you can also analyze the data on previous equipment checks

Miriam,

Your article is outstanding, but I disagree with some subtleties. I don't know what the current wording in ISO 9001 is, but as I recall the 1994 version said that it was necessary to analyze the degree and kind of calibration nonconformance, impact on past product acceptance, and whether product disposition was required. You are implying that a recall or reinpsection is mandatory:

PP "• If the product has already been shipped, the standard requires that you have a process in place for how you are going to recall it. If it was already delivered to the customer, the question is: How you are going to ask for the product back to inspect or test it again?"

PP I think the analysis has to be done carefully and ethically and the product disposition plan executed well - be it a full recall or no action at all. A company that goes beyond the binary mentality of inspection and truly controls its processes will probably not face this calibration scenario, but if they do they will likely have (a) processes that conform to standards regardless of inspection issues and (b) the knowledge required to (hopefully) arrive at the conclusion that the product is fine despite the calibration nonconformance. For instance, if one knew that a process was incapable of producing nonconformances (perhaps limited to the range of the equipment nonconformance), no action other than reporting would be necessary. Examples would be use of robust design practices like design for assembly, classification of characteristics, failsafing, consideration of the out-of-tolerance range on the gaging vs. the range at which the product inspection was done (forgive me if you already mentioned that).

PP Also, 10 years ago I found the QS-9000 MSA manual to be an excellent guide to running a strong measurement system, as it describes ways to assess gage capabilities that are overlooked by ISO 9001. I imagine the current version is even stronger.

I was wondering if you could share your take on broken or damaged equipment. For instance, you are 4 months into a 12 month cal cycle of an electrical meter and the inspector drops it and breaks it, maybe your 10 months in.  Regardless, for the sake of this discussion lets pretend that the inspector was dilligent and honest and submitted a written account of the occurance when it happened.  Maybe e-mailed quality or used an internal form, whatever the process is.  The bottom line is you have a clear record of when and how it occured.

In my view you should use the nonconformane process to document the removal of the asset from your system and the replacement if necessary but then you are left with a situation where you cannot validate previouse measurements through As-Found data at the closure of the calibration cycle.  

I work for a calibration company so every measurment is clearly documented and I have the ability to look for trends in data suggesting it may have been producing false readings.  I can look at the history of the instrument and see if there are accounts of as found failing conditions or reason to believe that it would have been out of tolerance.  To me, that is the deviation from program, failing to compete as calibration cycle.  I've worked for a few different companies, and they all seem to have an approach to this matter however no one has ever really stated their method was based on any kind of official guidance beyond an internal procedure.

It doesn't seem realistic to take a guilty until proven innocent approach so to speak but on the other hand there should be some level of confidence the instrument was believed to be accurate during its usage.  In my case the last time the instrument was used it might have been thousands of miles away or in a different country even so I find myself searching for a cost effective approach to resolution without physical validation, or a process defending that physical validation was not deemed necessary.  

I look forward to your reply,

Jason

This article is a few years old, but people are still linking to it as it was e-mailed to me today.

Some good points are made, but the terminology can get slightly confusing.  Out of tolerance and out of calibration are two separate issues that don't necessarily overlap.  Out of calibration means that the M&TE was not recalled when it was due for calibration.  While there is some risk this equipment will be recovered and be out of tolerance, this is a small risk that is mostly independent of the equipment's due date.  Out of tolerance can occur any time the equipment is used but is more likely the result of damage, misuse, wear, or other mishap, assuming the equipment was stable to begin with.  To that end, the data absolutely helps in making that determination.  Data ultimately determines whether calibration frequencies are adequate to reach a desired reliability goal.  A popular target is 95% reliability.

Ultimately I'm not sure if out of tolerance conditions can be entirely prevented by data, but the risk can certainly be minimized.  A calibration system should be robust enough to handle this kind of "what if" even though it is unlikely to occur.