**It is no secret that any equipment that either heats or cools is coming under great scrutiny as governments increasingly legislate efficiency standards. In particular, washing machines, air conditioners, and refrigerators are under a legislative microscope. Variable-speed control is one well-known means of reducing both the peak and average energy consumption of motors as used in compressors and fans. This is especially true for air conditioners as standards now require a reduction in the seasonal average energy consumption. So even the energy consumed by the evaporator and condenser fans becomes important.**

The situation is a bit different in washing machines and dish washers where energy efficiency efforts mainly focus on hot water consumption. Nevertheless, there are motor-related efficiencies to be had in washers when extracting water during the spin cycle as a way of minimizing the energy consumed by the dryer. Standards are also in force pertaining to lower power appliances such as refrigerators and pumps because they operate almost continuously and thus consume a significant amount of energy.

One way of addressing efficiency demands is with permanent magnet synchronous motors (PMSM). Compared to conventional induction motors, PMSMs contain less iron and copper, a plus in an era characterized by gyrating commodity prices.

Applying variable-speed control to PMSMs economically demands an approach that avoids rotor position sensors typically used in industrial drive applications. A sensorless field-oriented control algorithm permits variable-speed control of PMSMs using only measurements of motor current for feedback. The design challenge is to optimize the algorithm so it meets performance and energy efficiency requirements in many different applications.

A highly integrated mixed-signal controller makes it possible to implement complex control algorithms that maximize the efficiency of permanent magnet ac motors. Field-oriented control (FOC) with current-angle phase advance maximizes the efficiency of interior permanent magnet motors. The efficiency gain is almost 5%. This mode of control also implies a switch from trapezoidal to sinusoidal waveforms driving the motor windings. This minimizes torque ripple to reduce acoustic noise in the fan motor.

In the case of a whole-house air conditioning system, a mixed-signal control IC can simultaneously run both the compressor and fan motors. A combination of highly optimized hardware control blocks and a configurable control sequencer enables rapid execution of complex motor control algorithms.

The IC is one element in an appliance design platform that includes all the power and control silicon needed to drive the fan and compressor motor in an air conditioning system. The same IC can also be applied to the latest energy efficient laundry systems that use an energy saving heat pump in the drying cycle.

The accompanying circuit schematic includes the major components in an outdoor unit controller for an air conditioning system. The mixed-signal motor control IC drives the compressor motor, the fan motor, and the power factor correction circuit. The IC integrates three major functions: a motion control engine (MCE), the analog signal engine (ASE) and an 8-bit microcontroller. The MCE executes the motor control algorithms while an independent 8-bit microcontroller core runs the application.

The sensorless field-oriented control algorithm derives all the required motor information from the currents flowing in the dc link shunts. This avoids the need for position sensors on the motor shaft and isolated current transducers in the power inverter circuit. The Analog Signal Engine includes a fast A/D converter, multiple sampling circuits, and differential amplifiers needed to extract the motor winding current from the dc link current signal.

The Motion Control Engine executes the sensorless FOC algorithm. The algorithm includes a reverse rotation function that transforms the measured stator currents into a reference frame synchronized with the angle of the rotor magnet flux. The transformed currents have two quasi dc components: a direct-axis current aligned with the rotor flux and a quadrature axis component that generates motor torque because of interaction with the rotor magnet.

The d and q-axis current-loop compensators calculate the stator voltages to force the currents to track the set point values. The forward rotation function transforms these voltages to sinusoidal ac voltages in the stator reference frame. A space vector PWM generator uses these signals to derive transistor switching signals for a three-phase inverter.

When driving a classical permanent magnet synchronous motor with surface-mounted rotor magnets, the d axis reference current is set to zero to maximize the torque-per-amp. However, when driving an interior permanent magnet motor, the d-axis current will generate a reluctance torque component to augment the torque that the rotor magnets produce. The IPM control function is the key control element that enables the operation at a higher efficiency when driving the IPM motor.

The other key feature of the sensorless FOC algorithm is that it does not need to use high-resolution rotor angle sensors as typically found in industrial drive systems. In appliance drives, where low-speed performance is not important, the rotor angle can be derived from the winding back EMF signal. Brushless dc drives frequently employ a six-step commutation sequence. Here the back EMF is available directly by sampling the voltage on the unconnected winding. However, when driving the motor with sinusoidal currents, the back EMF must be calculated indirectly from the motor circuit model.

