Metrology and Calibration (Page 2 of 2)

<Back to Page 1>

Periodicity

Periodicity refers to the specified frequency of calibration or the regular interval between calibrations defined for an equipment or a part thereof.  The periodicity depends on how quickly the equipment drifts out of spec, what the equipment is for (is the application critical or not?), and what the company can afford to do, but normally it ranges from 3 months to 2 years. Periodicity may also be expressed as the calibration cycle, which is the number of calibrations required per year. 

Characterization

Aside from calibration which involves comparison of results against a standard and possibly adjustment to nominal settings, a measuring equipment may also be characterized. Characterization pertains to the evaluation or study of parameters or properties which do not have specifications. Characterization may be useful in determining the over-all limitations and capabilities of the equipment. 

Uncertainty

In quantum physics, the 'Uncertainty Principle' states that the exact position and exact velocity of a particle can not be determined at the same time. Because of this, one can not measure something without changing it. This uncertainty principle very well applies to metrology and calibration, i.e., every metrologist knows that there is no perfect measurement of anything. An excellent metrologist would therefore consider all factors affecting measurement uncertainty, and ensure that the total uncertainty or inaccuracy of the measurement is less than the tolerance of the specification for the parameter being measured. 

Test Accuracy Ratio

Test Accuracy Ratio, or TAR, is the ratio of the specification spec of the parameter being measured to the uncertainty unc of the measuring instrument, or TAR = spec/unc. Thus, the higher the TAR the better, but a higher TAR will also cost more. The metrologist should therefore aim for the most cost-effective TAR, i.e., the lowest TAR that will meet the company's quality objectives.

Guardbanding

Low TAR's don't mean that the company has to erroneously accept actual rejects as a result of measurement errors. Guardbanding is the process of tightening the pass limits of the specification to account for the uncertainty of the measuring equipment and other factors.  Guardbanding may reject some units that would otherwise be marginally acceptable, but this is still better than shipping 'bad' products to the customer.

Basic Terminology

Accuracy - how close a measurement reading is to the 'true' value of the parameter being measured

Precision - how repeatable or closely-grouped the measurement readings are

Resolution - the level of discrimination that the measuring equipment can show; the smallest unit change that it can discern or detect

Sensitivity - the smallest change in the input (stimulus) that causes a discernible change in the output

Stability - the tendency of a measuring equipment not to 'drift' or degrade over time and usage

<Back to Page 1>

See also:  Gage Repeatability and Reproducibility

See Also:  Quality Systems;  Monitors and Controls;  Document Control;  The ISO9000 Standard;

Gage Repeatability and Reproducibility