August 2003
Table of Contents

Detecting Out-of-Balance Conditions with
MEMS Technology


A vigorous spin cycle in your washing machine makes less work for the dryer, but high spin rates can shake your washer to pieces. To keep that from happening, this MEMS accelerometer detects off-balance load conditions and helps corrrect the problem.

Christophe Lemaire, Analog Devices, Inc.

In an environmentally conscious world, domestic appliances are easy targets for efficiency improvements. Clothes dryers are high on this list, consuming up to 3.5 × more power than washers. One of the most effective ways to cut down overall power consumption is to reduce the time it takes to dry the laundry. Higher spin speeds in washers mean a drier wash, shorter dryer cycles, and lower power consumption. The demand for fast spins, coupled with a need to reduce machine weight and operating noise, is driving research into better out-of-balance detection and improved vibration control.

Today’s Washing Machines
High rotational speeds and off-balance detection are now being marketed as key attributes and differentiating features among washing machines. Speeds >1500 rpm (25 Hz) achieved by some front-loaders create strong centrifugal forces that extract more water out of the laundry and thus shorten the spin cycle. These speeds are possible only if the load inside the drum is well balanced; if it’s not, there will be noisy rotational and translational vibration that could cause catastrophic failure. Load balance at the end of the wash cycle is a product of the type and amount of fabric being washed, so the only way to prevent mechanical problems during spin is to detect, assess, or even predict out-of-balance conditions in real time.

Detecting Out-of-Balance Conditions
A spectrum of methods is currently being used to detect out-of-balance conditions. At the low end is a combination of mechanical switches that sense when the drum displacement exceeds an established safety threshold. If excessive motion is generated, the drum itself will activate the switches and turn off the machine. A more sophisticated approach consists of monitoring variations in the machine’s electric motor torque, rotational velocity, or the amount of power drawn.

Figure 1. In the presence of an unbalanced load, the tach signal generated by the motor of a front-loading clothes washer exhibits a ripple at a given rotational speed. The machine can be programmed to shut down or take corrective action if and when the ripple exceeds preset threshold levels symptomatic of an excessive out-of-balance condition that could lead to undesirable effects at higher spinning speeds.
In a horizontal-axis machine (front loader), an unbalanced load will generate a gravity-induced torque or velocity ripple during each revolution. This condition can be detected by looking at the motor currents and/or voltages, or the rotational speed (tach) signal available on most washing machine motors. This signal is generally a pulse train whose frequency should be constant at steady speed. The signal is fed into the main control unit microcontroller and its frequency is monitored for any variation. A velocity ripple will reflect load imbalance (see Figure 1).

At the end of the wash/rinse cycle, just before the spin cycle, the water is usually drained, and the machine enters a cycle of slow and constant rotation during which the signals are monitored. If the ripple exceeds a set threshold, the cycle is interrupted and the rotational speed is reduced or varied to promote repositioning of the load inside the drum. This method is most effective only at low spin rates (<100 rpm). Even when algorithms are used to extrapolate the machine’s behavior at higher speeds, the accuracy of this technique is limited and the effect of an unbalanced load at higher speeds is unpredictable. A load that is well balanced at low speeds can become unbalanced at higher rates because different fabrics allow different degrees of water extraction.

In addition, when monitoring currents and voltages to sense fluctuations in drawn power, measurement accuracy is limited by the tolerances of the associated circuitry’s passive components. This limitation is acceptable in many cases, but as machines are made to spin faster and faster, improvement in accuracy is desirable.

Manufacturers have used a variety of other ways to detect and eliminate unbalanced loads. A European manufacturer combined motor current sensing with drum displacement sensing; another used several sensors, including a piezofilm accelerometer, together with mechanical damping to detect and control unwanted vibrations. Yet an- other’s use of a vibration sensor eliminated one of the three suspension springs and downsized the shock absorbers.

None of these solutions can handle vibration detection over the entire rate range of high-speed washing machines. The only accurate and effective way to identify a potential balance problem might be to directly measure the accelerational forces exerted on the drum/tub assembly throughout the spin cycle. But this technology was slow to arrive because:

• There were no sensors available that addressed the price/performance requirements of extremely cost-sensitive appliance manufacturers.

• Measuring out-of-balance conditions by monitoring the tach signal for velocity ripple worked fine for most washers because their spins rarely exceeded 1000 rpm.

Direct Sensing with MEMS Accelerometers
The ADXL210E iMEMS accelerometer (see Figure 2, and sidebar), a low-cost, dual-axis, ±10 g sensor with analog and digital outputs, is capable of resolving a few milli-g’s of acceleration.

