Registrar said that if a company has the computers (for office purposes only!!), they should be calibrated as if they referred to 7.6.! Has anyone else experienced that?

Calibrated computers???? Nope... I most definitely have not.

If the computers were used to run software for surveillance or measuring I would consider the last clause of 7.6 applicable, but I would dearly love to hear your registrar explain how to calibrate a computer....:ko: Came to think of it: That would be a mortal blow to Bill Gates.. :vfunny:

Are you certain you understood each other correctly?

/Claes

Never heard of calibrating computers used for office purpose only.

As Claes said, you need to calibrate/validate the software you use for measurement or data analysis purpose. That would involve using known/published data and comparing the results with known/published/manually calculated results.

Came to think of it: That would be a mortal blow to Bill Gates.. Mortal blow you say? How about Microsoft Calib-Agency Inc.? :biglaugh: Only Bill knows how to 'calibrate' MS Office! :rolleyes:

Dont look at me (and talk to me) like if i asked to calibrate computer :))))
I dont even know the auditor who asked it. I knew there was lot of uneffective (resultless) discussions on it in IRCA training courses... So i asked to ask you... And i am not surprised with results :)

Calibrated computers???? Sounds like something directly from a Dilbert cartoon. I'd have about as much respect for whoever suggested such a thing as a car mechanic who would suggest I need to have my turn signal fluid replaced. Jeez.

Mike S.

Yes, computers do need to be calibrated! Not the software mind you, but the computers themselves. It is a requirement! :eek:

Now, how is a computer calibrated? Computer outputs are a function of the inputs and the software. The hardware has little to do with the output. (Although I have a computer that loses exactly 1 hour each day - is that a software or hardware thing?) :confused:

Hmmmmmmmmm

Where is the shall on that one db ?

Calibration is only applicable to test hardware and test software, if together they make up a computerised system then you might inappropriately say that the computer is calibrated - this is however nonsense.

Martin, I must confess, I lied. The whole idea was so preposterous to me that I replied in pure sarcasm! I know, I really shouldn’t do that, and I apologize to the entire Cove!

Sorry db - missed the sarcasm.

Didnt think it sounded like you ;)

Need some advised on this. During audit we found some of the test station calibration was over due becos the cabilbrator has to be send out to US for calibration atleast 1 month. However the daily correlation is done without fail and the reading show no drift. Now is this an NCR? Can we use the correlation records as evidence to conformity?

Sounds like you could be in trouble.

If you need a specific piece of equipment to calibrate another piece of equipment you should schedule the calibration of the 'master' equipment so that it doesnt upset the scheduled calibration of other equipment, or have a back up piece of equipment.

Hi,

Well, if it's overdue it's overdue.

And if it takes a month (?) to calibrate that thing there is all the more reason to make certain it's done on time and plan for it. How do you cope without it for that long? Have you got another one? What does your procedure say?

It looks like a NCR to me...

/Claes

ooops.. Simultaneous posting again... Hi Marcus.. :)

Hey, let's get back to calibrating computers for office use! I haven't had a laugh like that in awhile! What's next, calibrating pens and pencils!!?? Or should I say, establishing a procedure to calibrate writing implements?! Oh, where do they get some of these auditors?

Johnny,

Sounds like a NC to me also (depending on what your procedures say you are supposed to do), but it also sounds like you can prevent this easily. Look over past cal records and see if you can get by with an actual "calibration" once every 2-3 months or more. If so, schedule a cal just before the calibrator goes out for its cal for 1 month. Between cal's. use your "daily correlation" -- which I assume is a part that is checked to verify a consistent reading from day-to-day -- to give you confidence in the measurement.

Or, consider another option. We have a critical measurement system made-up of 5 pieces of electronic equipment, 3 of which readings are taken from to calculate test values. Plus several cables, connectors, etc. to hook it all together. If any one of these 3 pieces of electronic equipment or cables, etc. goes bad or out-of-cal. on us we could get "bad" readings. This could happen one day after an outside cal. for all we know. The equipment manufacturers recommend yearly cal's. but tearing the system down for cal and then re-assembling and tuning it back in to get good results was a real pain, not to mention lost work time. We decided to make our calibration procedure an internal one where we look at the performance of the entire "measurement system" once each day by measuring "standards" of known value. If those "standards" fall within the required values, we say the "measurement system" is "calibrated" (maybe "verified" is more correct?) and go on from there. It saves money (no outside cal costs), and time and it works for us. Perhaps you can do something similar?

