If an item (or system) is to be tested with an instrument, and a decision is to be taken on the results/ readings from the it, there should be a good level of confidence in the accuracy of the instrument used. This is normally achieved by calibrating the working instrument against another device of known accuracy, traceable to National Standards, for example:
 |
National Standard Accurate to 0.001 % |
|
Calibration Laboratory Accurate to 0.01 % |
|
Company "Master" Item Accurate to 0.1 % |
|
Company Working Equipment Accurate to 1.0 % |
|
Any item against which an acceptance/ rejection will be made; or from which a linear measurement will be taken (e.g. tape measure) |
|
Other items of equipment may be used to obtain an approximate value. These need not be accurately calibrated, and should be marked to avoid any confusion, e.g. "For Indication Only". |
This will vary depending upon the criticality of the items to be tested, the frequency of use of the test instrument, industry standards of acceptability, etc. The company must decide on the maximum tolerance of accuracy for each item of test equipment.
Normal Methods of Re-Calibration
|
Send all working equipment to an external calibration laboratory |
|
Send one of each item (a "master") to a calibration laboratory, then sub-calibrate each working item against the master |
Alternative Methods of Re-Calibration
|
By Attribute - Using a known "faulty" product, and a known "good" product; test each one to ensure that the test equipment can identify the faulty and good component correctly. |
This again depends upon the amount of use the instrument is subjected to, its ability to retain its accuracy (from historical records), the criticality of the items to be tested, etc.
Infrequently used instruments could be calibrated prior to their use. Frequently used items would normally be checked/ re-calibrated at a regular interval decided by the company (based upon product criticality, cost, availability). It would be wise to view 12 months as the maximum calibration interval when commencing a calibration system.
|
Each instrument should be uniquely identified, allowing it to be traced. |
|
The calibration results should show the "master" item reading, and the "working" item reading (both prior to calibration and following calibration), to allow a calculation of the extent of inaccuracy. |
|
The calibration results should be retained for reference. |
|
The instrument should be labelled to show the next "calibration due" date to easily avoid its use outside of the period of confidence. |
|
Any means of adjusting the calibration should be sealed, allowing easy identification if it has been tampered with (e.g. a label across the joint of the casing). |
If the instrument is found to be outside of its tolerance of accuracy, any items previously tested with the instrument must be viewed as suspect. In these circumstances, it would be wise to review the test results obtained from the individual instrument. This could be achieved by compensating for the extent of inaccuracy to decide if the acceptability of the item would be reversed.
If the item is subject to re-calibration at regular intervals, it would be wise to shorten the interval in these circumstances as the confidence of accuracy will have been reduced.
Editor's Notes
Peter Griffin is managing consultant of P Griffin & Associates (PGA); one of the leading quality management consultancies operating in Europe and the USA.
P Griffin & Associates have assisted 260 companies (ranging from 3 employees to multi-national corporations) to achieve improvements in quality and ISO9000 Registration.
While the majority of companies fail to achieve ISO9000 certification at the first attempt, 98% of PGA's quality assurance clients have achieved first-time success.
Please pass your comments/ suggestions/ requests to
help@quality.co.uk
Quality Network Home 
P Griffin & Associates Index 
