We have an ISO17025 vendor that calibrates our pipettes. We are currently wanting to change our acceptance tolerance to model ISO 8655 tolerances.

The 17025 auditor is questioning the move to use the tolerances specified in ISO8655, because they are wider than the manufacture specifications for that device. Essentially, the auditor is questioning the ISO standard.

My point is this: it is an observation that mfg. specifications should be used, but certainly not a violation of 17025, or even bad metrology, to use a tolerance other than mfg. specifications. This is even more so, given that the tolerances come from an ISO standard.

Obviously this is an issue between the AB and my vendor, until the problem becomes my problem. Essentially having the AB auditor telling me what tolerances I will use becomes my problem.:tg:

Any thoughts or opinions on this would be greatly appreciated.

We have an ISO17025 vendor that calibrates our pipettes. We are currently wanting to change our acceptance tolerance to model ISO 8655 tolerances.

The 17025 auditor is questioning the move to use the tolerances specified in ISO8655, because they are wider than the manufacture specifications for that device. Essentially, the auditor is questioning the ISO standard.

My point is this: it is an observation that mfg. specifications should be used, but certainly not a violation of 17025, or even bad metrology, to use a tolerance other than mfg. specifications. This is even more so, given that the tolerances come from an ISO standard.

Obviously this is an issue between the AB and my vendor, until the problem becomes my problem. Essentially having the AB auditor telling me what tolerances I will use becomes my problem.:tg:

Any thoughts or opinions on this would be greatly appreciated.

It shouldn't make any difference if the ISO tolerances are more generous than the manufacturer's; what's important is that the tolerances (regardless of the source) are what's right for your purposes. You can say that a new tolerancing scheme came to you in a dream and the auditor shouldn't care so long as the scheme works for you.

That is an interesting one. My reply doesn't claim to be an answer, but more conversation from another who finds this befuddling.

An analogous situation I've had is calibrations to torque wrenches. There is an ISO procedure (ISO6789) with prescriptive requirements to cal torque wrenches of many types, and generic specs by type. It doesn't always match OEM specs.

I just re-read the SCOPE section of my applicable ISO, and it said it applied in particular to "...torque wrenches in accordance with numbers 258 and 259 of ISO 1703:1983."

I don't know the wording of your applicable ISO, but it raises the question (presuming the scope in yours is similar to mine), is calibrating in accordance with an ISO procedure (such as yours or mine listed above) as opposed to calibrating with an OEM procedure/specs (or other legitimate procedure/specs) something that should be done only when it is a requirement of another ISO? And conversely, unless there is a specific ISO requirement to calibrate something to an ISO procedure, should it not be used unless it is either the only legitimate available procedure, or it is not in conflict with OEM specs?

In our Torque wrench procedure circumstance, there were and (more currently) are some conflicts and disagreements. I can't think of the particular model, but our very astute senior torque person at my current lab (when presented with the ISO to cal torque wrenches) told me that he would have to cal to different specs if he used the ISO.

I'm not certain what the presumed real world scope of ISO cal procedures is, but I wonder if it is supposed to be for either ISO prescriptive environments, or where no other legitimate procedure exists. I am thinking that Torque wrenches are so commonly used globally, and perhaps a LOT of them globally with no legitimate specs or procedures, that its intent was to create some level of order, but perhaps not to replace legitimate OEM specs and procedures.

Just a few questioning thoughts. I'll be curious to see some more expert replies.

It shouldn't make any difference if the ISO tolerances are more generous than the manufacturer's; what's important is that the tolerances (regardless of the source) are what's right for your purposes. You can say that a new tolerancing scheme came to you in a dream and the auditor shouldn't care so long as the scheme works for you.

Thank you, Jim. This is along the lines of what I was thinking about.

An analogous situation I've had is calibrations to torque wrenches. There is an ISO procedure (ISO6789) with prescriptive requirements to cal torque wrenches of many types, and generic specs by type. It doesn't always match OEM specs.

I just re-read the SCOPE section of my applicable ISO, and it said it applied in particular to "...torque wrenches in accordance with numbers 258 and 259 of ISO 1703:1983."

