Automatic calibration of balances

[Dominique]

Hi everyone!
One question: How do I proof the reliability or validity of the automatic calibration I do with the analytical balance? I never questioned it, but now I am being told that the only calibration is the one done by the certified metrologist every six months because he presents the whole set of statistical data and everything. Apparently I can only do verifications....
Any advice?
Thanks!
Dominique G.

[tom jupille]

I suspect this is a matter of terminology. My suggestion would be to write an SOP stating that the balances will be calibrated every six months and the calibration verified daily (or weekly, or whatever), with the "autocalibration" used as verification.

[lmh]

LIke Tom, I suspect this is a matter of terminology, but I almost feel the terminology is the wrong way round (sorry Tom!).

To me, verification implies that no actual change is made, you just check that the instrument is giving the correct results.

Calibration implies that the instrument (or some internal calibration curve within it) is changed in some way to give the right result.

As such, when you do an automatic calibration, it is indeed a calibration. Results afterwards will not necessarily be the same as results before. When your certified meteorologist casts his cloud over your instrument, he/she may or may not actually change it in some way (in which case they are calibrating), but they do, in any case, produce a lot of statistical data verifying its performance. Because you don't do all the statistics from multiple weighings, your automatic calibration is not a verification (although I'd be deeply surprised if anything were going wrong).

But one can get too bogged down in terminology. The main thing is to have procedures in place so you know how the instrument is likely to have been performing on any particular day. If you don't run the automatic calibration routine as often as the manufacturer suggets, then the instrument may drift, and give the wrong answers. If you do, the results are only as good as the instrument's internal calibration mass, and the quality of its calibration routine (i.e. the results are probably the most reliable thing in your lab; analytical balances are pretty good!). If you're really fussy, you could buy yourself a high-quality (traceable?) mass to use to verify the automatic calibration after it's finished.

Incidentally, a personal soap-box theme of mine: I really like it when labs do their own verification in addition to, or instead of external metrologists. I really want to know that when I use my pipette in my lab, I get what I expect. That's a very different question to whether an expert from a calibration company can get the right result when he/she uses my pipette in his/her temperature-controlled super-environment. Frankly, it's bad science to assume that because something has been externally verified, it's also correct internally.

[Dominique]

Hi, thank you so much for your answers!
I came from a lab with procedures in place and traceable masses (a land of pink elephants and bunnies in that sense), and here at my new job, they are not very sure what is the right thing to do (auditors say one thing and instrument manuals say other, etc.), so I will just tell them. First thing I said was that we needed to do the automatic calibration everyday, to avoid wrong readings. The poor balances are doing as much as they can... I will do some serious SOP checking and implementing now!!!
Thank you so much for you insights!
I will get back to my silica drying now: it was PINK! argh...

[Csaba]

Hi,
Just a comment: Although many modern (top-range) scales has autocalibration with a built in weight, for some reason you rarely (never?) can use the built in weight to do a "manual" reading of the built in weight to verify that "nothing" has happend before doing an autocalibration. Instead you must have a separate control weight to do a separate verification - with all the issues that it may create.

[lmh]

A follow-up comment for Csaba,

Some years ago I worked in a lab that had a balance working just as you described. It had a simple (mechanical?) sliding knob at the bottom which applied the internal calibration mass. To calibrate the balance you pressed a "Cal" button and then slid the knob to apply the calibration mass when it asked for it.

This meant you could apply the calibration mass outside the calibration routine, just to check that the reading was correct.

On the other hand, the balance had a weakness. People with fiddly fingers who just push things would tend to slide the knob while they were standing next to the balance chatting... sometimes someone would slide it accidentally while moving things nearby. It was still possible to tare the balance, but it's obviously not a good idea to have it weighing 101.352g when you wanted to weigh 1.352g. Occasionally people would call me complaining the balance wasn't working, and it'd turn out the calibration mass was half-on, the knob slid half-across, and the readings would be all over everywhere.

It was a nice balance, but like so many things in a lab, needed its operators to think while using it... Perhaps that's why modern ones don't work that way?

[krickos]

Hi

Just to share how basicly do it for analytical balances:

Daily/before analys the "internal calibrationn" is made.

Weekly the analyst uses 2 certified weights after internal calibration to verify that the balance shows the correct weight.

Every 12 (or is it 6 months?) weights and balances are recertified/checked.

[bgranot]

I suspect this is a matter of terminology. My suggestion would be to write an SOP stating that the balances will be calibrated every six months and the calibration verified daily (or weekly, or whatever), with the "autocalibration" used as verification.