Mike S.

johnnybegood said:
Need some advised on this. During audit we found some of the test station calibration was over due becos the cabilbrator has to be send out to US for calibration atleast 1 month. However the daily correlation is done without fail and the reading show no drift. Now is this an NCR? Can we use the correlation records as evidence to conformity?

To me this is why a deviation system exists. For unavoidable problems with calibration at our site we have written deviations (approved by senior management) to cover the time period affected. This only flies if you actually have another means of checks/balance to use in the interim. Then once the problem is solved we use a sampling method for final verification that our stop-gap measure was effective.

If your team did nothing to plan ahead and put in place a stop-gap measure, ensuring that materials are identified for follow-up verification, then it is definately a NC. It sounds like they did what needed to be done (monitoring for variation) but didn't estabish a plan to notify/document/verify. Even if product needs to be released prior to equipment return you can set up another verification as needed to ensure only good product goes out the door.

We've never had any problem with our auditors on this. Our problematic calibration area is scales not the calipers, voltmeters, etc... so this is what we've established to be used in emergency situations.

E Wall

Do you have any corrective action system connected with the failures of your system that you have to write deviations for ?

M Greenaway said:

E Wall

Do you have any corrective action system connected with the failures of your system that you have to write deviations for ?

Yes, but only in some situations...not all (1 deviation does not always produce 1 RCA). In the case of calibration of equipment specifically - there hasn't been an RCA written for instances when budget constraints precluded our ability to meet the requirement if sub-contracted services were needed (which has happened a few times - I did mention our company is in Chp 11 processing, didn't I?), but when there was a disconnect in the internal process itself yes a RCA was written up.

While I personally believe even the budget constraints warranted an RCA on management for availability of resources - I'm told to drop it, be realistic...I'll never get a response on it. If this was found by a 2nd or 3rd party auditor...I'm sure they would push the issue which would hopefully force mgmt to address it...but has never happend.

To resurrect this post, I have another computer/software related question for Clause 7.6. In our case, we develop software and software only. Our software does have a measurement function, but this is a 'feature' of the product, not a product specification.

During the software lifecycle, there are several levels of testing. Once the product is 'completed', the QA team have a go at it before it is released or made available.

My question is: if we only produce software, does 7.6 still apply to us, or can we take an exclusion under 1.3 or can we just mention in our quality manual that the software itself does not require calibration?

Thanks for your help.

- MtlGuy

MTLGuy:

Clause 7.6 is about measuring and montioring devices. Measuring devices produce a descrete value. Monitoring devices may not.

7.6 a) states that only measuring equipment used to ensure valid results must either be calibrated or verified. If you have equipment used to ensure valid results of your software, then that equipment either needs to be calibrated or verified.

If you use software to ensure your software is vaild, then the last paragraph of 7.6 applies.

My guess is that 7.6 applies to your organiztion but that is only a guess. You must have equipment or software that needs to be verfied or confirmed. How else would you know if your product (software) meets specified requirements?

Regards,

Dirk

To resurrect this post, I have another computer/software related question for Clause 7.6. In our case, we develop software and software only. Our software does have a measurement function, but this is a 'feature' of the product, not a product specification.

During the software lifecycle, there are several levels of testing. Once the product is 'completed', the QA team have a go at it before it is released or made available.

My question is: if we only produce software, does 7.6 still apply to us, or can we take an exclusion under 1.3 or can we just mention in our quality manual that the software itself does not require calibration?

Thanks for your help.

- MtlGuy

If the software does not require calibration, then it does not apply. However, before you get eager, does the software calculate any values? Does it report any results? Have the formulas, if any, been tested to verify they yield mathmatically accurate results. The Intel IC chip "floating point problem" a few years ago would be an applicable example.

If absolutely none of this applies, and you have no gages whatsoever, then perhaps 7.6 does not apply.