I don't know the wording of your applicable ISO, but it raises the question (presuming the scope in yours is similar to mine), is calibrating in accordance with an ISO procedure (such as yours or mine listed above) as opposed to calibrating with an OEM procedure/specs (or other legitimate procedure/specs) something that should be done only when it is a requirement of another ISO? And conversely, unless there is a specific ISO requirement to calibrate something to an ISO procedure, should it not be used unless it is either the only legitimate available procedure, or it is not in conflict with OEM specs?

In our Torque wrench procedure circumstance, there were and (more currently) are some conflicts and disagreements. I can't think of the particular model, but our very astute senior torque person at my current lab (when presented with the ISO to cal torque wrenches) told me that he would have to cal to different specs if he used the ISO.

I'm not certain what the presumed real world scope of ISO cal procedures is, but I wonder if it is supposed to be for either ISO prescriptive environments, or where no other legitimate procedure exists. I am thinking that Torque wrenches are so commonly used globally, and perhaps a LOT of them globally with no legitimate specs or procedures, that its intent was to create some level of order, but perhaps not to replace legitimate OEM specs and procedures.

Just a few questioning thoughts. I'll be curious to see some more expert replies.

As always, thought-provoking statements. Here are some additional thoughts:

1. Does the mfg. always know better how to calibrate their device than any other entity? My point is this: mfg. companies make things; that's their business. The process of setting tolerances, determining suitability and such, may not be their primary strength.

2. If there was not a need for the ISO standard for pipettes, why was one written? Especially with pipettes, there are a lot of variables that come into play, probably as much as most other instruments. I believe that the ISO specification allows for more sources of error before stating that an instrument is in/out. Many times it is unclear how the mfg. calculates the acceptance tolerance for equipment.

As I have stated before, mfg. do not have a business interest to develop "realistic" tolerances. They are in the business of selling:); that requires having same/lower tolerance, and same/lower price than the competitor. Too, if they have a fair amount of failures requiring calibration, that means more service business for them. :tg:

Essentially, we have pipettes that have mfg. specifications of .01, ISO tolerance of .1, and a use requirement of 1. Obviously I made those numbers up, but not too unrealistic to demonstrate how much tighter the mfg. tolerance is to what we require the instrument to perform. I would think as long as the procedure/tolerance used meets the need of the process, that should be OK within 17025.

Hi Brad,

If I understand right, ISO 17025 pertains to General requirements for the competence of testing and calibration laboratories - so its more like a QMS for testing labs, while

ISO 8655 gives specific technical requirements needed for calibrating pipettes, dilutors and dispensers.

Thus there should be no issues while following ISO 8655.

Update: I would appear that the auditor for the Accreditation Body is highly insistent that the work is to be done to mfg. specifications, and essentially, not recognizing the ISO specification.

I have let my vendor know I am not happy with it,:mad: as it will create huge work on our end, with no value/process improvement that I can see.

As always, your thoughts/ comments are welcome.:)

Update: I would appear that the auditor for the Accreditation Body is highly insistent that the work is to be done to mfg. specifications, and essentially, not recognizing the ISO specification.

I have let my vendor know I am not happy with it,:mad: as it will create huge work on our end, with no value/process improvement that I can see.

As always, your thoughts/ comments are welcome.:)

What is the auditor citing as the requirement? What is he basing his "ruling" on?

What is the auditor citing as the requirement? What is he basing his "ruling" on?

That's a good question. Essentially I have not received any citation from them on why to perform it this way. This individual comes from a testing environment where everything was done to mfg. specifications, and thus cannot comprehend something being verified outside of the specification.

The issue is we have every mfg. known at just about every range; thus tracking them individually is a nightmare. Hence the reason we wanted to use the ISO specification; to standardize our approach.

Given the fact that I cite from 17025 customer requirement sections, and the preference to use ISO specifications, I am at a loss why such an intepretation is being so heavily relied upon. I think as long as the lab and I agree upon the tolerances, it should be fine.

This plays into why I have always taken the other side of things also, as the auditee. Having someone tell you to challenge the auditor/ appeal decisions is one thing; doing them is another. People spend a huge amount of time and effort getting accredited; and are reticent to "rock the boat". Too, given the expenditure of time and money, they may not be interested in switching accreditors.

