Method for measuring and testing using electronic sensors

Method for measuring and testing using electronic sensors

1. What is temperature?

Heat is a form of molecular motion. The hotter an object is, the faster its molecules move. Absolute zero is defined as the temperature at which all molecular motion stops. But how do we measure temperature since we can't see molecules moving? The basic standard used by the National Institute of Standards and Technology (NIST) is based on the ideal gas law, which states that as temperature rises, the pressure or volume of a gas must increase proportionally. This number is expressed as P×V=KT, where P=pressure, V=volume, T=absolute temperature, and K is a constant. Doubling the speed of molecules in a fixed volume will double the number of molecular collisions per second, or double the pressure. At absolute zero, an ideal gas shrinks to zero volume and pressure. Figure 1 shows a fixed-volume gas thermometer. Ideal gases do not exist, but helium is close to an ideal gas. Mercury manometers are used to measure the pressure of gases and vapors and adjustable containers are used to measure the gas pressure of a glass ball filled with helium. When the temperature changes, the plunger in the container is adjusted to keep the left arm of the manometer at a fixed height, thereby keeping the helium at a fixed volume. When the right arm is evacuated, the temperature of the mercury column shows the pressure of the gas, and thus the temperature of the helium. The principle is simple, but it is difficult to measure accurately. The temperature will affect the volume of the glass ball, and the temperature of the connecting tube and the glass ball is not the same. In addition, the small change of the mercury column and the meniscus on the surface of mercury also limit the accuracy of the measurement. More important than these problems is the need to correct the degree of deviation of helium from the ideal gas law. Therefore, the method of measuring temperature using the gas law is mainly used by national standard experiments, such as NIST.

2. Temperature scale

Companies or laboratories that manufacture or standardize thermometers need more practical standards. Thus, the International Temperature Scale (NIST) was created. It was previously called the National Practical Temperature Scale to distinguish it from the basic gas law temperature scale. International conferences involving several national standards laboratories have frequently reviewed and revised this scale. The latest revision was published in 1990 and was changed to ITS-90. Temperature scales begin with a series of recognized basic temperatures or fixed points. The participating laboratories agree to specify the freezing or melting points of certain high-purity materials. Sometimes the triple point is used as the exact temperature value. Figure 2 shows a typical temperature fixed element. A graphite crucible containing high-purity metals is sealed in a graphite container, which is also filled with argon or some other inert gas. Table 1 lists several temperature fixed points. For example, the freezing point of silver is specified as 1234.93 degrees Kelvin or 960.323 degrees Celsius. The triple point of water is more easily controlled than its freezing point. It is specified as 273.16K or 0.01℃. The triple point is very similar to the freezing point, except that the material is sealed in a glass container that has been evacuated to a vacuum. The water is subject to only its own vapor pressure instead of atmospheric pressure.

Because the freezing point is affected by air pressure and contamination, the triple point can be obtained more easily and repeatedly. The triple point refers to the material being in three-phase equilibrium, gas phase, liquid phase and solid phase. To make the temperature scale effective, sensors can be inserted between the specified fixed points. ITS-90 stipulates that several such sensors are responsible for measuring various parts of the temperature scale. The middle of the temperature scale is between the triple point of hydrogen and its freezing point. Advanced resistance thermometers are inserted in between, called standard full resistance thermometers. SIRT and SPRT are carefully manufactured and fired with high-purity platinum wires, and assembled with minimal supports, so they are not strained. After the thermometer is standardized at three or more fixed points, the thermometer can be used between these temperatures. The RT column of the thermometer is very complex. It must be processed by computer. Figure 3 is a SPRT sealed in a Pvrex glass sleeve. The lowest end of the scale is 0.65K, which is defined by the helium law thermometry method. Several overlapping ranges are defined by their own complex columns and charts. At the high end of the scale, temperatures above the freezing point of mercury are defined by radiation thermometry, which is based on the principle that infrared or optical radiation increases with increasing temperature. The old IPTS also used thermocouples made of platinum alloy to define part of the temperature scale, but this was cancelled in the 1990 revision.

