The Manned Octocopter Project

It had always been my dream to fly on something I built with my own hands. I had been pursuing my high school to fund such a project since I was in the 8th grade. When I was back college during my sophomore year, my high school finally agreed to fund this project and instructed me to mentor a group of 4 students along the way and teach them about the design and construction of such a monstrocity of a machine.

There were many hurdles along the way such as the phase wise disbursement of funds, finding tools and resources locally to help make this project into a reality and the fluctuating stock of items on the HobbyKing website. Not to mention the hassle of convincing the customs authorities and running back and forth from one office to another to get signatures and letters to let us import all the parts required.

Not only did I learn how to design, build and fly the manned octocopter, I also learned how to communicate effectively and the importance of sharing knowledge.

I would like to give a special thanks Capt. Randhawa for his support in the design and testing and to all those who helped make this project turn into a reality.

1. 8x Turnigy CA170 Motors ~ 350$ each

2. 4x Turnigy Fatboy 300A ESC ~ 250$ each

3. 4x Turnigy 200A HV ESC ~ 180$ each

4. 1x Hitech XG11 Tx/Rx Set ~ 800$

5. 1x DJI A3 Pro ~ 1300$

6. 1x DJI Naza M V2 ~ 300$

7. 16x Zippy 22.2V 5Ah 40C LiPo ~ 140$ each

8. 1x 4 Charging Port 6s LiPo Charger ~ 200$

9. 50x XT150 connector pairs ~ 2$ each

10. 10m 8AWG Sillicone Wire ~ 50$

11. Assortment of servo extensions ~ 40$

12. 4x Squash Balls ~ 2$

13. 25cm x 50cm 2mm MDF Sheet ~ 2$

14. Assortment of tools such as drill, angle grinder, screwdrivers, spanners, crimping tools, pliers etc

15. Access to a laser cutter or band saw

16. Assortment of 4mm, 6mm and 8mm nuts, bolts and washers ~ 20$

17. Propane Torch ~ 30$

18. Soldering Paste and Led ~ 20$

19. 20m of 2mm and 3mm braided steel wire ~ 40$

20. Aluminum Crimps ~ 2$

21. 40x Good quality turnbuckles (100mm-200mm size) ~ 30$

22. 40x Hooks and Eye Bolts of suitable size ~ 10$

23. Turnigy RPM Tachometer ~ 20$

24. Turnigy Wattmeter/Ampmeter ~ 20$

25. Battery Health Checker ~ 20$

26. 20m long 50mm x 50mm Square 2mm thickness Aluminum Channels ~ 200$

27. 5m long 25mm x 25mm square 3mm thickness Aluminum Channels ~ 50$

28. 5m long 50mm x 50mm Square 1mm thickness Aluminum Channels ~ 25$

29. 4mm thick Aluminum plate 6' x 4' ~ 100$

30. Adhesive Velcro ~ 10$

31. Tig Welder

32. Measuring Tools

33. Fusion 360

34. 11.1v 2200mAh LiPo 30C ~ 20$

35. XT60 Connector ~ 2$

36. 1.2m x 1.2m Heavy Duty Canvas ~ 10$

37. Heavy Duty Thread and Sewing Kit ~ 20$

38. Safety Harness ~ 100$

I can't stress this enough, for any sort of project, planning is the most important part!

In the manned octocopter project, planning consisted of -

1. Getting the team up to par with the knowledge of multirotors, structures and metalworking.

2. Getting the funds and meeting with the sponsors on a regular basis and providing them documentation of the updates.

3. How to import all these parts into a country like India without being questioned too much about the legitimacy of the project.

4. Where to source the components and how to take into consideration the fluctuations in prices over time and the possibility of the items going out of stock.

