Rotor position sensing plays a critical role in electric vehicle (EV) traction inverters by providing the angle feedback needed for precise motor control.
Salik Ahmad
Applications Marketing Engineer, Automotive, Allegro MicroSystems
As inverter systems move toward higher efficiency and power density, enabled by higher voltages and faster switching speeds, maintaining accurate sensing becomes more difficult in an electrically noisy and space-constrained environment.
Salik Ahmad, applications marketing engineer for automotive at Allegro MicroSystems, recently discussed these challenges with EV Engineering. In this interview, he explains why rotor position sensing is essential for synchronous motors and how engineers can keep that sensing accurate and safe in modern traction inverter designs.
This Q&A is the first of a two-part series. Part II will take a closer look at the design and reliability side of inductive sensing.
Here’s what Ahmad had to say…
What are the biggest engineering challenges in rotor position sensing for EV traction inverters?
Four main challenges tend to drive the design. The first is noise immunity. The position sensor sits in an electromagnetically hostile environment with high-voltage switching from 10 to 100 kHz. The angle measurement must stay clean and accurate in that environment.
The second is accuracy at high speed. EV motors can spin at more than 20,000 rpm. At those speeds the angular error budget shrinks to less than one degree.
The third is functional safety. The traction motor torque path is typically an Automotive Safety Integrity Level D (ASIL D) function under ISO 26262. Every component in that signal chain, including the position sensor, has to carry an ASIL rating that meets the system-level safety goal. What that means for the architecture is more complicated than a datasheet alone would suggest.
The fourth is size, integration, and cost. Inverters keep shrinking at the same power, increasing power density. Every millimeter and every dollar counts. These four challenges are not independent. A redundancy structure may help solve the safety challenge, but it also increases size. Solving one problem can therefore work against another.
On a system level, inverters are expected to become more efficient, drive motors smoothly, and operate efficiently as generators to recover kinetic energy during braking.
Why is absolute rotor position sensing necessary in EV traction motors?
The dominant drive motors in today’s EVs are synchronous machines. They’re generally permanent magnet synchronous motors or electrically excited synchronous motors. Synchronous means the stator’s rotating field has to stay aligned with the rotor’s field.
If that alignment is lost entirely, commutation is no longer possible, and the inverter cannot drive the motor. Even a partial misalignment causes torque to drop and efficiency to fall. The motor can also cog or stall. Absolute rotor position sensing provides the feedback needed to prevent that from happening.
Asynchronous induction motors are commonly used as secondary motors such as on a secondary axle. Those motors operate on slip. They need speed feedback rather than absolute position. That’s why we focus here on synchronous motors.
Sensorless approaches also exist in general. However, in traction motor applications, where full torque and accurate control are required at or near zero speed, a dedicated position sensor remains the standard approach.
Where does the position sensor fit within the traction inverter control system?
The sensor feeds the closed-loop field-oriented control algorithm. That algorithm continually adjusts the phase currents to maintain the commanded torque. The control loop typically runs at 10 to 100 kHz. At those rates the latency and noise in the position measurement contribute directly to the control error.
Simplified EV traction inverter architecture showing the rotor position sensor within the motor-control loop.
The sensor also operates within a much larger system. Gate drivers switch the silicon carbide (SiC) or insulated-gate bipolar transistor (IGBT) power transistors. That switching stage is the primary source of the fast voltage and current transients that create electromagnetic interference. The isolated bias supplies for the high-side drivers float at the switching-node voltage. That is one of the most electrically hostile points in the system.
Current sensors provide the phase-current feedback. The position sensor and current sensor work together as the two essential feedback inputs for field-oriented control. The microcontroller or digital signal processor takes that position and current data, computes the next gate-drive command, and closes the loop in microseconds.
The quality of the control loop is limited by whichever input degrades first. As a rough estimate, during normal operating conditions, about two-thirds of torque ripple comes from the current-sensing side and one-third from the position-sensing side.
How do high motor speeds affect the accuracy requirement?
EV motors need to spin at more than 20,000 rpm to meet their efficiency, size, cost, and performance targets. At that speed even one degree of angular error represents a significant misalignment between the stator field and the rotor field. Torque drops. Energy is wasted as heat. Vibration develops and audible noise can reach the cabin.
