3D Printed 50mm F-104 Starfighter

[telnar1236]

I designed the nose gear, nose gear mount, and nose retract control linkage over the last few days. Because I am using a mechanical retract, the top of the nose gear must be fixed, so the rotation point and control horn must be integrated into the strut further down which introduces complexity. In addition, because the retract is in front of the servo driving it, but also retracts forward, the control linkage for the servo has to reach around the retract.

The control horn on the nose gear is just a placeholder currently and would not work in its current configuration and the control rod from the 3d printed retract linkage to the servo is not modeled. Steering will be driven by a pull-pull connection to a bell crank which will be driven by the rudder servo, which avoids all the concerns about the steering linkage preventing retraction and the control horn on the retract unit moves inward when retracted so it won't interfere with the steering control horn.

Finally, I found some 30mm wheels and they fit better for the main gear. The nose gear wheel is 25mm and I may change out the main gear wheels to be the same.

 

[XSrcing]

I was going to say. Dubro makes a lightweight 25 mm wheel/tire.

https://www.dubro.com/collections/tires-wheels/products/mini-lite-wheels

 

[telnar1236]

I was going to say. Dubro makes a lightweight 25 mm wheel/tire.

https://www.dubro.com/collections/tires-wheels/products/mini-lite-wheels

Yeah, I have a couple of those sitting around. They're good wheels. Unfortunately, I want a 2mm shaft to make sure nothing breaks and 1.2mm is just too small.

I ended up using this: Amazon.com: MroMax RC Airplane Wheels 0.98",1Pcs Foam Wheel Replacement for DIY RC Model Plane Aircraft : Toys & Games

 

[XSrcing]

Yeah, I have a couple of those sitting around. They're good wheels. Unfortunately, I want a 2mm shaft to make sure nothing breaks and 1.2mm is just too small.

I ended up using this: Amazon.com: MroMax RC Airplane Wheels 0.98",1Pcs Foam Wheel Replacement for DIY RC Model Plane Aircraft : Toys & Games

A 1.2mm hole is just a hole that hasn't hit puberty yet.

Mtsooning 10PCS 2mm Mini HSS Drill Bit Straight Shank Twist Drill Premium High Speed Steel Jobber Length Drill for Electrical DIY Rotary Hand Tools https://a.co/d/1kNdjqb

 

[telnar1236]

A 1.2mm hole is just a hole that hasn't hit puberty yet.

Mtsooning 10PCS 2mm Mini HSS Drill Bit Straight Shank Twist Drill Premium High Speed Steel Jobber Length Drill for Electrical DIY Rotary Hand Tools https://a.co/d/1kNdjqb

You could drill it out, but it definitely messes with the bearing surface and increases friction. It wouldn't really be a problem, but it's also just easier to use a wheel that fits without modification.
For bigger 3D printed planes, though, you don't want to manually drill wheels. I actually have had problems with the heat from a wheel heating the metal axle enough to melt the PLA gear struts (it's very hot where I live already, and it was a hot summer day so the heat from the asphalt certainly didn't help).

 

[XSrcing]

Drill the hole oversize and press in a bronze bushing.

 

[telnar1236]

I designed the canopy and the control horns for the nose gear. There is a version of the canopy that can be 3D printed in clear PLA and a version of the canopy that is vacuum molded. I plan on going with the vacuum molded canopy, but the PLA canopy is simpler to make and install.

The nose gear assembly should be pretty straightforward using the pictures as reference. The control horn is inserted and glued in place after the lower part of the gear assembly is installed into the upper part and holds the lower part in place. The two pins for the nose gear are both 2mm and the lower pin is glued into the lower part but not the upper. The pin for the wheel is also 2mm while the pin that connects the retract linkage to the retract is 1mm.

 

[telnar1236]

The nose gear assembly is now printed and assembled. The upper half of the strut (Strut1) as printed interferes very slightly with the control horn for the retract unit when the gear is up, so I shave a bit off of it and it now fits well. The STL file is also updated to prevent this problem. Otherwise, everything fits together and seems to work nicely. The Y-shaped NoseGear_RetractLinkage cannot be glued to the metal pin (paperclip) that connects it to the retract control horn until the fuselage sides are assembled since it needs to pass through a hole in each side. It will probably also be a good idea to make collars to prevent the linkage from moving side to side relative to the control horn.

 

[telnar1236]

The halves of the plane are joined together! This is one of the biggest milestones on the way to completing the aircraft, and I had hoped to get here about two months ago. Mostly everything works fine together, with the rudder being the major exception. I think the internal control horn was just two small, so it has far too much play and I'll probably just pin it in place and just have a steerable nose gear if I can't fix it pretty easily. At this point there are just a few minor connections to make in the gear bay between control rods and the nose gear steering mechanism and canopy to install, and then I'm ready to try and fly!

