baseline drift on CM5 chip during first few injections

Hi all,
I recently switched from using a SA chip with biotinylated protein to a CM5 chip and amine coupling my protein.
My protein loads well onto the CM5 chip but when titrating my control peptide I see baseline drift during the first few injections of the compound and then around the 4th injection I start to get normal looking curves. 
Normally this peptide binds very well during the SA chip experiments so im not sure what is going on, any suggestions?   

Baseline drift is usually a sign of non-optimal equilibrated sensor surfaces. Drift is often seen directly after docking a (new) sensor chip or after the immobilization of the sensor surface. It can be the result of the wash-out of chemicals used during the immobilization procedure or the adjustment of the bound ligand to the flow buffer. Therefore, include at least three start-up cycles before injecting the analyte. The start-up cycle is the same as the analyte injection cycle, including regeneration when necessary but without analyte. Including double referencing should compensate small jumps and drift.One comment on the association phase which seems initially very fast and then some slower binding?
Kind regards
Arnoud