EV inverters, also known as traction inverters, convert the DC electricity from the battery into the 3-phase variable frequency AC that is needed to drive the motor at a given speed. EVs use asynchronous (induction) motors, or sometimes synchronous motors. These are both types of AC motors, with the speed of the motor determined by the AC frequency. The inverter can therefore control the speed and torque of the motor independently. A traction inverter is a type of motor drive or motor controller, and can also be referred to as a variable frequency drive (VFD).
Development of inverters is being driven by the need to improve vehicle efficiency and performance, reduce cost, and create more compact drivetrains. Improved efficiency extends vehicle range while reducing running costs. Cost is the main barrier to EV adoption — and cost reduction is therefore critical.
Traction inverter fundamentals
The basic principle of all rotary electric motors is that there are fixed magnets called stators and rotating magnets called rotors. When the stators and rotators have their opposite poles aligned, they repel each other, causing rotation. However, if all of the magnets maintain the same polarity, the motor could only rotate through a maximum of 180°. In order to reverse the polarity of either the rotor or the stator, all electric motors in practical use are fundamentally AC motors. What are referred to as DC motors, simply use mechanical switching within the motor itself to alternate the current flowing through the magnets. So in a sense, the brush and commutator arrangement within a DC motor can be considered to be a very simple inverter, which produces a square wave input.
The highly efficient motors used in EVs, as well as most large industrial machinery, are 3-phase AC motors. These require an input of 3 separate sinusoidal currents, with a 120° phase shift between them. The inverters used to create this supply are much more sophisticated than the simple brush/commutator arrangement that produces a single square wave AC in a DC motor. However, they still work by using switches to turn the current on and off, and to reverse its direction. By switching on and off quickly enough, power can be regulated without any noticeable vibration in the motor. Semiconductor switches in inverters switch on and off thousands of times per second to produce smooth sinusoidal AC waveforms. This approach of varying the duration of off and on periods is known as pulse width modulation (PWM).
Switching is a very energy efficient way of regulating power, without the inherent resistive losses of linear regulators. However, as the switch is transitioning between the off and on states, there is a very brief period where resistance does occur, leading to a little power loss and some heating. Faster switching is therefore better both in terms of efficiency and from a practical perspective, with resistive heating requiring thicker conductors and thermal management. Inverters with faster switches are therefore generally more efficient, lighter and more compact.
Practical AC motor drives that use PWM require two things. Firstly, they need semiconductor switches that can turn the circuit off and on at the required frequency (orders of magnitude higher than the AC frequency). And secondly, they need micro processors able to calculate the switching pattern required to produce a sinusoidal waveform at the required frequency.
The switches in EV inverters mush handle currents of hundreds of amps per phase and voltages of up to 800V DC without overheating. Since semiconductor based inverters became established in the 1980s, MOSFETs have been used for voltages below about 150V and IGBTs for higher voltages. However, more recently wide band gap semiconductors such as silicon carbide (SiC) and gallium nitride (GaN) have allowed MOSFETs to be used in high voltage applications. MOSFETs switch more quickly and therefore SiC based inverters offer efficiency, mass and compactness advantages. All of the switches used are three-terminal devices, with two terminals forming the power connections on either side of the switch and the third acting as the control input. While in a transistor the control input would be called the gate, in a MOSFET or IGBT it is known as the gate.
Heat is generated by inverter switches by both the switching loss mentioned above and resistive losses in conducting elements. The challenge with cooling these increasingly compact and high-powered systems is that materials that are good conductors of heat tend to also be good conductors of electricity, and electrical isolation is required. Ceramics are therefore widely used, since they are good electrical insulators with reasonably high thermal conductivity. Boards of copper sheet bonded to a ceramic substrate can be etched to form printed circuit boards with good thermal dissipation characteristics. However, the very rapid thermal cycling caused by switching losses is a change in terms of fatigue.
A major trend is increasing power density, with inverters achieving the same power output with both reduced mass and reduced volume. EVs are moving from inverters with silicon based IGBTs to those based on Silicon carbide (SiC). SiC can handle higher voltages, temperatures and frequencies, leading to smaller and lighter inverters. There has been a progression from MOSFETs, to Insulated Gate Bipolar Transistors (IGBTs) which are now widely used in production EV’s and are starting to be replaced by SiC.
The ability of SiC inverters to operate at higher frequencies means they can be used to control higher speed motors. Motors running at higher speeds can be smaller and lighter, giving perhaps the greatest reduction in overall system size and cost.
Continued development of SiC inverters is expected to reduce system weight and size, increase efficiency and reduce cost. Larger SiC fabrications with higher throughput will achieve lower costs, while increasing wafer sizes to 200 mm and beyond will also reduce cost. While these changes are expected within 10 years, SiC development is challenging, with new products taking several years to reach production.
