HTML Help compiler is rather old product. It does not support Unicode even it allows that topic files use UTF-8 encoding. TOC and index files are not Unicode. When HTML Help compiler compiles topic, TOC and index files into CHM it assumes that the system code page is same as the code page of the target CHM. This might not be the case if you have a Western OS but are localizing CHM into Asian languages such as Japanese. In such case the binary index of the CHM get invalid and when CHM is viewed the index data shows up as mojibake.
when the correct one should be
To avoid this Sisulizer by default does not create binary index when creating localized CHM file that would require different code page. Not having a binary index does not prevent the user to use index, it just makes index little bit slower. If you want to turn off this default and force Sisulizer to always use the binary index (if the original CHM uses binary index) you have to uncheck Use compatible index check box from the Options sheet of the HTML Help source dialog.