Powered Ornithopter · IAbstract This report summarizes our semester-long study of ornithopter design. We incorporated and built upon our previous apping mechanism, wing, and body

Powered OrnithopterFinal Design Report

ENGR 3345Mechanical and Aerospace Systems

Professor

Christopher Lee

Design Team

Daniel Dichter

Chris Joyce

Helen Lyons

Elliott Wyse

May 6, 2015

I Abstract

This report summarizes our semester-long study of ornithopter design. We incorporatedand built upon our previous flapping mechanism, wing, and body designs to design acomplete powered ornithopter. In order to make our ornithopter capable of flight we madenumerous changes to our previous iterations of all the body parts. The final ornithopterwas a qualified success: it flew in powered flight for approximately 2 seconds, though itsuffered from several design and fabrication shortcomings.

II Design

Mechanism Design

The mechanism of an ornithopter must produce a reciprocating flapping motion of twowings, in order to produce forward thrust, which allows the wings to generate lift. Ourflapping mechanism can be divided into three sections: the gearbox, the linear cam mech-anism, and the linkage that forms the wing roots.

Gearbox DesignThe gearbox reduces the speed of our two motors and increases their torque. Our motorsproduce roughly 4 watts of power each and can achieve speeds of 10,000 rpm even whenlightly loaded. Our targeted flapping frequency was roughly somewhere between 10 and20 flaps per second, which meant that we needed a gear ratio of roughly 50:1. With thisin mind, we designed a two stage gearbox with a final reduction of roughly 47:1. This iscomposed of one stage with a ratio of 48:9 and a second stage with a ratio of 70:8. Wechose to use gears with a modulus of 0.3 in order to achieve this final ratio in the smallestpossible gearbox. Our final gearbox produces roughly 214 rpm when powered with a 3.7 Vsupply.

II Design 1

Figure 1: CAD of our redesigned gearbox

The shaft positioning is accomplished by a single 3D printed part, which allows for excellentshaft alignment (because relative hole positioning is very accurate in 3D printing). Gearsare press fit onto their shafts, which is sufficient for torque transfer because most of thegears are operating at high speed and low torque. The gear that drives that cam is alsomechanically linked to the cam in order to provide increased torque transfer.

Cam MechanismThe cam mechanism consists of the cam, cam follower, and rails for the follower to slidealong. The cam profile is designed such that it provides a substantially more rapid down-stroke than upstroke. This is done to increase lift and thrust. 3D printing was used tomanufacture the cam as well, because it allowed easy manufacturing of complex profiles.

Figure 2: CAD of Cam

The cam follower consists of one 3D printed part in addition to a very small ball bearing.The ball bearing interfaces with the cam surface, providing a contact with negligible friction,which improves the efficiency of the system over other contact types. The 3D printed partof the follower provides holes, through which the rails of the cam mechanism fit, in addition

II Design 2

to attachment points for the ball bearing and wings. The rails themselves are very simple1mm driveshafts. We chose to use driveshafts because they are very hard and straightrelative to other types of rod available.

Wing MechanismWing roots for this design were also 3D printed, and provide attachment points for thegearbox, cam follower, and wing rods. As the cam follower moves vertically, the wingsare actuated through a flapping motion. The mechanism was designed such that the wingswere biased approximately 5 degrees upwards throughout the course of the flap. The reasonfor doing this was to increase the stability of the system by effectively introducing dihedralto the flapping wings. The total designed range of the flapping motion was roughly 50degrees, however in practice it is substantially reduced due to poor slot tolerances.

Wing Design

Figure 3: Wing design

The wings are constructed from carbon fiber rods, which mount to the wing roots and a thinlow density polyethylene (LDPE) film, which forms the wing planform. LDPE providesboth very low weight and very high flexibility, while retaining sufficient structural strengthto be suitable for this size of ornithopter. The planform was based on the planform of asimilarly sized RC ornithopter, and is a semi-elliptical shape with an aspect ratio of roughly3:1. No reinforcing battens were added due to the small size and large aspect ratio.