Increasing integration of the inverter with other electric drive unit components, such as the gearbox and motor, will increase power density further. It will also reduce material use and manufacturing cost.
All of the power that reaches the motor must flow through the inverter, meaning that the efficiency of the inverter has a direct impact on the efficiency, and therefore range, of the vehicle. Greater efficiency clearly means that a vehicle will travel further with the same battery, but it also means that a smaller battery may possible, with shorter charging times, lower cost and reduced weight. Reduced weight further increases range and vehicle performance.
EVs are moving from inverters with silicon based semi-conductors to those based on Silicon carbide (SiC). This is largely due to an increase in efficiency from 97-98% for silicon to 99% for SiC. With efficiency already so high, there is little scope for significant further improvement.
There may be more potential to increase the overall vehicle efficiency by more deeply integrating the inverter with the motor to precisely control individual motor coils. This could improve the efficiency, torque and power output of the motor itself.
Exro have developed an inverter that switches the motor between different coil configutations. Their CTO, Eric Hustedt, explained this: “Ultimately what the call driver does is it gives you huge constant power speed range which is to say we hit peak power very early on and then that peak power is maintained all the way out to maximum rpm and that’s actually quite difficult to do with the normal three-phase mode… so fundamentally it’s a drive that that switches the machine between two different coil configurations; in our case series and parallel but principally what that means for the machine is it changes the current density and the voltages produced by the machine to allow us to optimize it for different operating regions. For example at low speed where you want a lot of torque we would switch into series mode where we’re producing a lot of current density in the machine which is generating a large amount of low end torque and then as the machine speed increases we switch into parallel which allows the machine to continue operating at high speed without sacrificing performance so it gives you the benefit of a very torque dense machine without sacrificing high-speed power.”
There is a trend toward higher DC-link voltage in electric vehicles, driven primarily by increased charging speed. 800V is replacing 400V as the standard. Charging anxiety is becoming more important for consumers than range anxiety. This is partly about being able to quickly and easily find an available charger when needed, and partly about the waiting time to complete charging on a long journey. Reducing charging times is therefore very important for adoption of EVs.
Higher voltages makes some multilevel structures a feasible and efficient option for replacing two-level inverters. This may lead to higher efficiency, higher power density, better waveform quality, and inherent fault-tolerance.
While GaN inverters over some advantages over SiC, GaN is not really suitable for voltages over about 650 V. With EVs moving to 800V to enable rapid charging, use of a GaN inverter would require additional conversion, adding expense and complexity.
Thermal management and material selection
The semiconductor switches used in EV inverters must be mounted in such a way that heat can be dissipated, electrical isolation is maintained, and the thermal cycling caused by the high frequency intermittent heating of switching losses does not cause a mechanical fatigue failure. This presents challenges for the laminated substrate and the attachment of the semiconductor to the substrate. A substrate cooling system is also required. The manufacturing cost of these measures is often considerably higher than the semiconductor switches being mounted and cooled.
Solder is generally not suitable for the die attach due to insufficient fatigue resistance. Sintered attachments, using a paste of silver and copper power with an organic filler, are much more durable. These sintered attachments not only have far higher fatigue resistance but also around three times the thermal conductivity and a much higher melting temperature allowing higher temperature operation and therefore even more efficient heat dissipation. Despite these advantages, sintering is a very expensive process with significantly increases the cost of an inverter.
A common substrate used in inverters is direct bond copper (DBC), which consists of a thin layer of copper bonded to a ceramic substrate using a high-temperature oxidation process. A common and low cost ceramic is alumina (aluminium oxide). More costly ceramics such as Aluminium Nitride (AlN), Silicon Nitride (SiN) and Zirconium-doped Alumina can improve heat dissipation and fatigue resistance.
Substrate cooling systems generally consist of a heat sink soldered, clamped or glued to the underside of the substrate, with a fluid cooling system. Sintering is increasingly being used in high-performance inverters due to the same advantages that make it suited to the die attach. The heat sink, or ‘cold plate’ is often pre-attached to the substrate to ensure good thermal contact.
With inverters already achieving efficiencies of up to 99%, there is only minor scope for improvement in this area. Inverters are also relatively light and compact — perhaps 5 litres and 10kg for an EV weighing over 2,000 kg — meaning that further reductions will have a negligible effect on overall vehicle performance. However, changes that facilitate improvements in other areas may have more impact. For example, SiC inverters enable higher DC voltages than Gan, meaning faster changing times and smaller conductors. Faster switching also enables higher-speed motors, which significant potential for cost and mass reduction.
You may also like:
Filed Under: Inverter