I dissagree. The 'verification' daily or weekley need to be done with NIST tracable wt. The 'auto calibration' is the balance internal weights. These balance internal weights need to be verified with the NIST traceble weights, and periodically (FDA anually) need to be tested and actual calibrated with the external NIST tracable wieghts doing linearity etc...
We on a daily basis do three NIST weight checks from 3/4 of the max scale to the 'regulatory minimum weight'
It is all pending the regulatoy level your lab has to comply with, and how paranoid is your supervisor.

[DSP007]

1) As we say in Russia, "if you do not touch shit, it will not smell." The current balance is automatically calibrated when turned (ON) on and make additional recalibration them - only to spoil.
2) In order to show the regulator that "in the period between checks your balance scales worked fine" - make for yourself a quick procedure for checking the scales.
Buy a benchmark set of weights for your class (you can take from the old weights with a yoke), and eath day weigh two weights benchmark before work, for example 1 g and 100 g. The results are recorded in special paper form with the calendar dates or print and glue to the strip tape.
Then you will have proof that your scales for six months did not change the evidence on the g and 100 g of weight and work all the time.

[Peter Apps]

Of all the instruments and procedures in an analytical lab, weighing on a modern analytical balance is by far the most precise and repeatable.

When the metrologists come with their boxes of weights they determine the performance of the balance at that position on the bench, on that day, at that time, with whatever calibration its electronics has at the time. Then they issue a certificate that is an official record of how they found things. They do not adjust the balance.

Strictly, as soon as you press calibrate, the certificate does not apply any longer, because the balance has now been adjusted. In practise, which is actually the only thing that matters, the balances are so well designed and built that their performance is very very stable. Autocalibrating mainly allows the balance to adjust for changes in temperature and humidity.

Checking the balance's performance with a certified mass piece is assurance that it is working correctly. If weighing two or three chunks of metal once a day really disrupts your schedule, then you need to look at how difficult you are making it for yourself, and find a simpler way.

To my mind there is little value in checking that the balance can weigh 100g accurately when you are using it to weigh milligrams - rather check its performace with a certified mass piece that has a similar mass to what you actually want to weigh.


Peter

[WK]

Nicely worded post, Peter.

[DSP007]

Of all the instruments and procedures in an analytical lab, weighing on a modern analytical balance is by far the most precise and repeatable.

When the metrologists come with their boxes of weights they determine the performance of the balance at that position on the bench, on that day, at that time, with whatever calibration its electronics has at the time. Then they issue a certificate that is an official record of how they found things. They do not adjust the balance.

Strictly, as soon as you press calibrate, the certificate does not apply any longer, because the balance has now been adjusted. In practise, which is actually the only thing that matters, the balances are so well designed and built that their performance is very very stable. Autocalibrating mainly allows the balance to adjust for changes in temperature and humidity.

Checking the balance's performance with a certified mass piece is assurance that it is working correctly. If weighing two or three chunks of metal once a day really disrupts your schedule, then you need to look at how difficult you are making it for yourself, and find a simpler way.

Yes of course 100%+ . Its "regulator paranoja" , but with this nonsense ( " prove that your balance is not brought down between government verification" ) we encountered

To my mind there is little value in checking that the balance can weigh 100g accurately when you are using it to weigh milligrams - rather check its performace with a certified mass piece that has a similar mass to what you actually want to weigh.
Peter

Small weights (bencithes) have the disadvantage that " react to every sneeze" - scratched the surface, got a speck of dust, and their weight has changed.

[Peter Apps]

Small weights being more difficut to weigh accurately (responding to ever sneeze and scratch) is exactly the reason why small weighings should be checked with small weights. If human error is such that you can't get an accurate weight with a little piece of metal I doubt that you are going to be accurate enough with small quantities of chemicals.

And, although I have a profoundly sceptical view of external regulation, I do not see checking a balance's performance as paranoia. In the first place it is quick and easy to do, and in an instrumental lab weighing and precise volumetrics are the foundation of every result - because they are used to make up calibration standards.

Peter

[DSP007]

Well, you a normal person (man) and sense of paranoia ( the paranoid ill style of thinking) is not understood.
Meaning something like the following - what do you prove that, between the government checking you are not forced down the scale of balance ?

[krickos]

Nicely worded post, Peter.

Indeed, anyone at a hightthrougput sample lab who have faced the issue of a balance failing a check and trying to trace back any possibly effected tests since the last control do not see frequent control with certified weights covering the working range as a big deal.