Calibrated computers???? Sounds like something directly from a Dilbert cartoon. I'd have about as much respect for whoever suggested such a thing as a car mechanic who would suggest I need to have my turn signal fluid replaced. Jeez.

Mike S.

Hey, that's my mechanic you're putting down! :nope:

i don't how do we calibrate computers in ISO 9001. I beleive that its talking about the software.
However, in ISO 17025 asks for the validation of software (use for aquisition, processing, recording, reporting, storage or retreival of tests data).etc etc Clause 5.4.7 Control of Data- ISO/IEC 17025
Clause 5.5.4 talks about the caibration of equipment & its softwares.

Hey, let's get back to calibrating computers for office use! I haven't had a laugh like that in awhile! What's next, calibrating pens and pencils!!?? Or should I say, establishing a procedure to calibrate writing implements?! Oh, where do they get some of these auditors?

:applause:

Hi,

Well, if it's overdue it's overdue.

And if it takes a month (?) to calibrate that thing there is all the more reason to make certain it's done on time and plan for it. How do you cope without it for that long? Have you got another one? What does your procedure say?

It looks like a NCR to me...

/Claes

ooops.. Simultaneous posting again... Hi Marcus.. :)

I beleive that in this paricular case when you know that your equipment takes a month to calibrate & you need that equipment (can't spare it to calibrate) than you definetely need to buy its replacement.

Hey, that's my mechanic you're putting down! :nope:

Gert
Don't worry
American vehicles have extended fluid change intervals for their turn signals.
Denmark still has some way to go to catch up.:)

Hello All:

7.6 includes the requriement to verify - not just calibrate. There are many posting that only apply to calibration of computers and software. Look for the word "verified" in the text (7.6 a) )

Also, the point of 7.6 is to have control over measuring AND monitoring devices. The clause is NOT about calibration only.

Last paragraph of 7.6 states that the ability of software to satisfy intended applications must be confirmed.

Also, the measuring and monitoring equipment used to provide evidence of the product (in this case - software) conformity must be calibrated or verified - not just calibrated.

Regards,

Dirk

By the way, the Auditing Practices Group documents, available on the TC 176 / SC2 website may be helpful for many of our questions.

Here is the website:

http://isotc.iso.org/livelink/livelink/fetch/2000/2122/138402/755901/1069636/customview.html?func=ll&objId=1069636&objAction=browse&sort=name

Attached is the APG document on clause 7.6.

Regards,

Dirk

Check out the sanctioned interpreations too.....

There are 2 of them for 7.6.

http://www.tc176.org/Interpre.asp

Regards,

Dirk

...
I disagree with the following statement posted by hjilling:

"If the software does not require calibration, then it does not apply."

Last paragraph of 7.6 states that the ability of software to satisfy intended applications must be confirmed.

Also, the measuring and monitoring equipment used to provide evidence of the product (in this case - software) conformity must be calibrated or verified - not just calibrated.

Regards,

Dirk


Perhaps a misunderstanding (out of context). The poster claimed there was nothing in his software that needed to be verified. I explained some examples, and then said if there was nothing in the software that required to be verified, then it does not apply.

For example, a log sheet or list with no formulas, probably does not need to be verified. If there is something in the software that calculates, then it must be verified.

Hey all!

Marc has a thread for calibration vs. verification:

Calibration vs. Verification - A Good List of Definitions (http://elsmar.com/Forums/showthread.php?t=945)

IMO, the correct interpretation of this would depend on the application. I have seen software programs that have internal verification algorithms they refer to as calibration.

I haven't ran across much calibration of computers. I have utilized computers as part of a system for calibrating sensors connected to the computer. Sometimes software on the computer allows for software adjustments to the input signal.

1. If the system will be producing meaningful (significant/ critical/ etc.) data, the system needs to be validated.

2. If there are measurements being taken/obtained with the system, a N.I.S.T traceable comparison should be made. Usually this occurs with an external sensor of some sorts.

Oh,it makes no sense.

Hello Hjilling:

I misunderstood. Thank you for the clarification. I deleted my posting stating I disagree with you. I agree with you.

