When BeckmanCoulter's XLI system was launched in the early 1990's, digital photography was not yet common,
and besides the limited scope of available CCD sensors themselves there were also limitations in regard
to transfer speed, frame acquisition, and file size – on a Windows 95 computer... It was regarded sufficient
to capture a 200x2000 pixel image, showing only four fringes in the center of the interferogram, and indeed,
as all fringes are parallel, they contain basically the same information. Fig. 1 shows a typical interferogram
made with an XLI.
However, there is a great advantage in capturing more than a few fringes: as all fringes are redundant, noise can be minimized drastically by using more fringes. This can be understood by the mechanism of image evaluation: the only information required is the phase shift along the radial axis. This is calculated via Fourier transform of all pixel columns. Looking vertically, the fringes represent a squared cosine function (enveloped by an intensity profile that is not apparent within the small range shown in Fig. 1, but well in Fig. 2). Fourier transform will isolate the frequency of that cosine function (often referred to as "ppf" – pixels per fringe) and yields the phase shift to a reference point. For each pixel column, one phase shift value is obtained as a single value – alltogether, they form a line of fringe displacement in units of one fringe width. Considering that there is noise in the image, it is apparent that Fourier transform gains precision as to the exact shift of each pixel column the more vertical fringe range is available.
Aida will capture more than 50 fringes on a KODAK 4008x2672 CCD array, as shown in Fig. 2.
In addition to capturing more fringes in the vertical direction, Aida's sensor also features a higher pixel resolution – and it is broader, allowing for a larger magnification factor. The optics were recalculated in order to exploit the full width of the sensor, resulting in a 2.1fold magnification.
Both sensors are shown in Figure 3.
A high level of data redundance considerably enhances the signal/noise ratio and allows for measurements at extremely low concentration. However, Aida's enhancements also entend the accessible concentration range towards high concentrations: extremely steep gradients can be resolved, and due to the large number of fringes, Fourier transform will succeed also for faint images and poor contrast, as we encounter within steep gradients. An example is shown in Fig. 4.