 

[Crash Test RC]

That is impressive.

 

[telnar1236]

That is impressive.

Thanks! Honestly it's been too much of a pain for what I'm getting out of it, so I'll probably go back to slightly easier to build projects in the future

 

[telnar1236]

Since I've already spent 3 months on this project, what's a bit more time? I really want to have a working rudder, and the control horn does not provide enough leverage even after about an hour and a half of increasing the stiffness and constraint on the control rod (as a plus, the elevator is a bit stiffer now, though it was fine initially), so I decided to redesign the control horn and put it in a fairing. It goes a bit against my plan of no external control horns, but at least it is faired so the aerodynamic penalty should be small.

 

[Piotrsko]

If it's any consolation, the 104 on a stick with the reaction controls outside of KEDW test pilot school had such a fairing on the rudder for what I understand to be the same reason

 

[Inq]

@cdfigueredo reminded me of these things that have been around for decades that eliminate the horns completely. I'd imagine you could come up with a 3D printed version fairly easily. https://forum.flitetest.com/index.p...ear-driven-control-surface.71204/#post-734046

 

[telnar1236]

If it's any consolation, the 104 on a stick with the reaction controls outside of KEDW test pilot school had such a fairing on the rudder for what I understand to be the same reason

I had forgotten about that one. Is this the plane you're talking about? I think that's a rocket in the tail. It would be pretty easy to make an attachable fairing in that shape, so I might do exactly that. Thanks for pointing out the similarity.

 

[telnar1236]

@cdfigueredo reminded me of these things that have been around for decades that eliminate the horns completely. I'd imagine you could come up with a 3D printed version fairly easily. https://forum.flitetest.com/index.p...ear-driven-control-surface.71204/#post-734046

I'd never seen those before. Because of the way my F-104 is designed, I would need to drastically change both the location of the rudder servo and the design of the tail to implement those, but I'll probably use them in future designs if I can make a reliable 3d printed version.

 

[Piotrsko]

Yup that be the one, but I see they moved it. There's also thrusters in the nose and a couple of side thrusters just because surfaces don't seem to work above 100,000 msl.

 

[telnar1236]

Yup that be the one, but I see they moved it. There's also thrusters in the nose and a couple of side thrusters just because surfaces don't seem to work above 100,000 msl.

That's why I like building and learning about aircraft from the 1950s and 1960s so much. They were still working stuff out, so if, for example, you needed to figure out reaction controls, strap a bunch of them to a fighter jet, strap a rocket on the back, and try them out. Definitely a different way of doing things than today.

 

[Inq]

That's why I like building and learning about aircraft from the 1950s and 1960s so much. They were still working stuff out, so if, for example, you needed to figure out reaction controls, strap a bunch of them to a fighter jet, strap a rocket on the back, and try them out. Definitely a different way of doing things than today.

No kidding... $283/month!

...these days... about a $billion in computer time, another in wind-tunnel time and another in drone tests.

 

[telnar1236]

I had a ton of time sitting around away from my workbench and not able to go do stuff, and I discovered that a professional CFD software, Sim Flow, has a free version that can evaluate up to 200,000 nodes, so I decided to run CFD analysis of my F-104. I won't get overly technical here, but 200,000 nodes was barely enough to get an ok model and run it (ideally, I would refine the mesh a decent bit more). To keep the node count down, I just ran half the airplane with a symmetry boundary condition, and then doubled the values which is ok across the allowable range of pitch angles but not ok for anything but 0 degrees yaw. The good news is that it agrees very well with the prediction from XFLR5 which I used when designing the airplane (XFLR5 is a great design tool and requires much less technical knowledge to use well than full-on CFD). I knew from past experience that for low aspect ratio wings, XFLR5 stops converging well before the wings actually stall, that it over-predicts lift by a bit for mid-wing aircraft, and that it badly underestimates flap performance, all of which were true in this case. For example, with full flaps, the wing stays effective up to about 10 degrees angle of attack in the CFD model where convergence fails at 5 degrees in XFLR5 (drag gets high enough that the plane cannot sustain level flight before 10 degrees angle of attack, though), and at any given angle of attack, the wings produce at least 10% more lift using CFD.

I stopped running CFD at 15 degrees angle of attack, because pitch-up gets pretty bad by that point both in the CFD model and from experience, but I know from experience (and repeated crashes) that the F-104 wing continues to produce additional lift to around 20 or 30 degrees angle of attack.

Another big advantage of CFD is that it gives much more accurate drag numbers than any hand approximation. From XFLR5, I knew that my design could turn at at least 2g at 50 mph, but from CFD I get that my plane can turn at up to 2.6g and sustain a turn at about 2.4 g at this same speed.

Now the obligatory pretty CFD pictures (pressure is in m^2/s^2 and velocity in m/s):