See the June report for the previous update. The Interim Report provides further details on work accomplished in July. The report here focuses on progress made on the prototypes.
In the former structural analysis, the effects of centrifugal force were not considered. In this month, the centrifugal forces are calculated and included in the bending and torsion analysis. In particular, the 4-blade rotor was first analyzed. As a result, we could understand the dominant loading situation in this scale.
The blade was divided into 18 cross-sections, and each cross-section was further divided into 20 segments. The centrifugal force of each element can be calculated by knowing the angular velocity and radius in the vector form. The centrifugal force in each element also contributes to the moments in x, y, and z directions at the centroid of the cross-section. The moments from centrifugal force do not influence the bending of blade too much, but dominate the torsion. The following figures show the comparison of bending in both x-x and y-y directions and the torsion from different loading sources.
The ability of the supercapacitors to power the mesicopter model was verified. The supercapacitors provide very steep discharge characteristics. A milligram resolution scale is used to measure the lift. This scale uses averaging to reduce the readout noise. The drawback of the averaging is introduction of a settling time in the readout. This settling time is comparable to the time constants of the capacitor discharge. Additionally manual sampling of the readout is required. Therefore direct measurement of mesicopter lift, as a function of time, was not possible. The relation between the controller voltage and mesicopter lift in a steady state (constant propeller speed) can be easily measured. The inertia of the motors and propellers and the resulting time constants are small compared to the capacitor discharge time. Therefore it can be assumed that during the Supercapacitor discharge (at least after few milliseconds) the motors are operated in steady state and the generated mesicopter lift can be determined measuring the voltage at the capacitors.
The current Supercapacitor configuration consists of four 1F supercapacitors stacked in series. The individual capacitors are rated for 2.5 V each. Therefore theoretically the stack can be considered as a 0.25 F capacitor with maximum rated voltage of 10 V.
Two experiments were performed. First the Supercapacitor stack was disconnected from the motor controller then charged to a specific voltage and then connected to the controller. The capacitors were charged to 12 V and after the electrical contact to the controller was established the voltage dropped immediately to approximately 7-6 V and after that declined at a constant rate of app. 0.5 V/s. A high current spike was observed in the first milliseconds after the contact (in the non steady state regime). This startup delay is caused mainly by the time constants of the adaptive algorithm in the back-Emf motor controller. In order to avoid the start losses a second measurement was performed. This time the Supercapacitor stack was connected to the controller during the whole measurement. First an external voltage was applied to the capacitors and controllers for time long enough to assure steady state operation (few seconds). Then the external voltage was disconnected from the capacitors/controller. This time no start up was needed as the propellers were spinning at the instant of disconnecting from the external power supply. Similarly to the first experiment the voltage dropped immediately to app. 8-7 volts and then declined at a constant rate. In both cases the maximum mesicopter lift delivered by the capacitors was about 6 g (corresponds to 7-8 V at controller).
Several candidate motor/gearbox combinations are being evaluated for use in a larger 150 to 200 gram rotorcraft. This vehicle size should permit flight times greater than 10 minutes and allow for 20 to 40 grams of useful payload. The ability to carry a variety of sensor packages and have a reasonable endurance create improved experimental capabilities for applications and control research.
The two power plant packages being evaluated are a 9.8g, geared conventional DC motor from a Keyence remote control helicopter and a 12.8g, coreless motor/ gearbox package from Astroflight. This motor is sold as the Firefly and comes with an inline 16:1 gearbox. The target thrust was 50 grams and a variety of commercially available rotors were used. All were between 6 and 10 inches in diameter. The power supply is constrained by the capabilities of the Tadrian LI-Ion cells being used for power. A three-cell pack has a maximum voltage of approximately 9 volts. This value drops appreciably with increased current draw due to internal resistance and the cells are fused at around 3A to avoid dangerous internal pressures. The Keyence package was capable of generating sufficient thrust with 6.5V, but draws 1 amp of current. This would be a 4A total for the vehicle and is beyond the current limitations of the batteries. The Firefly package requires approximately 8V to generate 50 grams of thrust, but draws only 400mA. The total draw of 1.6A is reasonable for the Tadrian cells. Is also interesting to note the very large difference in efficiency between the two packages. The Keyence package requires over 6 Watts to generate approximately the same thrust as the Firefly package operating at only 3 Watts; beyond the issue of feasibility, the Firefly package provides roughly twice the endurance of the Keyence package.
A systems integration testbed has been constructed for exploring various approaches to communication, sensing, stabilization, and control. This system consists of a custom transmitter and a PCB frame flyer. The custom transmitter shown below uses a Great Planes Real Flight Futaba controller with a joystick port output plugged into a board containing a Microchip PIC17C756 and a Linx Technologies transmitter. The PIC samples the joystick port, converts roll, pitch, yaw, throttle, to motor controls and transmits them to the PCB flyer.
The flyer shown below uses 20mil printed circuit board for the frame which supports the chips, motors, and battery. The surface mount PIC17 reads the incoming motor controls from the Linx receiver and drives the motors via L293DD power chips. The vehicle with battery weighs 65 grams and generates 80 g of total thrust. The PICs are programmed in C giving the capability for expansion via the A/D inputs, I2C interface and many I/O pins.
An attempt is underway to use the Photobit PB0100 camera to perform vision stabilization. This chip is shown below compared to the PCB of the flyer.
The feedback system designed in June to control the vertical position of the 60 gram mesicopter has had a number of modifications. The plant model was modified, and the controller was changed to reflect this modification. The result was elimination of sustained oscillations in the mesicopter response and instead a improved response with ample dampening.
Another LED was attached to the front of the copter for a total of two. Instead of strictly controlling vertical position of the mesicopter, we instead are attempting to control both vertical position and yaw. The yaw angle will be calculated from simple trigonometry based on calibration and knowledge of the system. The non-linear trigonometry is neccessary because of the 3D rotation partially normal to the vision plane.
A new controller will have to be created based on a new height/yaw plant model. Study is underway to identify/calculate this new dynamic model. Implementation of a new controller around this model will give us complete autonomy of the mesicopter on the shaft.