Design of Heterogeneous Drive System Based on Object-oriented Devices and Speed Sensor

We present a smart motor architecture that incorporates a motor, driver, controller, and communication interface into a single control unit and implement virtualized motor driving. A new pulse width modulation (PWM) technique, called PK modulation (PKM), which is traditionally used as a power supply and is equipped with a communication functionality, is proposed. With this new technique, a multimotor driving system capable of driving heterogeneous motors (such as an AC motor and a DC motor) simultaneously is implemented. The driving system using this smart motor architecture can freely accommodate any change in the number of motors driven, relying solely on the power bus. Other advantages include a simplified driver, effective control of the power signal interference, and reduced number and weight of control lines. In this work, MATLAB/Simulink® is used for signal analysis. MCU ATMEL328PU is chosen as the central controller for hardware implementation. The feasibility of the system architecture is verified as well.


Introduction
Multimotor driving applications such as a multiaxis moving platform and a robotic arm are becoming popular in the field of automation in recent years. In addition to the issues of mechanism weight reduction and simplification of system architecture and control methods, compatible environments, miniaturization, and smart technologies will be in the highest demand in the future. (1)(2)(3) The traditional multimotor driving system has a limit on the number of motors it can drive owing to the constraint of controller performance. When the number of motors to be driven by the controller exceeds the limit, a communication interface, such as the Internet, RS485, or the etherCAD, must be used to achieve coordinated operations. Figure 1(a) shows the multimotor driving application to a robotic arm. Most of the robotic arms have six axes. Multiple motors are used to form multiple robotic arms (R 1 -R n ), which are individually controlled by different controllers (controllers A, B, C, and so forth) via different communication systems and media. As shown in the figure, the robotic arms R 1 , R 2 , and R n are driven and controlled by controller A. Using current technologies, controller A is equipped with three control boards, each of them capable of controlling the robotic arm with, at most, six axes. Communication and coordinated control are carried out through a communication interface. Figure 1(b) shows a multiple-axis moving platform (made up of Motors A, B, C, and D) commonly used in glass factories, steel plants, plastic film factories, and LED TV assembly plants to deliver large glass sheets, steel plates, and plastic films. The multiple-axis motors must be controlled and driven synchronously, individually or in groups. Constrained by hardware functionalities and driving methods, traditional multimotor drive controllers are only capable of controlling and driving homogeneous motors (motors of the same type, e.g., all DC or AC motors). (1)(2)(3)(4)(5) The so-called AI motor that combines a single DC motor, a driver circuit, and a serial communication interface into a single unit was previously presented. (6) In this design, the power is supplied through a DC bus while commands and data are transmitted using a serial communication interface so that the limit on the number of motors a traditional controller can handle is removed and the control method is simplified. However, the DC bus has a potential problem that the current harmonics can cause interference with the serial communication signals, giving rise to the issue of limited power of the single motor. For the solution, the use of a power line carrier (PLC) communication interface to transmit the DC-brushless-motorrelated data (e.g., current and rotational speed) to a PC at a remote location to achieve remote speed control has been proposed. (7)(8)(9)(10) However, a specially designed PLC communication interface is required. The fixed communication channel increases the hardware cost. The signals transmitted over the power line are frequently subject to the interference caused by the filter circuit and ambient magnetic field owing to the properties of the channel. Other than that, the DC motor offers the advantages of stability, controllability, low cost, and the need for only voltage regulation to control the rotation speed of the motor. It is perfect for applications that require high torque. However, the DC motor requires the feedback of the speed signal to control its speed and position. On the other hand, an AC motor has no brushes and therefore does not spark, making it perfect for use in a dusty environment. No additional speed signal is needed to perform speed and position control. Even though an additional expensive controller is needed, AC motors are still widely used in various driving applications owing to advantages such as high efficiency, energy saving, and other application features. Motors of various types each have inherent advantages and disadvantages. As shown in Fig. 2, it is hoped that the heterogeneous motors (whether DC or AC motors) incorporated in the control units can be connected by the DC bus. However, because driving strategies are different, heterogeneous motors cannot be controlled and driven at the same time owing to the differences in driving methods. The issues arising in driving heterogeneous motors are therefore seldom addressed in research papers. Driving theories related to both AC and DC motors are introduced below. Figure 3 shows a generic motor-driving architecture, in which the voltage of the DC bus is obtained by rectifying the AC power source from the grid and then used to power all units. Let ẋ be the state variable and Eq. (1) can descript the dynamic characteristics of the induction motor in the synchronous d-q two-axis rotating reference frame.

