Comment

1 May 2005

ZERO tolerance

By Pete Shelquist and John Zwak

By recognizing error areas, you can ensure accurate measurements.

Temperature is a critical measurement in the tires we trust when we put our family in the car, the production of energy to heat and cool our homes, the production of medicine we take to make us healthy, and the food we produce to feed the world.

These industries rely on accurate, reliable temperature measurement to help them produce quality products, safely and efficiently.

The importance of temperature measurement for personal safety, quality product, and efficient operation is paramount. However, there are errors in the temperature measuring system that can threaten the results we desire. Fortunately, there are known and predictable error sources in temperature measurement that we can analyze and minimize to help us reduce error and insure an accurate measurement.

To reduce the possibility of confusion, it is important to review terms commonly used in temperature measurement.

• Error: The difference between the "true value" and the "measured value."
• Uncertainty: The measurement uncertainty is a parameter associated with the result of a measurement, which characterizes the dispersion of the values that you could reasonably attribute to the measurand. In its simplest form, uncertainty is the known given acceptable range for a component or system. Components have unknown error; the known deviation of readings in a system is the system's uncertainty.
• Calibration: The process of verifying sensor performance with respect to temperature. Calibration uncertainty consists of temperature distribution and stability of the calibration media, the calibration standard, resistance measurement device and interpolation errors, and sensor stability. High quality calibration at proper intervals can recognize and minimize error.
• Hysteresis: Hysteresis is the difference in resistance at the same temperature as a result of only the thermal history of the sensor. Hysteresis errors occur when a sensor faces exposure to the same temperature from an increasing direction verses approaching from a decreasing direction. The inherent physical characteristics of an element will determine its hysteresis.
• Interchangeability: This is variation of a sensor to the nominal resistance temperature (R/T) curve as a result of manufacturing tolerances. Once you know this deviation to the curve, you can match the transmitter to a sensor. The advantage is highly improved system accuracy.
• Time response: Time response is the time required for a probe to indicate 63.2% of a step change in temperature. Depending on the application, an unknown time response or longer than expected time response can create error due to the temperature sensor not being able to react to the changes in the sensed medium.
• Self-heating: The current that measures sensor resistance can also heat the sensor. The amount of error created by this is variable based on the sensed temperature and the corresponding amount of resistance in the element. You can reduce this value by minimizing the sensing current, using a sensor with a low thermal resistance, and maximizing the thermal contact with the process.
• Leadwire resistance: The resistance of the leadwires can create error when not eliminated from the circuit. Three and four wire configurations can minimize this additional resistance.
• Stem conduction: This is error caused by heat conduction along the sheath and thermowell. Though usually a small part of the system error, it can be a major factor as line sizes and corresponding insertion lengths become smaller. The other contributing factor to stem conduction is the difference between the ambient and media temperatures. Minimizing stem conduction error can happen by using a sensor/thermowell that has sufficient immersion into the process and insulating the exterior portion to isolate it from ambient conditions.
• Repeatability: The ability to give the same measurement under repeated matched conditions.
• Stability: The ability to give the same measurement over a "long" period of time.

Hysteresis errors occur when a sensor faces exposure to the same temperature from an increasing direction

Probing error

There are multiple ways an error can occur with a probe or temperature measurement element or system. In the perfect world, for the longevity of a probe we would subject it to one continuous temperature. There would be no cycling, no extreme temperatures—only one steady, moderate temperature. The longevity of the probe and the accuracy over that time would be significant.

This is not always the case, as in probes used in batch processes with intense cleaning sessions. The environment that elements face is very diverse daily, if not hourly.

Consequently, over time the elements may begin to give bad readings or no readings at all. What are the common causes of error and what are the resulting problems with the probe? This can vary with your application, but the most common are: moisture or contaminating materials internally, exposure to high temperature over a long period of time, temperature cycling, and exposure to mechanical environmental conditions such as vibration.

Depending on the application, an unknown time response or a time response that takes too long can create error because the temperature sensor is not able to react to changes.

Depending on the amount of moisture contamination within the sheath and the normal operating temperatures, you can create a resistance drop or a simple shift in the resistance. You can demonstrate the ability to withstand moisture contamination by exposing the probe to 20° C and 100% relative humidity for a period of 30 days. Testing for moisture can occur with an Insulation Resistance (IR) test. This test measures the resistance between the element leads and the sheath. Typically, a minimum insulation resistance value of 100-500 megohms is acceptable.

High temperature exposure creates it own issue not only for the element, but also for every component of the temperature probe. Melting and distortion can occur if the temperature is beyond the design parameters of the probe. The result of this exposure to the element is typically a shift in resistance or an open circuit. You can test a probe for its resistance to this type of exposure with a Long Term Stability test. This entails exposing the probe to its maximum rated temperature for a minimum of 1,000 hours.

One area that can lead to error is the resistance of the leadwires if not eliminated from the circuit. One way to cut out the problem is using three and four wire configurations.

Temperature cycling, or wide ranges and rapid swings in temperature, can induce a shift in resistance or an open circuit. A temperature cycling test can show the repeatability of a sensor in these conditions. A typical test includes a minimum of 10 cycles from the sensor minimum to maximum rated temperature.

The vibration that a probe undergoes can cause a shift in resistance or an open circuit. Vibration is one of the more difficult parameters to quantify due to the difficulty in measuring.

Unlike temperature readings you can easily track over time, the frequency and amplitude of vibration is much more random, potentially more severe, and with no record. Typically, the operator at a facility has no warning it has happened. Probes should be able to withstand the specifications of ASTM procedure E644. The unit of measure for vibration is in "g's," a unit of acceleration relative to the force of gravity. It's not uncommon for a probe to be able to withstand 10, 15, or even 20 g's of vibration.

By identifying that error and uncertainty are present, you can list some of the factors that contribute to the error. With diligence, you can anticipate certain error is present and hopefully reduce factors that can contribute to error.

Behind the byline

Pete Shelquist is a business development engineer, and John Zwak is a design engineering supervisor for Minnetonka, Minn.-based Burns Engineering, Inc.