The requirement is sub-one-degree mechanical accuracy across the full speed range and across temperatures from -40° to 160° C. That budget applies to the entire signal chain and not only to the sensor. The printed circuit board (PCB) layout and target geometry contribute to the error. The cable run and signal processing also consume part of the budget. There are ways to compensate for latency in the signal chain. That can help recover some of the available error budget.
What does an engineer gain by replacing a resolver and resolver-to-digital converter with an integrated inductive sensor?
The switch mainly provides benefits in size, weight, cost, and integration simplicity. A resolver is a wound electromagnetic component. It’s relatively bulky, heavy, and expensive compared to inductive sensing solutions. It also needs a dedicated resolver-to-digital converter along with an excitation and conditioning chain.
An inductive sensor replaces those elements with typically three flat PCB coils and a single application-specific integrated circuit (ASIC). One coil transmits and the other two receive. The ASIC handles excitation, demodulation, and angle calculation on the chip. The design eliminates the winding and reduces the number of assembly steps and bill-of-material components. The sensor is also immune to dc stray fields because it operates using its own generated ac field rather than a permanent magnet.
A digital output, such as serial peripheral interface (SPI), Single Edge Nibble Transmission (SENT), or pulse-width modulation (PWM), also typically comes directly from the ASIC rather than through a separate resolver-to-digital converter. The catch is on the PCB coil design side.
Accuracy is now defined by the layout rather than by a precision-wound part. The receive-coil geometry and stack-up directly affect the error, and the air gap has to be controlled tightly. Designers also have to manage tilt, eccentricity, and concentricity, and use harmonic shaping and calibration to reach sub-degree accuracy. The engineering work moves from the resolver supplier onto the system design team.
Comparison of an inductive sensor PCB (left) and a conventional resolver (right), illustrating the potential reduction in size, weight, and system complexity.
However, that work also gives the designer more flexibility and can produce meaningful cost and integration benefits.
How do position sensors reject stray fields from high-current busbars?
Modern automotive position sensors reject stray magnetic fields through the sensing principle and the system design. Inductive sensors are inherently immune to dc and low-frequency magnetic fields because they aren’t a magnetic-sensing approach.
They measure inductive coupling between coils rather than sensing a magnetic field. They detect changes in an ac field at megahertz frequencies. That’s far above the dc fundamental and harmonic content of the busbar currents. Hall and magnetoresistive (xMR) magnetic sensors are more exposed to those fields.
Hall and xMR sensors therefore use differential and gradient-based architectures. Instead of measuring an absolute field, they measure the difference between two closely spaced sensing elements. A uniform stray field from a distant busbar adds equally to each element and cancels out of the difference. The local field gradient from the rotor magnet remains.
The layout and qualification are also important. Magnetic-field immunity can be evaluated using testing such as ISO 11452-8, the component-level standard for immunity to magnetic fields.
At the system level, the designer still needs to route high-current busbars away from the sensor. Current loops should be kept tight and symmetric. Shielding may also be needed where the geometry forces the busbars close to the sensor.
For ASIL-related applications, the field strength can also be monitored. A stray field large enough to corrupt the reading can then be detected and flagged rather than silently producing an incorrect angle.
Does the move to 800-V architectures create new challenges for the position sensor?
The move to 800 V is felt most directly by the gate driver and the isolation. However, it still affects the position sensor indirectly. The rotor position sensor sits on the low-voltage mechanical end of the motor shaft. It does not see the dc bus voltage directly. Its insulation rating is therefore not the main issue.
The real challenges come from the surrounding environment. At 800 V, SiC switching produces faster edges and higher dv/dt and di/dt. That increases noise coupling into the sensor signal path and supply. Larger and faster-changing busbar currents also produce stronger and more dynamic fields near the sensor. Faster switching and higher-efficiency control also create tighter timing requirements. Accurate low-latency feedback becomes even more important.
In practice, the sensor routing and grounding should be kept away from the switching node. The supply and signal lines may also require filtering or shielding. A digital interface with cyclic redundancy check (CRC) protection can help because coupled noise can be detected rather than interpreted as valid angle data.
The change is not that the sensor suddenly sees 800 V. The change is that faster switching and higher currents make electromagnetic compatibility and sensor robustness more important.
Look for Part II of the Q&A next week.
Filed Under: Featured Contributions, Inverter, Q&As, Sensors