5. Designing the physical machine while ensuring there is enough space for a human.

6. Designing a series of tests to ensure the machines safety for manned flight.

While designing the manned octocopter, we weighed out the pros and cons of an X8 configuration and a traditional octocopter configuration and found that a significant amount of thrust would be wasted if we went with the X8 configuration, however, we would be able to make the multirotor a lot smaller and save on weight. After doing the calculations we found that the loss in thrust outweighed the advantage of lower weight of the frame so we went with the traditional octocopter configuration. We also thought about other configurations such as the H, T and others, however, we figured tuning the machine would be a challenge as the non-symmetric design would require different gains for roll and pitch

We then had to figure out what the size/wheelbase of the octocopter had to be. Before we did this we ordered a bunch of propellers (28" to 32") from Xoar Propellers (SO to Xoar for giving us constant support and discounts!) and thrust tested them. The wheelbase was then figured by making CAD models on Fusion 360 while keeping the dimensions of the propellers used constant at 32" (based on the data that the CA120-70 motors produced roughly 25kgs of thrust at full throttle with 32x10 props) and giving each propeller a tip to tip clearance of 4". We found the dimension for the wheelbase very close to 3m hence we just sized it up to 3m for ease in calculation and measurements. We also did some basic measurements for how high the octocopter frame would have to stand from the ground in order for a human to be contained within while not stressing the landing gear by putting his entire weight on the frame/seat. We found the average measure for the potential pilots to be 1m.

We then figured out the basic shape and sizing of the containment and designed the rest of the frame around this dimension. We found a 55cm x 55cm inner containment to be the right size to accommodate a human.

After measuring the potential pilots and deciding on the size for the containment to be 50cm x 50cm on the inside of the containment, we had to decide what size of channel and what thickness we needed to use to construct the frame.

After doing some calculations we decided to use 50mm square aluminum channels with 2mm wall thickness. While the 1.5mm wall thickness channels would have sufficed, we figured it would melt and distort while tig welding the arm mount plates to it hence decided to use the 2mm.

To make the containment we cut 4 x 65cm long sections using an angle grinder (of course after putting on safety glasses and tying up all loose clothing) and on each end made markings with 45 degree angles giving us a measure of 55cm on the inner edge end and 65cm on the outer edge. We then used an adjustable miter saw and adjusted the angle to 45 and cut the edges.

*Its a good idea to pull the miter saw down while its is off and align it to ensure it will cut the section where you want it to*

After obtaining these sections, we placed them together to form a square with the inner dimension of 55cm and outer of approx 65cm. Some edges had imperfections so using a grinding wheel and files, we grinded it to a straight edge and remove burrs. After all these corrections had been made we tig welded the sections together and sanded the top and bottom surfaces to give us a smooth finish in order to give us a level surface onto which we could weld the arm mounting plates.

Using Fusion 360 I designed DXF files like the ones attached with this step such that each plate had 15cm of span on either end to distribute the load to the containment frame coming from the motors and 15cm of span on the arms side to effectively take the load coming from the thrust generated by the motors.

Once these files were made we took them to manual lathe and got them cut on 4mm thick 6061 Aluminum from the 4mm plate and obtained 8x edge arm plates and 8x face arm plates. After the cutting these arm templates the edges were grinded to a smooth finish. 3 holes of 6mm dia were also drilled on the arm end of the plates like in the DXF to enable assembly and disassembly of the arms for transport and repairs.

These plates were then arranged on the centre containment frame on the top and bottom like seen on the picture and tig welded in place. These provided a snug fit on arm channels and we were very happy with the outcome.

With the design we were going for, with the square containment frame in the centre, we had two different lengths of arms. The arms that mounted to the edges of the square will be referred to as the edge arms and those mounted to the side of the square will be referred to as the side mounted arms.

For the side arms we cut 125cm long sections of 2mm thickness 50mm square channels. The additional 5cm was given so that the axle of the motor would more or less be centred at the 1500mm mark from the centre of the multirotor or 1000mm mark from the inner edge of the square containment.