The equations below describe the two-phase equivalent circuit for the permanent magnet synchronous motor. The two phase currents are derived using the Clarke transform that calculates the currents in two quadrature windings that will produce the same field as the currents in the three phase windings. The two phase winding currents are the outputs of the forward rotation function that drive a space vector PWM generator:

Here *Ψ _{r}* is the rotor flux vector peak value,

*θ*is the rotor flux angle,

_{r}*vα*and

*v*are quadrature voltage values derived from motor voltage vectors rotated into a stationary reference frame using the transformation angle. The two phase currents

_{β}*i*and

_{α}*i*are derived using the Clarke transform. They represent currents in two quadrature windings that will produce the same field as the currents in the three-phase windings. RS and LS are motor winding resistance and inductance respectively. The two-phase winding currents are the outputs of the forward rotation function that drive the space vector PWM generator.

_{β}The important aspect of the two-phase circuit model is that the back EMF terms are time derivatives of cosine and sine flux functions so they can be determined through integration. The accompanying control schematic for an angle and speed estimator has two major subsystems. During the first stage of the calculation, the flux estimator derives the rotor cosine and sine flux functions. The flux integrators include low-frequency gain compensation to avoid dc saturation.

At the second stage, the rotor angle phase-locked-loop (PLL) forces the error between the rotor angle and the estimated angle to zero. The error is calculated using a vector rotation function whose quadrature output will be zero when the rotation angle input matches the angle of the cosine and sine flux functions.

### The Motion Control Engine

The second-order feedback loop in the PLL generates both angle and velocity signals. A further feature of the PLL is the start up sequencer that is necessary at low speeds when the winding back EMF signal is swamped by circuit noise. The first part of the start up sequence is a parking function that drives dc current into the stator windings to align the rotor at a known angle. Then the motor is driven with a constant current to generate a constant torque.

The PLL speed integrator is fed by a motor mechanical model that estimates the motor speed from the accelerating torque and system inertia. Once the motor reaches a certain minimum speed, the PLL switches to a closed-loop mode and tracks the rotor flux angle.

When designers implement FOC with a traditional DSP or RISC processor, they start by first translating the motor control schematic into state equations. These get written in C code. Then software development tools can generate machine code for the processor.

In contrast, FOC implemented with a motion control engine (MCE) eliminates some of the steps in developing motor control algorithms. It incorporates a graphical compiler, an algorithm development tool that transforms the control schematic directly into MCE sequencer code, thus avoiding all intermediary steps. This lets the developer modify the reference algorithm directly making use of a library of optimized control blocks such as a PI compensators and vector rotations.

The reference algorithm includes basic ac motor control functions. They include the FOC current loop with sensorless rotor position estimation and an outer velocity loop. The inner current loop is a well-established FOC algorithm used for PMSM control. It can typically be used “as is” but the optimal algorithm for the outer control loops will vary with application. Fan or pump controllers, for example, may need to regulate torque to maintain pressure while compressor controllers may just regulate speed.

In compressor control, load torque ripple may prevent a simple velocity loop from properly regulating at low speeds. A feed-forward algorithm can compensate for the load torque to eliminate mechanical vibration at low speeds. In washing machines, the controller can detect a wash load imbalance by analyzing the ripple signature in the motor speed and torque before entering the spin cycle. The designer edits the control algorithm schematic using a Matlab Simulink graphical user interface. Through Matlab, the designer and can add control blocks such as comparators, summing junctions, switches and integrators. The digital control IC executes the algorithm using hardware blocks from an MCE control library that are based on MCE sequencer code generated by the graphical compiler.

### Resources

Optimized control blocks reduce execution time significantly relative to software implementations. One example of an optimized control block is a vector rotator depicted in an accompanying diagram. The Cordic vector rotation has been developed specifically for ASIC implementation that relies on a series of add, subtract and shift functions that yield 12-bit accuracy in only 13 cycles. This calculation is 10 times faster than the calculation using Taylor expansion on a 32-bit RISC processor.

The digital timing circuits that generate the inverter PWM signals also generate the sample timing signals that let the analog signal engine (ASE) extract the motor winding current from the inverter dc link. This optimized combination of analog and digital signal processing circuits can simultaneously control two permanent-magnet synchronous motors.

Finally, there's enough processing capacity in an MCE to support the execution of a power factor control (PFC) algorithm. So the air conditioning control IC can handle the input power factor as well as the fan and compressor motor. Traditional RISC processor-based systems, in contrast, require separate fan and PFC control ICs.

Ray Andraka, “A survey of CORDIC algorithms for FPGA based computers”, Proc. of ACM/SIGDA Sixth International Symposium on FPGAs, 1998, Monterrey, CA, pp. 191-200. http://portal.acm.org/citation.cfm?doid=275107.275139

International Rectifier pages on MCE, http://www.irf.com/product-info/datasheets/data/irmcf343.pdf

MCE basics, http://machinedesign.com/article/servo-on-a-chip-1209