Reason #100,000 why a good auditor is worth more than what they are paid.:agree1:

One key here is the decision by the customer (BradM) regarding the specs.....

ISO/IEC 17025:2005 Clauses 5.4.1 and 5.4.4 allow the laboratory and the customer to agree on the requirements.....the laboratory has to perform any validation that may be necessary.....and also other Clauses such as 4.4.1

If the assessor has valid reason for objecting to the lab agreeing to the new specs, then that should be laid out in a concise manner with the applicable references (if any) cited.....

If the assessor refuses and has no identifiable basis, then the laboratory can appeal the assessor's decision to the AB.....

Hope this helps.....

One key here is the decision by the customer (BradM) regarding the specs.....

ISO/IEC 17025:2005 Clauses 5.4.1 and 5.4.4 allow the laboratory and the customer to agree on the requirements.....the laboratory has to perform any validation that may be necessary.....and also other Clauses such as 4.4.1

If the assessor has valid reason for objecting to the lab agreeing to the new specs, then that should be laid out in a concise manner with the applicable references (if any) cited.....

If the assessor refuses and has no identifiable basis, then the laboratory can appeal the assessor's decision to the AB.....

Hope this helps.....

As always, it's helpful. :agree1:

Update:

My vendor came to me and proposed using two different tolerances. The ISO tolerance would be used for incoming instruments, and the mfg. specifications used for outgoing (as-left) assessments. I said no.:mad:

My thought was this: why should my process have double the work added to it, without any value that I could see? I told them I would have a problem with it.

I got a phone call a few days later that stated the ISO standard would be fine to do the work against. :D;)

After attending some other training, I was able to discuss more pipette calibrations with people in the industry. Seems that people who strictly go by mfg. tolerances have extremely high failures!!:mg::) Several with more mature processes have established their own tolerances for the pipettes.

Saying there is a lot of Science and a lot of Art in calibration. As calibration professionals, we need to use avenue/ forums like the national organizations that Hershal provides information on, and of course, forums like the Cove, to discuss things.:agree1:

This could be an instasnce where the Method 6 of the newer ANSI/NCSL Z540.3-2006 guardbanding approach may have value.....it helps with the approach known as 2% Probability of False Accept (PFA).....

Hi,
The middle test point inaccuracy % is the difference between manufacturer's specification and ISO 8655 standard tolerance.
I believe the ISO 8655 standard tolerances only takes into account for full scale reading and 10% of full scale.

Tell your auditor even the manufacturer have no clue what should be the standard tolerances.:frust:
A good example is Thermo Scientific. They will publish ISO 8655 tolerances on their website, but do not actually use them if you send in pipettes for them to calibrate. Found out when I saw the calibration report, I just looked for the middle test point.

Hi,
The middle test point inaccuracy % is the difference between manufacturer's specification and ISO 8655 standard tolerance.
I believe the ISO 8655 standard tolerances only takes into account for full scale reading and 10% of full scale.

Tell your auditor even the manufacturer have no clue what should be the standard tolerances.:frust:
A good example is Thermo Scientific. They will publish ISO 8655 tolerances on their website, but do not actually use them if you send in pipettes for them to calibrate. Found out when I saw the calibration report, I just looked for the middle test point.

Excellent information. Thank you. And welcome to the Cove! It's great having you here.:agree1:

Hi All,

I'd like to bring up another issue here. Is there a difference between conformity evaluation and user testing?

To be more exact: ISO 8655-6 says that the systematic error is e=100*(Vm-Vs)/Vs where Vm-mean volume, Vs- selected volume. Section 8.4.2 says that for variable volume pipettors, and if the systematic error is used for conformity evaluation, Vs in the denominator shall be replaced with the nominal volume Vo (maximum selectable volume). The equation then becomes e=100*(Vm-Vs)/Vo
Well, let's say I have a pipettor with the range of 10uL-1000uL, and I am verifying it at 50uL (selected volume, Vs). If I calculate the systematic error dividing by 1000 (Vo) as opposed to 50 (Vs), I would get an artificially low error.
Practic example: the average of 10 measurements is 45.7uL; if the error is calculated by dividing with 1000, it would pass the ISO 8655-2 max permisible error criteria; if the error is calculated by dividing with 50, it would not pass the max permissible error criteria.

