AUGUST/SEPTEMBER 2011

www.automotive-electronics.co.uk

DRIVER ASSISTANCE SOFtWare TOOLs INFOTAINMENT SECURITY AUTOSAR

MICROCONTROLLERS Simplifying HEV and EV design

PRODUCTS + NEWS + AUTOSPORT SHOW REPORT

10 MICROCONTROLLERS

MCUs simplify HEV and EV motor control designs

Vitor Ribeiro and Maik Strietzel explain how efcient software algorithms and the latest microcontrollers can improve motor control on electric vehicles
ar manufacturers and their suppliers are working hard to develop economical new vehicle models with the aim of reducing both fuel consumption and CO2 emissions. Hybrid vehicles of every kind currently serve as a bridge technology until high-capacity electrical energy storage systems are available at reasonable prices, enabling the production of long-range allelectric vehicles. The electrication of motor vehicles is advancing rapidly and the efcient control of electric motors is becoming ever more important. Hybrid electric vehicles and electric vehicles (HEV and EV) have the great potential for achieving the reduction in greenhouse gas emissions demanded by legislators. The CO2saving capacity of a hybrid electric vehicle, for example, is approximately 30%. Further measures, such as

Fig. 1: Motor control according to the Clarke/Park principle

electric power steering, provide an additional 5%. A central element of both approaches is an electric motor which, in the case of the powertrain, is used either in combination with a conventional internal combustion engine (in hybrid cars) or as an independent source of power (in allelectric vehicles). Deciding factors in the selection of the motor include dimensions, weight, reliability, robustness, necessary torque and efciency. Both synchronous and asynchronous motors can be used, the former as a permanent magnet synchronous motor (PMSM) featuring high torque coupled with compact dimensions and high efciency (roughly 94%). However, the benets that this offers are offset by a higher price tag, as expensive materials from rare earth elements are usually required for the permanent magnets. Efforts are being made to use

asynchronous motors. These are robust and reasonably priced (they do not require magnets made from rare elements) and the dynamic properties can be easily controlled using suitable software algorithms. Disadvantages include the slightly lower efciency (around 90%), greater weight and lower torque in the start-up phase. Not only do they not require maintenance, but brushless versions of both motor types mean that brush loss is not an issue. PMSMs have a better dimensionsto-torque ratio and higher efciency and are currently the rst choice for use in the powertrains of electric and hybrid electric vehicles.

Control

Brushless motors require more effort for commutation, while safe and efcient control is a fundamental prerequisite for use in powertrains. The challenge is to nd the perfect balance of motor, power electronics, control unit (microcontroller) and control software. The algorithms used must be adapted to the respective motor and application so that the electronic controller commutates the motor optimally at all times. Failure to adapt these correctly may lead to undesired effects such as irregular running and excessive noise, which together have a negative impact on the degree of efciency that can be achieved. Motor control combines various control algorithms depending on the application. Fig. 1 shows eld-

automotive electronics | august/september 2011

www.automotive-electronics.co.uk

12 MICROCONTROLLERS
the detection of the rotor position is essential for precise motor control. As a key component, the rotor position sensor has signicant inuence on the performance and efciency of the motor system. is transmitted to the two sensor coils, set at 90 to each other, via the magnetic coupling (induction). Evaluation of the analogue sinusoidal and cosinusoidal signals returned by the resolver requires a resolver-todigital converter (RDC), which is used to determine the angular position and rate from the analogue data. Optical systems, such as the incremental encoder mentioned above, offer advantages in terms of performance and precision, but their decreased robustness and vulnerability to dirt and temperature conditions makes their use in motor vehicles rather difcult. In direct comparison with the other sensors, a resolver may be more expensive, but it is also more robust and reliable in return. It is also able to detect the absolute position of the rotor at any time even when it is at a standstill. This is not possible with either Hall sensors or incremental encoders. and ample communications interfaces such as Can, Lin and Flexray (Fig. 4). A particular highlight is the integrated RDC with the functionality described above. Integration of the RDC results in a signicantly simplied system architecture (Fig. 3b). Rotor position, sinusoidal and cosinusoidal values as well as the angular rate are available to the MCU at all times and can be read out from dedicated registers every 100ns. The block diagram in Fig. 4 gives an overview of on-chip resources that can be available. Motor controllers are very often developed by a model-based method, and there are now very powerful software tools that generally have one thing in common they work with oating-point numbers. To transfer the algorithms developed in this way to a conventional microcontroller, the oating-point numbers must be converted to integers. Since this is not always a completely smooth process, an integrated oating-point unit (FPU) is standard in some MCUs and is of great benet. In terms of results, this means an increase in computing power and less work for porting from models to the real world.
Fig. 4: Microcontroller block diagram