Regards,

Dirk

Thanks all for the feedback. Let me give my interpretation of 7.6 and then feel free to tell me if I'm way out in left field or not.

I thought 7.6 pertained to any device that is used to monitor or measure the final product itself (as opposed to the actual product itself).

At designated intervals in our software development process, we perform unit testing and regression testing to ensure the specific feature (and the entire product) works as expected. Once the final version is completed, our QA team has a go at it, also performing GUI testing, regression testing and the like.

Currently all testing is done manually, we do not have any automated testing tools. These steps satisfy Clause 7.3.5 and 7.3.6 which address Design and Development verification and validation respectively. During this phase, we ensure that "the resulting product is capable of meeting the requirements for the specified application or intended use."

That being said, we have no 'device' that proves we are 'within spec' (no calipers, no guages...). We also have no software that performs any monitoring or measuring of specified requirements.

So, can anyone see any reason that I can't claim an exclusion to 7.6?

-MtlGuy

So, can anyone see any reason that I can't claim an exclusion to 7.6?



I wouldn't phrase it that way - using the word "exclusion" - but I would put a paragraph in the quality manual detailing how 7.6 is not appropriate to your operation.

Thanks all for the feedback. Let me give my interpretation of 7.6 and then feel free to tell me if I'm way out in left field or not.

I thought 7.6 pertained to any device that is used to monitor or measure the final product itself (as opposed to the actual product itself).

At designated intervals in our software development process, we perform unit testing and regression testing to ensure the specific feature (and the entire product) works as expected. Once the final version is completed, our QA team has a go at it, also performing GUI testing, regression testing and the like.

Currently all testing is done manually, we do not have any automated testing tools. These steps satisfy Clause 7.3.5 and 7.3.6 which address Design and Development verification and validation respectively. During this phase, we ensure that "the resulting product is capable of meeting the requirements for the specified application or intended use."

That being said, we have no 'device' that proves we are 'within spec' (no calipers, no guages...). We also have no software that performs any monitoring or measuring of specified requirements.

So, can anyone see any reason that I can't claim an exclusion to 7.6?

-MtlGuy

7.6 states "The organization shall determine the monitoring and measurement to be undertaken and the monitoring and measurement devices needed..." I agree that you may not need monitoring and measurement devices, but don´t you need to "determine the monitoring and measurements to be undertaken"?
You stated that Once the final version is completed, our QA team has a go at it, also performing GUI testing, regression testing and the like. ... isn´t this your "determining the monitoring and measurement"?

What is involved in the GUI testing and the regression testing?

Just out of curiosity, if the final inspection of ABC Co. is a visual inspection based on color, does the final inspector need to have a color blindness test periodically? ISO 9001 states "Where necessary to ensure valid results , measurement equipment (the inspectors eyes) shall : a)be calibrated or verified (color blindness test)...

I hope I haven´t gone on a tangent here or "tripped" out to much...

There are certainly instances where a computer is part of a calibration, i.e. you're calibrating a sytem. An example of this would be our Image Analysis system, since the video input can vary depending on the camera, and the length of measurement is per pixel, this instrument needs to be calibrated as a system at a set resolution, camera mag, etc...once calibrated the software retains correction factors...if any change occurs either to the camera, software, display settings, or the display itself the sytem requires re-calibration.

It's not really calibration of a computer...which I agree is laughable, but the hardware (as far as the display) does come into play as a critical component.

:biglaugh:

If the software you designed is not performing calculations, then there is likely no need to "verify the calculation."

If that is the case, the design verification sounds like you are doing a good job.

Later, if you find you have gages, you can calibrate them. Odds are, there are gages somewhere in the building.

You don't have to exclude 7.6. You may just have a VERY SHORT list of gages!

MtlGuy:

How does your organization perform the following?

"we perform unit testing and regression testing to ensure the specific feature (and the entire product) works as expected."

I am going to guess that you run the software on a computer. If so, does the computer itself affect the results of the testing? How do you know?

Do you use any computer? Specific computers? Do you other software to test the software you produce?

