|
Weighing is a common form of measurement in commerce, industries and households. Weighing instruments are often highly accurate, but users, i.e. their customers and/or regulatory bodies, often need to know just how inaccurate a particular scale may be. Originally, this information was obtained by classifying and verifying the equipment for type approval. Subsequently, the equipment was tested or calibrated on a regular basis.
Typical calibration procedures
Calibrating scales involves several different procedures depending on national- and/or industry-specific guidelines or regulations, or on the potential consequences of erroneous weighing results. One clear and thorough guide is the EA-10/18, Guidelines on the Calibration of Non-automatic Weighing Instruments, which was prepared by the European Co-operation for Accreditation, and published by the European Collaboration in Measurement and Standards (euromet).
Typical scale calibration involves weighing various standard weights
in three separate tests:
• repeatability test
• eccentricity test
• weighing test (test for errors of indication)
In the pharmaceutical industry in the United States, tests for determining minimum weighing capability are also performed.
Usually, the object being weighed is placed on the load receptor and the weighing result is read only once. If you weigh the object repeatedly, you will notice slight, random variation in the indications. The Repeatability Test involves weighing an object several times to determine the repeatability of the scale used.
The Eccentricity Test involves placing the object being weighed in the middle of the load receptor as accurately as possible. This is sometimes difficult due to the shape or construction of the object being weighed. Typical calibration procedures include the Eccentricity Test. You can determine how much the eccentricity of the load will affect the indication on the scale by weighing the same weight at the corners of the load receptor.
The Weighing Test examines the error of the indication on the scale for several predefined loads. This enables you to correct the errors and definitions for non-linearity and hysteresis.
If the scale’s maximum load limit is extremely large, it may be impractical to use standard weights for calibrating the entire range. In such a case, suitable substitution mass is used instead. Substitution mass should also be used if the construction of the scale does not allow the use of standard weights.
The purpose of the Minimum Weight Test is to determine the minimum weight, which can be assuredly and accurately measured using the scale in question. This condition is met if the measurement error is less than 0.1% of the weight, with a probability of 99.73%.
Knowing the error of the scale indication at the point of each calibration is not sufficient. You must also know how certain you can be about the error found at each point of calibration. There are several sources of uncertainty of the error, e.g.:
•The masses of the weights are only known with a certain uncertainty.
•Air convection causes extra force on the load receptor.
•Air buoyancy around the weights varies according to barometric pressure, air temperature and humidity.
•A substitute load is used in calibrating the scale.
•Digital scale indications are rounded to the resolution in use.
•Analogous scales have limited readability.
•There are random variations in the indications as can be seen in the Repeatability Test.
•The weights are not in the exact middle of the load receptor.
The values of uncertainty determined at each point of calibration are expressed as standard uncertainties (coverage probability: 68.27 %), which correspond to one standard deviation of a normally distributed variable. The combined standard uncertainty of the error at a certain point of calibration has a coverage probability of 68.27 % as well.
 |
|
Example: The calibration error and its uncertainty at the calibration point of 10 kg may be expressed e.g. E = 2.5 g and u(E ) = ±0.7 g, which means that the calculated error in the indication is 2.5 g and the actual error, with a coverage probability of 68.27 %, is between is between 1.8 g and 3.2 g. |
In practice, a coverage probability of 68.27 % is insufficient. Normally, it is extended to a level of 95.45 % by multiplying it with the coverage factor k = 2. If the distribution of the indicated error cannot be considered normal, or the reliability of the standard uncertainty value is insufficient, then a larger value should be used for the k-factor.
If you are able to use the k = 2 coverage factor, then the error and its extended uncertainty at the point of calibration are E = 2.5 g and U(E) = ±1.4 g. This means that the calculated error of the indication is 2.5 g and the actual error, with a coverage probability of 95.45 %, is between 1.1 g and 3.9 g.
The purpose of calibration is to determine how accurate a weighing instrument is. As the above-mentioned case indicates, you know that if you repeat the calibration several times, the indication of weighing an object of 10 kilograms will be between 10.0011 kg and 10.0039 kg 95.45 % of the time.However, the uncertainty of the results of later routine weighings is usually larger. Typical reasons for this are:
• Routine weighing measurements involve random loads, while calibration is made at certain calibration points.
• Routine weighing measurements are not repeated whereas indications received through calibrations may be averages of repeated weighing measurements.
• Finer resolution is often used in calibration.
• Loading/unloading cycles in calibration and routine weighing may be different.
• A load may be situated eccentrically in routine weighing.
• Tare balancing device may be used in routine weighing.
• The temperature, barometric pressure and relative humidity of the air may vary.
• The adjustment of the weighing instrument may have changed.
Standard and expanded uncertainties of weighing results are calculated using technical data of the weighing instrument, its calibration results, knowledge of its typical behaviour and knowledge of the conditions of the location where the instrument is used. Defining the uncertainty of weighing results is highly recommended, at least once, for all typical applications and always for critical applications. Calculating the uncertainty of weighing results assists you in deciding whether or not the accuracy of the weighing instrument is sufficient and how often it should be calibrated. However, determining the uncertainty of weighing results is not part of calibration.
CMX’s scale calibration enables you to uniquely configure calibration and test each weighing instrument. Correspondingly, copying configurations from one scale to another is easy. Error limits can be set according to OIML or Handbook-44. Wide variation in user-specific limits is also possible.
CMX calculates combined standard uncertainty and expanded uncertainty at calibration of the weighing instrument. It allows you to enter additional, user-defined uncertainty components in addition to supported uncertainty components.
CMX’s versatile calibration certificate and possibility to define a user specific certificate assure that you can fulfill requirements set for your calibration certificates.
Copyright © - Beamex 2006 |
