Accurate, reliable temperature measurement is particularly necessary when validating sterilisers and other thermal processes in healthcare and pharmaceutical applications
Accurate, reliable temperature measurement is particularly necessary when validating sterilisers and other thermal processes in healthcare and pharmaceutical applications. The relevant guideline of HTM2010 states that the repeatability of measurement should be +/–0.25°C or better and that the combined errors of the complete measurement system, including sensors, should not exceed +/–0.5°C.
Calibration of the complete measurement system of thermocouples, data logger and computer against a UKAS (or similar) reference thermometer is necessary to reduce systematic and random errors, but does not remove the data logger measure-ment uncertainty. The data logger measurement uncertainty is calculated using a statistical approach, with a combination of quadratic summation and reporting the final value at a 95% confidence level with a Gaussian distribution. The measurement uncertainty is a function of the equipment specification and characteristics and cannot be changed by any subsequent calibration.
Calibration may improve the repeatability of measurement by reducing systematic errors, such as the thermocouple error, but measurements subsequently made with the complete measurement system have to be qualified by a statement of measurement uncertainty.
Thermocouple temperature sensors: The thermocouple uses the Seebeck effect – when two dissimilar metal wires are connected together; if the two junctions are at different temperatures a voltage potential difference occurs and a current flows which is related to the temperature difference between the two junctions (see Figure 1).
For practical purposes, a thermocouple can measure only the temperature difference between the two junctions: the ‘hot’ junction, which is exposed to the temperature to be measured, and the ‘cold’ junction, which is where connection is made from the thermocouple sensor into the measurement system. The measured output voltage will then correspond to the temperature difference between the hot and cold junctions. The cold junction temperature has to be measured by either a thermistor or PT100 sensor; and this temperature has to be compensated in the measuring system to give the true hot junction temperature.
Internationally agreed tables (IEC 584-1:1995) show the millivolt output for each thermocouple type referenced to a cold junction temperature of 0°C and form the basis of this measurement technology. The thermocouple mV/°C relation-ship is non-linear and must be linearised in the data logger to convert the mV output into °C.
System errors
In any measurement system there are two types of error: systematic and random. Systematic errors can be reduced by careful calibration but other errors, mainly due to the limitations of the measure-ment system, are random in nature and cannot be removed by calibration (see table 1).
Calibration of the complete measurement system of thermocouples, data logger and laptop against a UKAS reference thermometer each time the equipment is used reduces systematic errors at the moment of calibration. Any measurements subsequently made with the complete measurement system have to be qualified by a statement of ‘measurement uncertainty’. Measurement uncertainty is the statistical error in the data logger (and calibration equipment) and cannot be removed by calibration.
Data logger error: In a data logger, after calibration, the main sources of error are: the difference between the thermocouple cold junction sensor temperature and the actual temperature of the input terminals, thermal uniformity across the input terminals, input switching thermal drift error, dc amplifier thermal error and dc amplifier time drift error.
The errors are usually combined in the manufacturer’s specification as ‘range accuracy’; ‘cold junction error’ and ‘stability error due to ambient temperature’. The range accuracy is of specific importance, as it cannot be improved by calibration.
These values can be obtained from the manufacturer’s published data sheets. These errors are random and vary with the ambient temperature of the data logger and the location of the data logger (particularly if it is near to a heat source such as an autoclave), making it essential that the effect of these errors is known to qualify the readings
Thermocouple error: Type T thermo-couples are available in two grades of thermocouple wire: Class 1 and Class 2. The difference between them is shown by the accuracy at 0°C; Class 1 accuracy is +/–0.5°C while Class 2 accuracy is +/–1.5°C. The systematic error in the Class 1 thermo-couples can be removed by calibration using the dry-block bath and an accurate UKAS calibrated reference thermometer.
Calibration equipment error: The complete system is calibrated by inserting the thermocouples in the dry-block bath and when stable measuring the temperature on an accurate UKAS calibrated reference thermometer. The calibration measurement error is defined by the UKAS temperature reference error and the temperature dry-block bath error.
