FPV Aircraft

A racer drone has a fundamentally different objective to the acrobatic freestyle drones, which is to be as fast and agile as possible. Depending on the size of the course this might not be just top speed, but more about acceleration and direction changes.
To achieve this performance having a high thrust-to-weight ratio is important, but as I was using my old motors this wasn't going to improve thrust, and so the main focus was on weight reduction.

A lot of components of the drone are essential for it to function, but one of the components which isn't is the GoPro. As the objective was to race instead of filming HD footage, not only could the GoPro be removed, but the mount could be removed, and a smaller frame could be used as there didn't need to be space to mount the GoPro.

A standard freestyle frame has an elongated middle section, which provides space on the top plate to mount a GoPro on the front and the battery behind that, but this means there is both an elongated top plate and bottom plate, and they require 8 standoffs to support them, which accounts for a lot of weight. As there is no need to mount a GoPro, and it's better to bottom mount the battery on a lighter drone, a racing drone frame has a much shorter middle section, a much thinner top plate, and less standoffs.

So a frame like this specifically designed for racing was chosen, which had a very short middle section and very thin width arms, so it was much lighter than the typical freestyle frame. Due to the arms and also for additional weight saving I chose a 4-in-1 ESC, which could be mounted on the same stack as the flight controller and minimise the amount of hardware used. The shortness of the middle section meant there was much less space for the analogue camera in the front, so a micro sized camera was chosen.

Significantly reducing the weight of the drone took a large amount of load off the motors, which combined with the lower pitched propellers typically used on the freestyle drones meant the drone was incredibly responsive in low end acceleration but lacked control in cornering. Reducing the load from weight saving gave some overhead on the motors, so I experimented with higher pitched propellers that give more top end speed and more control in cornering at the expense of a higher resistance on the motor.

Third freestyle Build

The second freestyle drone was a monumental improvement from the first, but having improved my piloting skills through extensive flying and also experiencing the speed of the racing drone, I wanted to push the performance of a freestyle drone.

To improve speed and agility a key factor is thrust-to-weight ratio, and for the racer I was able to drastically reduce weight due to the smaller frame and removal of the GoPro, but for my freestyle build I still wanted to have that larger sized frame to carry a GoPro. So unable to just cut weight drastically, I needed to try to cut weight, but also increase thrust.

To increase thrust I upgraded to a set of 2207 2700kv motors. As explained before, a higher Kv motor results in a higher rpm at the same voltage, so increasing from 2400Kv to 2700Kv provided more thrust.
A wider stator (23 compared to 22) gives the magnetic force more leverage over the motor bell, resulting in highly granular control on the bottom 25-50% of throttle. That made micro adjustments easy and helped me gain confidence flying lower to the ground on the previous build.
A taller stator (07 compared to 06) allows for taller magnets which provide the motor the torque needed to spool up faster from aggressive throttle changes, resulting in snappy responsiveness and explosive peak power. So now that I had gained confidence flying lower and wanted to incorporate more snappy moves experienced from the racer in my freestyle flying, a 2207 motor was better suited.

Similarly to the racer build, as there is kind of a minimum weight due to the necessity of some components, one of the main ways to reduce weight is by using a lighter frame. But as stated above, I couldn't select a racer frame because I still wanted to be able to mount a GoPro, and so I researched different frames to find a lighter freestyle one.
The frame still had the same elongated middle section of a freestyle drone, but was lighter from utilising thinner depth bottom and top plates, much thinner width arms, and generally being a more compact design. The way in which the drone was designed meant that it used less hardware, and the choice of material in the hardware and mounts was lighter, for example the rear standoffs were moulded plastic, which housed a moulded antenna mount, compared to the large 3D printed one on the previous frame.

Increasing the RPM of the motor means the propeller is going to be spun round faster, which creates more resistance and requires more torque to combat this. However, Kv (speed per Volt) and Kt (torque per Amp) are inverses of each other. So not only does a higher Kv motor require more torque to overcome the increased resistance, it requires more Amps per unit of torque. This massively increases the current draw of the motors, so I upgraded to a 60 Amp ESC. I chose a 4-in-1 ESC because I was impressed by this form factor on the racer build which also had thinner width arms, and its compact design also aids in weight saving.

3D Printed GoPro mounts

The second frame came with a 3D printed GoPro mount, but this frame had a moulded mount that didn't provide as much protection and was at a different angle to what I used for my analogue camera. And I also wanted to start using ND filters to improve the footage, but due to the form factor of the GoPro session the ND filters couldn't click onto the front, and so the only options were ones that stuck on, but then you could only ever use one strength of filter, or glass ones, but they needed some way to be attached.
So I worked with my cousin on designing and developing a 3D printed mount to solve these issues. I helped with the practical side of the process, including testing the mounts during flying and identifying issues with the fit, strength, vibration and camera view, which were then fed back into the design.

There were a few constraints to work around. The mount had to line up with the existing screw holes in the frame, as it attached through slots in the frame and into the existing standoffs. It also needed to be kept as light as possible without becoming too fragile.

The mount was split into three main sections. The first was the flat base that attached to the frame, following the shape and screw positions of the front section of the top frame. The second held the GoPro at the chosen angle, while leaving enough access to the buttons and screen and allowing the camera to be slid in and out. The third section formed a wedge between the base and the GoPro section, providing the support needed to hold the camera at the required angle.