MICROCONTROLLERS 13
easily. For example, the phase currents may not correspond to the expected values, or the expected speed may not have been reached. In these cases, the MCU can set the system to a dened state without compromising vehicle safety. But how is the resolver monitored and how are malfunctions diagnosed? All signals from and to the resolver can be monitored via integrated diagnostics and fault proles such as short-circuits to ground, cable breakages and interruptions or even short-circuits within the resolver windings can be identied quickly and reliably. If such faults occur, the MCU immediately receives an internal interrupt signal so it can react quickly and in a targeted manner to the situation. All this takes place internally and with minimal latency, in stark contrast to systems with external RDCs, in the case of which the failure must be signalled by means of a slow interface to the MCU.

Hall sensor

Fig. 2: Schematic and mechanical structure of a resolver Source: Tyco Electronics

orientated control (FOC) combined with PID regulation systems for controlling rotor speed, torque and ux. In FOC, the Clarke/Park transformation is used to transform the phase currents measured (at least two must be continuously recorded) from a stator-based three-dimensional system to a rotor-based two-dimensional system. These transformed variables, the current rotor position and a target position specied by the application, or a target rotational speed, are taken as starting points for the control algorithms. Due to their fast, precise and overshoot-free control characteristics, PID regulators are ideal for bringing complex systems into a stable state. By means of inverse Park/Clarke transformation, the system is again

transformed into a rotor-based vector system and the voltages to be set are transmitted to the motor via the power electronics by means of pulse-width modulation. The current rotor position can be either directly determined with a sensor or estimated using a complex calculation system. The latter method also known as sensorless angle detection is based on the logging and evaluation of two actively controlled phases. This method can be subject to deviation of several angular degrees and is not currently used for motor vehicle applications. Sensor-based rotor position detection can be conducted with various sensor systems. The following section gives a brief overview of the most common sensor types. However, it can generally be said that

Hall sensors are based on the Hall Effect, whereby a voltage is induced by changing the magnetic eld around a current-carrying conductor. With the help of a magnetic ring attached to the rotor and a sensor unit afxed to the stator, the Hall Effect sensor is a cheap and easy means of detecting angles. The higher the number of magnetic poles and Hall elements, the higher the resolution and accuracy. However, this sensor is susceptible to magnetic interference.

Incremental encoder

Figs 3a (top) and 3b: Motor control with external and integrated RDC

One frequently used sensor is the incremental encoder. This is available in a wide range of designs, featuring both mechanical and optical scanning to determine the current angular position. To measure an angle, an incremental encoder must be based on a zero or reference position. For the microcontroller (MCU), actual angle determination only involves detecting the direction of rotation and counting the pulses emitted. The angular rate can be calculated by simply measuring the intervals between two pulses. The insensitivity to magnetic interference is benecial here. By contrast, any mechanical friction losses and susceptibility to dirt in the case of optical systems are disadvantageous.

Inverters in vehicles

Resolver

One very robust sensor, often used in the automotive industry, is the resolver, which is neither at risk from magnetic interference and dirt, nor subject to friction losses during angle detection. The resolver consists of the rotor, which is permanently attached to the motor shaft (motor rotor), and the ring-shaped stator, which is permanently attached to the motor housing. The stator consists of at least one excitation coil and two sensor coils. Higher resolutions can be achieved by increasing the number of pole pairs. Fig. 2 shows a resolver. The excitation coil is fed with an analogue sinusoidal signal. The analogue signal

