MS Excel may not always be used for programmatic conversion. As a result, odd problems can arise. First things first: Have you tried directly using MS Excel to convert your file to PDF? If the problems exist there as well, documents4j is helpless.
Other than that, there might be a report-options set that is required for making the export successful. You can have a look at the VBS scripts that documents4j runs for triggering Excel conversion: https://github.com/documents4j/documents4j/tree/master/documents4j-transformer-msoffice/documents4j-transformer-msoffice-excel/src/main/resources.
These files can be opened locally and run. You must first launch excel start.VBS from the Windows command prompt. Then you execute excel convert.VBS input.xls output.pdf 999, which causes the specified file to be converted to PDF. After finishing, you can tidy up by running excel stop.VBS.
The two parameters that are interesting to you are:
excelApplication.Workbooks.Open(inputFile, , True, , , , , , , , , , , , 2)
And
excelDocument.ExportAsFixedFormat xlTypePDF, outputFile