I rarely have a single project going at a time, and sometimes a project has to be set aside because of price, lack of skill or lack of interest. I hope to come back to these projects some day but for now their documentation will live here.
DIY Inspire 2 knockoff drone
I love Aliexpress, I know some people think its just trash but there are diamonds in the rough. I am fortunate that my line of work requires me to qualify manufacturers, which gives me an edge when contacting sellers on Aliexpress to make sure their product meets my needs.
Years ago my Grandma bought me a DIY drone kit, it came with everything needed to put it together and fly. I loved it. I flew it around my highschool’s field way back before drones were popular or regulated. There was no FPV back then, I just had a friends GoPro strapped on and could watch the video later. After many crashes this drone was in shambles. I repaired it once a couple years later but eventually I lost this drone during one of my moves. 10+ years later I finally decided I wanted a drone again, except this time I wanted it to carry a lot more than just a gopro.
While browsing Aliexpress I stumbled upon these
If you know anything about DJI you know that their drones are expensive, and the Inspire 2 drone is especially so. This 4 pack of motors has a total lift capacity of 8 kilos or 17.5lbs, and coming straight from the manufacturer allows me to get them for far cheaper than if I bought them through DJI. Just for reference this is the drone they are used on.
After finding these motors I knew I wanted to build a heavy lift drone to carry a full frame camera for aerial photography, something I have never done before but I am confident I can achieve. I began research trying to find out what components I would need and what features I wanted to match. Using ChatGPT I discovered that I can match the DJI’s waypoint missions, auto level, auto pilot and auto land features by using an off the shelf flight controller call a Pixhawk. These are incredibly powerful and fully featured flight controllers capable of far more than I can tell you about here, I recommend watching some videos about Missionplanner and the Pixhawk flight controllers. I decided on the Pixhawk 6c mini due to its compact size, inclusion of parts like GPS, lower price, lower complexity and full access to Mission Planner and Ardupilot.
Along with this flight controller I also needed motor controllers for each motor, a GPS for location control, a remote and receiver and a battery, along with a frame of my own design to house it all. I continued using ChatGPT to help me research components that are compatible and meet my needs for price and performance which led me to the below parts list. I decided against including any cameras or first person view (FPV) headsets in my parts list as my initial goal was to build the drone, later I can add the camera systems.
For the battery I decided to temporarily use a Dewalt 8 AH 20V power tool battery. I chose this because I already have two and they match the voltage requirements of the motors. RC aircraft batteries are very expensive, and by temporarily using batteries I have on hand I can significantly reduce the upfront cost.
In order to design the body of the drone I started by modeling the motors based off of images and dimensions I could find online. I then downloaded CAD models of all other components from Grabcad and ended up with a digital parts kit I could use to assemble the drone.
In this parts kit you can see from left to right: ESC motor controllers, GPS, Pixhawk 6C mini, Dewalt battery with adapter, 25mm OD carbon tube, 10mm OD carbon tube, self designed 25mm T, and 4X reverse designed motors with 15 inch propellers and self designed motor mounts with provisions for wiring.
With my digital parts kit complete I began placing my components in their final locations, roughly basing my design off of the DJI inspire 2
I tried to place the battery in the center of gravity and place all the other components around it while making it as compact as possible. I also added a threaded insert to the bottom where I could connect add on’s at a later time. From here I designed a simple body around the components while making considerations for 3D printability, assembly and later expansion. I made the lid removable which allows the body to be printed without supports and allows the lid to be replaced with another design if you wish to expand its capabilities.
I then added a rear battery cover which will be on a hinge, this allows for quick change of the battery which uses the Dewalt quick release button. You can see this battery cover below in reduced opacity to allow sight of the battery inside
I next added a front casing to the drone body. Again I made considerations for later expandability and printability my making this piece removable by 4x m3 screws. This initial cover you see below was designed to fit two Gopro’s inside, one pointing forwards and another angled down for a view of the ground. Later front covers could be designed to hold a full frame DSLR camera if wanted
DIY Low Cost CANBUS Accessory Gauge
My goal is to build a gauge that can read a car’s CAN Bus data and display data like engine rpm, oil temp, coolant temp, air fuel ratio and many other pieces of information that you may want to see in a race car. I wanted this gauge to be very compact and affordable, allowing someone to purchase a single gauge at a time and as they build their car they can add more gauges. This gauge is built using a 1.28 inch circular display that has a built in Raspberry Pi pico, this allows me to use existing micropython, C, Arduino or Cpp libraries. In order to communicate with the cars CAN bus, I am going to use an MCP2515 which is a standalone CAN controller. There are rp2040 libraries for this MCP2515 written in both micropython and C, though after some research I have decided on using C as micropython may be too slow for this project (In this choice I lose the ease of programming that comes with Python)
My first task was to just learn how to program this display. To do this I downloaded the displays example code from its documentation page Here
Along with the displays documentation I had to follow the documentation for the rp2040 chip. This documentation provides a quick install link for the pi pico visual studio IDE and code compiler. Follow section 9.2 to install the IDE.
Now that I have a coding environment and example code, I started by changing small things to see how the display reacts and ensure that my understanding of the code is correct and that I am able to upload the compiled code to the display. Initially the display showed a Chinese logo followed by a bogus IMU readout, my first test was to replace the IMU readout text with my own text saying “Test”. After I was able to compile the code and upload it to the pico I confirmed that the second screen now said “Test”. This took me at least four hours.
From here I set out to replace the Chinese logo with my own logo. I found the Imagedata.C file and discovered that the logo was stored as an array of pixel data, this is nice because I can use online converters to convert my logo to a C array and paste that in place of the original array. It was not as easy as I had originally expected, it was actually much easier.
Above you can see that the logo is too large, the colors are off and the text is very grainy. After a LOT of struggling I made a discovery. The library that the display is using is the LVGL embedded graphics library, and the creators of this library also built an online conversion tool of their own that creates the entire C file used. Once I discovered this it was far easier to upload my image, though I did have to resize it a couple times to get it perfect.
Here is the link to the image converter, make sure to choose “CF true color” and “C Array”. Once you have the C file, copy and paste all the If statements that contain the image data, also copy and replace the .data-size at the bottom of the page. lv_confg.h has some settings you can change for image data types.
My next goal is to change use my logo as a loading screen that shows up on boot for 2 seconds then gives way to the data screens. The LVGL library calls each screen a “Tile” which can be scrolled through. The way the example code is set up is that the IMU detects a lift or shake of the screen which then gets turned into a simulated rotary encoder which acts as the input signal used to scroll through the tiles. In later versions of the LVGL library there’s a set tile function that would have been incredibly helpful but at the time of writing this I have no idea how to update a library. In order to show the logo then transition to the next tile I opted to stick with the simulated rotary encoder, after three seconds of booting up a timer sets the simulated encoder to position 2. Later on I can add a button on the back of the display which is used to set the tile, this way you can click a button to keep scrolling through the different tiles to show a different piece of data.