3. Commercial sensors

Next, let's look at and compare commercial temperature sensors: thermocouples, resistance thermometers, thermistors, and silicon IC sensors. Let's start with a quick look. Table 1 compares their characteristics, and Figure 4 shows their operating ranges and accuracies. A thermocouple is simply two different metals joined together. Once connected, it generates an electromotive force that increases approximately linearly with increasing temperature. The sensitivity, linearity, and temperature range of a thermocouple are related to the metals used. Over the years, several thermocouples have become standards. In the United States, NIST has published eight thermocouples, identified by letter codes on a millivolt-to-temperature scale. Five of them, J, K, T, G, and N, are made of alkali metal alloys and have different temperature ranges and uses. The sensitivity is generally tens of millivolts per degree Celsius. Three of them, R, S, and B, are made of platinum metal. Of course, this is the most expensive, most stable and repeatable thermocouple, and is most commonly used for high-temperature work, but the sensitivity is lower. Some manufacturers and distributors produce thermocouple wire and non-pointers to these standards.

In addition, some manufacturers produce special thermocouples for high temperatures, low temperatures, and other special purposes. The most common of these is the tungsten alloy thermocouple, which can measure temperatures up to 2015°C, or 4260°F. A resistor or thermometer has a coil of thin wire or metal film. The resistance of most metals changes with temperature, but platinum or nickel are most commonly used to make resistors or thermometers. In general, resistance thermometers are more stable, accurate, and sensitive than thermocouples, but only at lower temperatures. Resistance thermometers made of platinum are the most stable and accurate. And it is suitable for the highest temperature range, and the comprehensive price is relatively low, so it is suitable for medium temperature process purposes. However, recent technological advances in the manufacture of metal film components have offset the cheap advantages of nickel. This process is similar to the principle of metal film resistors. Sometimes people also use other metals, mainly copper and an alloy called Balco. Most readers may be familiar with thermistors. They are different from thermal resistors and non-resistance thermometers. They are highly sensitive, very nonlinear, and only applicable to a limited temperature range. Thermistors also have positive temperature coefficient types, but the most suitable for temperature measurement is the negative temperature coefficient PTC device. Its resistance decreases with increasing temperature, about 3%-5% of resistance per degree. Thermistors have various sizes, appearances, accuracy and prices of any commercial temperature sensor. Temperature sensor ICs are the newest and easiest devices suitable for most experiments. They are sensitive and linear, and easy to connect to operational amplifiers and analog-to-digital converters. In terms of disadvantages, these ICs are not as standardized as other sensors. Accurately calibrated ICs are relatively expensive. Their applicable temperature range is about the same as that of epoxy-coated thermistors.

4. Which sensor is best?

This depends on the temperature, application, and accuracy. At high temperatures, thermocouples may be the only choice. The most accurate is usually a platinum resistance thermometer, but precision thermistors are more accurate near room temperature. Because thermistors are highly sensitive, they are best suited for narrow range applications, such as medical thermometers. Thermistors and ICs are well suited for medium accuracy measurements and temperature compensation applications. ICs and resistance thermometers are available in a relatively small number of packages. For small sizes and quick applications, glass bulb thermistors range from 0.014 inches to 0.005 inches in diameter, and insulated thermocouple wire can be as low as 0.005 inches in diameter. For larger sizes, round thermistors can be up to 1 inch in diameter, and thermocouple wire can be as thick as 14AWG or thicker, with insulation ranging from PVC to ceramic, fiber, or ceramic beads. To measure surface temperature, you can use a ribbon thermocouple or directly find the thermocouple wire on the metal surface. Each of these devices is described in detail below.

