-axis balance
All antennas inside the shell since the material is RF transparent (thanks Jason for this idea)
I went with a 4" frame and 1000 mah battery to keep things smaller and more manageable. Laying out the components, it made most sense to angle the battery – high in front to accomodate the fpv cam, and low in back to keep the whole thing slim. This also aligned it with the direction of flight. I was messing around with angled struts for Goldberg at this time, so I angled the arms so they'd block as little thrust as possible. This was before my arm shape thrust stand tests, so I angled the arms the wrong way, but this is easily fixable by reversing motor direction.
I would've loved to make a really smooth and curvy body like Mifune, but the battery on top dictated a straight mid section. The fuselage ended up being a series of extrusions and extruded cuts instead of an organic lofted shape. The whole thing was a 2mm shell.
The elevated motor mounts were fun to draw up. They have a bit of motor protection at the ends, but are open at the sides so you can grab the bell for prop changes. The cowling below the motor mount reinforce the mount area, 'cause I was pretty sure a flat 4mm 3d printed mount wouldn't survive a crash.
The first one was a stretched frame, but things didn't quite fit as well as I liked. It was printed in PCTPE (because that's what I had on hand) and was extremely floppy. It was also just wide enough to fit the FC but NOT the 4 in 1 esc. Doh.
The stretched layout relied too much on the fuselage for torsional stiffness, so I went to a pure X. I thickened the shell to 2.5mm, and added a 4mm base layer. I made the fillets between the arms and fuselage much bigger for a stronger joint. After several iterations I arrived at this.
I printed this in 645, which was a bit stiffer, but still way flexier than carbon. The build was like playing Operation, with the internal wiring and the 4 in 1 deep inside the shell, but I got pretty good at it with a pair of needle nose pliers. I excitedly set it on the floor and armed it, and BOOM. The motors instantly went full throttle and the rig hopped. If I was any slower to disarm it would've flown into the ceiling. What was going on? Did the arms flex at some frequency that perfectly matched the motors?
I tried a few other materials, even brittle PLA. The PLA print was simply gorgeous and just as stiff as carbon, but it chipped when the cat knocked it to the floor (feed me!). Armadillo was somewhere between PLA and 645, and seemed like a good compromise. But whatever material I used the same thing happened. I even managed to drill the ceiling once when my disarming reflex was too slow. Here it is in Armadillo, doing a little dance with the props off.
I drew up yet another version with a wider body so I could stuff the FC in there with nothing but foam all around it, and that didn't help either. Now I'm guessing it can't be a resonant frequency thing, 'cause I've tried too many materials with different stiffnesses (Stiffnesses is also Gary Oldman's porn name). The motors spin up smoothly with Betaflight, so it's gotta be a FC issue. I didn't have another Radiance on hand so I threw a Kombini in there, and...same thing. By now it had been 3 months and the endless failing was getting to me. I sent it to Chris 'Patient 0' Rahn to see if a fresh set of eyes could find the problem.
Chris blackboxed it and it seemed like the Radiance had a bad yaw gyro. He put a new Radiance in there and lo and behold, she flew. I was ecstatic but confused – what are the odds that I had two FC's with the same exact problem?
I'd just discovered NylonX for Babar, and I tested some pods with it. It's as hard and brittle as PLA, but it survived repeated spikes onto a tile floor.
So naturally I had to print a Micheru with it. I stopped the first print when an arm lifted off the print plate, and look what I can do to that failed print.
I got the quad back from Chris and gave it a quick test. With the FC loosely in the frame it shook just like before. I wanted to shoot myself but I told myself the NylonX frame was just as stiff as carbon and HAD to work. I moved the parts over and hovered it, covered in cold sweat. It had a higher pitched sound from the frame material, but it behaved normally. I headed out to maiden it with the only 2 1000's I had, and a couple really really old 850's.
It flew fine, if badly tuned, on the 1000's, but a punchout on a crappy 850 caused my video to go black and resulted in a crash. The frame mostly survived except for the top of the battery strap slot, so that'll be reinforced.
Since the frame was so stiff I decided to see how it'd do .5mm thinner. I made a fresh print and moved the parts over with a new 30 amp 4 in 1 (the original 20a 4 in 1 wasn't doing so well, no doubt damaged by all the high vibe testing). The shakes returned with the thinner frame, so violent that a motor came loose despite the use of loctite in the screws. So it was back to 2.5/4mm, and now it had a mild shake that was cured with some Betaflight filtering help from @ledboy. The fpv maiden the next day was thankfully uneventful: after five frustrating months I can't believe it's finally flying!
@ledboy, aka Matt Brooks, joined our club FPVAddiction last year and puts in more batteries than anyone. His partner Michelle saw his dedication and decided to build some permanent gates for our field. She did such a great job that she transformed our field into a world class track. Look at this field!
When she was done with that she decided she wanted to livestream our events, so she learned how to do that and bought the equipment. I commented that I should name a frame after her to thank her for all her hard work, and she jokingly replied that there should indeed be a 'Micheru'.
Back to our regularly scheduled programming
I took Micheru to a race and had a pretty lousy day. I wasn't confident a 1000 could last the 2+ minutes with the MultiGP format, so I flew it with heavier 1300's. I got midair'd and this happened.
Would it have survived with a lighter battery? We'll see. I moved the parts over to a print with one badly printed motor mount (the arm lifted off the platform, the supports got smushed into the mount) and the vibes returned. With some help from Chris Griffin I got a nicer print and it was flying normally again. So sensitive, this one.
Confident that the frame was viable, I sent the first sample off to @ledboy and he built it up right away. Matt had the same vibration troubles that I did, and crashed it while tuning it out.
Meanwhile, I kept flying mine, and I noticed that it'd react differently from day to day, and shake at different throttle values. I tried lighter props one day and it was unflyable. Then I discovered that the frame had developed a crack and my patience finally ran out.
I probably should've blackbox'd it and really gotten analytical with it, but there's only so many times a guy can get kicked in the nuts. Besides, it responded so differently from day to day that it was impossible to hone in on the problem. There's clearly something about this design that the flight controller just doesn't like, and maybe one day I'll figure it out (Does the enclosure amplify vibrations? And if so, how do Phantoms fly?). Meanwhile, it's off to the tub with Micheru. I really want to move away from 2 1/2-d designs, but for now carbon is just too damn good. Maybe someday there'll be a filament that can handle this.