Section 1: The newer prototype.
Introduction
The failure of the Singlecopter brings back the idea of s quadcopter. This time, rather than making it large, bulky, and full of "DIY" components, the quadcopter is fitted with more "Plug-and-Play" components such as an ELRS receiver.
Component Changes
It's determined that, despite being cheap, using LoRa for the command & control link of the UAV is just too heavy, pin-consumptive, and has a much steeper learning curve that might lead to a blow-up in the budget and will require much more time. LoRa was scrapped along with this iteration of the frame, and ELRS was ultimately chosen for the next iteration despite its larger price per unit, the frame was also changed into a much simpler X-frame. The rest of the components stays intact however.
To reduce the weight, the frame, which was initially made of plywood, was changed to a purely balsa-based frame. According to various tests, a single 720 coreless motor with a 65mm propeller should be able to generate up to 25 grams of force, this number is a maximum which should not be reached, that means the UAV needs to be below 50 grams to ensure a safe flight.
The total weight – excluding the battery – is then calculated to be 48 grams, this is excluding the battery. For a quadcopter to ensure a safe flight, a TWR of 2 is required, so this frame is still overweight for a safe flight. Despite the weight being severely lighter than the previous iterations, the frame was too fragile for the UAV to be stored in a container and then transported. The propellers were slightly skewed, which will affect its performance especially in stability.
Section 2: Flight Testing, Validation, and Closure
First Flight Test & Crash
This time, a flight test was finally able to be conducted, unfortunately the UAV instantly crashed and the frame snaps into three different pieces, one of the motors was also pronounced dead at the crash site, and with this, the first X-frame is scrapped, and a newer hybrid frame which is smaller, yet more structurally sound is used.
(left: older frame, right: newer, more compact frame)
Second Flight Test
In less than a day, the new frame was finally ready to set up for the skies, however, it again instantly flipped. This time, the test was conducted on a soft surface, and the UAV was saved. The root cause was then found to be a flaw in the PID logic, which rather than multiplying the error and then summing it up for the “integral” signal, sums up the integral gain of the controller. The project then proceeds to a third test flight, which the UAV managed to lift itself up to a height of around 15~20cm above the floor, however then it starts to flip again, and when a cyclic input was given, the UAV all of a sudden flips and falls. The UAV did not even manage to hover for a few seconds, and it ended up completely destroying the STM32F103C8T6 and the ELRS module.
The STM32 has component redundancy and, and so the next flight controller was easily built, however, the ELRS module is the most expensive component of this UAV, and the crash destroys it. As an interim means of RC Link, a single ESP-01 ESP8266 module was used; it receives JSON-typed data from a remote control software developed using MITAppInventor. The ESP-01 then reformats the data so that it imitates the signals ELRS modules usually use.
Fourth (Final) Flight Test
The third test flight was able to be conducted later on, however, the UAV crashed instantly. This crash leads to the discovery that the format was accidentally reformatted again so that the ESP-01 and the STM32 ended up mis-communicating. This made the development went stagnant for around a month, until the replacement ELRS receiver finally came.
While waiting for the replacement receiver to arrive, an inspection of the code for the firmware somehow shows one wrongly placed negative sign in the motor mixing matrix, this explains the reason behind the "flipping" of the UAV which was done due to a wrongly actuated feedback loop.
After the replacement ELRS had arrived however, a ground test was conducted, which ironically fried the STM32F103C8T6 for no known reason. The last remaining STM32F103C8T6 was then used to further test this UAV, which ironically, ended up similar to the previous STM32F103C8T6, but this time, the UAV seem to work just fine until it got stuck in 100% for no known reason. These 2 incidents leads to the conclusion that the STM32F103C8T6 was not adequate to act as a Flight Control Computer for UAVs, especially if the UAV is more and more complex in terms of coding. This clarifies the limits of the STM32F103C8T6 as a flight control computer, as the maximum limit for this particular microcontroller unit (MCU) is a simple quadcopter fitted with basic PID without any forms of sensor fusions, especially if the program was optimised enough.
The next action plan is to actually attempt in using something still considerably cheap, yet much more powerful, such as the STM32F401CCU6. However, based on experiences with the Coreless Quadcopter, the configuration is deemed to be too power consumptive, and so, for the next project, a coaxial-styled helicopter is used instead.
External Links
ESP01 to CRSF Code:
No comments:
Post a Comment