Sensors Suppliers Sensors 101 Sensors Articles

Input / Output Devices

Other Sensors

Many types of transducers are beneficiaries of IC technology. The benefits are twofold. There is the advantage of the chips themselves -- enabling manufacturers to build more intelligence and better noise immunity into their transducers. However, it has become apparent that spin-offs from ICs have an equally important effect. For example, the same basic technology used to make the miniature circuits is helping to improve the resolution of optical encoders.

Especially impressive is how IC technology has helped reduce the cost of some devices. Just a decade ago, many pressure transducers cost $300 or more. Today, with the advent of micromachining, devices with similar capabilities may only cost a few dollars.

Temperature sensors: Temperature sensors include thermocouples, resistive temperature detectors (RTDs), and thermistors. Each device is limited in terms of measurement accuracy. The amount of error associated with a given measurement approach must also be considered when choosing a sensor.

Thermocouples are widely used to measure temperature despite low sensitivity and moderate accuracy. They consist of two dissimilar metal wires joined at one end called the sensing junction. The voltage produced across this junction is proportional to its temperature. Because of a wide variety of usable wire materials, thermocouples collectively span an operating temperature range from -200 to 2,000°C. Typical accuracies are from 1 to 3%, depending on material type and manufacturing variations. Sensitivities range from 40 to 80 µV/°C. Also, the response time for most thermocouples is on the order of a few seconds.

Connecting a thermocouple to a measuring device creates a second dissimilar-metal junction called the reference junction. In a type-J thermocouple, for example, the reference junction is formed where the iron lead connects to a copper terminal or wire. This junction is in series with the sensing junction. Voltage generated at the reference junction must be subtracted from the overall measured voltage to determine the temperature of the sensing junction.

To simplify temperature calculations, reference-junction temperature is maintained at a level that produces a known constant voltage. The most common reference is 0°C, which is the temperature of an ice bath. A 0°C reference insures repeatability and accuracy because the ice point of water is a constant. The National Institute of Standards and Technology accepts the ice point of water as the standard reference temperature.

An ice bath may not always be convenient, however. In that case, the most common alternate method of determining reference voltage is with an integrated-circuit temperature sensor. The IC sensor is placed near the reference junction and measures local temperature. From its temperature reading, reference-junction voltage may be calculated.

Thermocouples age or drift when used at high temperatures. Aging is usually a much greater problem in the low-cost base-metal thermocouples than in those using platinum-based noble metals.

The upper temperature limit of base-metal thermocouples in air is generally determined by their oxidation resistance. This limit varies with wire size. The type-K thermocouple is the most widely used due to its oxidation resistance and high melting point. Even so, decalibration drifts of 10°C have been recorded with as little as 50 hr of exposure at 1,250°C.

A new nickel-base thermocouple called nicrosil/nisil has higher stability in air and better air-oxidation resistance at high temperatures than base-metal types. Its high-temperature drift is three times less than that of a type-K device. Estimates are that maintenance costs for the new thermocouple are 12 times lower than those of type K.

Most materials change resistance with temperature variations. Two types of temperature transducers use this phenomenon: RTDs and thermistors. Depending on the measured temperature, these devices are increasingly used in lieu of thermocouples in many applications. The reason is simplicity. Thermocouples need special wires and cold-junction compensation; RTDs and thermistors do not. Thermistors find use up to about 100°C, while RTDs are practical up to about 400°C.

The primary difference between RTDs and thermistors is that the former use a metal-sensing element while the latter use a semiconductor. As a result, RTDs offer better linearity and are more stable at high temperatures. But RTDs usually cost more and require more complex measuring circuits.


RTDs typically consist of a wound platinum wire inside a glass or stainless-steel package. Platinum is preferred for several reasons. It is malleable, linear, has a higher resistivity than more common metals, and resists corrosion. While some low-cost RTDs use nickel alloy or copper wire, these sensors are usually less sensitive and have lower temperature ranges.

