MEASURE FOR MEASURE
First Step Toward Disaster
Opting not to calibrate leads to bad decisions
by Christopher L. Grachanen
As a calibration professional, I’ve been involved in many assessments to determine compliance with calibration requirements related to the ISO 9000 series of standards. These requirements center around two key mandates in section 7.6 of ISO 9001:2008:
- Calibrate your monitoring and measuring equipment whenever necessary to ensure results are valid.
- Evaluate the validity of previous measurements whenever you discover your measuring or monitoring equipment is out of calibration.1
A precondition to the mandate for evaluating out-of-calibration measuring and monitoring equipment (M&ME) is that it has been calibrated. This leads to the rather loosely worded mandate that M&ME be calibrated to help ensure measurement results—and the corresponding decisions based on those measurements—are valid.
This requirement to calibrate as needed to ensure valid results, when interpreted by different users tasked with determining equipment calibration status, often results in nonstandard calibration statuses for similar equipment employed in similar applications. This is especially true in organizations required to comply with corporate directives to reduce discretionary spending, which typically includes calibration support costs.
The three most common user justifications to forego M&ME calibration are that the equipment:
- Is monitored by other equipment that is calibrated.
- Is used only to indicate go/no-go status.
- Is not used to make a final determination as to the satisfactory state of a product.
The first two justifications necessitate an understanding of the equipment’s calibration requirements and not just the applications in which they are employed. The following two examples will help shed light on this:
- A DC power supply is used to power a printed circuit board (PCB)
during testing. The power supply output is monitored with a DCV digital multimeter.
Calibration of the power supply requires period and random deviation (PARD) to be within published specifications. Excessive PARD can interfere with the PCB’s proper operation. A DCV digital multimeter cannot detect whether the power supply’s PARD is within accepted levels and therefore cannot alert operators about a potential PARD-induced problem.
- A spectrum analyzer is used to determine whether a product’s internal
harmonic exceeds a predetermined value—a go/no-go application.
Calibration of a spectrum analyzer requires measurement of known signals throughout its frequency range to ensure applied signal amplitudes are displayed correctly. A spectrum analyzer’s displayed response can be subject to such issues as excessive non-linearity, signal compression and signal dropouts, all of which can cause an applied signal to be displayed erroneously or, in a worst-case scenario, completely mask a signal’s presence.
The final user justification is truly subjective in nature due to the perceived non-criticality of interim measurements. The late ASQ fellow Phil Stein implicitly addressed this topic when he eloquently stated, “Does it matter if the measurement result is wrong? If the answer is yes, then calibrate. If the answer is no, then reevaluate why you are bothering to waste time making the measurement.”2
Don’t go for gold
In keeping with the theme of subjective calibration requirements, I remember attending an International Organization for Standardization briefing in which a lead assessor advocated foregoing M&ME calibration in lieu of a "golden standard" maintenance program.
He cited a machine shop maintenance program he was familiar with in which dimensional hand tools, such as calipers and micrometers, were used to periodically measure a calibrated gauge block. This was done to establish confidence in its performance and thus avoided the need and cost of calibrating it.
The lead assessor said this program could be adopted for electrical M&ME such that only designated golden standards would be calibrated while all other equipment would be periodically compared to these golden standards.
One of the proposed candidates for this program was electrical power analyzers. Periodically comparing and evaluating power analyzer responses to the golden standard analyzer response over the range of conditions they would be expected to encounter when testing products—such as different power levels, power factors and crest factors—would have required extensive user resources given the number of analyzers in the user inventory.
The reality of limited user resources would have undoubtedly limited golden standard evaluations to a few test conditions. As a result, analyzer responses for non-evaluated test conditions would have needed to be assumed.
Consequently, this situation would have increased measurement risk for the tests in which the analyzers were employed. Given this and other compelling reasons, the lead assessor’s "golden" program was not pursued.
Count on calibration
Obviously, the subject of measurement risk in the context of product risk assessment is beyond the scope of this article. Despite the complexity of this subject, it can be asserted that measurement risk is reduced for measurements derived from calibrated M&ME compared with non-calibrated M&ME. This is due to the increased probability calibrated M&ME performance is within published specifications—in other words, operating within established uncertainties.
It should come as no surprise that user interpretation of ISO 9001:2008 calibration requirements can result in inconsistent M&ME calibration statuses for similar equipment employed in similar applications. Common sense dictates calibration professionals should always be consulted in matters related to M&ME calibration requirements.
The bottom line is: Users who take it upon themselves to forego calibration of their active M&ME, without adequately evaluating the impact of their decision within a specific measurement application, increase the likelihood of making a bad measurement-based decision due to increased—or, in worst case, unknown—measurement uncertainty.
- Praxiom, "ISO 9001:2008 Translated Into Plain English," http://praxiom.com/iso-9001.htm.
- Graeme C. Payne, "Managing the Measurement System," Quality Progress, March 2007, pp. 77–78.
Christopher L. Grachanen is a master engineer and operations manager at Hewlett-Packard Co. in Houston. He earned an MBA from Regis University in Denver. Grachanen is a co-author of The Metrology Handbook (ASQ Quality Press), a senior member of ASQ, an ASQ-certified calibration technician and the treasurer of the Measurement Quality Division.