II Design 3

Tail Design

The tail consists of a square carbon tube, which fits a mounting hole in the rear of thegearbox, along with a 3D printed piece, which attaches the tail fins to the carbon tube. Wechose to use a v-tail, because it provides additional roll stability (due to dihedral), whilehaving low weight and incorporating both yaw and pitch stability into a two-fin form,providing lower drag. Additionally, relative to the wings the tail is pointed downwards10 degrees. This is to counteract the fact that the center of thrust from the wings ismuch higher than the center of mass of the entire system. While the wings will pitch theornithopter down, the tail will bias it towards pitching up.

Figure 4: Tail

Electronics

The electronics system has 4 components: the RC controller, the RC receiver, the battery,and the motors. Our RC controller and battery were lifted from the CyberDactyl toy,as they were known to function, already on hand, and quite light. The RC receiver waspurchased from MicronWings.com, model DT-Rx31d. We use it to control two brushedDC motors, also purchased from MicronWings.com. This electrical system worked quitewell, and was very light. It was difficult to balance the electronics on the gearbox, becausewe did not really account for it in the initial design.

III Fabrication Procedure

We used a similar fabrication procedure as we have when constructing previous sections.Our drive mechanism consists of five main 3D printed parts – the gearbox, two wing pivots,the cam follower, and the cam itself. We also have two 3D printed parts along the tail shaft,one to support the rear of the wing and another to support the tail fins.

III Fabrication Procedure 4

The gearbox has our two motors press fit into the gearbox, with shafts running through thegearbox and gears. We 3D printed a clock-cage style gearbox that also includes supportsfor the cam follower and wing root pivots.

Figure 5: Finished gearbox with cam, slider, motors, and gears attached

We attached the LDPE planform to the carbon fiber rod leading edges with tape as shownbelow. This turned out to be a much higher-functioning system than our previous designof Mylar film epoxied to the wing.

Figure 6: New wing material and fabrication

III Fabrication Procedure 5

When assembling the tail, we attached each of the two acrylic tail fins into their respectiveslots with epoxy. We used a tight fit to attach the tail mount onto the end of the squarerod so that we would be able to add or remove weight as needed.

Figure 7: Complete tail assembly fit onto the tail shaft

Most of these fabrication procedures are not markedly different from what was attemptedin past iterations. The main differences are the inclusion of a custom gear train ratherthan one integral to the motor, a switch in wing materials, and an overall downsizing ofthe ornithopter.

One key aspect of our design is that it must be well-lubricated with graphite to functionproperly. The action of the cam induces a torque on the cam follower which is transmittedto the rails, and the entire system must be very low friction for this to work efficiently. Thisproblem is exacerbated by the small scale and lack of precision available in 3D printing.This gearbox part, on a production ornithopter, would probably be better made as injectionmolded plastic. That would allow for much more precision alignment of components.

IV Flight Testing Results

Although our ornithopter did not generate enough lift to sustain flight, it was clear thatthe gliding performance was significantly improved after adding the flapping. We predictthat the major factors hindering our sustained flight were a lack of sufficient thrust as wellas limited pitch stability. The average flight time was about two seconds and it traveledroughly 15 feet. This flight time is an improvement of roughly 400

Figure 8: Flight path of our ornithopter

IV Flight Testing Results 6

V Diagnosis

One of our main issues was that it was difficult to place the center of gravity fall exactlywhere we wanted along the tail shaft. This was reasonably easy to compensate for bymoving the location of the electronics along the shaft and added ballast in the form of acoil of .22 gauge wire between the tail and electronics as shown below. We also increasedthe size of out tail fins to add weight (additionally because we were concerned that theywere too small to have a substantial effect) While this greatly improved the balance of ourornithopter, it is non-ideal as with an ornithopter all weight on the vehicle should have apurpose.

Figure 9: Complete ornithopter with ballast added

Another large issue that we encountered is that our desired cam profile was too steep todrive in the desired direction, so we had to run it in reverse. We discovered this problemtoo late to print modified cams with new profiles, though this problem is certainly notinsurmountable. This was an efficiency loss, not a show-stopping problem.