Figure 2. The ADXL210E, a dual-axis, ±10 g accelerometer, offers X- and Y-axis acceleration output in the form of PWM signals, making it easy to interface directly to the counter port of a microcontroller. The duty cycle T1/T2 is directly proportional to acceleration. The product requires a minimum number of passive external components.

It requires minimal external passive components and can be mounted on a small PCB affixed to the tub and directly connected to the microcontroller in the machine’s controller unit. The dual-axis configuration allows both rotational and axial vibration sensing.

The original sensor was introduced in 1999, but when a lower cost, small leadless chip carrier package version became available in 2002, the accelerometer became the first commercially available MEMS product that could both satisfy out-of-balance applications and approach the target costs set by appliance OEMs.

In most vibration sensing applications there is no need to detect static acceleration signals, so the ADXL210E can be AC coupled to the microcontroller and thus eliminate the need for calibration. Adding a capacitor in series with the analog outputs will effectively achieve AC coupling. When using the digital outputs, the same result can be achieved by comparing one group of samples to a previous group and thus determine whether any signal change is the result of measured acceleration or offset drift. Any change slow enough to be unrelated to vibration-induced acceleration can be ignored and eliminated by resetting the
0 g offset value.

If need be, users can maximize the accler- ometer’s resolution and dynamic range by adding a small capacitor between the analog outputs and ground. The combination of this capacitor with an onchip resistor forms an RC filter, which can be set to limit the bandwidth to the frequency of interest. This will effectively limit the amount of noise generated by the device.

MEMS Sensor Evolution
The ADXL210E has evolved from several previous generations of Analog Devices’ iMEMS sensors to become one of the first MEMS accelerometers to be considered for use in washing machine applications. The recently introduced ADXL311 reduces the costs still further. For applications requiring high accuracy and stability over temperature, the ADXL203 will soon be available. Other products featuring even lower cost or improved levels of performance are scheduled to follow in the relatively near term. Appliance OEMs will soon be able to choose from a wide selection of MEMS sensors designed for cost-sensitive applications, satisfying the requirements of out-of-balance detection in washing machines.

Appliance industry OEMs, and manufacturers of washing machines in particular, are being challenged by fierce competition, tough environmental and energy conservation standards, and an increasingly knowledgeable consumer. The result is a rising de- mand for washers with higher spin rates that require more sophisticated out-of-balance detection systems. While existing systems do provide some level of this for today’s washers, their accuracy is questionable over the entire operating range of the machine.

Economies of scale achieved in the production of tens of millions MEMS automotive accelerometers have contributed to significant reductions in manufacturing costs, encouraging their use in cost-sensitive consumer products. Despite the appliance industry’s traditionally conservative stance on adopting new technology, MEMS sensors have found their way into washing machines for direct, more accu rate, out-of-balance condition detection. In the near future, new MEMS sensors are expected to become available at a price/performance point that will lead to acceptance across the industry.

The ADXL210E
Figure 3. The ADXL210E is available in a space-saving, 5 by 5 by 2 mm, 8-pin ceramic leadless chip carrier package.
This low-cost, low-power, complete 2-axis accelerometer (see Figure 3) with analog and PWM outputs is contained on a single monolithic IC. It can measure both dynamic acceleration (e.g., vibration) and static acceleration (e.g., gravity) with a full-scale range of ±10 g. The sensor is built using the same iMEMS (integrated microelectromechanical systems) process used to produce more than 100 million accelerometers crucial to the airbag crash detection modules in today’s automobiles.

The outputs are either analog voltage signals or digital signals whose duty cycles (ratio of pulsewidth to period) are proportional to acceleration. A microprocessor counter without an A/D converter or glue logic can directly measure the duty cycle outputs. The duty cycle period is adjustable from 0.5 to 10 ms by means of a single resistor. The typical noise floor is 200 mg/Hz, allowing signals below 2 mg (at 60 Hz bandwidth) to be resolved.

Christophe Lemaire is Marketing Manager, Consumer and Industrial Markets, Micromachined Products Division, Analog Devices, Inc., Cambridge, MA.

For more information, contact Analog Devices, Inc., 617-761-7000.

For further reading on this and related topics, see these Sensors articles.

"Machine Condition Monitoring", Part 1 and Part 2, May and June 2003
"Controlling Vibration with Magnetorheological Fluid Damping," February 2002
"An Extremely Low-Noise Micromachined Accelerometer with Custom ASIC Circuitry," May 2001

Sensors Weekly
  What's New
  Product Picks

Questex Media
Home | Contact Us | Advertise
© 2009 Questex Media Group, Inc.. All rights reserved.
Reproduction in whole or in part is prohibited.
Please send any technical comments or questions to our webmaster.