For the edge arms we cut 112.5cm long sections of the same channels. The additional 5cm on one end was given for the same reason above and the other 3.5cm was given so that two 45 degree cuts could be made like shown in the pictures so that arm could slot in the edge. This slot was a bit tricky to cut and required some practise on aluminium scrap before we were able cut it on the real deal. We cut it with an angle grinder and it required a steady hand and a lot of patience. A good idea while cutting such a profile would be cut a little less and file it down as required to fit well.

The channels were then fitted in between the plates and filed down when and where required for a snug and flush fit. After the channels were fitted well, marks were made through the holes made on the arm mounting plates on the top and bottom of the channels using a centre punch. While making these marks it is important to have someone holding the end of the arm parallel to the ground as it could deflect the plates due to the moment it is imparting on them leading to inaccurate markings.

Once these markings were made, holes were drilled and filed with a round file to remove sharp edges and improve fitting.

The leftover aluminium 6061 4mm plates were then cut into 10cm x 10cm edge plates to be used as motor mounts. Once these were cut on the lathe, they were ground down to smoothen out the rough edges and the hole pattern of the motors were marked using a centre punch. After this 4mm holes were drilled and plate was checked for accuracy of fitting. It seemed to fit well so we repeated the same for 7 more plates.

Now in order to mount the motor mount to the arm, we needed to figure out how to secure it while being able to take the 27kgs of thrust the motor would produce. We decided that the same mounting holes could be used to mount ~25mm aluminium L frames on either side giving a 50mm central gap just good enough to fit the arm and on the motor arm side, 2 x 4mm holes could be drilled through and through such that they would be locking pins and would only take shear load distributed over 4 contact points. The placement of the side holes was arbitrary and a 6cm separation was chosen between the two holes. It was tough getting the holes to be concentric on the arm, however, on the L frames, it was easy to just disassemble them from the mount, put them together forming a T and drill through.

After all the holes were drilled, they were filed to remove any burrs. We also realised this assembly would be useful if in the future we found the manoeuvrability of the drone to be sluggish, we could just slide all motors inwards, re-drill the holes and take advantage of a smaller wheelbase.

This is the one part I was not very proud of since its functionality was sub-par and entire landing gear was weak, however, its design aided it in doing a very good job of absorbing oscillations while landing and take off as it allowed the frame to pivot on the roll axis.

The landing gear was constructed out of 3mm thick, 25mm x 25mm square aluminium sections. All parts were tig welded and their dimensions can be seen in the image. The idea was to have the landing gears spread outwards 30degrees from the vertical, in order to ensure this, the ends were cut at 30degree angles so that it would rest like so. The landing gears also need to be prevented from spreading too far apart and this was done by using eye-bolts, 2mm steel cable and turn buckles. The picture does this subassembly a lot more justice than my explanation in words does.

At the end of the landing gear legs we had initially bolted on castor wheels, however, they seemed to not be effective in grass and started to break off just after a few test landings so they were removed.

Since all the wires were 8 and 10 AWG, a conventional soldering iron just wasn't doing it for us and we figured we probably need to outsource this job, however, my mentor Captain, suggested we use a propane torch to heat the bullet connectors inside the XT150's and solder it. I am attaching with this step an interesting video which helped us learn how to solder with a propane torch.

Since the wires were so thick, melting the solder led at the top of the bullet connector while the wires were positioned would just lead it to overflow and not wick in the wire and make a good joint hence we had to drill tiny 2mm holes on the side and after heating the connectors, feed the solder led in from there. It is also easier to pull the connector housing onto the connector while it is still hot so we did so shortly after letting it cool a little.

We saw that with the weight of the electronics alone the arms started to flex downwards, which made us think whether with the upward thrust from the motors would flex it too much to a point of failure during dynamic loading. We tested the side and edge arms at full throttle and weren't too happy with the flexing and hence decided we needed to either reinforce the arm mounting plates or find a way to anchor down the arms using cables.

