Last week (20.08.2009) we decided to have a go with a new set of foils. Fully submerged T foils this time, with two flaps per foil. Altitude, roll and pitch are controlled by the “pilot” with an old RC transmitter. The potentiometers of the transmitter are directly wired to a micro controller which processes the three inputs and with the aid of mixing functions generates four outputs.
The foils are 140mm chord and 1300mm span, airfoil is a Naca 63412. Flap is at 70%.
Construction started, as usual, with a set of CNC cut foam cores. This time though the negatives were not thrown away but sanded and covered in an adhesive film. This gives some sort of mould. The foil was then put under the mill and a cut out was made for the upper flange of the spar. In there went 10 full length basalt tows and 14 carbon tows. When it had fully cured the part was put in the mill again, this time making the cut outs for the web and the lower flange. The technique worked very well, no deformation of the air foil was noticeable. Special care was taken to saturate the wood and surrounding foam with epoxy resin to avoid the two flanges separating from the web because of shear forces.
With the spar in place it was time to start work on the skin. The skin is composed of three layers of 220g/m^2 of basalt fiber. Between the first and the second layer there is a ribbon of peelply, which will act as hinge for the flap. The three layers were saturated with resin on one side, the mould was put on, the foil turned and the process repeated. With a bit of care it is no problem to make the basalt follow the leading edge. Finally we had our foils in the moulds, under some weight we let them cure for 20 hours at 27°C.
As is visible from the pictures the surface quality is far from perfect. We decided to ignore this and have a filler n’ sanding party later if we felt like it. Quality was already much better than our last foils so we hoped they would work! After the exciting de-moulding moment the foils were put in the mill once again. The trailing edge and the flap to foil gap was cut on the lower side. There is an interesting detail here: the gap on the lower side is not cut below the hinge line but the centre of cut (with a 3mm mill) 5mm forward. This was guesstimation and worked very well!
Now the struts were made. Nothing spectacular: some 15mm marine plywood covered with a layer of 220g/m^2 basalt on each side.
At this point it was time to free the flaps. Wit a ruler and a hobby knife the upper skins of basalt were cut a bit, then with a diamond cutting wheel and a Dremel the lower gap was extended till the upper skin. Now comes the interesting part: you have to bend the flap, delaminating some of the remaining basalt without breaking the skin off the core. The idea is to only have the peelply and a few delaminated basalt fibers hold the flap. Finding the right amount of violence to do this is not at all easy. In the end it worked even tough we realised we cut a little too much through the skin on the upper side in a few spots.
Struts and foils were now glued together, epoxy with tome thixotropic agent did the job well.
Servos and pushrods were added next. We used four Hitec HS-645MG. This sounds easy but took quite a bit of time…
We operated the servos at 5V, they are rated at 7.7kg.cm (106.93 oz/in.) at 4.8V. Since the distance between hinge line and flap horn is twice the distance between servo rotation point and the hole used in the servo horn we have a lever arm of 2:1. That makes a torque of 14.4kg.cm available at the flap. As it later turned out this seems to be more than sufficient, from the quantity of energy drawn from the battery the servos are operating within their capabilities.
Flap control is done with a spring steel pushrod that moves in a plastic sleeving. Notice that the sleeving is fixed to the strut in several points with glass fibre. The last part of the sleeve is curved because we did not have the servo straight overhead, the rest however is firmly bonded to the strut. The pushrod takes tension loads while the sleeving prevents the pushrod from buckilng under compression loads. This kind of technique is used extensively in model aircrafts.
As for control we are using an Axon microcontroller with four HS-645 RC servos. A member of our team will do a bachelor thesis about an automatic control system next semester (Spring 2010).
The big moment arrived and everything was mounted to the surfboard, the blocks where the struts are bolted to are the same ones we used at Easter. This time kevlar replaced nylon as guy-wires.
Notice the difference in area between the new and old foils! The first day of tests was quite exciting. On the roll axis the craft was very very twitchy. As a result flying for more than a few meters was a real challenge. We set our software man to work and by the next day the control curve on the roll mix was no longer linear but exponential. The amount of expo is adjustable with a potentiometer.
During the tests we made the following observations
- As predicted by xfoil the flap only produces more drag quite soon, only small positive deflections are necessary.
- The changes in chamber are best kept small, once up to speed the craft can be flown like an aeroplane with the pitch and roll controls.
- Even though it is possible to take off quite soon, at higher speeds (about 10 km/h) the craft becomes much easier to control on the roll axis because of the smaller deflections of the flaps.
- Prop wash from the tow boat is felt as a low vibration. The flaps however seem to be solid enough, we had no flutter issues.
Here is the video of our Summer Project 2009: