Solid-state liquid-level and flow detection are boosting the reliability of smart appliances.
A new platform of electronic sensors is under development for the appliance industry. In particular, two types of heat-transfer-based sensors target the sensing, metering, and controlling of water volume. One directly measures water level inside appliances while the other monitors the flow rate of water dispensed through a valve.
Water-level sensor development is currently driven by applications in the dish and clothes-washer industry where the U.S. Dept. of Energy has mandated conservation of energy and water. The water flow-rate sensor targets the refrigeration industry to more accurately fill ice-cube trays. There it's expected to reduce the number of over and underfill service calls.
The liquid-level sensor (LLS) measures water level in the tub bottom of a clothes washer or sump section in a dishwasher. Prototypes measure water level over a 1 to 2-in. (25 to 50-mm) range with a resolution of ±0.1 in. (±2.5 mm.)
The LLS employs multiple resistive heaters spaced vertically on a printedcircuit board. The temperature of each heater depends on whether it is exposed to air or immersed in water. Spacing of the heaters determines the resolution of the level measurements.
A unique feature is the copperconstantan, or T-type, thermocouples that sense the temperature of each heater. The thermocouple junctions are formed by soldering a thin constantan wire directly to the copper pad on the PC board. The PC board copper becomes one of the junction metals while the constantan wire forms the other. Cold junctions locally compensate for fluctuations in air and water temperature. Temperature fluctuations in a dishwasher or clothes washer range from 50 to 175°F (10 to 80°C). A thin coating impervious to water protects the components and PC substrate material against corrosion.
Thermocouple output signals get conditioned and converted to an analog 0.5 to 4.5-Vdc signal fed directly into the a/d of a microprocessor. Power consumption during measurement is typically in the range of 250 to 500 mW. The actual power dissipated depends upon the resolution and level range. A surface-mount thermistor near the bottom of the PC board reads out water temperature continuously.
Difficulty in visually determining water levels accounts for some of the variations from probe to probe. There is nonlinearity at the highest and lowest level because the temperature profile tapers off at the top and bottom-most heaters where edge and other boundary effects come into play. Designing the sensors for a specific level range takes these effects into consideration. Test probes responded linearly over a liquid-level range of approximately 1.5 in. (38 mm.)
Level sensors placed at the bottom of a dish or clothes washer see a number of conditions that create inaccurate readings. In particular there are water "sloshing" and spray effects. A protective probe base and shield combination mitigates these conditions. In addition, the shield minimizes fluid flow in the immediate area of the PC-substrate stabilizing-sensor output.
Substrate material options include rigid glass epoxy such as FR-4 or a ceramic like alumina. More flexible polyimide-based films like those used in flexible circuits are also possibilities. Substrate selection impacts parameters such as overall probe size, probe response time, reliability in harsh environments, and the probe's ability to physically conform to contoured shapes.
Some appliances must meter water as it flows past a valve for accurate dispensing. One such application is in the automatic icemakers of refrigerators. Traditionally, an electromechanical timer controls the fill of the ice tray. The timer dispenses water into the tray over a fixed period. The problem is that the flow period is independent of the incoming water line pressure. Low pressure, clogged filters, and other water-line obstructions all result in ice trays that are underfilled. On the other hand, high line pressure may make the ice tray overfill and perhaps flood the freezer compartment.
Accurate metering avoids these problems. But metering based on mechanical sensors, such as those employing paddle-wheel designs, poses long-term reliability issues. Gradual wear and mineral build-up on the bearings may eventually jam the sensor and generate a service call. An all-electronic sensor with no moving parts is far more desirable.
Thermal-based flow sensors are often candidates for accurate, no-moving-parts monitoring of fluid flow. They're more technically known as thermo-anemometers. One design suspends a heat source and temperature sensor within the fluid. Fluid flowing past the heat source cools it through forced convection. The rate of heat loss is proportional to the rate of flow. Of course the fluid ambient temperature affects the amount of heat the source gives up for a specific flow rate. So a second temperature sensor mounted upstream compensates for fluid-temperature changes. Both temperature sensors are NTC thermistors.
It's best to minimize the thermal mass of the heater and sensors to get a quick response time. Towards this objective the resistive heater and both temperature sensors reside in a four-wire bridge configuration. The entire assembly is screen-printed onto a thin ceramic substrate. The result is a very low thermal-mass design.
Development work on the sensor includes the use of a virtual icemaker control. Here a laptop PC, data-acquisition hardware, and National Instruments LabView programming environment analyzes the sensor's differential output signal. A control algorithm triggers the resistive heater on and off while opening and closing a relaydriven water valve. In one case such a virtual control tested various parameter settings to get a desired fill level based on the flow rate. Preheating the resistive heater 1.5 to 2 sec before opening the valve produced a faster measuring response.
Development efforts currently focus on optimizing the polymer encapsulation and improving sensitivity of the device. The goal is to get ±5% measurement accuracy over the entire 0.15 to 0.75-gpm range.
A ceramic-filled thermoplastic encapsulates the sensor to protect against moisture. Its high thermal conductivity insures good heat transfer between the ceramic sensor substrate and the water.
It has been observed that the level sensor responds up to 10X faster during the fill cycle than the drain cycle. Whether this affects the accuracy of readings strongly depends on how fast water flows in and out of the system. Similarly, instantaneous spikes in line pressure during the fill time of the ice tray may throw off accuracy. But the likelihood of these spikes during a 20-sec fill cycle may not be worth worrying about. In any case, a more sophisticated control algorithm could handle even this problem by continuously monitoring flow once the valve opens. Other limitations involve application-specific factors such as the presence of foam and bubbles in soap-saturated water.
A few modifications can let these sensors serve in other areas beside appliances. Packaged differently, the same sensor can handle other areas of level and flow sensing. Through such means both the LLS and WFS can measure over extended ranges, better withstand harsh environments, and incorporate multiparameter sensing.
The LLS technology, for example, can use flexible substrates. This lets the sensor physically conform to the wall of a tank or reservoir. High-pressure or vacuum environments may find the in a threadedhousing and a hermetically sealed glass interface. Signal-conditioning electronics are in a reliable, leakproof package.
A flow-bypass extends range of particular designs. Adaptations measuring air or flow may find use other appliances such clothes dryers.
Finally, use of modular-design may let sensors sit in probe package. As example, a single probe protruding through sump of the appliance may sense water temperature, level, and possibly turbidity. This not only reduces sensor installation cost, but minimizes the number of potential leak points in the sump.