Reinforcing the plates would add too much weight so we decided to make V like structure on either side that would help anchor the arms down using cables. At the edge of the V, we mounted two 30mm, 2mm thick L brackets put together like a T and drilled holes through the leg of the T for the cables to run through and the top of the T to mount it to the V with 4mm bolts.

With some leftover scraps of the 4mm plate, we cut an 'A' like shape and bent it using a hydraulic press to give us the L as seen in the pictures. The square side of the L had dimensions 50mm x 50mm and was mounted to the arms (about 20-30cm from the motor ends) using two bolts and the triangular side had a height of 75mm and the hole in it was arbitrarily placed for the hook/turnbuckles to anchor onto.

After all the assemblies were made, the arms were mounted on the frame and the 3mm steel cable was looped and routed through V with turn buckles anchoring on the arm ends. The aluminium crimps were used to join the steel cable in a loop and crimped using the crimping tool. This was repeated for each arm. The forward and rearward arms were anchored using both V's on either side for stability as they were in plane with the V and equidistant from both.

POWER: BATTERY => ESC => MOTOR

CONTROL: RECEIVER/SERVO TESTER => ESC

Each motor had to be tested for its direction of rotation and based on what flight controller you are using, corrected for the spin direction too. This can easily be done by switching 2 of the 3 wires coming from the ESC to the motor. After all motor directions were tested and corrected, the wiring diagram from the manufacturer was followed and connections made to the flight controller, in my case the DJI A3.

The DJI A3's IMU's were very sensitive to magnetic interference from metals and despite constructing a vibration dampening mount with MDF and the squash balls as a barrier between the metal frame and FC, we were unable to calibrate the system due to interference so we swapped out the A3 for the Naza M V2 which had a similar wiring diagram. The Naza M V2 worked very reliably. The wiring diagrams for the FC's can be found online with a simple google search.

Each arm had two 22.2v batteries connected in series to give 44.4v with a capacity of 5Ah. The XT150 connectors allowed for ease of series connection due to their design.

We used zip tie guides to serve as channels to guide the ESC to FC wires (3 small wires, red black white) and not get caught/pulled out in any case.

After a couple of unmanned tests and calibrating for the manned test using a hung sandbag with the weight of the pilot, we had to design the seat for the containment of a human payload. We did so by cutting heavy duty canvas to fit in the containment and cut out holes for the legs while leaving arms/flaps that would wrap around the frame.

Once the canvas was cut, its seams had to be stitched with heavy duty thread and velcro attached at the bottom of the canvas and on the arms that would wrap around the containment frame, providing them an anchor to the frame.

The pilot was also made to wear a rock climbing/safety harness and was strapped in the containment frame using steel cables anchored at 3 points in the case the seat failed.

I understand that everyone would have safety concerns regarding strapping a human payload in such a machine but I can guarantee everyone that we had conducted multiple tests, loaded and unloaded, tethered and untethered simulating wind gusts, a moving suspended human and many other possible scenarios after which we were certain that the manned flight would be uneventful and predictable.

The first and only manned flight was conducted on July 22nd, 2017. After this manned test, a few demo flights were conducted for the press with a mannequin/dummy payload with the weight of the test pilot and the multi rotor was retired and set up as an exhibit at my high school CHIREC.

The multi rotor is now set up as an exhibit as can be seen in the picture. Below were the procedures we would follow for its operation.

SET + ASSEMBLY:

Attach the 8 arms to their respectively numbered positions using m6x75 bolts with regular washer, regular washer, spring washer and lock nut.

Attach propellers to motors (if removed); tighten bolts in opposite pairs to avoid wood compression.

Attach V channels to central square in the correct orientation using m6x75 bolts with regular washer, regular washer, spring washer and lock nut.

Attach Landing gear with spring washers on both sides + locknut m6x50.

Use the respective cabling harness and to connect the two landing gears.

Hook cables to respective arms.

Tighten all cables using turnbuckles.