Thank you,

Dirk

Hi Dirk,

Yes, we do run and test the software on a computer (several computers, actually). The computer itself may affect the performance of the software if there is not enough memory, but we set minimum computer requirements (with regards to memory, O/S, etc...) to effectively run our software. These requirements are made known to the client during the contract phase and are documented in the user manuals.

The QA team does their testing manually on several computers in the QA department (clicking the open button to make sure the application opens and so on). No software is used to test the software.

Thanks

MtlGuy

Hello MtlGuy:

If the computer itself may affect the test results, then in some way you have to verify that the computer does NOT affect the testing results.

In my opinion, the following is evidence of verifying monitoring devices:

"The computer itself may affect the performance of the software if there is not enough memory, but we set minimum computer requirements (with regards to memory, O/S, etc...) to effectively run our software. These requirements are made known to the client during the contract phase and are documented in the user manuals."

Some how you have to know if the computer (monitoring device) has enough memory. You even define them as requirements. How do you know you have met the requirements to ensure the computer is not affecting the results of the testing? You know this becaue you have verified your monitoring / testing devices.

7.6 seems to apply, in my opinion.

Why not use any and all computers available to you? Becasue they have to be verified to have enough memory, O/S, etc.

Regards,

Dirk

Hi Dirk,

I agree that we must verify that the computer NOT affect the testing results, but we do not have a tool for this (well, we have a QA tool, but it isn't nice to call him that :lol: ).

We, or the user knows that the computer has enough memory or the correct operating systems by either checking the 'properties' of the computer (manually, by right clicking on 'my computer' and selecting 'properties'), or by opening the computer and looking at the size of the ram connected to the mother board.

Our verification is done visually/manually as is our testing. Therefore, I don't see how it would apply since we have no device that performs this task.

Before I posted, I re-read your reply. Are you saying that the computer that actually runs the software is the measuring device? If so, how do we ensure that the computer is calibrated against measurement standards traceable to international or national measurement standards?

I'm starting to feel dizzy.

Thanks.

MtlGuy

You can:

Have a computer as a component in a measuring system that is part of the calibration process.
Calibrate a measuring device connected to a computer.
Validate a computer for it's intended process/use.
Validate a program on the computer to assure that it is performing the required algorithms accurately and (if applicable) the data is secure.Calibration involves verifying uncertainty using a traceable standard. You cannot do this to your computer in your game room at home (unless you want to calibrate clock speed, processor time, power supply voltage, etc. which seems a bit silly).

However.. if you have temperature probes, PLC inputs, imaging equipment, etc. connected to the computer, and you make acceptance by the display/printout of the computer, then the computer becomes part of the loop.:2cents:

Many years ago I had a NUPIC auditor write a finding that we had no documentation to support the fact that the the built in "calibration" software in one of our instruments had been validated. No amount of explanation could sway him. After months of wrangling, a new auditor scheduled a follow up visit and closed this finding (and another 15 similar BS issues).

We injected a signal from NIST traceable waveform generator into our box. The software inside the unit would then apply a variety of offsets and multipliers to the variety of signal analysis circuits in the box to make everything come out right. After that, we would inject a variety of signals from the NIST traceable waveform generator and verify that each circuit in the box produced a reading within specification before issuing a calibration certificate. In fact, we had over 5,000 instances of test reports and calibration certificates that validated the software.

Why does the word "calibration" keep coming into this thread? Why? 7.6 is about measuring and montioring devices and includes calibration and verification.

calibrate a measuring device
calibrate a monitoing device
verify a measuring device
verify a monitoring device

Please forget calibration for this thread.

The computer IS the device used to monitor or measure your product. Why do you test your software on the computer? Becasue it montiors or measures your product's conformity to requirements. You have no other way to know if your product is any good.

You already verify your device "by either checking the 'properties' of the computer (manually, by right clicking on 'my computer' and selecting 'properties'), or by opening the computer and looking at the size of the ram connected to the mother board." With those actions, you have verified your montioring or measuring device.

Verifying a measuring or monitoring device visually or manually doesn't count? If verification is done visually, then it is NOT verification?

MtlGuy - You have to break free of the traditional view of what people think 7.6 is about. Calibration is only a part of clause 7.6.

Regards,

Dirk