Ok...I found at least one problem, thanks to that file. Not only was I being too sensitive, I was being *way* too sensitive. When you were binding a single axis I detected movement on all 4 axes beyond my threshold (One was 1000 times higher, and the others merely 4 times higher), so I was always just taking the first axis. See if this works. It still generates the debug.bin file, as I might need one for your d-pad or if my fix doesn't work.
<Attachment outdated>