Scratch Built Balsa F-104

[telnar1236]

The fuselage is now partially assembled and the ribs for the wings have now been cut. This is the first attempt I am making to use the laser cutter to manufacture something that could not practically be made by hand. The thickness of some of the parts of the ribs is less than 1mm. Despite this, the ribs are quite strong in the axes in which they need to support a load. I have also added geodetic ribs to the original structure to increase torsional stiffness and help ensure the airfoil is easy to cover. The fuselage now has the alignment jig for the nose integrated into it's structure. This speeds construction and provides a better mounting point for the nose gear. The shape is also slightly changed to improve appearance. I changed the routing for the stringers so that the main load bearing ones would follow a more straight path to reduce the chance of buckling, although the thinner stringers meant to support the skin still follow curved paths to achieve the correct contours. It may not look like much now, but the result is a much neater fuselage.

 

[Taildragger]

A few days ago I decided I wanted to build an F-104 EDF. Because I don't have too much space to fly, it needed to be as light as absolutely possible, which meant balsa and plywood construction. At this point I've designed everything in CAD and laser-cut and assembled the fuselage. The fuselage formers and some internal decks are cut from plywood while the stringers and airfoil formers are balsa. The internal duct is made out of resin impregnated painters tape.

Some specs (I used metric because it's easier to do the math with):
1.1 m length, 47 cm wingspan (not quite scale, but it will help the plane slow down and you won't notice the difference)
Hyperflow 56 mm fan w/ 8.6 N (880 g) thrust, 40 A ESC
Airfoil will be Clark Y for wings, NACA 16009 for the vertical stab, flat plate for the horizontal stab
On paper top-speed in SLF is about 38 m/s (85 mph) which is about what I'm aiming for
On paper stall-speed is about 10.5 m/s (23.5 mph) which is a bit fast, but not too bad for an F-104.

View attachment 185495 View attachment 185496 View attachment 185497 View attachment 185498

Oooh, pretty laser... I just want to touch it...OW! Do. not. touch. the. pretty. purple. laser.

 

[DinosEatPeople1]

Oooh, pretty laser... I just want to touch it...OW! Do. not. touch. the. pretty. purple. laser.

lmao

 

[telnar1236]

Unfortunately the current wing design is too fragile to assemble. Guess there are other reasons that you don't see many planes with such thin members in their wings. Based on this I'm forced to change the wing design yet again.

 

[telnar1236]

Here's the new F-104 wing design, now with more structural support to help it deal with bending loads during assembly. I learned the lesson that you need to design for the largest load a component will see at any point in its life and not just during intended use a long time ago, but I keep getting reminded of it. The wing still makes use of a geodetic structure, but rib layout is simplified to make construction easier. I have also included two image files with the plans, one with labels and one without. Each is 300mm x 180mm.

Plans:

 

[2jujube7]

Here's the new F-104 wing design, now with more structural support to help it deal with bending loads during assembly. I learned the lesson that you need to design for the largest load a component will see at any point in its life and not just during intended use a long time ago, but I keep getting reminded of it. The wing still makes use of a geodetic structure, but rib layout is simplified to make construction easier. I have also included two image files with the plans, one with labels and one without. Each is 300mm x 180mm. View attachment 195326
Plans:
View attachment 195327 View attachment 195328

wow, that's a lot of spars and different airfoils that you've CADed out!

 

[telnar1236]

wow, that's a lot of spars and different airfoils that you've CADed out!

Using devWing means the work isn't too bad since there are a fair number of tools to streamline the workflow. You define panels in the wing, then once you have the .dat files for the root and tip sides (you can download these from airfoil tools), the software interpolates and defines any intermediate ribs for you. It is still a fair bit of work making sure everything will properly fit together and devWing is horribly buggy, but with all the tools it only takes maybe an hour and a half to get a complete wing CAD model and another hour to clean everything up enough and combine everything into one sheet for laser cutting. Most of that work is checking part alignments, though, and making sure that devWing hasn't done anything too stupid with locating spars inside of each other or generating some strange geometry that doesn't make sense. Best part is that the demo version, which is what I use, is free, which makes designing very accessible, even if you don't have an engineering background. In terms of the FreeCAD model, the demo version of devWing doesn't let you export geometries, so I just modelled it as a solid without any of the ribs or spars since I didn't want to spend hours modelling the the individual ribs when I already had adequate .png files to cut.

