During Winter Project 2010 we tested an electronic stabilization system for our surfboard with T-foils. The system used four wands attached to potentiometers, with the aid of basic trigonometry the distance to the water surface was calculated in four points. The approach worked, the four wands were however fragile and not very elegant.
The objective of the Easter Project 2012 was to eliminate all but one wand and, for the orientation in space, use an Inertia Measurement Unit (IMU). The Razor 9 DOF IMU from Sparkfun was chosen. The sensor package was completed by a GS407 Helical GPS Receiver, a Regal 5321 rotary sensor (for the wand), a Mettler SS1321 load cell attached to a Tacuna Systems strain gage conditioner and an OpenLog data logger. The trusted Axon microcontroller served as main processing unit.
Since the foils, built for the Summer Project 2009, had a terrible surface quality, these were treated with filler, sanded and covered with adhesive film. EPP Trailing edges were also fitted. To reduce drag even further, thinner guy wires were used. In 2010 2.5mm Kevlar cord was used, this time 0.8mm Spectra was chosen. The Spectra line was rated at 490N ultimate load, this seemed sufficient. As it would turn out later, this rating was likely very optimistic.
A new controller was developed. A first set of matrices for the LQR controller were calculated using the in house developed simulation tool and MATLAB Simulink. The most important difference with respect to the previous controller was the omittance of an observer, since all states could be directly measured, and a different approach for the pitch-height controller. During the tests of the old controller we had to deactivate the pitch regulating part, it would saturate the flaps and prevent the height part from having any effect at all. The new controller used pitch to regulate the flying height and not chamber. A too low distance from the water therefore resulted in a pitch up correction and not in an increment of chamber on both foils.
The moment of truth came and we hit the lake. During the launching of the craft the wand made contact with the lakebed and parted company with the rotary sensor assembly. Luckily we found it again, a quick repair with Kevlar and CA made sure everything would stayed together.
The surfboard flew with the initial controller setting, it was however very nervous and, especially on the roll axis, it tended to have a large overshoot. With a little tuning things got much better. The new pitch-height controller performed as we had hoped: in 2010 manual pitch corrections were periodically necessary, this time no manual input at all was given. Standing on the board proved to be very challenging, however no one of us has ever tried real surfing.
After about half an hour of testing the rear support block shifted towards the stern, as a consequence one front guy wire snapped, this resulted in a sudden negative angle of attack of the rear foil, in the rupture of several other guy wires and ultimately in the loss of the rear foil. This put an end to the tests and we headed back to base. Even though the test session was much shorter than we had hoped, it proved possible to stabilize the surfboard with our new array of sensors and the new approach to pitch-height control was shown to be more valid than the previous.
A quick analysis of the logged data underlined the necessity for further test sessions. As can be seen from the above plots the controller response is not optimal. Sadly the load cell enclosure was not entirely waterproof, no valid force measurement was obtained.
This project also showed the limitations of our current platform. The setup procedure with guy wires is lengthy and troublesome: it is not possible to set the angles of incidence with a deviation under 0.5° from the desired 6°. The struts have a lot of surface area and low stiffness, a bad combination. Because of the large surface area, and therefore high generated lift, it is only possible to tow the board on a straight course, yaw stability is a constant problem and small miss alignments result in always present transverse velocity components. Furthermore attaching the strut support blocks to the board with ratchet lashes and anti-slip mats is far from optimal.
The collected data will now be analyzed in detail and development of a new platform will be started. Video coming soon!