5. IC sensor

A silicon diode in forward bias and a base-emitter junction can often be used to measure temperature. At room temperature, a forward biased junction drops a maximum of 0.7V, and has a large negative temperature coefficient of -2mV/°C. The exact voltage and temperature coefficients are dependent on the geometry of the junction, the current density, and other factors. Accurate calibration requires measuring each diode or transistor individually at a known temperature. The basic equation for a PN junction is I=IO(eqv/KT-1), where q is the charge of the electron, K is a physical constant called the Boltzmann constant, and T is the absolute temperature in degrees Kelvin, which is a constant that is essentially equal to the reverse bias leakage current. At room temperature, KT/q is about 26mV. Under normal forward bias conditions, this term is tiny and insignificant and can be ignored, so I=IOeq-1. v/KT, so I=I/Io=V. The working principle of the temperature sensor IC is based on the difference between the two base-emitter voltages. At this time, the current at the node maintains a fixed ratio of I2/I21. A little algebraic operation on this equation can give the voltage difference. The circuit in the figure uses this voltage difference to produce an output voltage or current that is proportional to the temperature. Table 3 lists four ICs. The AD590 and AD592 perform the same, but the newer AD592 is cheaper and uses a TO-92 package. It is suitable for the temperature range of the classroom. Beyond this range, the accuracy is more stringent. National's LM34/LM35 is a three-terminal device with a zero output at 0°F or 0°C, while the LM135/235/335 is a device similar to a Zener diode, and its output is proportional to the absolute temperature. Let's take a look at the AD592/590. AD592 and AD590 are two-terminal regulators with an output of 1μA/K and 272.5μA at 0°C. The manufacturer calibrates this at 5G, and guarantees that it will work between 4G and 3G, but be aware that increasing the voltage will increase power consumption and introduce slight measurement errors. Figure 5 shows their use in a simple circuit that can give a temperature reading of 0°C or 0°F from a digital meter in volts.

When a current of 1μA/K flows through R1, R1 divides the current value into a voltage value with a sensitivity of 1mV/0°C, 1.000K or 1mV/0°F, 1.8000K. The voltage across R1 is proportional to the absolute temperature. The compensation provided by resistors R2, R3 and R4 is equal to the voltage between R1 and 0℃ or 0°F. This compensation is adjusted using a digital voltmeter. To obtain the reading of the Celsius table, R3 must be adjusted to an output of 273.2mV. To obtain the reading of the Fahrenheit table, the output should be adjusted to 459.7mV. If R1 is originally ±0.01%, or a digital ohmmeter is used for fine-tuning, temperature calibration is not required to achieve the accuracy specified by the IC. If you want to use a lower-level IC, it is easy to achieve a high accuracy.

R1 can be replaced with an adjustable resistor. Keep the IC at a known temperature and place a digital voltmeter across R1. And adjust R1 to the correct reading of lmV/degree. It is recommended to put the IC in a closed sheath and put it in a uniformly stirred body of ice and water to achieve balance. Fine-tune R1 so that the voltage across it is 273.2mV at 0℃, or 491.4mV at 320F. Adjust R3 according to the above method. AD593 has children's models, from ±5℃ at 25℃, AD590J to ±5℃, AD590M, AD592 obtains a guaranteed 25℃, accuracy is from ±2.5℃, AD592AN to ±0.5℃, AD59ICN, AD590 is packaged in T0-52, transistor case or flat package, and AD592 is packaged in T0-92 type when shipped. National's LM34/35 series is easier to use. This three-terminal IC outputs 10mV/0F, LM34 or 10mV/℃. To read the temperature, all you need is a digital voltmeter and a battery or voltage source, anywhere from 4V to 30V. Figure 6 combines an LM34 or LM35 with a high voltage/frequency converter, the LM351, to produce a frequency proportional to temperature. The component values ​​shown produce an output with a precision of 100. At 100°F or 100°C, the output is 10kHz. To calibrate it, you can temporarily remove the sensor, provide a precise 1000V input, and adjust R3 for a full output of 10.00V without zeroing. If you want to improve the accuracy of a looser tolerance IC, you can place the IC at a known accuracy temperature near the upper end of the range and adjust R3 to get the correct output. The LM34/35 needs to be negatively biased to track subzero temperatures. Figure 7 shows the basic principle. The IC is powered from the power rail, but a bias current of about 50μA is applied to the output. The applicable temperature range of LM35 is -55 to 150℃, -40 to 110℃ is LM35C, and 0 to 100℃ is LM35D. The accuracy at 25℃ is ±1℃ and ±0.5℃ is LM35A. The Fahrenheit model of LM135 has similar series. Its package is T0-46 metal type and T0-92 plastic type. The last row of IC in Table 3 is National's LM135/235/335 series.