Wire-type industrial RTDs wrap the sensing wire around a bobbin. But for high-precision or laboratory measurements, the wire is left unsupported, a type of construction called a birdcage element. Because birdcage elements are free to expand or contract with temperature changes, they experience less strain. This, in turn, minimizes strain-induced resistance changes but makes the devices susceptible to failures caused by vibration.

While wire-wound bobbin RTDs are strong enough for most applications, a more rugged version is the thin-film type. Here, manufacturers screen a metal film (slurry) onto a ceramic substrate. The cured film is laser trimmed and sealed.

One problem with RTDs not made of platinum is limited temperature range. For example, copper types can measure only up to about 120°C. To increase range, some manufacturers use permalloy (Ni-Fe), which has a resistivity several times greater than that of many other metals.

In picking temperature sensors, it is important to realize that packaging considerations limit RTDs to below about 400°C unless specifically designed for higher temperatures. The basic problem is iron poisoning of the platinum sensing element. At elevated temperatures, small amounts of iron impurities can leach out of the stainless-steel package and corrode the platinum. Two methods can prevent this. One is to use higher quality stainless-steel packages (324 or 327 alloys). The other solution is to place a quartz sheath between the detector and stainless-steel package.

Acquiring data from RTDs involves resistance measurements. Platinum RTDs start at 100Ω at 0°C and vary by 0.0385Ω for every 0.1°C change in temperature. Unfortunately, package leads made from copper wire also have a measurable temperature dependence. As temperatures increase, the resistance of the copper leads also increases. A lead resistance variation of as little as 1Ω may cause measurement errors of a few degrees.

To avoid errors caused by lead-resistance variations, several measurement schemes use more than two lead wires. In multiple-lead RTDs, a constant-current source excites the sensor through one pair of leads, while the voltage drop is read from a separate lead pair. Because extremely small currents are in the voltage measurement loop, resistive drops are negligible.

One type of multiple-lead RTD has three wires and is commonly interfaced to measurement systems through a bridge-completion circuit. However, several weaknesses are associated with this technique. In order to provide repeatable measurements, lead resistances must closely match and the bridge must remain balanced. Although a potentiometer may be used to balance the bridge, it also causes major inconveniences. For one, the electromechanical device is unstable over time and requires frequent adjustment. Also, an adjustment is necessary every time the RTD is replaced.

A four-wire ohms-measurement circuit is a better way to cancel the effect of lead resistance. A stable power supply forces a known current through the RTD and the primary lead resistances. Because the input impedance of the measuring circuit is usually greater than 10MΩ, virtually no current flows and no voltage is dropped across voltage sense leads. The resistance of the RTD is found by dividing measured voltage by source current.


Thermistors, ike RTDs, are temperature-sensitive resistors. As temperature increases, thermistor resistance decreases, but at a much faster rate than that of RTDs. As a result, thermistors can sense minute changes in temperature that are otherwise undetected by RTDs and thermocouples.

Temperature coefficients for thermistors are typically as large as -2 to -6%/°C, compared to about 0.4%/°C for RTDs. Most thermistors have a resistance of 5,000Ω at 25°C and sensitivities between 100 and 300Ω/°C, much larger than the 0.4-Ω/°C sensitivity for 100-Ω RTDs. Bridge-completion and ohms-measurement circuits are not needed when connecting thermistors to acquisition equipment because lead resistances are negligible with respect to sensor variations.

An average thermistor with a temperature coefficient of 4%/°C for example, changes 200Ω/°C. At this level of sensitivity, a measurement lead with a resistance of 10Ω only contributes a 0.05°C error. The same lead causes an error of 25°C in a 100-Ω RTD, 500 times worse.

Compared to other sensors, thermistors have a limited measuring range, typically from -80 to 150°C. Also, because they are often made from semiconductors or sintered mixtures of metal oxides, they can sustain permanent damage at temperatures above their specified operating range.

