One Calibration company in particular has provided me with some great background info. I don't think it is improper to post a document he sent me as long as he gets his name in print.
Please let me know if I am out of line by posting thisKenneth Bonnel has written a number or studies regarding Pipettes and I believe he played the lead role in writing and set up of Troemners Metrology SOP's. The table of data included with his report may not come out in this post very well but it's pretty easy to understand the final conclusions. While I understand some companies only calibrate because they are required to. I just don't feel satisfied with collecting numbers to have on file. Why not do it right and get real data?
Pipette Calibrations – The Correct Number of Test Volumes
Kenneth Bonnell
President
Alpha Omega Calibrations
kbonnell alphaomegacal
December 2002
Abstract
There has been much debate about the correct number of volumes that should be tested in order to accurately characterize the performance of any liquid handling device. Some manufacturers define the performance of these devices at only two volumes while others publish specifications that detail the performance at three or more points. In order to deduce a statistically justifiable answer, multiple pipettes must be tested at the two-end volumes (high and low) and then retested incorporating the center point. A gravimetric analysis will be performed using the procedures outlined in ISO 8655.
Introduction
There are numerous manufacturers of pipettes, which produce hundreds of different makes and models. The performance of these instruments is tested using different methodologies and employing different techniques, which makes a specification comparison between devices extremely difficult. Due to this variability, some manufacturers only define the performance at one or two points while others characterize the performance at three or more volumes. Obviously, this poses several problems for any metrologist interested in calibrating these devices and reporting meaningful data.
There are many calibration companies that simply follow the manufacturer’s recommendations, while others are uncomfortable reporting only two (2) volumes simply because the adjustment is based on only one other point. These devices are manufactured and reported to be linear, so it would seem to be essential that they are assessed at three points. This is due to the fact that the metrologist performing the calibration would have two (2) other points to reference while performing the calibration and, of course, two points determine a line.
Experimental Design
In order to validate this assertion that three volumes are a requirement in order to accurately characterize the performance of any liquid handling device, the following experimental design was employed.
- Gravimetric assessment according to ISO 8655
- Grade 3, bi-distilled, degassed water
- Standardize on only 1 technician that has been properly trained and proficiency tested
- All testing will be performed using the forward mode
- All pipettes will be assessed using a 10-10 measurement structure initially and then reassessed using a 10-10-10 measurement structure in order to incorporate the center volume
- All pipettes will be assessed 10 times by the same technician in order to collect a sampling of at least 100 aliquots per volume
- The Z-factor correction will be calculated using all of the requisite environmental inputs as well as the appropriate cubic expansion coefficient
- All instruments are new, manufacturer’s tolerances will be utilized
- The addition tare method will be employed
- The instruments will be allowed to thermally equilibrate to the laboratory for at least 4 hours before any data is ascertained.
- A wide variety of instruments were selected
o Different manufacturers
o Different models
o Different volumetric ranges
- The center volume was calculated using a linear extrapolation technique for any instrument defined at only two points.
All of the pipettes were initially tested at only the low and high volumes and adjusted if any of the instruments were out at either volume. We calibrated these instruments towards the center of the tolerance band, so that when the center volume was tested it should have passed. The pipettes were tested and adjusted multiple times using the aforementioned experimental design in order to provide at least 100 aliquots per instrument and volume, which, of course, offered enough data points in order to statistically justify the results.
Data
Mfg Model # of Times Tested Adjusted (Y/N) 10-10-10 Failures Failure %
Biohit Proline 0.1-2.5ul 10 Y 2 20%
Biohit Proline 2-20ul 10 Y 1 10%
Biohit Proline 100-1000ul 10 Y 2 20%
Eppendorf Reference, 2-20ul 10 Y 1 10%
Eppendorf 10-100ul 10 Y 1 10%
Eppendorf 100-1000ul 10 Y 2 20%
Gilson P2 P2 10 Y 4 40%
Gilson P20 P20 10 Y 1 10%
Gilson P100 P100 10 Y 2 20%
Gilson P200 P200 10 Y 3 30%
Gilson P1000 P1000 10 Y 2 20%
Nichiryo Nichipette 2-20ul 10 Y 3 30%
Nichiryo Nichipette 20-200ul 10 Y 5 50%
Nichiryo Nichipette 100-1000ul 10 Y 2 20%
Mean = 22.14%
Standard Deviation (based on experimental design, 100 aliquots per volume) = 1.27%
Uncertainty (k=2) = 2.54%
Analysis
The data mean in terms of failures was 22.14% with an uncertainty of + 2.54%, which, of course, gives us a 95% confidence interval in terms of the results. This means that a pipette that was adjusted at the low end and high end using only the other point as a reference will normally fail when assessed at the center volume between 19.60% and 24.98% of the time.
Conclusion
It is abundantly clear that all instruments should be assessed using at least three (3) points, since there is at least a 1 in 5 chance that the instrument will fail when calibrated using only two (2) volumes. This is simply due to the fact that these devices are designed to be linear and it takes at least two (2) reference points to adjust the instrument properly besides the one point that is being tested. The data clearly substantiates and supports this supposition.