The operation of LM135 is a two-terminal regulator IC similar to a Zener diode. It is similar to the standard of LM185. It has a third terminal for users to connect to power, so that the standard, bias current or Zener current can be any value between 400μA and 50mA. Its output is l0mV/K. It is 273V at 0℃, which is proportional to the absolute temperature. The most stringent 25℃ guaranteed accuracy without user calibration is ±1℃ (LM135A and LM235A), and the loosest is ±6℃. The rated temperature range of LM335 and LM135 is a continuous range of -55℃ to 150℃, and LM235 is -40 to 100℃. Its package is T0-46 metal and T0-92 plastic.

6. Thermistor

Negative temperature coefficient thermistors are best suited for measuring temperature. They are narrow-range, highly sensitive and nonlinear devices. Their resistance at 25°C can range from less than 100Ω to more than 1MΩ. Their general sensitivity is -3% to -5%/°C. Therefore, their resistance can change from tens of ohms to tens of thousands of ohms per degree. To make negative temperature coefficient thermistors, metal oxide powders are required, usually nickel oxide and manganese oxide, and sometimes other things are added to mix. These powders are made into a slurry with water and adhesive, and then pressed into the required shape, such as discs and cylinders, and then dried. The dried thermistors are then burned at a temperature above 1000°C to form a refractory ceramic-like structure. Figure 8 shows some common thermistors. The most commonly used thermistors for measuring temperature are disc thermistors coated with epoxy resin, usually with a diameter of less than 0.1 inches. At higher temperatures, disc thermistors of similar size and sealed with glass are used. Bead thermistors with or without glass sealing have smaller sizes and faster responses. Thermistors are available in sizes from about 0.005 inches to 0.0005 inches in diameter, and in larger sizes, in cylindrical, disc, and ring shapes up to 1 inch in diameter. Some manufacturers also produce thermistor sensor assemblies, ranging from straight-bar pointers to components that can be fixed to transistor housings and surface mounted. Thermistors have always been inaccurate or unstable. This is the case with the cheapest devices. The typical resistance tolerance at 25°C is between 5% and 20%, which is equivalent to an accuracy of 1 to 5°C. This tolerance is looser at high and low temperatures. At least three companies, YSZ, Fenwal, and Thermomet-rics, offer replaceable precision disc thermistors coated with epoxy resin. The temperature range is from -80°C to 150°C, and the tolerance at both ends is about 1°C. It achieves accuracy and stability by grinding the disc resistors to precise values ​​in a heat treatment cabinet where the temperature is strictly controlled, and by aging tests and individual tests.

The resistance range at 25°C is from 100Ω to 1MΩ, but there is a value of 2252Ω at 25°C that has become a similar standard used in medical and laboratory thermometers. YSZ's 400 series has various probe types. This 2252Ω device is 1.66MΩ at -80°C and 41.9Ω at 150°C, which shows how sensitive this thermometer is. The manufacture of small glass thermistors is slightly different. It is to coat two high-temperature wires, usually platinum wires, with a drop of mud containing oxides, and then dip them into molten glass after baking. The resulting high-temperature device is generally more stable than the disc resistor with epoxy resin, but it cannot be adjusted. Manufacturers supply thermistors suitable for precise purposes through unit measurement tests.