https://drive.google.com/open?id=1j7faVSYusqAtOf6j0K5guSZnvAAybGwt
Link above is a google drive link to download the custom built VESC controller. (Note that I have designed using Altium designer, you may need to use altium designer or circuit maker to able to view it)
Hi, I wish to share my work which I have done for my final year project is to design a electronic speed controller for UAV application. The purpose of this design is to reduce the size of ESC to equivalent electronic speed controller found in the market. The video of the testing of custom Built VESC is in the link below.
https://drive.google.com/open?id=1C9Z8OBv1kUSU0SGAbeJvfhMMl3X7-sxU - link for video
As can see from the video, the motor I am using is T-motor 170KV fitted with a 28 inch propeller. I am able to turn the VESC to start the motor at 8 A but if I control it from DJI A2 flight controller the tendency for motor to jam at start with a high pitch sound is high. The control mode is in duty cycle control mode. It is possible to run at PID speed control mode.
Here is the list of 2D view and 3D png view
https://drive.google.com/open?id=12iCWiM4dinO5eCjTpMFo0dHwpXSkizvF -power board
https://drive.google.com/open?id=1ufgMx54i3iKiHnsTx0UpRPaPskoCkia7 - control unit
https://drive.google.com/open?id=12iCWiM4dinO5eCjTpMFo0dHwpXSkizvF- control unit 3D
https://drive.google.com/open?id=1hC3zveAax8t0LDM6rIDTZ5Xnh-i9eU0J- power board 3D
https://drive.google.com/open?id=10eLOHAkvbW3f1B8ccPYusECfrd1no4E6 - Assembly of power board and control unit
The problem with difficult startup might be linked to inertia of the motor is low and possible low inductance value of the motor winding. I have tested on Tmotor u8 and Tmotor U7 pro v2. (I might not be correct)
The performance I have tested is by run the custom built ESC for 10 minutes and record the temperature of the VESC using VESC tool and the temperature will rise but will keep steady at 50 degree celcius. If kept at constant speed the increase of temperature will be linear with time but if dynamic speed- such as sudden high and low speed with active braking the VESC will be heat up faster. To compare with other ESC running with the motor such as Tmotor80A, the motor would heat up if using Tmotor80A but it less likely to heat up when using custom-built VESC. To make up of the space available, I have remove CAN communication, Analog control port and other non necessary communication ports .
The cost to built 5 of this custom built VESC cost around RM750 which about 179 USD. The cost might be reduce if is in mass production. The most expensive part would be the 6 layer PCB on the control unit which I have ordered from PCBway with cost around 190 USD excluding shipping and import duty tax. There are certain improvement needed as I done my final year project and it would be costly to do a version 2 without funding and more analysis.The improvement would be placement of shunt resistor between inverter and motor phases and heatsink casing for it.
I am sorry if the details of the custom built VESC such as how to perform configurations, performance and other criteria is not provided completely. I would like to share my project as and appreciation of existing community and open source platform that allow me to understand more about AC motor control as a beginner. I do not have intention to patent it for production, you may use it and further improvement if possible. Thank you
Nice work!
Questions:
1 Why have the gate driver on the top board
2 Why 6 layers ? could be done with 2/4
Hi,
1. The gate driver on top of the board instead of the bottom of the board or any other location is to prevent heat from phase current transfer to the gate driver. Additionally, I have not enough time for prototypes as i am given at most 6 months to complete the prototypes with proves of it is working and test analysis being done. Note that the connectors between the board is direct soldered instead of using plug/receptacle connectors to mitigate vibration of the body that will cause noises to the signal received by microcontroller or MOSFET.
2. The more layer of PCB will provides better return path adjacent to incident path of signal. Given such small constrain size of 3*6 cm there would not be enough return path (ground signal if you may understand different way) and return paths will crossed which will create crosstalk which is not desirable in PCB designs. Additionally, having more PCB layer will act as shielding to improve against EMI. The MOSFET driver alone has multiple frequencies and voltage levels flowing through, which some of them is sensitive such as SPI port, VSense of switching regulator, and voltage level to the MOSFET gate with high frequencies at least 12V. Different voltages will present an aggressor and victim which you would know aggressor is the high voltage and victim is low voltage. Provided that the PCB which act as ground plane is properly grounded or else will make things worst due to it has become antenna which to amplify the noise.
Jordan
Hi Jordan,
cool project! It is impressive to see that big propeller spinning by this tiny board. What is the max current you measured - 25A @ 25V = 625W?
For me, the most valuable in this modular design (Control PCB + Power PCB) is to be able to switch the power stage for high power systems and keep the same control board (MCU + Driver). This is also popular design on the market. Especially, a few years ago when you needed 50+ MOSFETs to drive 50A. I am still using 100A Red Brick ESC with 4 PCBs - 1 control board and 1 power board per each phase.
As for the soldering vs relying on the standard connectors - vibrations would not be an issue. Especially, if your concern is mechanic noise causing electric noise. This is my experience with electric and internal combustion engine (combustion == vibration) UAVs.
The 6 layer PCB seems overkill to me. However, I am more software than hardware guy. But one of the first things they teach you in CS class is "If you can't measure it, don't start to optimize it." So to start with 6 layer PCB this early in the project you should either have a lot of experience with measuring EMI or as in your case overkill the prototype to meet the deadline and ignoring the manufacturing costs. I think both PCBs can have 2 layers each and will work just fine. How the EMI will look like in this case I have no idea :)
Btw do you know what is the max static thrust of your BLDC and propeller assembly?
Thanks for sharing and greetings
Kiril