More and more appliance manufacturers are now adopting variable-speed permanent magnet (PM) synchronous motors to improve energy efficiency and enhance product features. Industrial drive manufacturers recognized the high energy efficiency and power-to-weight ratio of PM motors long ago, but recent advancements in control electronics have led to their widespread adoption by appliance manufacturers. Directional control (FOC), utilizing DC-link current feedback technology, minimizes system costs and is highly attractive for appliance drive applications. The sinusoidal control of the motor produces smooth torque output with low acoustic noise. Therefore, FOC is suitable for use in devices where low noise and high energy efficiency are critical, such as fans, water pumps, washing machines, and dryers. Many appliances, such as air conditioners and washer-dryers, require independent control of the speeds of two motors to optimize their performance. The traditional approach for these systems is to use motor control processors with serial communication links for synchronization. However, controlling two motors with a single device greatly simplifies hardware and system design. Recently introduced appliance control platforms can simultaneously control two PM AC motors using only DC bus current feedback technology. These chips employ embedded FOC algorithms, reducing development time and driving rapid adoption of this technology by appliance manufacturers. Sensorless motor control (FOC) algorithms are common in industrial drive systems, typically employing encoders or resolvers to measure rotor position. Closed-loop current control algorithms utilize the rotor's angular coordinates to correct motor current and rotor flux, thereby maximizing torque output. Advanced rotor position estimation algorithms in electrical control systems eliminate the need for high-resolution position sensors (Figure 1). Figure 1: The Oriented Control (FOC) system eliminates the need for position sensors by using DC-link current feedback technology. The estimation algorithm obtains the rotor flux position of the PM based on a motor model employing winding voltage and current. This method is well-suited because the magnets on the rotor determine the position of the rotor flux. Winding current measurement data is obtained from the DC-link current using correct ADC sampling timing based on the power converter's state knowledge. The system block diagram shows that the winding current measurement data and winding voltage drive values ​​are inputs to the rotor flux model used to calculate the rotor angular position and angular velocity. The torque and flux control loop not only achieves maximum torque output at lower speeds but also enables high-speed operation under field weakening conditions. In first-generation FOC motor drive systems, these functions were implemented using a combination of analog and digital circuitry. Today, most of these motor drive systems utilize high-speed DSPs or RISC processors to implement relevant algorithms on a single device. Software implementation not only offers advantages in flexibility and hardware simplicity but also presents significant software development challenges for drive system developers. Generating the software code for the control algorithm requires many steps. First, system engineers convert the control schematic into a set of differential equations representing various control functions. Second, software engineers convert these differential equations into C code representing instructions to be executed by the processor. Errors in this process can prolong development time unless the code is well-structured and documented, and a long-term software maintenance team is available. RISC processor and DSP vendors can provide a complete set of FOC software examples to help motor drive companies accelerate the development process. This is quite possible because FOC control technology is very mature, and the algorithm structure can be very well defined. However, software implementation currently offers no particular advantage because the flexibility of the algorithm structure is not essential. Control system design engineers can utilize digital ASIC or FPGA technology to implement FOC algorithms in hardware. The first step of the development process is not significantly different from the software approach, but the second step no longer uses C code; instead, hardware engineers convert the differential equations into Verilog code representing logic gate interconnections. This design allows for the definition and storage of control parameters in control registers, providing flexibility, while the algorithm structure is hardwired in a digital ASIC. This approach is common in telecommunications systems requiring high-speed processing, and many motor control ASICs can implement FOC and other motor control functions. The advantage of this approach lies not only in its execution speed but also in its significantly reduced system development time. Motion Control Engines (MCEs) offer an alternative approach that combines the high-speed performance of dedicated ASIC hardware with the flexibility of programmable processors. This approach is particularly effective because the FOC algorithm uses many standard functions, such as error amplifiers, proportional-integral (PI) compensators, and vector rotators that appear repeatedly in control circuits. MCEs consist of a hardware library of motor control functions, and the interconnection between the motion control sequencer and these functions is efficiently implemented by assigning input and output memory addresses to the corresponding system variables. Control system engineers do not need to convert the control schematic into a system of differential equations