Here's my suggestion to find out where could be the source of the problem.
Obtain these files from Flexcolor station.
(a) scanner raw image in tiff (scanner raw: RGB data in tiff before profile being assigned)
(b) converted image with Adobe RGB in tiff, from Flexcolor.
(c) scanner profile for (a)
Test to be conducted with CS5
(1) Import the scanner raw image (a) into CS5, assigned with scanner profile (c), converted to Adobe RGB
Do it for both Adobe ACE (with RelCOl) and Microsoft ICM (with RelCol and AbsCol).
(2) Import the tiff image with Adobe RGB (b) into CS5
(i) Compare images, are (1) and (2) same?
(ii) If they are not the same, is the neutral in (1) correct or still too warm? Which is closer to the "preview" in Flexcolor?
(iii) Is (c) a the custom profile, scanner generic profile, or a dedicated profile (for the negative) that comes along with Flexcolor?
Alternatively, if the files can be made available for investigation,
drop (a), (b) and (c) here. < CLICK
Do not drop (a) and (b) together if huge files are expected.