Calibration seeks a higher power

Every calibration should meet a specified tolerance. The terms “tolerance” and “accuracy” often lead to some confusion, and it is important to define them such that we are all on the same page.

Accuracy is the ratio of the error to the full-scale output or the ratio of the error to the output, expressed in percent span or percent reading, respectively.

Tolerance is the permissible deviation from a specified value, and it may be in measurement units, percent of span, or percent of reading.

There are subtle differences in the terms. 

We recommend the tolerance, specified in measurement units, also serve for the calibration requirements performed at your facility. By specifying an actual value, mistakes caused by calculating percentages of span or reading do not occur. 

In addition, tolerances should be in the units measured for the calibration.

For example, you must perform the calibration of a 0-to-300 psig pressure transmitter with a specified calibration tolerance of ±2 psig. The output tolerance would be: 

(2 psig) / (300 psig) × (16) mA = 0.1067 mA

The calculated tolerance rounds down to 0.10 mA because rounding up to 0.11 mA would exceed the calculated tolerance. We recommend both ±2 psig and ±0.10 mA tolerances appear on the calibration data sheet if recording the remote indications and output milliamp signal. 

Note the manufacturer’s specified accuracy for this instrument may be 0.25% full scale. Calibration tolerances rest on the manufacturer’s specification only. Calibration tolerances should be determined from a combination of factors.

These factors include:

• Requirements of the process
• Capability of available test equipment
• Consistency with like instruments at your facility
• Manufacturer’s specified tolerance

For instance, say a process requires ±5°C; available test equipment is capable of ±0.25°C; and manufacturer’s stated accuracy is ±0.25°C. The specified calibration tolerance must be between the process requirement and manufacturer’s specified tolerance. 

Additionally the test equipment must be capable of the tolerance needed. Assigning a calibration tolerance of ±1°C for consistency, with similar instruments and to meet the recommended accuracy ratio of 4:1, is acceptable and typical.

A good rule of thumb is to ensure an accuracy ratio of 4:1 when performing calibrations. This means the instrument or standard used should be four times more accurate than the instrument undergoing calibration. 

Therefore, the test equipment (such as a field standard) used to calibrate the process instrument should be four times more accurate than the process instrument, the laboratory standard used to calibrate the field standard should be four times more accurate than the field standard, and so on.

With today’s technology, an accuracy ratio of 4:1 is becoming more difficult to achieve. Why is a 4:1 ratio the rule? Ensuring a 4:1 ratio will minimize the effect of the accuracy of the standard on the overall calibration accuracy.  

Using our previous example of the test equipment with a tolerance of ±0.25°C and it is 0.5°C out of tolerance during a scheduled calibration. Since we took into consideration an accuracy ratio of 4:1 and assigned a calibration tolerance of ±1°C to the process instrument, it is less likely that our calibration performed using that standard errs by an unacceptable amount.

The out-of-tolerance standard still needs to undergo reverse traceability of all calibrations performed using the test standard. However, our assurance is high that the process instrument is within tolerance. This brings us to traceability.

All calibrations should be traceable to a nationally or internationally recognized standard. In the U.S., the National Institute of Standards and Technology (NIST) is the nationally recognized standard. 

Traceability is “the property of a result of a measurement whereby it can be related to appropriate standards, generally national or international standards, through an unbroken chain of comparisons.” 

Traceability is accomplished by ensuring the test standards we use are routinely calibrated by “higher level” reference standards. Typically, the standards we use from the shop periodically go to a standards lab, which has more accurate test equipment. 

The standards from the calibration lab are periodically checked for calibration by “higher level” standards and so on, until eventually the standards are tested against Primary Standards maintained by NIST or another internationally recognized standard.

Finally, uncertainty is the parameter associated with the result of a measurement that characterizes the dispersion of the values for which the measureand could reasonably be responsible. 

The expression of uncertainty of a measurement is a range of values, which are likely to enclose the true value. This range can be error bars on a graph or as a ± value. 

Source: Mike Cable, Calibration: A Technician’s Guide, ISA Press 2005.