Appliance technology has evolved significantly over the years, and two trends are merging to fuel the next round of appliance evolution — digital motor control and digital sensor processing. Digital Signal Controllers (DSCs) are at the confluence of these trends, and enable digital sensor processing, digital motor control and power factor correction to be implemented on a single chip.
DSCs have hardware accommodation for efficient DSP processing on-chip. While many advanced sensorless algorithms do not require DSP, most benefit from on-chip DSP resources. For example, the highly efficient Field-Oriented Motor Control (FOC) algorithm benefits from a single-cycle MAC, accumulator saturation and other features common to DSPs and DSCs, but not typically found on MCUs.
While the real world is analog, digital techniques have progressively augmented traditional analog design implementations. At the point where the balance is tipped, where the digital approach cannot be effectively matched with analog wizardry, digital control and processing become mainstream tools in the designer’s tool bag.
Digital motor control | | | Fig. 1. Resource utilization for an advanced sensorless FOC algorithm on a 28-pin, 128 KB Flash DSC. | |
The ongoing trend toward electronic control is driving the venerable commutator out of its DC motor shell, as evidenced by the migration from Brushed to Brushless DC (BLDC) motors. Another disruptive shift was the replacement of Hall Effect sensors with resistor dividers to measure EMF feedback, through the main phase connections, to determine rotor position and velocity.
Now, advanced algorithms such as FOC are considered highly desirable, due to the resulting excellent torque response, improved motor efficiency and lower audible noise obtained through sinusoidal drives. The FOC algorithm asymptotically decouples the rotor torque and rotor flux, making the speed linearly related to torque current. This permits, for example, an induction motor to possess the same behavior of a separately excited DC motor.
All of these motor transformations have one thing in common: they are enabled by progressively more sophisticated electronic control, coupled with an economically viable control solution. The primary impediments delaying these dramatic shifts have been the availability of cost-effective control, and the time needed to understand and refine the technology.
Many underpinnings of those impediments have been removed with modern DSCs, which offer cost-optimized and application-optimized performance coupled with robust applications support. Regulatory add-ons, such as active power factor correction, exacerbate the control workload and accelerate acceptance of the digital solution. As DSCs continue to add capability and performance while prices continue to decrease, new possibilities for product development emerge. This can be illustrated with the example of the merging of digital motor control with digital sensor processing onto one DSC.
Digital sensor processing | | | Fig. 2. Sensor signal plus noise in the time domain. | |
Today, a sensor signal within a typical motor-driven appliance is filtered, level shifted, amplified and fed to the Analog-to-Digital Converter (ADC) of an embedded controller. The performance of the embedded application depends on accurate information coming from the sensor. Sensors can degrade with use, which may accelerate product failure. Analog-filter and gain characteristics can change over time, temperature and manufacturing variations.
Digital sensor processing can be employed to provide advanced compensation for a variety of sensor conditions, including sensor degradation, performance reductions in age-sensitive or temperature-sensitive components, failure prediction, substituting a less expensive sensor without performance loss, or improving sensor accuracy. Since DSCs are increasingly being used for motor control, they provide an opportunity to utilize excess resources to also handle digital sensor processing tasks. Fig. 1 illustrates the resource utilization for an advanced sensorless FOC algorithm on a 28-pin 128 KB Flash, the dsPIC33F128MC202 DSC. (Note: dsPIC is a registered trademark of Microchip Technology.)
There is richness to sensor data in the frequency domain that is not apparent in the time domain. Sensors are often remote from signal conditioning circuitry, and their signals are subject to environmentally induced noise. A low signal-to-noise ratio can adversely affect the performance of the end application.
| | | Fig. 3. Sensor signal plus noise in the frequency domain. | |
Take, for example, the thermocouple, which is a common sensor used for measuring temperature. Thermocouples are 2-wire sensing elements that use the Seebeck effect to measure the temperature at the junction of the two wires. The Seebeck effect creates a small voltage across the junction of two dissimilar metals that correlates to the temperature at the junction. Because of low signal amplitude and low current drive, thermocouple signals are highly susceptible to power-line contamination. For slowly moving temperatures, a low pass filter may satisfactorily reject power-line noise.