Another problem is called self-heating error. As thermistors dissipate power they may warm slightly. Because they are so small, typical power dissipation constants are about 1 mW/°C. When measuring cold temperatures, for example, it is possible for the device to be warmer than the temperature being measured. Thermistors also are more nonlinear and curve-fitting polynomial equations or resistance-temperature lookup tables are often used to approximate temperature.

A significant advantage of thermistors, however, is that they easily interface with PCs, requiring very little circuitry. In many cases a simple voltage-divider network is sufficient. A dc supply produces a voltage Vs that is shared between the thermistor and the fixed resistor R. The voltage Vo across the fixed resistor is measured to calculate thermistor resistance RT.

The best value for the fixed resistor is determined from several temperature curves. The resistor value is chosen based on the specified thermistor, resistance ratio, temperature range, and required linearity. Most thermistor manufacturers provide literature that simplifies the complicated process.

Manufacturers sometimes place thermistors in two groups determined by lead attachment. The first classification is the bead type. These have platinum wires sintered into a ceramic body (bead). Beads are sometimes left bare or have an organic coating (epoxy), but those sealed in glass have the best stability (temperature drift over time). Stability is directly proportional to glass thickness because thicker glass helps prevent changes caused by oxidation. For instance, a bare bead drifts ten times more than a glass-coated one.

Metallized surface-contact thermistors form the second group. These are called chips or flakes. In contrast to bead types, leads are not sintered directly into the ceramic. Instead, the sintered ceramic is coated with a metallic contact. Either the chip manufacturer or user attaches leads to this contact.

One advantage of chip thermistors over bead types is that the chips are easily trimmed by cutting or grinding. Thus, they are easy to match and, therefore, are interchangeable. While matched bead thermistors are available, they cost more than interchangeable chips.

Interchangeability is one of the major advantages of thermistors. The 2,252Ω elements have a unit-to-unit variation of less than ±0.2°C from 0° to 70°C. Decalibration drifts are on the order of less than 1°C per year if used constantly at 150°C. Drifts of less than 0.3°C can be expected at temperatures below 100°C

Temperature switches

These devices typically comprise sensing elements and switching contacts housed in a single mechanical assembly. The sensing element measures temperature and actuates the contacts in response to thermal variations. Switches may open or close on temperature rise depending on their internal construction.

In most applications, temperature switches provide one of two functions: control or cutoff. Temperature switches operating in the control mode are called thermostats. They are used to maintain the temperature of a system within a specified range. Their contacts either pass or block control currents as a function of system temperature.

In the cutoff mode, temperature switches protect equipment against over or undertemperature conditions. In many cases, temperature cutoffs are a backup for a primary temperature controller. Once temperatures return to normal, some cutoffs automatically reset, while others must be manually reset. One type of cutoff, called thermal fuse, must be replaced following actuation.

Electromechanical temperature-actuated switches are available in several different forms. Designers can choose from reed switches, liquid-filled thermostats, mercury-in-glass tubes, and a family of differential-expansion devices.

Differential-expansion thermostats are the most common ones used today. They operate on various switching mechanisms including bimetallic discs, fused-bimetal elements, and mechanically linked assemblies. Some switch with a snapping action, while others, called creep-type switches, change states gradually.

Differential-expansion thermostats make use of an interaction between two metals with widely varying coefficients of thermal expansion. The interaction causes a movement that actuates a set of contacts, opening or closing them.

One type of differential-expansion switch is the fused bimetal thermostat. In bimetal thermostats, the sensor is made of two strips of dissimilar metal bonded into one element. When the temperature changes, unequal expansion of the two metals causes the strip to bend into an arc. The movement either makes or breaks an electrical contact.

In some cases, the bimetal element is the current-carrying conductor, while in others it only pushes the conductors together. In either case, the movement of the bimetal strip is gradual, and the amount of movement is proportional to the temperature. The make temperature is calibrated with an adjusting screw that varies the position of the fixed contact.