Three-phase AC induction motor vector driving principles
Here, ˆe θ is the position of the magnetic field; ˆs ds φ and ˆs qs φ are the d-and q-axis magnetic components in the static d-q two-axis rotating reference frame, respectively. The position of the magnetic field ˆe θ can also be directly estimated. For stator flux orientation vector control, the speed of the rotor can be both determined with speed sensor hardware and estimated with current sensors. Since the d axis rotates along the reference magnetic field, the q-axis rotor flux q is zero. In this manner, the slip speed ˆs e ω and rotor speed can be estimated by transforming Eq.
Here, e ds i and e qs i denote the d-and q-axis current components in the synchronous frame, and are used to control the stator flux and electromagnetic torque, respectively. The theory of fluxoriented vector control (FOC) gives the induction motor the torque and flux decoupling control that the DC motor has and the architecture is shown in Fig. 4. The variable-frequency drive controls the induction motor, which is always employed with pulse width modulation (PWM) power signals. However, Eq. (8) Indeed, an AC motor can be driven by an open loop using a six-step driven scheme that is simpler but less efficient. Figure 5 shows the open-loop four-quadrant driving architecture for a traditional permanent magnet DC (PMDC) motor . In this actual circuit of the driving system, the PWM signal shown in Fig. 6(a) is used to drive the DC motor.

Driving theory of traditional DC motor
The gate-driven DC motor is shown in Fig. 6(b) and the armature current i a can be obtained from (1 ) (1 ) a   a  a   T  T  T  T  T  T   a  a  a  a  a  T  T  a  a  a   v  v  e  e  e  e  e  e  E  i t  where V d , v d , τ a , R a , T, i a (t), δ, E, Δi a , and I a are the DC bus voltage, the input voltage, the time constant of the armature circuit, the armature resistance, the modulation period, the armature current, the duty ratio, the inductive voltage in the armature, the AC component of armature current, and the DC component of armature current. The perturbing torque ΔT e calculated from Eq. (10) is a function of Δi a , which is the cause of vibration and noise. It is known from the equation that the vibration depends mainly on the frequency f c of the carrier Vc. Although choosing a higher f c may reduce the vibration, it increases the switching loss of power switches (11) and reduces the system efficiency. However, the final conclusion is that the PWM power signal is still not equipped with communication capabilities.
