The Quest for Lift

There are many kinds of propellers available for multi rotor applications, and while I started with the basic APC props shown bellow, they are not enough for the speed and weight of the hexacopter.

Thus I not only needed to go for larger propellers, but also for stronger ones. Carbon fiber propellers were almost the obvious choice.

The third and forth option is Graupner props, these are professional propellers, and since they are only made in Germany, they are very expensive. 

So there are still plenty of options. Wood was not used because its the heaviest of the three. 

Flashing Custom Firmware: OH THE HORROR!

   Well I knew this would have to be done sooner or later but what I had to realize is that pretty much every micro-controller needs to run on custom firmware. The 6 Atmega 8P based ESCs run on SimonK firmware, the ATMega2560 on the APM 2.5 runs on custom hexacopter firmware as well as the camera boards I am planning to add. This means a lot of manual flashing for me, so here is some details on how that was done.

Come to think about it, the haxacopter is like an Octopus in that it has multiple brains, one central main brain and several smaller brains in the arms.  So yeah, a cute little analogy here.

This is the main section of 1 of the ESCs, yu can see the voltage regulators as well as the Atmega 8P micro controller, a crystal oscillator, etc. Its really a small, full featured micro controller.

This is the back of the ESC, it reveals a large discharge capacitor as well as an array of 12  high current MOSFETs for a total peak continuous AC driving power of 30Amps with a peak power of 40Amps for 10 seconds on average cooling

To note here is that what the ESCs do is primarily convert DC voltage from the battery into AC voltage for the brushless motor. Secondly, they regulate a steady 5V @ 2A power supply for the APM and the rest of the system, assuring operation of the device. Since they are used as drivers for the motors, they will obviously also get throttle input from the APM. 

Normal ESCs will take the average PWM control single from the Rx and average the values to get a good overall speed for something like a plane, on a multi rotor on the other hand you need to have rapid and accurate throttle response. The SimonK and similar ESC firmwares bypass some of the features on the ESCs such as voltage protections and other unnecessary add-ons, favoring throttle response and thus allowing Multirotors to "Lock" in a direction, thus removing the whole that comes from the flight controller trying to stabilize the multi rotor.

First of all all flashing require a USB ASP or similar device .

The first method of flashing is smothering 6 wires to incredibly small and fragile pads on the controller, just for reference, the picture above is using 16 gauge if not 24 servo wire.

This method is slow and annoying. I've decided to not go for it.

The good folks over at Hobbyking have come up with this idea. Instead of having to solder to the actual board, you can simply touch the pins of the ATmega 8 and program it that way. Its fast and works wonders. 'Nuff said. Its about 20$ for the tool but it will save you a lot of defective ESCs, talking first hand here :)

On the software side I used the KK Flash tool, it works great and its kept up to date, I have registered some errors but overall its fine. 

A USB ASP device

I have made a tutorial on flashing using the KK Multicopter Flash Tool on actual KK 2.0 boards, although the software is compatible with many more boards and ESCs. 

Problem with the Xbee Adapters

The APM takes data over a protocol called Mavlink and Xbees are one of the supported (and cheaper) communication methods. Mavlink was designed to be a very fast data protocol as to allow for actual live feed of data from fast moving vehicles or sensor arrays, thus it requires a high nitrate of about a 11000 baud rate. The Xbees are capable of that but are set to 6200 baud by default. Thus we need to change the baud rate in the XCTU utility through the terminal. This implies reflashing the firmware to the Xbee (I'm using Pro modules but its the same for standard series 1 modules).

    The Xbee pro modules run over  900mhz radio waves. Thus they do not interfere with the main 2.4Ghz control line of the Hexacopter. On the other hand they might disturb the 860Mhz radio signal coming from the Video receiver. I'm going to need to mess around with cloverleaf antennas and the channels used on the two devices later.

     Comming back to the main problem, in order to connect the Xbee to an X86 PC, an adapter board must be used, I chose the once form Paralax as it was good quality and a proven board.

   What becomes immediately obvious upon examining the USB adapter is that there is no reset button present. This is a major problem as the Xbee modules will be in a soft-brick after flshing, meaning that I will have to reset the xbee while powered. To do this, I have to "short" the RST pin to one of the ground pins. For some reason, the Xbees are not posting at all after doing this, possibly being Hard-bricked. I will have to look further into this but I am not making it a priority as it is more of a side/safety feature that can be added anytime later. 

Once functional it will enable me to see live telemetry from a laptop as well as control the APM 2.5 through something like a joystick or a gamepad, thus making it far more accessible for beginners, and funner with something like the Wii mote ;)

Battery Changes

     Today I realized that two 2200mah 3S batteries are not going to be enough for any susteined flight.  Thus I am moving towards bigger batteries, much bigger.

This is also going to address the weight problem as the hexacopter is currently too powerful for the amount of power its producing. Thus the batteries will get it above the current 17% hover throttle to at least 30+%, an ideal value being between 40-60%.

I am thinking of the following batteries:

Or possibly something like this:

So these options should be fine in series with XT60 connectors, will probably have to use XT90 at the endings as XT60 is only good up to 90Amps and I could be drawing 180Amps between the two batteries. XT90s are generally good for 220Amps each. They are huge though. 

I'm still looking at a better power delivery. There are different options

Ardupilot based Hexacopter Workflow

Andrei Aldea

DeadCat MK1 Project Documentation

These are the project notes for my Quadcopter build. I will try to include as much information regarding everything that I have done as well as provide the best and most pictures that I possibly can.

