Hi Vini,
Welcome to the forum.
The most common cause of NSB is electrostatic attraction of a positively charged analyte or other component to the negatively charged binding layer. Molecules that naturally bind polysaccharides might exhibit NSB to the alginate-containing binding layer, such as lectins. It can be difficult to identify the type of NSB. A deactivation of the reference spots can lower NSB.
What you can try is to adjust the flow buffer by adding more salt (200 mM NaCl), more detergent (0.02% Tween 20) or a non-related protein such as BSA to block all non-specific interaction sites.
Also check:
www.sprpages.nl/experiments/troubleshooting
In addition, check the raw data if it is really NSB and not a processing error during the curve fitting procedure.
kind regards
Arnoud