Improvement of signal-to-noise ratio of electropherograms and analysis reproducibility with digital signal processing and multiple injections

Poor signal-to-noise ratios and analytical reproducibility in capillary electrophoresis were improved by application of multiple injections of the same sample at (pseudo) randomly chosen time moments. One order of signal-to-noise ratio improvement was achieved in practice using an in-house developed computer-controlled pneumatic sampler. However, it is demonstrated here that better implementation of the capabilities of multiple input requires that the detector signal noise be "pre-whitened", i.e. spikes, baseline drift, oscillations etc, must be removed from the detector signal before transforming it to the electropherogram in the common form. Such filtering of the detector signal was realized by implementing iterative, menu-driven software written in Matlab, a high-level programming language.