In my lab I have a Waters HPLC/diode array/MS system running MassLynx/FractionLynx 4.0 in the following configuration:

1. Operating System: Microsoft Windows XP
2. Waters ZMD 2000 Mass Spectrometer
3. Waters 996 PDA/Waters 600 HPLC Pump
4. 2 Analog detectors - ELSD and fluorescence
5. Waters 2700 Autosampler
6. Gilson 215 Fraction Collector
7. Dell dual Pentium 3 processor computer,
Processor speed = 1.4 GHz
RAM = 2 GBytes
Bus Speed = 133 MHz
8. We are not using the Waters Millennium software in this application.

In a typical preparative HPLC run, the 2700 autosampler draws 5 mL of a pre-processed solution of a crude extract and injects onto a semi-prep C18 column. Through a series of flow splitters, appropriate volumes of the column eluent is directed to the 4 detectors, with ~95% of the flow directed to the Gilson fraction collector. The
Gilson fraction collector is set up to collect
88X8 mL fractions each of 35 seconds duration. The delay between the Mass spec signal and time of elution is 18 seconds. There is a 3 minute solvent delay before we initiate fraction collection.
The trigger for fraction collection is an ELSD analog signal threshold >0. Since we set the ELSD to a baseline of 10 intensity units, collection should be continuous as long as the ELSD signal never dips below zero. Collection should
halt once the 88th well is full. This is a time-based collection method.

So that's the system and the method: Here's the problem: Once a run has started, the Gilson waits its 3 minutes, receives the analog trigger from the FractionLynx and initiates fraction collection. But before the 88th fraction is collected, the Gilson randomly stops collecting fractions. An error message pops up that reads “FractionKernel error”. We did not experience this problem when running the hardware under
Windows NT/FractionLynx 3.5. It has only just occurred following the upgrade to 4.0, and it is not now possible to go back to the old system.

We have frequently been in contact with Waters technical service and tried some of their suggestions such as disabling Dr Watson
and using other triggers such as ES+ TIC>0, but to no avail. Their most recent suggestion is a bigger, faster computer with a new, proprietary board, a faster buss and more memory.

Has anyone else experienced this problem? Even better, has anyone solved this problem? Any input would be greatly valued. Thanks for the help.
Tom McCloud in Frederick, MD