In the simplest case, the motor controller consists of a microcontroller, a power output stage, the motor in conjunction with a rotor position sensor (resolver) and the RDC (Fig. 3a), which is usually implemented as a discrete circuit. Resolver diagnosis This generates the resolver signal An MCU can identify a failure or and determines the rotor position malfunction of the motor relatively and rate as quickly and precisely as possible based on the sinusoidal and cosinusoidal information returned. This information must be forwarded to the microcontroller so that it can be taken into account in the motor control algorithms as described. External RDCs are usually linked to the MCU via a serial peripheral interface (SPI). Depending on the systems design and the manufacturer of the RDC, this can also take place via other serial or parallel ports. These suffer from a serious disadvantage in that the MCU does not have constant access to the rotor data, and must instead always request it from the external RDC. Not only is this relatively slow, but it is also a potential source of errors that can have a negative impact on the functional safety of the entire system.

Motor control MCUs

Applications for motor control MCUs with integrated RDC include the aforementioned inverter for the electric drive motor. The block diagram for an inverter is shown in Fig. 5a. For reasons of functional safety, the motor control MCU (master) is supported by a second, smaller microcontroller (slave). Both microcontrollers maintain

Motor control MCUs

Some of the latest MCUs take an entirely different approach. They can generate all the motor control signals

automotive electronics | august/september 2011

www.automotive-electronics.co.uk

www.automotive-electronics.co.uk 

august/september 2011 | automotive electronics

14 MICROCONTROLLERS
a constant dialogue and, as soon as abnormal behaviour is detected, the slave MCU can initiate a system reset or cut off the power supply to the motor. The master MCU generates the commutation signals for the motor, measures and monitors the phase currents, generates the resolver signals and determines the rotor position and angular rate via the sine-cosine feedback. All this information is used by the motor control algorithm for targeted control of the rotor speed and torque. The inverter communicates with the battery management system (BMS) via the internal network (Can or Flexray) Figs 5a (top) and 5b: Diagram of an HEV to ensure that the energy required can and EV inverter and also be provided. EPS control Economical use of energy is essential in an HEV or EV, and systems that require a constant supply of energy, such as hydraulic power steering, would over the long term put excessive strain on the energy storage systems. Hydraulic systems run continuously to build up the required pressure by means of a servo pump. By contrast, a purely electric power steering (EPS) system only needs energy when it is being used. In principle, the design is similar to that of the inverter for the powertrain. Once again, master and slave MCUs are used, while the power element is different mainly in terms of the power to be regulated. A PMSM is used for this application as well, while the resolver functions as described (Fig. 5b). However, the master MCU must still evaluate a steering input device and a torque sensor. In order that the correct steering assistance can be given at all times, other information, such as speed of travel, is required. These data are provided via the internal network, for example the Can bus. This information is used to calculate the extra steering torque required and transmit it to the electric motor on the steering column. The aim is to minimise the effort expended by the driver and provide dynamic support for steering movement. For example, steering movement requires more support when the vehicle is almost at a standstill, such as during parking manoeuvres, than at high speed on the motorway. EPS is intended to increase both driver comfort and safety. It is entirely possible for the electronic stability control (ESC) to regulate vehicle stability actively by means of steering intervention. It is also conceivable for EPS to be part of a driver assistance system that could park a vehicle automatically without driver input. development is marching ever onward, albeit not quite as quickly as many would prefer. Until then, maximum use must be made of the energy available in the form of fuel and/or electrical power in hybrid vehicles. Efcient software algorithms, powerful microcontrollers and highly efcient electric motors will make this possible. A higher level of integration leads to leaner and cheaper motor control systems that can be used in applications above and beyond those presented here. l Vitor Ribeiro is senior product marketing engineer and Maik Strietzel is application engineer for Fujitsu Semiconductor Europes automotive business unit

Summary and outlook

More and more hybrid vehicles are coming onto the market, and while they are initially mostly simple startstop systems, the number of full hybrids (HEVs) and plug-in hybrids (PHEVs) is expected to increase in the years to come. All-electric vehicles that function entirely without an internal combustion engine need safe, cost-effective and high-capacity energy storage systems. Here, too, technological

automotive electronics | august/september 2011

www.automotive-electronics.co.uk