Application of single chip microcomputer in permanent magnet brushless DC motor drive system of electric golf cart

introduction

The first decade of the 21st century has almost passed quietly, but the era of electric transportation that people longed for has not arrived as expected. When many electric vehicle projects led by the government and actively participated by enterprises and research institutions, such as PNGV, Freedom CAR, and PREDIT111, ended with the roar of engines, people began to realize that the huge inertia and strong vitality of the traditional automobile industry far exceeded their imagination. For a long time in the future, electric vehicles can only stay in the laboratory.

At present, the application research of pure electric vehicles has turned to fixed-point and directional vehicles, mainly buses, community vehicles and micro vehicles for specific purposes. These vehicles have some common characteristics, such as being managed by institutions, operating in specific areas, and not running at high speeds. We can optimize the design and management of vehicles based on these characteristics to reduce costs and improve performance, compete with traditional internal combustion engine vehicles, and create an energy-saving and environmentally friendly image, which is important for institutions and enterprises [1].

Project and system introduction

Golf carts are a type of micro-car with a specific purpose. They are operated on golf courses. The different purposes of the drivers and passengers and the road conditions of the course reduce the vehicle's mileage, but place relatively high requirements on the power performance of the drive system. As we all know, golf courses are undulating, which requires the golf cart drive motor to have excellent overload performance; the low vehicle speed means that the golf cart drive motor does not need a wide speed regulation range. To meet these requirements, the use of permanent magnet brushless DC motors (BLDC) is the best choice: BLDC can achieve extremely high efficiency over a wide load range as long as its speed remains below the base speed. In addition, it is rugged, reliable, and easy to adjust. If the reliability of the position sensor components can be improved, it will be maintenance-free throughout its operating life, which makes it even more attractive [2].

We have examined a variety of similar (two-seater) electric golf carts. They all use traditional DC motors, mostly externally excited. The rated power of the motors ranges from 2 to 3kW. They are all equipped with lead-acid batteries with a maximum capacity of 150AH and a nominal driving range of 150km. Our prototype vehicles were tested before modification, and their maximum efficiency did not exceed 70%. But there is one very important thing in common: their power voltage level is 48V. This value may be determined from the communication power supply system or from the requirements of safety voltage, but in any case it has become the de facto standard. It restricts the establishment of our entire drive system.

Key points and difficulties in system design

Since BLDC has many advantages, people certainly have reasons to apply it to mini cars such as golf carts. But why do all the electric golf carts on the market use traditional DC motors? There may be many answers, but there are two points that cannot be avoided, that is, cost and reliability. First of all, cost. BLDC with similar parameters is more expensive than traditional DC motors, mainly because permanent magnets are expensive. However, the price of permanent magnets is now on a downward trend [3]. The drive of a separately excited DC motor requires the main circuit to have three bridge arms, but two bridge arms are located in the excitation circuit and have a small capacity. The drive of BLDC requires the main circuit to be a three-phase bridge drive circuit, and armature current flows through them, which greatly increases the investment in power switching devices. As for reliability, the use of Hall position sensors to detect the motor rotor position to guide the power device to perform appropriate commutation has low cost and simple detection circuits, but low reliability [4]. Of course, even if other types of sensors are used, the reliability is not much higher. I personally think that this is as troublesome as the brushes and commutators of traditional DC motors. I will explain how to solve these problems, as well as some common problems that other motor drive systems have, in the following content.

Lower voltage levels bring challenges in handling high currents

At the designed maximum power, the peak current handled by the power switch device will reach 100A. The high current will impose strict requirements on the parasitic parameters, distributed inductance and other issues caused by the device layout, and of course heat dissipation. Under the same circumstances, the BLDC drive requires more power switch devices, but we still hope not to increase the size of the controller. Due to cost constraints, it is impossible to use integrated or intelligent power devices (IPM) with excellent performance but expensive. The only possibility is to try to improve the heat dissipation conditions to reduce the number of power MOSFETs. Here we introduce a heat dissipation method called "aluminum-based copper clad laminate" [5], which is inspired by IPM. In this type of power device, the power chip is directly surface-mounted on the aluminum substrate without even being packaged. Then we also found that it is also widely used in high-intensity LED light sources, automotive ignition systems and other occasions. By adopting this heat dissipation method, we successfully reduced the original seven parallel groups to three parallel groups, and the effect is gratifying. By using the surface mounting method, the parasitic inductance of the pins of the power switch device can also be greatly reduced, which can be said to kill two birds with one stone.

