• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.

Scratch Built Balsa F-104

I just can't justify keeping it together since I'd never fly it. It's just too hard to launch. It is coming back though. Here's the base CAD for V2 with the 64mm fan. I'm also changing out the airfoil for a Drela AG37 at the root and a Drela AG35 at the tip. Should help avoid the tip stalls and reduce drag. The larger size means I can manage a full 6ch setup so the slightly lower lift airfoil can be offset by flaps.

Whit Armstrong

Well-known member
I just can't justify keeping it together since I'd never fly it. It's just too hard to launch. It is coming back though. Here's the base CAD for V2 with the 64mm fan. I'm also changing out the airfoil for a Drela AG37 at the root and a Drela AG35 at the tip. Should help avoid the tip stalls and reduce drag. The larger size means I can manage a full 6ch setup so the slightly lower lift airfoil can be offset by flaps. View attachment 189209
What cad is that? looks good!
DevFus4. You can download the trial version for free from their website and it really streamlines the design process for balsa planes. The trial version won't let you export anything, though, and while it's good, it's also pretty limited in what it can do, so I tend to take screenshots of the formers and then make a proper model in FreeCAD to use for the rest of the design process. Same deal when designing the wings using devWing 3. From FreeCAD you can then generate G-code to cut the formers. For simpler designs, you can also just take the screenshots, edit them a bit with paint, and use the png file directly in LaserGRBL to cut the formers with a laser cutter. Using that set of software in that way means you can design the planes using only free software and end up with kit quality planes for the cost of some sheet balsa.
For anyone interested, here are the png files for the formers after some processing, meant to be cut out of 1.5mm plywood, or to be made of 3mm balsa for a lighter weight plane. Each image is 290mm x 160mm, sized to fit on a 300mm x 180mm bed. The base version is for a lightweight minimal detail plane (a friend asked for it to make a stick and tissue prop powered plane and it was easy to do) while the B version is what I will be building and has provisions for a scale cockpit.


Good news and bad news on the F-104. I finally got the chance for a test flight and she does fly, and exceptionally well! The VGs do their job so you retain aileron control even close to a stall, and she slows down surprisingly nicely for her small wings. However, I now understand why the phrase "banking with intent to turn" was coined for the F-104. The high wing-loading means that I have flown 70mm EDFs that can turn in a smaller space and the lack of a rudder is very noticeable. More unfortunately hand-launches are a real problem. The launch for the maiden went smoothly and the flight went well, but on the second hand launch, I didn't quite get enough airspeed and she tip-stalled into the ground so much of the nose was destroyed and the rest of the airframe suffered some minor damage. I'm considering rebuilding the whole design with more weight savings and scaled up for a 64mm edf to reduce the wing-loading and give me room for gear. I selected the 55mm EDF because of its small profile and because I wanted an easy to transport plane, but a 64mm setup on the same battery weighs less while achieving almost 300g more thrust, and scaled up plane would give me the wing area and thrust to add all the features I missed in this one.
Maybe try making the wings a bit thicker so you don't lose much on scale looks but get a lower stall speed?
Maybe try making the wings a bit thicker so you don't lose much on scale looks but get a lower stall speed?
It's great to hear she flew well! (y)

Have you thought about a bungee launch system to help with keeping that initial speed high and thumbs on the sticks?
Thanks for the recommendations. I have looked at both of those possibilities.

In terms of making the wings thicker, I used a Clark Y airfoil on version 1. It has an excellent lift coefficient and a very gentle stall character. There are higher lift airfoils out there, but going to a much higher lift airfoil would either mean incurring a significant drag penalty, or using an under cambered airfoil which would make inverted flight a pain. I've attached the airfoil profile and lift curve for the Clark Y airfoil I initially used from Airfoil Tools.
Clark Y Airfoil.PNG
Clark Y Lift Curve.PNG

As can be seen, version 1 used a pretty thick airfoil already. I'm actually going to a thinner lower lift airfoil for version 2. Having a larger design that can accommodate retracts and flaps means I can increase the lift coefficient well beyond what could be achieved with a flat bottom airfoil by dropping the flaps while simultaneously meaning I have to worry less about takeoff speed. Mixing in maneuvering flaps should also improve turn rate. The better L/D ratio should enable a better top speed as well. Because of the location of the horizontal stabilizer, a stall angle beyond about 10 degrees isn't too useful. The wing can blank out the stabilizer and result in a deep stall. With the Clark Y airfoil, I had to be very careful to avoid that, but with the new AG37 and AG35 airfoils, it should be much more difficult to put the airplane at a high enough angle of attack to get into that situation.

To avoid tip stalls, the airfoil at the wing root is different from the airfoil at the wing tip. When the airfoil is the same throughout, as it was on the version 1, the spanwise flow and lower Reynolds number at the tip means the tip stalls first. If one wing tip stalls before the other, the asymmetric lift results in a tip stall. The easiest way to avoid this is with wash in the wingtip, but it's hard to get that precise without a complicated jig. The change in airfoil from root to tip results in a bit more drag since the lift distribution is a bit less ideal, but it makes the construction process much simpler, and the increase in drag is pretty tiny.