because a fully optimized ASIC implementation is available in the MCE library. Instead, control system engineers use schematic editing tools to graphically define the control schematic by interconnecting the standard functions in the motion control library. The graphical compiler translates the control schematic into MCE sequencer instructions for interconnecting the hardware control functions. The compiler assigns addresses in the MCE's shared RAM area to each algorithm variable defined in the control node. The MCE sequencer instructions define the memory addresses for each control function block and its input/output variables. Because the MCE stores these instructions in memory, it offers the same flexibility as RISC processors and DSPs. The PWM switching frequency sets the timing of algorithm execution, the ADC sampling rate, and the update rate of the output voltage. MCE library elements represent space vector modulators and ADC inputs, but they appear only once in the control schematic because they correspond to physical input/output pins. On the other hand, MCE library control functions such as vector rotators or PI compensators can appear multiple times in the control algorithm because the MCE stores their inputs and outputs in data memory. Each instantiation of a library function occupies data memory space to store variables and MCE instructions, thus memory capacity limits the complexity of the algorithm. Each library function occupies a certain number of system clock cycles each time it is executed, therefore the total number of clock cycles in the control loop must be less than the number of clock cycles in the PWM cycle. The rotor angle estimator and current control loop consume approximately 1,400 system clock cycles, equivalent to 11 μs at a maximum system clock frequency of 128 MHz. This allows for simultaneous control of two motors within a 50 μs PWM cycle equivalent to a 20 kHz switching frequency. Of course, controlling two motors requires two sets of space vector PWM modulators and additional analog inputs for current sampling. Figure 2 shows the block diagram of a dual-motor control IC, which includes an embedded motion control engine and an 8-bit microcontroller core for executing application-level code. The advantage of this approach is that it separates the slow system-level functions executed on the microcontroller from the high-speed motor control algorithms executed by the MCE. Figure 2: Dual-motor control chip MCE library functions including an 8-bit microcontroller for executing application-level tasks The key to high-speed execution of the control algorithm is the efficiency of the MCE library functions implementation in the ASIC. Two important feedback control units (PI control compensator and vector rotation block) serve as typical examples of library functions. ASIC implementation requires optimizing silicon and clock cycle usage without sacrificing robustness and reliability. Figure 3: Proportional-Integral (PI) Control Compensator Efficiently Utilizes MCE Hardware Resources and Clock Cycles. A well-known ASIC implementation of a PI control compensator (Figure 3) is based on a continuous-time transfer function: Transforming this expression to the discrete-time domain yields a set of differential equations defining the ASIC implementation: To maintain accuracy at low input levels and recalibrate the output to a 16-bit variable, the sum of the integral terms has 32-bit resolution. An anti-saturation module prevents integral saturation when the output reaches the system's physical limits. The vector rotation module is a two-dimensional matrix function used to implement voltage conversion between rotating and fixed reference frames. Forward rotation has sine and cosine terms: Several trigonometric identities can simplify the operation of sine and cosine terms to a sine function operation in the range of 0 to 90°, but the operation will vary depending on the available hardware. In some microcontroller implementations, the lack of fast multiplication functions forces software developers to rely on simple lookup tables. In DSPs or RISC processors with single-cycle multiplication instructions, the sine function can be calculated using Taylor expansion. To achieve 12-bit precision in just 13 cycles using an ASIC implementation based on a series of addition, subtraction, and shift functions, a vector rotation function known as the CORDIC algorithm was developed (Figure 4). This operation is 10 times faster than using Taylor expansion on a 32-bit RISC processor. Figure 4: The CORDIC algorithm calculates vector rotations an order of magnitude faster than the Taylor expansion method , simplifying motor control. While many configurations may exist, the configuration shown in Figure 5, which uses a single control IC to operate a dual-motor platform, is the most efficient. This configuration not only eliminates the need for a second IC to control the second motor (which leads to unnecessary redundancy), but also enables complex interface designs for the two motors. For example, when one motor fails (such as a short circuit or lockout), the second motor can be immediately de-energized as a reflex action, reducing latency associated with communication with the main control system. In air conditioning applications, for instance, the compressor motor speed and evaporator fan speed need to be tracked to optimize efficiency. The control system sets the motor speed by directly writing to the MCE register, avoiding complex communication between multiple ICs. Figure 5: Dual-motor control provided by a single platform can eliminate redundant hardware and the need for complex communication links between two separate controllers, and can realize complex interfaces between the two motors.