Here, ϕ and f c are the magnetic field of the stator and the frequency of the carrier Vc. When the stability is reached, v a can be obtained using Eq. (11) and the electromagnetic torque T e after  stability can also be calculated using Eq. (12). Figure 7 shows a DC motor driving architecture diagram. In any case, the two driving strategies are obviously different, so they cannot be mixed or operated together in the same multimotor driving platform. However, it can be seen in Figs. 4 and 6 that the PWM signal is used to control the input power whether the DC or induction motor is to be driven. The modulation scheme is employed and shown in Figs. 8(a) and 8(b). According to the theory, the harmonic components of the PWM power signal can be calculated as Equation (13) is verified using the rated voltage as the base value, as shown in Figs. 8(a) and 8(b). The spectral analysis of the PWM power signal also reveals that the main harmonics of the PWM power signal are determined by the carrier frequency f c . It can be seen that the above principles of the methods used to drive heterogeneous motors are different, but the motors are all driven by PWM power signals. However, PWM power signals have widely distributed harmonics and have no communication capabilities. PWM techniques are commonly used to drive AC and DC motors. One of the most researched topics in recent years is the use of technologies to improve efficiency and eliminate harmonics. References 7 and 11 to 14 are mainly on how to eliminate harmonics by controlling the conduction angle and extinction angle of the power switch. References 11 and 13 are focused on the control of carrier frequency in PWM. This is the simplest and most effective PWM technique, and a simple first-order low-pass filter is used to remove all the harmonics that are multiples of the fundamental frequency. The space vector has also been used (11) to control both the PWM signal and the conduction of power switches in order to eliminate the ramp and improve overall efficiency. References 12 and 13 present a technique for eliminating specific harmonics by calculating and determining appropriate switching angles. However, one disadvantage of this method is that only a small number of specific harmonics can be eliminated at the same time since the calculation consumes excessive resources. The existing papers reveal that the PWM strategy is not equipped with a communication function. In this study, a PK modulation (PKM) technique is therefore proposed to add a communication capability to PWM technology and the control of harmonics. The concept of building a customized motor is also proposed to solve the problems of the limited number of controller axes and the mechanism weight. The wiring and the control method can also be simplified. Above all, the main contribution of this study is the discovery that heterogeneous motors can be driven using the same platform.

Proposed Methods
To solve the problems mentioned above, in this study project, we present a smart motor architecture and propose the introduction of the object-oriented device (OOD) concept (15) to the smart motor architecture, which combines the controller, gate drive, and communication interface, and incorporates the motor into a single unit. The two smart motor architectures using the OOD concept and incorporating the DC and induction motors are shown in Figs. 9(a) and 9(b), respectively. The purpose is to achieve virtualized motor driving and control. The controller of the smart motor is responsible for receiving commands from the communication interface and driving the motor by the close-loop control method. Figure 10(a) shows the architecture of the smart-motor-based driving platform. To be compatible with the traditional method of driving a motor with PWM power signals, the traditional DC bus Vd circuit must be redesigned, as shown in Fig. 10(a). By adding the front controller, the communication is enabled for the PKM method, which is used to modulate the DC output of the power amplifier into the PKM power signal. Basically, the PWM power signal uses the PK encoding scheme to modulate and transmit the data, (15) as shown in Figs. 10(b) and 10(c). The PKM power signal already contains communication data. Therefore, the PKM power signal supplies power to the motor while performing communication at the same time. The motor to be driven, whether it is a DC or AC motor, is powered by the modified DC bus. This new bus design is aimed at replacing the traditional PWM power signal and with one that has a communication capability. Figure 10(a) shows that the smart motors are connected to the newly designed DC bus V d circuit.
Both return-to-zero (RZ) and not-return-to-zero (NRZ) signal formates can be used. Figures  10(b) and 10(c) show that the communication interface embedded in the controller is responsible for receiving the PKM power signal in the RZ signal formate, performing PK decoding, and sending the data to the controller. Then, the controller either drives the motor in accordance with the command and internally defined control strategy, or sends the data back. The function of the full-wave rectifier is to convert the PKM power signal into DC current and to power the incorporated motor. With the new DC bus, two power lines are needed to accommodate the free change of the smart motors to be driven. A direct result is that the problems of controlling heterogeneous motors with a single control board and the limited number of motors to be controlled are solved. Driving multiple motors with only two power lines reduces the weight of the power lines required. The wiring complexity is also reduced since the smart motors are connected serially to the DC bus. Furthermore, the PKM power signal carries data and thus reduces the chance of interference with the small signals transmitted over communication interfaces such as RS-232, UART, RS-485, EtherCAD, and PLC. Parameter setting for the communication interface is no longer necessary. The communication speed and data length can be adjusted automatically. Other advantages include improved control of harmonics, reduced interference to the environment, low hardware cost, and good communication quality. In this study, the front controller and PK encoding are used to modulate the output DC voltage V of the power stage into the PKM power signal v p . In the communication process, the smart motors need only to count the pulses in order to do PK decoding. The communication process relies on only a series of pulses. (15) Therefore, the PWM power signal v p can be in one of the various forms. The communication speed is also variable. However, a periodic NRZ signal is also used to concentrate the harmonics and avoid interference from other frequencies in the environment. Then, the Fourier analysis of the harmonic components can also be carried out using Eq. (14).
