A look at the end result of phase 1:
All faces have been defined, resulting in a complete looking model. True, I did not put any ‘make-up’ on the model, so it looks rather unassuming. But the purpose of phase 1 was to fully define all faces that will be used in phase 2 for sizing/cutting the internal structure.
And as an aside, it was also intended to familiarize myself with Fusion 360.
To see it from all sides, see this video. (It does contain a few “print-throughs” and glitches)
With phase 1 out of the way, its time to finish the CNC macine. Once that is done phase 2 will be started. Though I have to say that the shed is quite cold now… so I may occasionally be unable to resist the temptation to stay warm and start on phase 2 before the CNC machine is fully operational.
Just in case you are interested, the CNC build is blogged about on http://starbase55.com/cnc.
Another quick look at the model in Fusion360:
As you can see, this time the floats and the nacelle were added.
The float were surprisingly easy, but the nacelle proved to be a very stubborn subject.
The problems with the nacelle are three fold:
- I do not have a good drawing of them. The seawings drawings are ok, but because the wing is different, it cannot be used as is. Besides, I had to make a decision about the spinner size. I will be using the ramoser vario-prop and the biggest spinner size for them is 50mm. Hence I had to adjust the size of the prop opening this.
- The ‘loft’ tool of fusion360 hides a lot of complexity that one does need to know -at least in its effects. This took the most time. Time and again the lofts I produced did not fit/match what was needed. In the final model shown above, there are still some imperfections. But these are of the sub-millimeter kind that will be completely corrected/obscured while building. They should be impossible to see in the final product.
- It was necessary to go through a series of iterations until I achieved a look that was the same as in the pictures. Trouble was, these iteration take about 4 to 6 hours each before the result could be compared to a picture. Doing just a couple of them easily costs days of work.
On the risk of becoming too Fusion360-technical, I do want to summarize the loft experiences I made:
Always try to create a loft adjacent to a face, and use the edge of the face as profile or rail. Oftentimes it is easy to use a sketch line as profile or rail, but these do not allow the fine-grained adjustments for G. If necessary, create a helper face by extruding a line.
When creating a loft as a fillet between two faces, try to use a rail in the middle. If the faces are flat, use the G1 (tangent) on the face edge (either profile or rail).
On complex shapes, create a number of planes & sketches at right angles to the shape. After the shape is ‘lofted’, go back to the sketches, intersect with the shape, and then improve the rails/profiles as needed. With little experience, three iterations should get you to the desired end result.
If there are a lot of spline profiles/rails, make sure the splines handles are of a fixed length & orientation, or in a steadily (i.e. mathematical formula) increasing cq decreasing sequence. This produces in my experience the smoothest resulting lofts.
And don’t be afraid to experiment…
Next up: I have to add lots of detail to the model. Things that I glossed over in my rush to get the overall shape correct. Things like: flaps, elevator, rudder, ailerons, wheel bays and the nacelle is not entirely complete either.
So from here on the model will probably remain fairly much the same, even though lots of work is still necessary.
Another quick look at the model in Fusion360:
If this looks like a small change to you, then you probably have no experience modelling a CL-415 in Fusion360 ;-)
But joking aside, I think this is a good deal of progress. The fuselage itself was improved, the vertical and horizontal stabelizers have seen several iterations, and the wing plus wing-tip was added.
Also, the wing profile was redone. It turned out that the profile used in the first iteration (see previous post) did not have a real flat underside. Even though the Clark-Y profile is supposed to have a flat base. True, the difference was very small, but even small deviations can make modelling in Fusion360 (and probably all CAD systems) very difficult.
As said, the main wing uses the Clark-Y profile with 11.7% thickness. The stabelizers use NACA-0012, fully symmetrical 12% thickness.
Of course the same kind of rudder/elevator hinges are used as on the real model. Note that the model is skin only and does not really show the hinges just yet.
Onward to the floats and engine nacelle…
Just a quick update of what has happened so far:
It does seem rather simple, but it took me about two weeks to get sufficient experience in Fusion360 to get to this point.
The most difficulties I had with the nose curvature, the top of the cabin, the side panel (with windows) merging with the side and the dome that connects the fuselage with the wings.
The boat keel was surprisingly easy.
This is of course just the shape of the plane that I am implementing now. I gave up on trying to work my way outwards from simple parts (i.e. the parts that are assembled into a plane) as I kept getting in situations where I had to retrace my steps, sometimes to the very beginning.
Next up are the stabelizers, wings, floats and nacelles.
One of the most difficult decisions one must make when trying to create a scale model of an iconic plane, is how much to deviate from the original.
Ideally, everything should be exactly to scale. But this will often result in a plane that is difficult -if not impossible- to fly.
For the CL-415 this is most obvious in the tail section: The two vertical fins in the horizontal stabeliser are slightly angled and will force the plane to the right. This was apparently done to compensate for the increase in engine power in the design upgrade from CL-215 to CL-415.
To me, this will lead to an unacceptable flight performance in the model. Placing them straight and have the propellors counter rotate to remove the tendency to “turn left” seems like a no-brainer.
Two smaller issues are the non-symmetrical wings and horizontal stabilizer. As you may have noticed, the wing and stabilizer are not the same on the left and right side. The difference is not big, but it is there. Again, this would not fly well on a model, hence on the model both sides of stabilizer and wing will be symmetrical.
But the biggest worry is the main wing. Or rather the wing load. The total wing area is approximately 100 dm2 when designed to scale. To get a realistic flight experience the wing load should be as low as possible. If build as lightly as possible, I estimate that the total weight would be around 6-7kg mark. That gives a wing load of 60..70 gr/dm2. Which is not all that bad, but does not leave much room to the upside. Especially if the landing gear should be fully operational, or when a water load/drop mechanism must be added. A model this size should be able to drop upwards of 1 litre, which would of course add another kg (or more) to the flying weight. Adding these extra’s would easily bring the model to 8-9kg range, and hence a wingload of 80..90 gr/dm2.
It will fly with that wingload, only I would expect that the scale flying impression will be negatively impacted. The rule of thumb is that the stall speed is srq(21 * gr/dm2), which translates for 60gr/dm2 to a stall speed of 35km/h and for 90gr/dm2 a whopping 43km/h. Take-off is of course above the stall speed, and a takeoff at 50 or 60kmh is not really “scale look”. Even 35km/h is pushing the limits of scale feeling.
I am thus contemplating to increase the wing area by 10% to get some “elbow room”. This would mean adding 15 mm to the leading edge and 25mm to the trailing edge of the wing. Adding this area would bring the wing load for an 8kg flying weight back down to approximately 75 gr/dm2 and the stall speed to 40km/h for a fully loaded model.
This would have the added optical advantage that the wings would seem less “flimsy”. Very often when a real plane is build to exact scale, the wings end up looking flimsy, “to thin/small”.
Also, there is room to do this, it would not impact the fuselage very much, and in fact it would be easy to make this reversible. In other words I could start with implementing a scale version, and if it becomes necessary simply upgrade to a larger wing. For this to work, I have to size the fuselage/wing coupling for the slightly larger wing. This should not create any problems.