Am I right in saying that conformity evaluation applies to the manufacturer to demonstrate that the pipettor is working within manufacturer's specifications (therefore divide by nominal volume, 1000 when calculating the systematic error), whereas when the user is testing, they should use the selected volume Vs (50 in this case) to calculate the systematic error.

Sorry for the length of the post, I wanted to give all details for an easier understanding.

Ed.

Welcome to the Cove :)

Thanks for the valuable post !

Ed, that's a good question. I'll need to do some looking and research on it.

Hi All,

I'd like to bring up another issue here. Is there a difference between conformity evaluation and user testing?

To be more exact: ISO 8655-6 says that the systematic error is e=100*(Vm-Vs)/Vs where Vm-mean volume, Vs- selected volume. Section 8.4.2 says that for variable volume pipettors, and if the systematic error is used for conformity evaluation, Vs in the denominator shall be replaced with the nominal volume Vo (maximum selectable volume). The equation then becomes e=100*(Vm-Vs)/Vo
Well, let's say I have a pipettor with the range of 10uL-1000uL, and I am verifying it at 50uL (selected volume, Vs). If I calculate the systematic error dividing by 1000 (Vo) as opposed to 50 (Vs), I would get an artificially low error.
Practic example: the average of 10 measurements is 45.7uL; if the error is calculated by dividing with 1000, it would pass the ISO 8655-2 max permisible error criteria; if the error is calculated by dividing with 50, it would not pass the max permissible error criteria.

Am I right in saying that conformity evaluation applies to the manufacturer to demonstrate that the pipettor is working within manufacturer's specifications (therefore divide by nominal volume, 1000 when calculating the systematic error), whereas when the user is testing, they should use the selected volume Vs (50 in this case) to calculate the systematic error.

Sorry for the length of the post, I wanted to give all details for an easier understanding.

Ed.

Ed, here is an interpretation put together by one of my associates:

The ISO tables were abbreviated and dealt specifically with the nominal volume defined as full volume or the greatest possible user-selectable volume.

Dr. Burkhard Winter (ISO/TC 48 Laboratory equipment pro) indicated in a reply that "The % values given in Table 1 of ISO 8655-2 are correct. They are related to the nominal volume. Manufacturers often relate their values to the "selected" volume in the case of variable-volume pipettes (see 7.4 in ISO 8655-2). To avoid ambiguities and misunderstandings, it is better to use the absolute values given in µl."

"Narrower mpe's than given in the Table 1 are possible but greatly depend on the air volume in the pipette tip and the handling and lab conditions
of the user. See Table B.1 for details."

"It is recommendable that manufacturer claiming narrower mpe's proof it, e.g. by a calibration certificate."
In the posting above the maximum permissible error for a 1000µl is ±8.0µl is 0.8% or at 50µl 16% or 8.0µl. Would suggest that the user be more aware of narrowing the selection of pipette when sampling at 50µl (almost 9% at 45.7µl). Ideally (midrange) would be a 100µl nominal volume pipette with an accuracy of 0.8µl much more in line with the manufacturer's stated accuracy as Dr. Winter indicated when the "mpe's proof it by a calibration certificate". That would be a much better choice. One should not get in the habit of trying to make an abbreviated table fit all applications but be alert to the potential performance, good or bad.

Thank you Brad for your post. Very well documented.

However, when I looked at the manufacturer's calibration certificate (BIOHIT, ISO 17025 accredited), they have calculated the maximum permissible errors by dividing with the selected volume (50µL), as opposed to the nominal volume (1000µL). Their values were: 50.3, 50.2, 50.3, 49.9, and 49.9; with an average of 50.1, calculated inaccuracy of 0.24%, and calculated imprecision of 0.41%. Their maximum permissible inaccuracy is 3.00% while the maximum permissible imprecision is 1.5%, which are different then those in ISO 8655-2 for 1000µL (0.8% and 0.3% respectively), or for 50µL (1.0% and 0.4% respectively). It looks to me like Biohit is applying their max permissible errors to the selected volume, not the nominal volume as per ISO 8655.

Any other thoughts?

Thanks.

Ed