Total system error: Calibration of the complete measurement system of thermo-couples, data logger and software against a UKAS reference thermometer each time the equipment is used reduces all errors, systematic and random, at the moment of calibration. Any measurements subsequently made with the complete measurement system have to be qualified by a statement of measurement uncertainty.
HTM 2010 states that the repeatability of measurement should be +/–0.25°C or better and that the combined uncertainties of the complete measurement system, including sensors, should not exceed +/–0.5°C.
System calibration
To comply with the requirements of HTM 2010, the complete measurement system must be calibrated. There are two methods that are currently used. One method uses a millivolt source to simulate temperature, while the other uses a heat block as an actual source of temperature.
Temperature calibration with a millivolt source: Many users still calibrate their equipment using a millivolt source that can generate simulated thermocouple outputs. In this mode, type T thermocouple cable is connected from the copper terminals of the millivolt source directly into the data logger. Selecting an equivalent output of, say, 134°C should enable the data logger to be checked at that temperature.
After noting any offsets, the thermocouple is disconnected from the millivolt source, the ends twisted together and then later used to validate an autoclave. Apart from the fact that this does not truly calibrate the complete measurement system, this approach also has several sources of error.
To enable a millivolt source to output simulated temperature for type T thermocouples in °C, the millivolt source must have an internal temperature measure-ment for cold junction compensation. When used to inject the simulated temperature into the data logger an error will occur if the millivolt cold junction and the data logger cold junction are not at exactly the same temperature. This is obviously not possible to ascertain and will result in unquantifiable errors.
Furthermore, the copper output terminals on the millivolt source will be wired to type T thermocouple cable resulting in yet a further cold junction error (Figure 3).
Even the best process calibrators that have simulated thermocouple outputs have limited accuracy. It is possible to accurately calibrate a temperature data logger using
a millivolt source. Again this does not calibrate the complete measurement system but does reduce the errors noted above.
Temperature calibration with a millivolt source and ice point (0°C) reference: In this mode the millivolt source is used with no internal cold junction compensation and with copper leads connected from the copper terminals of the millivolt source. The leads are individually welded to type T thermocouple wire; copper to copper and copper to constantan. These junctions are inserted into an ice point reference with the type T thermocouple wired to the data logger.
Selecting the correct millivolt value of, say, 5.91mv for 134°C will enable the data logger to be checked exactly at that temperature. After noting any offsets the thermocouple assembly is removed from the bath and the ends of the type T thermocouple twisted together and then later used to validate an autoclave (Figure 4).
As before, this does not truly calibrate the complete measurement system, but this approach is the correct way to check the accuracy and to calibrate the data logger.
This method also allows the easy use of the IEC 584-1:1995 emf/°C thermocouple tables as they are referenced to a cold junction temperature of 0°C.
Temperature calibration with a temperature bath: To comply with the requirements of HTM 2010, the complete measurement system comprising of thermocouples, data acquisition unit and computer with suitable software should be calibrated against a reference thermometer with UKAS calibration. This is achieved by inserting the thermocouples into a dry-block temperature bath and when stable, measuring the reference temperature and the thermocouple outputs on the data logger. The system software should provide multi-point calibration at ‘low’, ‘high’ and ‘check-point’ temperatures. Suggested calibration temperatures are: 40°C for low, 150°C for high, and 121°C/134°C for the check point.
This method correctly calibrates the complete system over the whole operating temperature range (Figure 5). After calibration the software automatically calculates the scale and offset values for each sensor.
Modern dry-block baths provide good stability and fast heating and cooling, enabling rapid setting of the three temperature set points. During calibration the dry-block temperatures are measured by an accurate UKAS thermometer as the calibrated temperature reference. This method ensures that the complete measurement system of thermocouples, data logger and software is calibrated against a UKAS reference thermometer each time the equipment is used. With this method it is only necessary to annually re-calibrate the UKAS reference thermometer.
No comments:
Post a Comment