Link to devWing and devFuse:
devCad, devWing, devFus, devFoam, devCnc Foam, devStl Tools, devFont Download

 

[telnar1236]

I finished the wings' main structure and redesigned the vertical stabilizer to incorporate the necessary changes to allow construction. The wing is very light with the structure weighing in at only 50g now that it is assembled while also being exceptionally strong.

Pictures of the wings:

CAD and plans of the tail:

I decided to bite the bullet and take the time to actually model the wing using Xfoil to analyze flap performance and wing characteristics. This let me compare the Goettingen 286 airfoil and the AG35-AG37 wing. While the Goettingen airfoil without flaps produces much better lift and offers a much wider angle of attack range without flaps, the AG35 airfoil (the root airfoil) with flaps offers only a slight loss of performance when compared to the Goettingen with flaps. Thus, I was able to make the choice to go with the thinner Drela aifoils with confidence. Xfoil is a great tool from Mark Drela at MIT for airfoil analysis, and it is free, but using it can get quite technical quite quickly, so just be aware of that. However it is not too difficult to use and it provides useful tools for analysis, design, and flow visualization.

Pictures from Xfoil:

 

[telnar1236]

The vertical stabilizer is now cut and assembled.

 

[telnar1236]

With the wings, tail, body, and EDF unit integrated, the F-104 is starting to take shape. The thrust tube is a new fiberglass layup. Based on the weight of all the individual components, the newest version of the F-104 is on track to weight in just under 800g ready to fly.

 

[TooJung2Die]

That's looking beautiful.

 

[Piotrsko]

Wash in on the wing tips?

 

[telnar1236]

Wash in on the wing tips?

There isn't actually any wash in on the wing tips, but it does look like that in the photos for some reason. The entire wing has a slightly positive (2 degree) incidence angle relative to the fuselage which optimizes for minimum drag in SLF (including the fuselage and empennage) at 38 mph with the heaviest battery I plan to fly with. As I will typically be flying faster than 38 mph, this lets me fly in a consistent region of velocity stability without compromising too much on top speed. With flaps down for takeoff or landing, the effective AOA and CL of the wing increases which lowers the best L/D to ratio to 29 mph for landing and 31 mph for takeoff settings which is about as low as practically possible. I think that the positive incidence angle combined with the 8 degrees anhedral makes it look it has wash in in the photos. I debated trying to build in 1 degree of washout in the wingtips, but I wasn't confident in being able to make the wings precisely enough without a complicated jig, and asymmetric washout could cause very bad stall tendencies. To control the stall and help avoid tip stalls, the airfoil changes along the span of the wing from an AG37 airfoil at the root to an AG35 airfoil at the tip. The AG35 airfoil has a higher stall angle, but since it has a smaller chord than that at the root, the stall angle is only barely higher than at the root meaning that the whole wing stalls fairly smoothly. The stall is not all that gentle, but lift doesn't fall off too badly for a couple of degrees which gives the pilot a few moments to recover. The choice to go from AG37 at the root to the higher lift AG35 at the tip does mean that the lift doesn't follow the Prandtl distribution quite as well as it might, but the wing is already highly tapered and with the high wing loading, induced drag is going to be a problem no matter what I do. I know this is a very long answer to a pretty simple question, but since I spent far too long designing the wing, I figured I might as well explain the whole thought process.

 

[telnar1236]

Assembly continues. All the electronics are integrated, so it can now rest on its own gear. I am far happier with the strength of the gear and the gear stance than I was with V1. The new flap design is also attached. I modified my previous design for slotted flaps to make it far neater and more efficient, but the downside was that the flaps need to be integrated prior to covering the wing. It's a bit heavier than I expected, but the remaining components are a known quantity, so I can say with confidence that the final version will weigh in at between just under 900g and 1000g, depending on the flight battery, which is right where the original design was meant to weigh in. I'm a week or two out from finishing now, depending on how busy I am with life.

At this point I am debating the merits of balsa and covering film against tissue again. I think the most likely path will be balsa sheeting on the fuselage and covering film on the wings and empennage, which is what the weight numbers I gave are for, but I could still choose to go either way. With balsa covering, the stall speed is 28.5 mph (calculated with improvements in lift from flaps, but not accounting for the lift enhancing effects of the slots), and with tissue, it is 28 mph, so the reduction in weight only gives a tiny improvement in handling characteristics, meaning it is largely an aesthetic decision. Depending on how I model the slots, that could drop the stall speed as low as 25 mph, so I'm hopeful, but I've been practicing extremely hot landings in case they don't offer much improvement, and worst case I can belly land in the grass.