The Frame: I decided to go for a frame that had not become very popular yet, inspired by the Orvillecopter.

What makes this type of frame unique is the angle at which the arms are spaced. In the Dead Cat type configuration (the origin of the name is obvious), the front arms of the quadcopter are pushed back, allowing for easy mounting of camcorders since the propellers won’t get in the shot. The rear arms are at the same angle as a standard X configuration quadcopter.

The frame I used is a modified XK450 frame, similar to the DJI Flamewheel. It is shown below with the optional extended legs, which I used. Its a standard X configuration, but I wanted more.

So this is the SK450 with the Dead Cat modification. The Long legs above are used on the arms.

The Motors:

For Motors I decided to go with Turnigy NTM Prop Drive 28-26 1350KV Motors. These are AC Motors, so they are very efficient, they will pull a max of 30Amps each, they also have a very good thrust to weight ratio with 8 inch props. According to the website, their specs are as follows:

Poles: 3

Motor Wind: 14T

Max current: 18A

Max Power: 227W @ 11.1V (3S) / 302W @ 15V (4S)

Shaft: 3mm

Weight: 56g

ESC: 25~30A

Cell count: 3s~4s Lipoly

Bolt holes: 16mm & 19mm

Bolt thread: M3

Connection: 3.5mm Bullet-connector

Prop Test Data:

8x3.8 - 11.1V / 158W / 14.2A / 0.736kg thrust

8x3.8 - 14.8V / 310W / 21A / 1.1kg thrust

.

This means that the motors will spin at 100% throttle 16200 RPM and 22005 RPM depending on the battery voltage

The Batteries:

Specifications:

Capacity: 2200mAh

Voltage: 3S1P / 3 Cell / 11.1V

Discharge: 35C Constant / 70C Burst

Weight: 199g (including wire, plug & case)

Dimensions: 115x35x27mm

Balance Plug: JST-XH

Discharge Plug: XT60

The Flight Controller:

Other Components:

How They Work:

High strand-count silicone insulated average Amperage tolerance:

• 8AWG 200 amps
• 10AWG 140 amps
• 12AWG 90 amps
• 14AWG 60 amps
• 16AWG 35 amps
• 18AWG 20 amps
• 20AWG 12 amps
• 22AWG 10 amps

Update October 17 2013: The Xbee Pro modules are refusing to change baud rate and have been soft bricked, I’m on the phone with Digi and Adafruit to see what can be done. Until I get them working, telemetry will not be active.

Update October 19 2013: A shipment from Hong Kong will arrive at JFK with the parts I need to finally assemble the quadcopter, it includes some video equipment as well as 10-14AWG wire and high diameter heat shrink tubing.

This is how my frame was destroyed… proving it was not a very good frame:

http://www.youtube.com/watch?v=jhNKEC8GqtI&feature=youtu.be

Now I’m waiting as of October 31 2013 for new parts from Hong Kong. Its extremely frustrating…

Update November 11 2013: I’m encountering some technical problems. I am going into deeper issues.

I have done some calculations on the motors, assuming Linear results rather than curves. They should provide a fairly close match to real life results:

Model: NTM Prop Drive Series 2826 1350kv

Kv: 1350rpm/v

Poles: 3

Motor Wind: 14T

Max current: 18A

Max Power: 227W @ 11.1V (3S) / 302W @ 15V (4S)

Shaft: 3mm

Weight: 56g

ESC: 25~30A

Cell count: 3s~4s Lipoly

Bolt holes: 16mm & 19mm

Bolt thread: M3

Connection: 3.5mm Bullet-connector

Prop Test Data:

8x3.8 - 11.1V / 158W / 14.2A / 0.736kg thrust

8x3.8 - 14.8V / 310W / 21A   / 1.1kg thrust

This results in the following assuming linear curve (might differ slightly):

9x4.5 - 11.1V / 211W / 18.9A  / 0.980 kg thrust

10x4.5 - 11.1v / 234W /   21A   / 1.098 kg thrust

11x4.5 - 11.1V / 258W / 23.1A / 1.198 kg thrust

12x4.5 - 11.1V / 281W / 25.2A / 1.307 kg thrust

Note: We see that 4S 8x3.8 is equivalent in thrust with 3S 10x4.5 in amperage.

This is all assuming linear thrust, real data might be different as the motor has in reality a curved thrust.

Max Current for the motor is rated at 18A so it might be pushing it...

For the Sake of it 4S (14.8V) configurations. They are not faisable!

9x4.5 - 14.8V / 413W / 18.9A / 0.980 kg thrust  //The calculations are not made yet. Results are 3S!!!!

10x4.5 - 14.8v / 234W /   21A / 1.098 kg thrust

11x4.5 - 14.8V / 258W / 23.1A / 1.198 kg thrust

12x4.5 - 14.8V / 281W / 25.2A / 1.307 kg thrust

Now to compare to predicted weight of the hexacopter:

Lift capacity = (Number of motors * thrust/motor) / 2 (50% capacity or less is ideal)

Overall theoretical lift by 3S:

 9x4.5 -11.1V  / 5.88Kg Overall - 2.94 Kg Ideal

10x4.5 -11.1V  / 6.588Kg Overall - 3.295 Ideal

11x4.5 -11.1V  / 7.188Kg Overall - 3.594 Ideal

12x4.5 -11.1V / 7.842Kg Overall - 3.921 Ideal

10x4.5- 11x.4.5 Ideal Overall