where η denotes the efficiency of the signal after PKM and can be calculated using Eq. (16). Figure 11 shows the spectral distribution of the PKM power signal v p used to transmit "K" in ASCII code. The DC bus provides the rated voltage as the base value. The frequencies ( f s ) of the pulses are 1 and 3 kHz. The correctness of the spectral distribution characteristics of Eq. (14) is verified. The direct comparison of the results with those in Fig. 11 reveals that the frequency distribution is more concentrated than that of the PWM power signal. It is known that, while the PKM power signal is used to transmit data, the DC component can be obtained from Eq. (17) when n = 0. When v p approaches 0, the DC loss is reduced and thus the efficiency is improved, as shown in Fig. 11. This also implies that the harmonic distribution of the PKM power signal changes depending on the communication pulse f s . However, changing the communication pulse f s does not affect the communication capability. This means that when the PWM power signal is used for communication, the interference to bandwidth caused by the electromagnetic wave in the space can be controlled by changing the communication pulse f s at the transmitting end. By changing the position of the spectral distribution, we can control the source of interference by electromagnetic waves effectively and avoid the frequency bands that easily interfere with the surrounding environment. Therefore, the communication pulse f s can be changed in the implementation to avoid the frequencies that might cause interference to adjust the switching loss of power switches. Although PKM uses f s to modulate communication data, changing f s does not affect the reliability of the communication. (15) This is an outstanding property of PK communication that existing communication technologies do not have. v p can be rectified by the rectifier into a square wave and used as a power supply to smart motors. To achieve the goal of bidirectional communication of smart motors, an identical communication channel of opposite direction for signal feedback must be created within the power stage and central control units. This also implies that each smart motor must have the same rated voltage, which is the only limitation.
In addition, since the PWM power signal is digital, no floating-point computation is needed for PK encoding and decoding. A single inexpensive chip is sufficient for encoding and decoding since the demand for computing resources is low. Other advantages of the PK encoding scheme include improved coding/decoding efficiency and low computing/hardware costs.

Measured Results
To verify the correctness and feasibility of the theories proposed in this study, a hardware platform is implemented, as shown in Fig. 12. The platform adopts the architecture presented in Fig. 10(a) and simulates the multiaxis moving platform shown in Fig. 1(b). Smart motors A-D are respectively used in place of respective motors A-D in Fig. 1(b) with speed sensors as servers. Consequently, each smart motor receives only the command and can independently control the monomer itself. This novel and smart motor drive architecture realizes the virtualization of motor operation. In this platform, two smart motors (i.e., three-phase induction motors A and B in Fig. 12) are used to simulate the conditions of no load (motor B, also called ACM2) and a load of 0.1 N‧m (motor A, also called ACM1). The other two smart motors (i.e., DC motors A and B in Fig. 12) are used to simulate the conditions of no load (motor C, also called DCM2) and a load of 0.1 N‧m (motor D, also called ACM1). This platform uses a front controller with a Bluetooth communication interface. This interface is used to receive commands from a cellular phone and to transmit data to the phone that is also included in this test platform. The waveform data are sent to the PC via the front controller simultaneously and recorded, displayed, and analyzed using Matlab ® /Simulink ® . The digital oscilloscope is used to display the real-time electrical signals of a single machine. Figures 13(a) and 13(b) show the actual circuits used to drive induction and DC motors, respectively. The tests of single driving, group driving, and synchronous driving are carried out. The four motors of the moving platform are simulated to verify the functions and stability. The output voltage of the power amplifier is controlled at 24 V. To provide a universal voltage V d on the DC bus and the rated voltages to various kinds of motors, a step-down regulator is added to the rectifier in the package circuit. In addition, to be compatible with the traditional driving methods, the format of the PWM power signal is slightly modified so that the output of the PWM power signal after the rectifier still contains the DC value of the power amplifier output. I. Single-axis drive test: The purpose of this test is to simulate the driving of a single axis of a multiaxis motor driving system for heterogeneous motors.  