Regarding the current sharing problem of multiple tubes in parallel, the worst case method [6][7] is used to analyze the steady-state current sharing problem of multiple tubes in parallel. We use this method to determine the derating factor adopted when multiple tubes are in parallel. However, there are too many factors that affect the dynamic current sharing problem, which makes it inconvenient to analyze. Analyzing the influence of multiple parameters from a statistical perspective is a direction worth considering.

Torque control strategy brings "closed loop failure" problem

The torque control strategy is used to control the golf cart drive system, which has many advantages such as large starting torque, fast response, and good current limiting effect. However, the torque control strategy brings the problem of "closed loop failure[8]": because the designed drive system has a double overload capacity, when the load torque cannot reach the given torque of the accelerator pedal, any change in the accelerator pedal position between the load torque value and the maximum given torque value will not cause any change to the vehicle's operating state. This is different from the drive response of traditional internal combustion vehicles. [page]

In a large number of actual debugging, our team has summarized an effective method: the idea is very simple, that is, let the accelerator pedal position correspond not only to the given torque, but also to the maximum given line voltage of the motor winding. At this time, any change in the accelerator pedal position will inevitably lead to a change in the maximum given line voltage and will inevitably change the motor speed. This can be derived from the voltage and speed regulation characteristics of the brushless DC motor. Here I call it the "maximum torque control strategy". Corresponding to different types of motors, this strategy may need to be adjusted as necessary.

Simple and novel position-free sensing strategy

It is very important to find a reliable and low-cost strategy for position acquisition without position sensors in the full speed range. Thanks to the working characteristics of permanent magnet brushless DC motors, which only require discrete position signals and the mutual inductance coupling effect between phase windings, our research team has developed a position sensorless algorithm called "indirect inductance method". Through analysis, we found that the mutual inductance coupling effect will cause the difference in phase terminal voltage during the effective and invalid periods of PWM modulation to have a fixed relationship with the rotor position. Theoretically, as long as the accuracy of the voltage sensor device meets the requirements, reliable position information can be obtained. In the low speed range, this method is more effective and can effectively make up for the shortcomings of the back-electromotive force method to obtain rotor position information in the full speed range. Due to progress, this method is not reflected in this design. At present, the algorithm implementation of this strategy is still proceeding in an orderly manner.

Characteristics of Microchip chips and their application in projects

The main control chip is the core of the control system. It provides drive signals to the inverter, processes power drive protection, samples and converts analog signals such as current in real time, collects position signals, receives external information or controls the outside through switch input and output, exchanges information with other external systems through the CAN bus, analyzes and processes various information, coordinates the work of various parts, etc.

The main control chip dsPIC30F4011 used in the permanent magnet brushless DC motor drive system for electric golf carts described in this design comes from Microchip and is designed specifically for the motor control field. The dsPIC30F chip is known as an MCU with DSP functions. It has both strong control functions and strong digital signal processing capabilities of DSP. These features make it simpler and cheaper than general DSP hardware development circuits, and more adaptable to digital signal processing requirements than similar single-chip microcomputers. In the design of the controller, the following peripheral module resources of the chip are mainly used [9]:

① Motor control PWM module (MCPWM): PWM works in the middle alignment mode, and the modulation frequency is selected as 10kHz. Literature [4] believes that this frequency can achieve the best energy density, noise and electromagnetic interference at the same time; the output is configured as an independent mode, and the special event trigger SEVTCMP is used to synchronize the A/D sampling at the middle moment when the duty cycle is effective. This moment is considered to have the smallest ground coupling interference and is expected to obtain accurate analog values;