Another efficiency loss came from the shafts running from the front to the back of theornithopter–the camshaft and gear shafts. The shafts tended to shift in their holes andbecome slightly misaligned. We used hot glue where possible to fix them in place, but itdid little to improve the problem.

While testing our ornithopter we found that mylar that we had been using in all previousiterations was too stiff and heavy. It did not deform enough while the wings were flappingto provide significant asymmetrical bending and provided little thrust. Switching from themylar to LDPE resulted in a visible thrust increase.

V Diagnosis 7

VI Reflection

This component of the project very effectively built on what we have learned in the preced-ing components. Our specific learning goals around this project were to determine whatwas standing between a flapping mechanism that would theoretically fly and one that wouldactually fly. We probably could have switched to lighter, smaller COTS components earlier– that was a slightly last-minute change, as we put too much emphasis early on on usingproven and easily accessible systems.

Had we switched to smaller motors earlier, we likely would have had an easier time ofsome of the late-stage debugging. While our overall system architecture remained verysimilar, switching to smaller versions of components was a more involved process than weanticipated.

The other potential option would have been to significantly scale up our gearbox andwingspan to accommodate the larger components that we started with, though this likelywould not have worked as well – wing area scales as the square of length, whereas massscales as the cube of length. We definitely made the right decisions, we just probably wouldhave been better served by making those decisions earlier.

An additional design improvement would be to increase the wingspan. This would haveafforded us more thrust, and given us a better chance at sustained flight. Also, increasingthe length of the tail would have given us more pitch stability.

A problem we mentioned earlier was the shifting of shaft causing efficiency losses. Thisshifting was caused by drilling out varying hole sizes for the same size shaft. To preventthat from being the case, we should have determined what the appropriate size to press fitthe shaft into the gearbox, and consistently used that standard for our of our hole.

an unanticipated design flaw, was that upon landing the ornithopter will crash directly ontothe cam. We thought that by only landing on the grass, our ornithopter would stand upto multiple flight tests. Unfortunately, the tail shaft would yielded every time, and duringtesting the shaft bent out of shape.

VII Conclusion

This project was a moderate success – getting an ornithopter that controls its fall to aslower rate is a crucial step to get to a flying ornithopter. As discussed in the reflectionsection, if we had switched to miniaturized components earlier, we likely would have hadmore success isolating the specific issues we faced during this phase from those that arisewhen switching components. However, we made great strides in minimizing the weight ofthe system and balancing it appropriately. We also seem to have structural support wherenecessary – we have not had component failures in this phase, unlike some other phases.

While we did not hit our target of a self-sustaining ornithopter, we did make a serious

VII Conclusion 8

attempt at every system that is necessary to create an ornithopter on this scale, and eachone works generally as designed. More time would certainly enable a vehicle that is moreflight-capable, but this one certainly taught us all a valuable lesson.

VII Conclusion 9

A Appendix

Bill of Materials

Part Manufacturer Unit Cost Quantity Total Cost

Gearbox 7mm Twin Drive ProfileMount for 3mm Depron

Micron Wings 28.00 1 28.00

Gear Mod 0.3 - 8 Teeth - 0.95mmBore


Gear Mod 0.3 - 70 Teeth - 0.95mmBore


Shaft - 0.97mm Dia Burred (Packof 5)


Brass Bearing 1mm ID with Flange(pack of 5)


DT Rx31d Receiver Unit for Servos(With ”Dual” Onboard BrushedESCs)


Nano-Tech 130Mah Battery WithJST Plug


Charger Lipo 325Mah Variable ForMCPX


Carbon Fibre Rod 0.5mm Micron Wings 6.50 2 13.00

Carbon Rod 1.0mm x 50cm Micron Wings 6.80 2 13.60

Carbon Rod Square 2.0mm x 40cm Micron Wings 5.70 2 11.40

Miniature High-Precision StainlessSteel Ball Bearings—ABEC-5

McMaster-Carr 7.50 1 7.50

A Appendix 10

Powered Ornithopter · IAbstract This report summarizes our semester-long study of ornithopter design. We incorporated and built upon our previous apping mechanism, wing, and body

Documents