Digital filters are preferable for filter characteristics requiring adaptation (such as sharp rejection of a 50 Hz signal versus a 60 Hz signal), sharp cutoff frequencies or stability over time, temperature and production variation. For complex filters, the traditional analog filtering approach requires more components, is subject to drift over temperature and aging, and requires component swap-out for filter characteristic changes.
The thermocouple example can be extrapolated to any other sensor with induced noise that degrades application performance. For sensor applications with higher rates of change, the power-line noise may fall within the spectra of the sensor signal. This may best be filtered using a notch digital filter. Fig. 2 indicates a time-domain representation of a signal buried in noise. Fig. 3 illustrates how the desired signal becomes easily discernable in the frequency domain.
Another illustrative example can be found in a different class of sensor processing; one where sensor accuracy or reliability is desirable, but economically challenging. In such cases, two approaches can be taken. One is to use a lower-cost sensor and augment the lost reliability with digital sensor processing. The other is to combine sensor functions to eliminate a sensor. Alternately, it may be possible to do both. Turbidity detection can be used as a platform to illustrate plausible concepts.
Turbidity sensing | | | Fig. 4. DSCs can optimally handle sensor processing and motor control, and may replace the user-interface MCU, depending on architectural preference. | |
Turbidity detection can be used to detect the level of particulates in dish or laundry water, to determine that the load is clean enough to terminate a wash cycle early. This helps to avoid additional tub fills, which saves machine energy and time. Industrial-grade turbidimeters, used at water treatment plants to assess water quality in the treatment cycle, are very precise and rugged, but are too expensive for appliance use.
Appliance turbidity sensors employ optical techniques, where a beam of light is passed through the liquid being tested. Two optical detectors — one positioned head-on to the light source, the other at an angle of 90 Deg to the light source — measure the transmitted and scattered light photons, respectively. The greater the concentration of suspended particles in the water, the less light gets through and the more it is scattered. The turbidity of the water is determined by analyzing the ratio of the scattered light signal, divided by the transmitted light signal.
Most dishwashers have removed the detector collecting scattered light information to save cost, which has resulted in reduced sensitivity. On the other hand, while this technique measures occlusion to a degree that is satisfactory for dishwasher applications, it may not be sensitive enough for clothes washer applications where the change in turbidity is relatively small. The frequency-domain analysis of transmitted light would not only provide the strength of the collected signal, but also spectral information that may yield sufficient information to enhance sensitivity for washing machines.
By implementing frequency-domain analysis of turbidity information on a DSC, dishwashers may be able to optimize detergent use and reduce rinse cycles by administering proper concentrations of detergent for local water conditions. Digital sensor processing may also be able to distinguish between particulates, normal turbulence, and the presence of detergent by examining frequency domain characteristics. This may permit the detector to have dual uses: both detergent detection and normal turbidity measurement.
An expensive element of turbidity detection is the material through which the liquid passes. Lower-cost material can succumb to abrasion following long-term exposure to particulates in the water flow. This causes reflected light that reduces the sensitivity of measurement. Digital sensor processing may permit lower-priced materials to be employed, by using digital techniques to improve signal selectivity — thus extending sensor life.
Practical considerations | | | Table 1. Disruptive technology shifts in DC motor control | |
Appliance engineers do not need to be steeped in DSP technology to take advantage of frequency-domain techniques. Low-cost tools can recommend digital filter types and generate coefficients based on the desired filter parameters. In fact, free, optimized DSP libraries and tools that are available today permit the examination of sensor signals in the frequency domain. DSCs are well-adapted implementation vehicles for this frequency-domain processing, since they have specialized hardware and addressing modes optimized for DSP. Additionally, DSCs incorporate hardware specialized for motor control and power-drive applications, so both sensor processing and motor control can be accomplished with a single chip. Training and vendor support can also accelerate time to market.
The use of DSCs to perform both digital motor control and digital sensor processing presents appliance designers with new opportunities to
RSS Feed
Twitter