② 8-channel 10-bit high-speed A/D conversion channel (AD): used to sample 8 signals including bus voltage, two sets of throttle settings, two sets of brake analog signals, two-phase current, and aluminum substrate temperature simultaneously in each PWM cycle, and the sampling is synchronized with the PWM time base;

③ Level change interrupt (CN): When the position signal from the motor Hall sensor changes level, a level change interrupt will be generated. In the level change interrupt service subroutine, the motor commutation, motor direction identification and speed calculation are implemented;

④ Timer 4 (TMR4): Timer 4 works in cycle counting mode to record the interval between two adjacent electrical cycles for calculating the rotation speed;

⑤ Controller Area Network (CAN) module: sends some information about the motor and vehicle to the upper instrument (LCD display) through CAN communication, and can receive instructions from the upper instrument (touch screen).

Even for beginners, you will find that Microchip's development platform is very easy to use. Its integrated development environment is completely free, and there are also some low-cost online debugging tools such as ICD2. Of course, to develop the system, you also need a target board. In addition, with the adoption of RISC, you will find that it is just as easy to program in assembly language. Of course, I still use C language in my design. In some places where high-quality target code is required, a hybrid programming method of embedded inline assembly is used to achieve a balance between code quality and efficiency. [page]

Microchip's technical support is excellent, and the network resources are quite rich, especially in the field of motor control, with a wide range of categories and fast update speed, and quite a few of the application notes have corresponding Chinese versions, which is a valuable asset for beginners. Mastering them well before designing can achieve twice the result with half the effort. But there is one flaw. I found that the source code styles of the application notes or examples written by different technical support engineers are very different, and some comments are not very standardized. It is better to unify them.

Figure 1 Product Selection Guide

We use Figure 1 to illustrate Microchip's rich product line, including 8, 16, 32-bit MCUs and DSCs, analog devices and interface products, memory and RF devices, etc. There are excellent products in each product line. In the design, in addition to the main control chip using Microchip's dsPIC30F4011, the bus driver MCP2551 of the CAN network and the rail-to-rail op amp MCP604 are also used. CAN is used to communicate with the upper LCD display instrument, transmit the vehicle status information to it, and can accept the instructions of the instrument to change the control parameters or respond to control instructions; MCP604 constitutes the main body of the analog quantity detection and active filter unit; the over-temperature and over-current hardware protection signals in the system are also from the output of the comparator composed of the internal op amp of MCP604, and these output signals are connected to the power drive protection pin (FLTA) of dsPIC30F4011 through line or. For the driver of MOSFET, we once planned to use TC4431, but it is only a single-ended driver, which requires an independent power supply to power the high-end driver. In order to reduce costs, we only designed a 15V power supply and finally chose IR2181 from IR Corporation, using bootstrap capacitors to power its high-end MOSFET driver. Of course, Microchip's power management devices are also very easy to use, and there is a complete power management solution on its website. In the next design, we plan to improve the battery management system of the drive system to increase the vehicle's mileage and extend the battery life

.

After submitting the entry, I felt relieved. Looking back on the days and nights of hard work in the laboratory in the past six months, I feel a lot of emotion: I would like to express my sincere gratitude to my teammates. We are a team, and everyone is the key to success. Fortunately, my teammates are very excellent and share the same beliefs with me. Without their efforts, it is impossible to achieve any results; participating in a project and fully participating in its entire process is very necessary for the growth of an electronic engineer. From project establishment, feasibility study, preliminary design of the plan, detailed design, trial production of the prototype system to experimental debugging, my learning ability, ability to analyze and solve problems, communication and coordination ability have made great progress; thank Microchip and the world of electronic products for providing us with a platform such as the "Microchip 16-bit Embedded Control Design Grand Prix": engineers and electronic enthusiasts from all over the country can exchange inspiration and creativity, share achievements and the joy of success through this platform! Thank you very much, I wish the competition will get better and better, and the scale will get bigger and bigger. If there is another opportunity, I hope to compete with electronic enthusiasts from all over the world and join us in the grand event!