No binding observed to domain

Hello,
I have been immobilizing a full length biotinylated transcription factor to an SA to measure binding affinity to small molecules. Recently, I have tried to switch to expressing only the DNA binding domain of the protein in order to improve my theoretical Rmax as well as the yield of the protein. My domain is biotinylated with a short, rigid linker between the domain and the avi-tag. I was able to immobilize the domain with a response of 10,000 RU, giving a theoretical Rmax of ~180. However, the maximum response I obtained was ~5 RU for small molecules which I know are strong binders to the full length protein and which we know bind to the domain of the protein. The binding curves were also indiscernible between binders and non-binding controls (non-binding to the FL protein). I think it is also pertinent to note that the domain was DNA binding as seen in an EMSA, so I am assuming it is folding correctly before being immobilized. I am wondering what steps I should take in order to troubleshoot this problem. The protein and small molecules used are proprietary so I can't give too much information, but are there any general issues I should be looking out for that could be preventing binding to the domain? Thank you in advance. 

It is always difficult to pinpoint the reason why an interaction does not work. From the sideline i can give you only some suggestions. Immobilize the single domain and full length molecule side by side at the same relative density (is immobilisation level compensated for size). Then run your compound over both in the same experiment. I looked up the EMSA and I am wondering how different the buffers are between EMSA and SPR. Could this give the difference in (non)binding?