In terms of a bungee launch system, that would have been ideal for version 1. I actually went as far as measuring out lengths on some spare PVC pipes, but the goal of the smaller plane was ease of transport. Moving the bungee launcher and plane is harder than moving the somewhat larger 64mm plane by itself, so I'm just increasing the size of the plane since I already had a spare 64mm fan, and otherwise the electronics are identical. Landing gear was always the goal and version 1 was just too small to accommodate reliable servo driven retracts due to their weight.
More progress. The EDF unit is installed, and the ducting is partially complete. To reduce weight, the ducting is partially made out of formed balsa this time. It is a tremendous pain getting it properly installed, but but it is significantly lighter than resin infused tape. I still used that for the more complex geometries, though. There is a also a cheater inlet designed in from the get-go this time. This means it can be much more efficient than what I got last time by just cutting a hole through the bottom of the plane as an afterthought. The next step after completing the inlet ducting will be building the wings which I plan to hotwire cut out of foam and then sheet with balsa. It's a slightly heavier technique than constructing a built up balsa wing, but it will make getting the necessary geometry for slotted flaps much easier, and sheeted foam wings always end up a lot prettier in my experience.
The wings are cut and sheeted. Balsa sheeted foam always turns out that little bit nicer than sheeting over a built-up wing. After some consideration, and many recommendations, including one here from @BoredGuy , I've gone back to thicker wings. The root airfoil is a Gottingen 286 and the tip airfoil is still a Clark Y. This should still have the same characteristics where the root stalls before the tip, and the Gottingen airfoil has a similar low angle of attack L/D ratio to the Drela, while having a much higher maximum lift coefficient and more drag at higher angles which is ideal for achieving higher top-end speeds while also keeping the stall speed low. Because the tip airfoil is higher drag, the wing is a bit draggier, but not actually that much more at low angles of attack.

20210120_193721.jpg 20210123_202424.jpg

Building slotted flaps is also easy with sheeted foam wings. You can cut out the slots while still sealing the upper surface to help maintain lift and avoid separation. I've put a few close-ups of how the slots are made.

20210120_193919.jpg 20210123_202525.jpg

Everything still needs some processing to get it ready to attach, but it's coming together well.
The tail surface has been cut and covered and all the flight control servos installed. The wings and tail are now mounted to the fuselage. I took the opportunity to redesign the tail from the version 1 airplane to make it more scale and use a thinner airfoil since the larger plane allows the servos to fit better. I think the tail redesign looks much nicer, but it shouldn't have a significant effect on flight characteristics.

I also finally got the mechanical retracts I ordered at the beginning of January for this plane. They will take some modification to fit an F-104, and I'll need to see how they hold up, but they seem remarkably good quality for the $10 price point.
The plane is now to the point where I can start covering it with balsa. I also made the necessary modifications to change the retracts from straight up and down to 50 degrees from the horizontal. Having disassembled and reassembled them now, I am deeply impressed with the manufacturing quality and strength, but a bit less so with the design. The nose gear has the servo mounted to the strut, which simplifies the push-rod setup, but means the gear must be deeply recessed in the fuselage to be flush when up. This is actually good for my design, because I was worried about the nose gear being too long after I had made the changes to the main gear, but for a low wing plane, it could cause some issues.
It's been a while since I have had any time to work on the F-104 since things got pretty busy for a while, but I did do a bit more looking at what went wrong with version 1 and figured I should share what I learned so that others can hopefully avoid the same issues.

Roll Instability:
  • The vertical stabilizer was too small. Directional stability was fine, but the plane was barely stable in the roll axis. Fortunately this is already fixed with the larger and more scale vertical tail on version 2.
  • The anhedral in the wings also hurt roll-stability. Instead of using the exact scale measurement, I went with what looked right since the wings had a much thicker airfoil and the correct scale anhedral did not look scale. This is also fixed in version 2.
  • The sweep of the wings on version 1 was increased from scale to help counter the increased anhedral, but this did not do enough, and I should have checked this mathematically prior to construction. In version 2, the wing sweep is approximately scale, but the decreased anhedral means a greater sweep is not needed.
  • The lesson to learn from this is to always check stability in all axes prior to construction. I don't think these issues with roll stability caused the crash, but they likely made an already difficult plane even harder to fly.
Vortex Generator Design:
  • The vortex generators at the wingtips did mostly work, but there were some issues. Most significantly, they made the already minimally stable aircraft less stable still at high angles of attack by increasing the effect of the anhedral.
  • There were some positives and some negatives to this. It made roll control a lot more responsive at high angles of attack where the ailerons would normally be ineffective as was intended, but it also made the plane more vulnerable to disturbances when flying like that.
  • I was very careful with my control inputs when flying slow and didn't experience this, but I think the use of the VGs may have increased the likelihood of the aircraft snapping in a stall. As best I can tell, they still would help in a spin, though.
I think roll stability issues are important to be aware of since they aren't as easy to visualize.
It's amazing that has enough wing to get airborne. It definitely looks cool.
Thanks. The wing loading is deceptively light, actually about 25% lighter than the Freewing F-5, for example, and with a much higher lift airfoil. The wings look tiny because of the body, but the interior is actually mostly empty space, so the wings are a lot more lightly loaded than they look. That lighter loading is the main reason for the larger scale of the version 2, and, with the larger scale, the wing loading is actually pretty typical of an EDF.