Our Inverters

Author: Jonn Dillon

2019

We are one of the few teams in the world developing and building our own inverters (Brushless DC Motor Controllers). Using cutting edge control systems, technologies and software to deal with truely unique use cases.

Why We Use  Custom Inverters:

  • Our motors are unique

    • Needed small motors for running them in the hub​

      • If using smaller motors, you need to main the flux​; this means that it either needs to be a longer motor, or has more magnets / poles​Consequently, we run 42 Pole motors

    • Conforming to university requirements:

      • Under 120V

      • Other teams use 400-600V as its significantly more efficent, easier and industry standard

    • To be somewhat competitive​:

      • Wanted higher kW

      • Had to use brushless DC motors, has the highest energy density out of other available motor types, this means we can exploit the drone industry​

  • Because our motors are unique​​​,

    • Very few inverters / motor controllers can deal with the ​current requirements as they are built for higher industry standard voltages​

    • Very few inverters / motor controllers that can deal with the motor pole count​

    • Almost all found, cannot do both, except manufacturers

    • Manufactuer's inverters have less than optimal control strategies and speed. Motors and inverters are built to go in forklifts and ultra lights, where the speed response is very low​

Inverter Design

There are two primary stages: the Power Stage and the Control Stage.

  • Power Stage:

    • Job: ​To take the DC supply from the Accumulator, and transform it into a 3-Phase supply for the motors

    • Contains 12 MOSFETs, 2 per normal gate

      • Due to the high amperage requirements, we could not find a singular MOSFET to perform the task of controlling a phase individually​

      • With the two, we could balance the amperage load and allow enough safety margin

    • Has a large number of capacitors​​​​ to deal with the complex power requirements and extreme in-rush and out-rush currents

      • 20+ Film Capacitors for large current draw

      • 50+  Multi-Layer Ceramic Capacitors (MLCC's) for extremely rapid current response​

      • Used many capacitors as using Aluminium Electrolytic is a very bad idea

    • Has a number of sensors​:

      • Current sensors​

      • Temps

    • Large copper busbars​

      • Nothing could withstand the high amperage draw​

  • Control Stage:

    • Job: To take the CANBUS communications and translate that into ​a control wave / signaling to turn the Power Stages MOSFETs on and off at the right times

    • Has two zones, control and isolated:

      • Control zone contains the processing and various communication / peripheral devices​

      • Isolated zone contains the gate drivers, power supply for the motor's electronics, and some input filtering and ADC's 

      • Between these, is isolated power supplies and signal isolators to ensure that if anything happens:

    • On board processor has special power stage controllers pretty much built for this exact purpose​

      • Reduces a large number of effects of poor coding​​

    • Gate drivers have on board hard wired protections​, specifically for dead time​​

      • Prevents short circuits between the full DC supply bus​

      • Imagine the full accumulator power allowing to short together

Inverter Control - Jonn Dillon & Zoe Goodward

Controlling The Motors

Since all inverters are built on very specific timings and triggering, an advanced control system has to be used. Failure to do so can cause disastrous effects, from causing a MOSFET to explode right up to melting motor coils and demagnatising the magnets inside of the rotor.

The control system we have chosen is known as Trapazoidal Control. This is the 'easiest' and far more importantly, the quickest to process/calculate. Due to the very high pole count we have (42), the slowest we can go is often the highest speeds other inverters have to run at. Additionally, the faster we can go, the more efficiency we can extract. 

  • What type we chose

  • What issues there are if it goes wrong (control specific)

  • Effects of the control:

    • Trapazoidal has greater effiency to a point

    • FOC has better efficenecy, but not possible on the hardware

    • FOC also has no torque ripple becasue it maps/models the flux fields inside the motor, never letting the rotor's angular momentum/pull 

    • Trapazoidal is very fast to calculate and process, where as the initial mathamatical transforms in FOC is very slow (relativly)

    • Trapazoidal requires Hall effects to sense the rotor position, FOC does not

      • Becomes unstable at higher speeds, due to noise floor and ADC quantization/sampling error

    • FOC uses the back EMF of the motor to figure out where the rotor is at

  • Chose Trapazoidal​

    • Our hardware chosen to effectively control the MOSFETs and interact with communication systems is:​

      • Not fast enough (16Mhz) to do FOC (specificly those mathamatical transforms)​

      • Not fast enough for the 50/100Khz golden speed range for making FOC more efficent than Trapazoidal

    • Easier to program, debug and teach​​

Future Work

  • Rebuilding the power stage:

    • Include more MOSFETs for better current and voltage distrubition​

    • Reposition current sensors to see real usage rather than capacitor in-rush / stabilisation

    • Design for 'blast proofing' in the first place, to save PCB and componentry on MOSFET failure

    • Reposition in-rush capacitor stage to be MLCC onboard and singular large film of board

  • Use Isolated Gate drivers:

    • To prevent isolated side of board being taken out each time​

    • And the time & cost it takes to replace parts

  • Use a gate driver per MOSFET

  • Add current dump/storage caps around gate drivers

    • Allows for greater current in-rush more often​

    • Specificly to support operating the MOSFETs at a higher speed (FOC...)

    • Allows use of 'harder' to switch MOSFETs

  • Explore use of Silicon Carbide MOSFETs (SiC MOSFETs)

    • They have much higher efficiencies, better thermal transfer, more robust switching and standard high wattage ability rather than using the specially designed ones we are using currently.

    • Harder to 'switch' them, 'new', have to buy in bulk and can be costly

  • Migrate away from microprocessors to an FPGA/s​

  • Facebook
  • YouTube
  • Instagram
Stay up to date with all of QUT Motorsport's latest developments and events in the 'news' page.' 
  • Facebook
  • YouTube
  • Instagram
  • LinkedIn Social Icon

2 George Street 4000 Brisbane City, Garden's Point Campus O-120

Photos for Website by Michael Hanau. Renders for Website by Jack Chilton