Fused-bimetal thermostats are typically used in electric blankets. Normally closed contacts pass current, allowing the blanket to heat. Electric blankets remain at a safe operating temperature with full heating current applied. If an overtemperature occurs, the bimetal strip deflects and breaks the circuit. When the cause of overheating is removed, the switch returns to a normal state.

A potential problem with fused-bimetal thermostats is contact arcing. Because the contacts slowly open and close, arcing may occur during intermediate states where contacts are close together or only lightly touching. Arcing wears out the mating surfaces of the contacts and raises contact resistance. As a result, cycle-life limits of creep-type fused-bimetal thermostats should be closely observed, especially in products where many actuations are expected.

In contrast, snap-acting bimetal thermostats open and close extremely fast with greatly reduced arcing. Switching times are as low as 0.1 msec. Rapid-contact separation insures long contact life as well as calibration stability. It also reduces the amount of radio-frequency interference that moving contacts often generate.

Most snap-acting elements are concave bimetal discs. When the temperature changes, expansion on one side is much greater than that on the other. The stress created by the unequal expansion increases until it overcomes the biasing stress. At that point, the disc inverts with a snap into a convex shape. When normal temperatures return, the process is reversed.

Bimetal-disc thermostats usually have a fixed (tamperproof) temperature setting and a relatively large operating differential. The large differential makes them useful when a substantial dead zone (10 to 20°F) is desirable. Some bimetallic switch makers recommend a minimum differential of 15°F, and 25°F for systems operating above 250°F. This optimizes the trade-off between the operating differential and the crispness of the snap action. Switching elements with small operating differentials have reduced snap action.

Bimetal-disc thermostats respond to temperature changes down to about 3 or 4°F. They can sense airstream or surface temperatures and radiant heat. Their operating range is from -20 to 1,490°F. They can switch dry circuit loads of 400 Vac at up to 50 A.

Another type of differential-expansion thermostat is a mechanically linked strut and shell. Unlike bimetal switches where two metals are bonded together into a single element, the active metals in mechanically linked switches remain separate. The more thermally active metal, usually brass or stainless steel, is the switch housing or shell. The less active metal, usually a high-nickel alloy, is a strut assembly on which the contacts are mounted.

As temperatures vary, the shell and strut expand or contract unequally, opening or closing the contacts. Actuation temperatures are adjustable with a screw that either compresses or stretches the strut assembly. Because their outer shells are active sensing elements, strut-and-shell switches respond extremely fast to temperature changes.

One advantage of strut-and-shell thermostats is that they typically have high-resolution sensitivity, down to 0.1°F. Because the switching mechanism makes or breaks slowly (creep action), almost any temperature change causes a corresponding change in contact spacing. That is, contact action can be produced by very small temperature variations. In contrast, snap-acting bimetal elements often have resolution sensitivities of several degrees because a finite amount of energy must be absorbed to overcome the restraining forces holding the contacts in place.

Strut-and-shell thermostats also have high vibration and shock resistance. Because the strut assembly is constantly under tension or compression, it resists mechanical shock and vibration. This allows it to operate reliably and accurately under physically harsh conditions. Such hazards can cause problems in other types of differential-expansion switches.

Another advantage of the strut-and-shell switch is that it "anticipates" rapid temperature changes as a result of an inherent time lag between the shell and internal struts. The lag allows the shell to lead the struts by a time interval proportional to the rate of temperature change. In the case of rapid temperature change, the shell exerts a larger net force on the struts and actuates the switch sooner than for a gradual change. This produces several degrees or more of anticipation, leading to a tighter control band.

Another mechanically linked type operates on a stainless-steel sensing tube. The tube contains a two-section inner rod whose active section expands at a much lower rate than the tube. The expansion difference is multiplied by a lever that actuates a snap switch or pilot valve. Sensing-tube thermostats operate at up to 2,000°F. 


Sensors: Sensors Information

Previous Page     Next Page