About System precision test procedere

<p>This posting refers to Waters Acquity UPLC System Installation Checklist (page 20 to 25).</p><p></p><p>Greetings :-)</p><p></p><p>Well, on 05. August this year I had gotten a new UPLC installed (uh, the english time grammar isn't my strong side, is it).</p><p>This was not "my" first UPLC, but it was the first time I was a little "wondering" about the test.</p><p></p><p>Lets view the uplc as a formula 1 racing car (or a dragster or something even more faster). It is very very fast and it could deal with a lot of pressure. <span __jive_emoticon_name="happy" __jive_macro_name="emoticon" class="jive_macro jive_macro_emoticon" src="/4.0.8/images/emoticons/happy.gif"></span></p><p>And then there is the installation test procedere, where it is stone written to use a 50x2,1 mm column and with the test application the average pressure is less than 6000 psi (iirc). Maybe it is a little above 6000 psi, but that doesn't really matter in this case.</p><p></p><p>The interesting part about the pressure is the simple fact that solvents were getting comprossed above around 4500 psi. And the higher the pressure the more compressible the solvents became. Once I run into trouble when I validated an uplc system under glp and meassured 0.945 ml/min instead of the adjusted 0.900 ml/min (at operating pressure high above 6000 psi). Had some discussions with the qm about the test. <span __jive_emoticon_name="laugh" __jive_macro_name="emoticon" class="jive_macro jive_macro_emoticon" src="/4.0.8/images/emoticons/laugh.gif"></span></p><p>Coming from this corner I view the installation test procedere like driving the formular 1 with wheels from an ordinary VW Beetle, only usefull to collect the daily milk from the shop (but I guess this is something nobody would do in generell). Okay, the uplc normally doesn't has to break within 2 to 3 seconds from over 100 km/h to full stop. And a formular 1 racing car is most likely not the lc of choice to seperate 100 peaks within 3 minutes. And both are most liekly not the first choice for shopping purposes.</p><p>But the accuracy of the injection volume isn't getting better when the injection valve is switching from (ultra)high to atmospheric pressure and back. One can view this like the wave of a little tsunami in the injector block. Rheodyne designed new valves to handle the challenge. And if someday an hyper-ultra-lc system is designed for beeing used with 1 µm particle I think that building an autosampler which could accurate inject volumes like 0.05 µl is more challenging than designing the pump or the detector.</p><p></p><p>Right now there is iirc only one manufacturer of u(h)plc systems who is specifying the injection accuracy of its system at 1000 bar (appr. 15000 psi) - make three educated guesses who it is not and the first two "Waters" won't count. <span __jive_emoticon_name="cry" __jive_macro_name="emoticon" class="jive_macro jive_macro_emoticon" src="/4.0.8/images/emoticons/cry.gif"></span></p><p>Sure the Acquity normally handles reproducibility tests at 14000 psi without sweeting at all - but I am afraid such a test is out of spec.</p><p></p><p>So how about having a second installation test at appr. 13000 to 14000 psi with 0.5 µl injection volume and maximal allowed tolerance of 0.5% at 6 injections?</p><p>For me and a most of my colleges such a test would realistic represent our daily work with the uplc.</p><p></p><p>Andreas</p>


  • lizh


    Firstly, we wanted to give you a little background on the reason for the election of the test, it is really to make the testing convenient and “testable” in the lab.

    Because ACQUITY (and similarly  the Alliance  system) is designed to deliver accurate flow at pressure, the solvents are compressed to the operating back pressure and then metered. The solvents are much denser at the high back pressures and so the volumes are larger when the solvents are allowed to "decompress" and return to atmospheric pressure.  The solvents, literally, expand and create a positive bias to the volumetric measurements.  The higher the back pressure, the more pronounced the bias.  One thousand psi is about as high as the back pressure can go before the bias would make it seem that we were not meeting our specifications at atmospheric pressure.  This is why we measure our flow accuracy at 600 psi.  We do have a leak test that we run at high pressure to assure that the leak rate is acceptable.

    Briefly I am told by RDE we considered offering a "look up" table to describe the expected measurements at higher back pressure.  We have considered this, but water is about the only fluid understood well enough for us to have confidence in such an approach. Plus, I am sure it would be regarded as a “fudge factor”. Was the competitive system offering this as a specification or as an installation test?

    I do not see any relationship between backpressure and injection precision or accuracy reported from the systems that we have in marketplace – Does anyone here?.

    Early on users kept their pressures more conservatively, but not more frequently the systems are being run as fast as possible. That said, we could not find a really good set of data that specifically compared precision at highest pressure vs. a lower pressure. When I have the data generated we will post.

    Finally, your request for a testing precision at higher pressure is not unreasonable as an installation test and we look into this farther. However, to set your expectations it would not be conditions that we would wish to promote in qualification testing, where we want the methods to be as redundant as possible.

    Liz and the ACQUITY Team