Page 461 - The Mechatronics Handbook
P. 461

0066_Frame_C19  Page 83  Wednesday, January 9, 2002  5:27 PM









                       Thermistors
                       Thermistors are bulk semiconductors made from an oxide of nickel, cobalt, manganese, or other metal.
                       The oxide is ground to a fine powder and then sintered to produce the actual thermistor material that
                       is then incorporated into a sensor. Thermistors are resistance temperature sensing devices with several
                       notable differences from RTDs such as their large negative temperature coefficients, and extreme nonlinear
                       response. The resistance of a thermistor is usually so large (several thousand ohms) that lead wire
                       resistance is rarely a concern. Thus, they are inevitably two-wire devices unless multiple thermistors or
                       components are included in the probe. There must be some means of electrical bonding between the
                       wire leads and the thermistor semiconductor. This bonding and the typical epoxy encapsulation places
                       a limit on the maximum usable temperature, even though the thermistor itself is a refractory material.
                         There are several schemes for dealing with the nonlinearity of thermistors, ranging from applying a
                       correction curve with a computer to having multiple thermistors with differing characteristics complete
                       with nonthermal resistors as a bridge within a single encapsulated probe. For moderate temperatures
                       spanning 200 K, a simple external bridge can be used to linearize the signal.
                         Although not normally called thermistors, germanium, silicon, and carbon are semiconductors that
                       can also be used to monitor temperatures by measuring their resistance. Germanium is used for very
                       precise measurements at cryogenic temperatures down to less than 1 K. The change in resistance can
                       be very large and very nonlinear, but still very repeatable with a typical unit going from 7000 Ω at
                       2 K to 6 Ω at 60 K (Doebelin, 1990). Silicon can be used at room temperature and, depending on its
                       doping, can have a very steep temperature curve. It is rarely used as a temperature sensor since other
                       methods work better over its useful range of –200°C to 200°C. Carbon resistors out of a parts drawer
                       found in any laboratory can be used for cryogenic measurements from 1 to 20 K, but they must be
                       individually calibrated.

                       Integrated Circuit Temperature Sensors
                       The base-to-emitter voltage drop of a transistor operating at a constant current is a simple function of
                       absolute temperature. Thus, any transistor can be used as a temperature sensor. In reality, this is much
                       more of a problem with building thermally stable electronics than a convenient means of measuring
                       temperature. Integrated circuits are available that monitor the collector current, amplify, and linearize
                       the base-to-emitter voltage to yield an output that is proportional to absolute temperature. Common
                       integrated circuit temperature sensors are available with outputs of 10 mV/K, or 1 µA/K. The temperature
                       range over which they may be used is limited to  −50°C to 150°C by the construction techniques of
                       integrated circuits. This makes them very useful for referencing one junction of the thermocouple and
                       most ambient temperature measurements. Although not intrinsically water proof, the ICs are small metal
                       cans or plastic cases resembling signal transistors and can be potted or used in thermowells.
                         The IC sensors with a voltage output are commonly two terminal devices, with a possible optional lead
                       for trimming the response. When a small current of about 1 mA is allowed to pass through it, it will have
                       a voltage drop directly proportional to the absolute temperature (National, 2000). Even simpler IC trans-
                       ducers are available with separate excitation and signal leads. These are usually calibrated to 10 mV/°F or
                       °C. These have an inherent limitation of not being able to measure below a few degrees above 0°F or
                       0°C unless both positive and negative power supplies are available.
                         Voltage output ICs are very convenient where the temperature being monitored is local to the readout
                       and the voltage drop across the lead wires is not a concern, but for remote sensors, which require long
                       lines, current sensors are preferred. Current sensors are also two terminal devices that behave as high
                       impedance current sources so whatever lead resistance present may increase the voltage, but will not affect
                       the current through the sensor (Analog, 1997). Both types can be individually adjusted by trimming resis-
                       tances on the chip with a laser during manufacture to provide the rated output or they can have an external
                       adjustment lead. Even with trimming and calibration, the accuracy over the entire span from −50°C to
                       +150°C is rarely better than two or three degrees. Several individual ICs may be hooked up to give
                       minimum or average temperature. Voltage ICs are placed in parallel for minimum temperature and in



                       ©2002 CRC Press LLC
   456   457   458   459   460   461   462   463   464   465   466