Ensure all the wiring is connected correctly.

PRE FLIGHT CHECKLIST

Check bolt tightness of the arms.

Check bolt tightness of the propellers and motor mounts.

Check tension of the cables.

Ensure the wiring follows the diagram .

Spin each propeller and inspect each motor to ensure there is no abnormal forces.

Ensure all the ESC’s are turned to the off position.

Ensure all the batteries are secured with velcro and zip ties and each of them are balanced and healthy.

Connect the 3s lipo to the flight controller.

Let it initialize and calibrate correctly.

After sufficient GPS signals have been established, switch to fully auto mode (Alt. + Att. hold).

Harness yourself securely in the middle of the octocopter facing arm no. 1.

Connect all batteries using the resistor wire to the ESC’s.

Connect the two batteries in series (ESC-RB-RB-ESC) with use of the resistor to eliminate spark.

Once the capacitors have charged (5-10s later) disconnect the resistor and connect to BATT-ESC directly.

Once all ESC’s have been plugged in, arm the motors by turning the switch to the on position.

Push both stick on the transmitter to the lower-most and outward positions to get the motors to idle.

Increase the throttle to little more than half and the flight controller will handle all the controls and lift you of the ground at slow upward velocity.

POST FLIGHT CHECKLIST

Once the propellers stop spinning approach the octocopter from below and turn all the ESC’s to off position.

Disconnect the BATT-ESC connections.

Disconnect the 3s from the flight controller.

Loosen the harness and deboard the octocopter from underneath with the help of 2 people to lift it off the ground.

Check the highest temperature of all motors and ESC’s immediately after.

Disconnect low-voltage alarms from the battery's balance plug.

Check the for the highest temperature on each battery pack and ensure there is no swelling.

In the case of swelling, unfasten the battery from the arm and let it aside to cool down.

Check for battery voltages and cell voltages and log it using battery monitor.

Reduce tension in cables using turnbuckles

DISASSEMBLY:

COMPLETE DISASSEMBLY FOR STORAGE

Disconnect ESC to flight controller connections.

Remove and safely store flight controller unit.

Detach propellers: untighten propeller bolts in opposite pairs to avoid wood compression.

Unhook all cables Detach arms and store all bolts, washers and lock nuts.

Detach V channels from central square and store all bolts, washers and lock nuts.

Detach landing gear.

PARTIAL DISASSEMBLY FOR TRANSPORT

Disconnect ESC to flight controller connections.

Remove and safely store flight controller unit.

Unhook all cables Detach arms and store all bolts, washers and lock nuts.

Detach landing gear if required for transport.

CLEANING

Dust in the components affects the efficiency of the aircraft. Hence it is imperative to keep the machine clean and away from possible weathering. WD-40 should be used to remove dust from the inside the moving components of the octocopter. To clean the outsides of these components a dry cotton cleaning cloth should suffice. The components which require this form of maintenance are the following: ESCs and Motors and landing gear. The wooden propellers can be cleaned using a dry cotton cloth to remove the settled dust.

STORAGE PROCEDURES

After the cleaning procedure disassemble the entire structure, store the disassembled parts and components in dry and safe location.

For long term storage, the batteries must be discharged using turnigy system.

The propellers must be securely stored in their respective box with no load above the box to prevent cracking and damage to the propellers.

MAINTENANCE:

SCHEDULING

If octocopter is not to be in use for over 60 days completely disassemble and follow storage procedures

Follow cleaning procedure every alternate flight, before and after storage.

Grease all bolt threads every 5 flights and replace all nuts and bolts every 6 months.

The manned octocopter was designed for an estimated lifetime of 50 flights.

BREAKDOWN

Check configuration of flight controller.

Check ESC to flight controller connections.

Check the battery to motor connection.

Evaluate condition of all components.

If issue persists isolate source component and replace. DO NOT REPAIR AS FUNCTIONING OF THE COMPONENT MAY BE COMPROMISED.