We inevitably ran into issues with the first versions. The borders around the front were either too narrow, becoming flimsy and not holding the GoPro in place, or they were too thick, supporting the GoPro but intruding into the video. The stiffness of the mount was either too flexible, causing the GoPro to move around under the loads of the flight, or too rigid, with the mount breaking from a hard impact. There was also the issue that when choosing to lower the angle of the GoPro, the gap between the front screw holes and the bottom of the GoPro section was decreased, which made it really difficult to access them. Additionally, the depth of the GoPro section was either too large, not holding the GoPro and ND filter in securely, or too small, putting lots of strain on the mount.

The shape of the border design was iterated to gain additional thickness in areas that wouldn't be in frame, to ensure the mount held the GoPro securely while not intruding into the video. Different materials and infill combinations were experimented with to find a balance between rigidity and durability. Additionally, holes were added to the bottom of the GoPro section to allow access for allen key to the screws.

Trend from 4s to 6s batteries (and failed test)

I flew this drone for an extensive period of time, and felt like the design was pretty refined for freestyle flying. The frame was on the lighter end of those of that form factor, the 2700Kv motors were really smooth and the Kv was at the top end of performance. I had experimented with different propellers, to find ones with the best balance of lightness, agility, top speed and durability. The GoPro session 5 was the lightest of the available action cameras to provide quality HD footage. And so the build was at a nice spot for freestyle flying.

But as time went on, there was a general trend in the industry from using 4s to 6s batteries. For the main reason of less battery sag. Battery sag is when at high current draw, the internal resistance of the battery causes voltage to drop, causing a temporary loss of performance (lower RPM/thrust) that recovers when throttle is reduced. This limits long periods of high throttle maneuvers and near the end of the flight (when the voltage is already low), the voltage can suddenly fall off a cliff, often resulting in a crash.

The relationship between electrical power, voltage and current is:

\[ P = VI \]

So using a higher voltage battery will provide the same power at a lower current, which significantly reduces battery sag, and as the starting voltage is higher, a temporary drop in voltage will feel less noticeable on performance even if the battery starts to sag. Both of these mean 6s batteries allow for more sustained periods of high throttle, provide better recovery from dives, and provide more consistent performance throughout the entire flight.

The theoretical motor speed can be estimated from the motor's KV rating and applied voltage:

\[ RPM = K_V V \]

So because 6s is a higher voltage you need to have lower Kv motors to achieve the same RPM.

My cousin had a 6s battery, and so to test out using 6s batteries before buying new motors, I used a setting on the flight controller that could effectively limit the Kv of the current motors. However, I hadn't accounted for the increased energy spikes, and so at low throttle the drone flew fine, but at higher throttle the voltage spikes exceeded what the capacitor was capable of absorbing, which fried the MOSFETs on the ESC. Due to the higher voltage of 6s, and the larger energy spikes and harsher electrical noise, both the voltage rating and capacitance of the capacitor needs to be increased.

From theoretical advantages, online reviews and videos, and the testing I had managed to do, I was impressed by 6s batteries. So I bought a new ESC, a higher voltage and capacitance capacitor, and a few 6s batteries. The flight controller setting to limit the motor KV was working smoothly, so I didn't need to purchase new motors.

Fourth freestyle Build (new frame)

Then the upgrade to using 6s was another leap in performance, enabling larger dives and longer periods of high throttle flying.

But although performance was improved, the build wasn't fundamentally designed around using these new components. The only components still being used from the initial design were the FPV camera and flight controller. The 6s batteries were larger and heavier than the 4s ones, the new capacitors were larger, and the new VTX was larger. This meant that the layout of the components in the rear was messy, and there were concerns about the strength of the top plate.

This highlighted a design choice that needs to be made. Do you try to optimise purely around weight saving, or also factor in things such as ease of use and durability? For example the previous frame's rear section tapered in, which saved weight, but made it really difficult to secure the VTX, RX and capacitor.

So I decided to put all of the components on a new frame, which was slightly larger. Two main parts of this were the increased space in the rear section, to more securely mount the VTX, RX and capacitor, and a larger top plate, for more secure mounting of the battery. The frame also has specific slots to mount capacitors, more durable front standoffs, a raised plate to reduce oscillations to the GoPro, and a split bottom plate which allows for easier replacement of arms.

So the design choice I made was to not purely optimise for weight, in order to have a securely mounted, durable setup, for smooth flying and video footage. There is also a case that depending on the flying style you want to have, a heavier drone will carry more momentum when off power, and so if this is advantageous there is no trade off to a heavier build. But if I also wanted a really lightweight build, the vision was to put a new set of really smooth motors on this heavier rig, and then use the previous frame and current motors to build a lightweight 6s drone.

Lessons from extensive flight testing

One thing that is inevitable from flying FPV drones, is crashing. Now if a crash happens in an environment that is visible and easily accessible, you can usually see the drone (or are told by the spotter), and walk right over to it. But sometimes a crash happens where the drone isn't visible or easily accessible, for example in a crop. If the battery has been disconnected, then there is not much that can help locate the drone other than the spotter, or watching recorded analogue footage to take more notice of the surroundings. But if the battery is still connected, there are a couple of things that can be done.

If you have a video signal, the video receiver on the goggles uses both an omni directional antenna, and a directional antenna. So you can unscrew the omnidirectional one, so whatever direction gives the best signal, is the direction of the drone. The signal quality can be gauged by either static, or the receiver indicates percentage signal strength.
Similarly, if you have a radio signal, the antenna used on the TBS Crossfire radio module is directional, and this module indicates signal strength. So a combination of scanning directional antennas for both the video and radio systems, to find the strongest signal, can help locate the drone.
If you have a radio signal you can also arm the drone for a split second, and it will make a noise, but this is only helpful when you are within close proximity to the drone, to locate it exactly.