Yes ISIS is not very good for comparing and examining spectra. I usually go to VSpec for this but I expect BASS also does this better than ISIS.
This peak in the instrument response can be a problem to correct for accurately for even using a A or B star. Here is the shape of my ALPY instrument response as generated using ISIS (ie including a correction for the lamp black body curve)
The key is to take care to remove the residual Balmer lines and not to over smooth when generating the instrument response. You can see the hump in the problem area in my instrument response here. It is important it is reproduced accurately in the instrument response. If it is flattened out too much when generating the instrument response it generates a hump in the corrected spectrum.