Almost every calibration certificate for a scale contains a reference to EURAMET cg-18. Anyone who has to assess such a certificate, for example a quality manager preparing for an audit, sees tables with repeatability, eccentricity, indication errors and uncertainties. What do those figures mean, and why is this particular method so widely used?
In this article we explain what the guide contains, how the three tests are carried out, how test points are chosen, how the measurement uncertainty is built up and how you read a report. If you want to know how we carry this out at your premises, see scale calibration according to EURAMET cg-18.
What is EURAMET cg-18?
EURAMET is the European association of national metrology institutes. Under the name Calibration Guide No. 18 it publishes a guide for the calibration of non-automatic weighing instruments, from analytical balances to weighbridges. The guide describes which tests you carry out, how you process the results and how you calculate the measurement uncertainty.
cg-18 is not a law, nor a standard that a scale must comply with. It contains no tolerances that determine whether a scale is right or wrong. It is a method for establishing how a scale weighs, in a way that is comparable between laboratories. Other documents apply to legal verification, such as OIML R76 and the European standard EN 45501; those do contain tolerances per accuracy class. The guide is revised from time to time. The current version can be found on the EURAMET website.
Incidentally, a calibration does not start with weights but with the conditions. The scale stands in its usual place, is level and has had sufficient warm-up time. The temperature is recorded and the test weights are given time to reach ambient temperature. It is also noted whether the scale was adjusted just before the calibration, for example with an internal adjustment function.
The three tests in detail
Repeatability
In the repeatability test, the same load is placed on the platform and removed again a number of times. In between, the zero indication is checked. From the series of indications follows the standard deviation s. It shows how well the scale displays the same weight the same way every time.
The test load is often in the upper part of the weighing range. For laboratory balances an additional measurement with a smaller weight, close to the quantities the user weighs out, is useful. For balances and bench scales, five to ten repetitions are usually carried out. For heavy instruments, where each placement is a lot of work, there are often fewer.
For fine balances, repeatability is often the most important source of uncertainty. Draughts, vibrations and static charge become directly visible in this test.
Eccentricity
In the eccentricity test, also called the corner load test, a weight is placed successively in the centre and at positions away from the centre. For a rectangular platform these are usually the centre and four points towards the corners. The difference between the indication off-centre and in the centre is the eccentricity error.
The test load is a substantial part of the weighing range, often around one third of Max. On a weighbridge the positions above the supports are loaded. On a crane scale the test usually has no meaning, because the load always hangs from the same point. A large eccentricity error in practice often points to something mechanical: a platform that is catching, dirt under a corner or a load cell that responds differently from the others.
Indication error
The core of the calibration is the determination of the indication error. At a number of points spread over the weighing range, the indication is compared with the conventional mass of the test weights. The difference is the indication error at each point. Loading is often done both increasing and decreasing, so that hysteresis also becomes visible.
For large weighbridges or crane scales there is not always enough weight available to reach Max. cg-18 then describes working with a substitution load: part of the load consists of other material, whose contribution is determined with the scale itself. That increases the uncertainty, and that should also be reflected in the report.
Choice of test points
cg-18 does not prescribe a fixed list of test points. A series of about five to ten points, fairly evenly distributed between zero and Max, is customary. In practice we supplement these with points that matter to the user, such as:
- the quantity that is weighed daily, for example 25 kg bags or a 50 mg sample;
- the range in which products are released or invoiced;
- for scales with multiple weighing ranges, points in each range;
- the point at which an internal tolerance or control limit applies.
Anyone who calibrates only at Max knows little about behaviour at small loads. And anyone who measures only in the middle of the range misses errors at the edges. A good choice begins with the question of what the scale is used for. That is why we ask about this beforehand.
Measurement uncertainty and uncertainty in use
An indication error without measurement uncertainty is a number without context. cg-18 describes how the uncertainty of the calibration is built up from separate contributions:
- the readability d of the scale, both at zero and under load;
- the repeatability from the first test;
- the eccentricity, because test weights are never placed exactly in the centre;
- the uncertainty of the test weights and their possible drift since their own calibration;
- influences of air buoyancy and temperature differences between weight and surroundings.
These contributions are combined into a standard uncertainty and multiplied by a coverage factor, usually k = 2. That corresponds to a confidence level of approximately 95 %.
In addition, cg-18 defines the uncertainty in use. This looks ahead: how large is the uncertainty of any weighing after the calibration? It also includes factors that played no role during the calibration, such as an indication error that you do not correct for, temperature fluctuations in the room and drift over time. The result is often presented as a formula in which the uncertainty depends on the load. The minimum weight also follows from this; read more in minimum weight of an analytical balance.
How to read a cg-18 report
- Identification and conditions. Check make, model, serial number, location and date. Look at the temperature and whether the scale was adjusted just before the calibration.
- Weights used. The report names the reference weights and their traceability to national standards.
- Repeatability and eccentricity. Compare the spread with the readability and with your own requirement. A standard deviation much larger than d points to environmental influences or a defect.
- Indication error per point. Assess not only the error but the error together with the expanded uncertainty. Only then do you know whether the scale falls within your tolerance.
- Before and after adjustment. If the report contains two series, the first says something about the period since the previous calibration. If it was outside your tolerance, assess the consequences for earlier weighings.
- Uncertainty in use. If it is included, you can use it to demonstrate that the scale is suitable for your smallest weighing.
A detailed explanation of the parts of a certificate can be found in the calibration certificate explained.
How MASSLAB helps you
MASSLAB calibrates scales of all makes according to EURAMET cg-18, on site or in our laboratory. We agree the test points with you beforehand to match your use, work with weights from our own laboratory that are traceable to national standards, and report the results before and after any adjustment. On request we include the uncertainty in use. The difference from adjustment and legal verification is explained in calibrating, adjusting and verifying scales. More about our working method under scale calibration.