shows the real-time display of the waveform on the oscilloscope and the data shown on the cellular phone when DCM1 is tested without a load. In Fig. 14(a), the yellow waveform (3) represents the rotational speed. The green waveform (1) represents the PWM signal and the pink waveform (2) is current. The blue waveform (4) is the PKM power signal sent to DCM1 from the front controller. Figure 14(b) shows the operational data received by the cellular phone. Figure 15 shows the driving results when open-loop speed control is used to drive ACM1 by the six-step square-wave control method (with 120° conduction angle). The PWM technique is also used to drive the motor during conduction. In Fig. 15(a), the yellow waveform (1) represents i U . The green waveform (2) is i V and the pink waveform (3)    The PWM technique is also used to drive the motors during conduction. Open-loop speed control is also used. Tests are carried out with and without a load. MATLAB ® /Simulink ® is used to collect and display smart-motor-related data such as current, formalized voltage, and speed. The data are also collected and displayed using the cellular phone with a bluetooth interface. Figure 16 shows the experimental results of the multiaxis forward rotation of the smart motors. DCM1 and ACM1 are driven without a load. DCM2 and ACM2 are driven with a 0.1 N‧m load. III. Multiaxis synchronization: The purpose of this test is to simulate the synchronous driving of multiple motors of the multiaxis moving platform driven by heterogeneous motors. The test is to drive two sets of smart motors, i.e., two 12 V DCM1 and DCM2, and two 24 V ACM1 and ACM2 at the same time. ACM1 still uses six-step square-wave control. ACM2 uses stator orientation vector control. Both ACM1 and ACM2 use close-loop speed control. The speed command for DCM1 and DCM2 is 30 rpm. The speed command for ACM1 and ACM2 is 2000 rpm. MATLAB ® /Simulink ® is used to collect and display smartmotor-related data. The tests are carried out with and without a load. Data and results are transmitted to and displayed on the cellular phone through the Bluetooth communication interface. Figure 17 shows the experimental results of the multiaxis forward rotation of smart motors. DCM1 and ACM1 are driven without a load. DCM2 and ACM2 are driven with a 0.1 N‧m load.
The implementation results verify that the smart motor architecture proposed in this study is capable of integrating and driving heterogeneous motors. The monitoring setup and the display of results can be performed remotely. The control method, either close or open loop, can be changed freely in the process, thus simplifying the control process.

Conclusions
Although the multimotor driving system is widely used in various industrial fields, the issues of simplifying the driving process and system architecture are seldom addressed. A smart motor architecture in which multiple motors can be driven with only two power lines is proposed in this study. The system hardware and wiring are greatly simplified. The number of motors to be driven can be changed freely as well. The most valuable contribution of this study is the proposal of a method of sending control and driving commands using a communication channel to drive heterogeneous motors of different power ratings and types. The virtualization concept offers the technology to enable heterogeneous motors to be driven simultaneously. The power lines, used as the power supply and for communication, do not need any expensive or complicated PLC communication interface. Other advantages include a simple architecture, low cost, and high system stability. The most important feature of this architecture is that the communication speed can be changed freely so that the interference caused by harmonics can be avoided. The switching loss of power switches and vibration can both be reduced, which is another major contribution of this study.