 

[telnar1236]

The fuselage sheeting is now about half finished. Obviously, once the whole thing is covered, it will need to be filled and sanded. Covering is going faster this time than with previous versions because of all of the thinner longerons. Other changes since the previous post are that the internal ducting is finished, and the vertical stabilizer's leading edge is properly sanded to shape.

 

[telnar1236]

And the fuselage is now fully sheeted. The nose cone, canopy, covering for the wing and tail, a few finishing touches to the same, the gear doors, control linkages, horizontal stabilizer, and some aesthetic touch-ups are all that remain before the airplane is flight capable. Depending on the weather, I may or may not paint her prior to the maiden flight.

 

[Piotrsko]

Interesting choice of airfoils, much better than the full size, still looks to have an interesting stall but I have no experience with that max camber location. Does it work like a LE flap or more like typical underamber wing? Xfoil plots didn't help me much (no suprise there).

 

[telnar1236]

Interesting choice of airfoils, much better than the full size, still looks to have an interesting stall but I have no experience with that max camber location. Does it work like a LE flap or more like typical underamber wing? Xfoil plots didn't help me much (no suprise there).

The airfoils are pretty typical flat-bottom airfoils. I'm not totally sure about what you're asking about for the leading edge flap, but the airfoils are slightly under-cambered. I chose them partially because of how low the drag coefficient they offered was, and partially because the two thicker airfoils I evaluated (Clark Y, and Goettingen 286) did not offer large improvements in the regime where I normally intend to fly. The F-104 has a dangerous deep stall because of the T-tail configuration, so exceeding about 10 degrees alpha is a bad idea no matter what airfoil you're using. Keeping that in mind, the max lift coefficients for many thicker airfoils are only a bit better (1.4 at 10 degrees alpha, vs. 1.2 at 10 degrees alpha for example, so stall speeds of 30.4 mph vs. 32.8 mph w/o flaps), and at 8 degrees alpha, the lift coefficients are pretty much the same. Moreover, the ability of the thicker airfoils to reach higher angles of attack dramatically increases the risk of accidentally going into a deep stall. The flaps are 75% of the way back along the chord line, and are slotted, which should dramatically improve performance on the wing I chose to use. With flaps, the Xfoil predictions for the max lift coefficient of the airfoils I analyzed also shows a limited benefit from going to a thicker airfoil (1.74 vs. 1.8 with 30 degrees flap deflection). The thicker airfoil can achieve this over a wider range of angles of attack, but using a slotted flap setup reduces this advantage while also increasing the lift coefficient, and with slotted flaps the thicker airfoil also only has a small advantage. The thin airfoil is also markedly lighter, which offsets some of the benefits of a thicker airfoil for stall speed and maneuverability, although it doesn't negate them fully. While a higher lift airfoil will typically trim at a lower angle of attack reducing drag, the F-104 is capable of going fast enough that the wing trims at a negative alpha, meaning the opposite is true, so the lower drag coefficient of the thinner airfoils is that much more pronounced (at the calculated top speed of a bit over 100 mph, the F-104 will trim just above the airfoils' zero lift angle).

In terms of stability below 8 degrees alpha, the lift slopes are pretty similar, and the aerodynamic center is only a bit behind the CG, so the changes are pretty negligible. Damping of pitch oscillations is a bit worse with the thinner airfoil, but not enough to matter.

To summarize, the thicker airfoil did not offer significant advantages in the angle of attack range where the aircraft was stable and the use of slotted flaps instead of conventional flaps meant that it didn't offer much of an advantage during take off and landing. However, the thinner airfoil reduces weight and drag, allowing a higher top speed, means that the wing will stall before the aircraft reaches an angle of attack where it is in danger of a deep stall, and takes up less space inside the fuselage. The stall character of the F-104 will be scary, but it was going to be scary no matter what I did, and at least this way stalls will always be highly predictable.

 

[telnar1236]

Interesting choice of airfoils, much better than the full size, still looks to have an interesting stall but I have no experience with that max camber location. Does it work like a LE flap or more like typical underamber wing? Xfoil plots didn't help me much (no suprise there).

Here are some much clearer graphs of CL vs. alpha with different flap deflections. These are without slots, so the actual wing will have a gentler stall character when the flaps are down. The results for the 50 degrees flap deflection of the GOE286 airfoil are suspect, since the Xfoil results did not seem to show proper flow separation, so it's likely the solution it arrived at was not correct.

 

[Piotrsko]

Then I shall await the test flight post report.