TechEmporium: the MMC5 duty is a bug, apparently I was using the VRC6 square duty lookup for the MMC5 channels by accident.
Also, the DPCM notes start in octave 0. That's why it looks like there's none (default view is at octave 3).
Ah; I see now with the DPCM note assignment, but that MMC5 bug was quite the catch. It's pretty minor, though, considering that there's only one instance of the Vxx effect at the top of each channel in this case.
Should I try an MMC5 NSF with alternating duty cycles, or wait till you do the necessary fix?
As for multi-expansion NSFs, now I'm curious as to why the Wolverine NSF has so many if only one is in use, according to rainwarrior. And why would the importer generate a VRC7 track & instruments only? I've decided to export track 24 & found that it does work, so something odd's going on here.
_______________________
Technology: the one thing that's hated & cursed at by all engineers, technologists, scientists & technicians!
No, I don't need any additional MMC5 duty testing; I've found the bug and it will be fixed for the next version of this.
Wolverine doesn't use any of its flagged expansions. The person who made the NSF was just being silly. The importer only tries to import VRC7 because FamiTracker can only select one expansion and there's an arbitrary choice of which one if there's multiple.
As far as "hints", don't be cryptic. Post the original NSF or FTM that doesn't import correctly, describe what is wrong, and I will figure it out. I'm not going to waste my time on guessing games.
Zxx effects don't appear in imported NSFs. Compare track 1 of the NSF of TMNT 2 and the resulting FTM. The DPCM samples are a lot clickier in the NSF than in the FTM.
Edit: Whoops, the files didn't even attach. Here they are if anyone still cares.
I didn't bother to capture DMC Volume register $4011 writes, because they are primarily used for raw PCM playback. I actually hadn't run into any games that were using it for something else yet, so thanks for pointing this one out.
In this case, it is being used to set a starting PCM value on the DPCM sample; because their data is relative, they normally just start playing at whatever volume position the unit was last left in, and continue from there. Resetting it at the start of the sample isn't normally something you'd want to do because of the pop, which you've noticed in this case.
So... I will try to support this. It gets a little more complicated because I'm probably going to want to suppress picking up raw PCM playback, which uses the same register (I don't think anybody wants to hear 60hz clicking in lieu of raw PCM).
Oh, hadn't noticed you asked for the file I used! Sorry, rainwarrior!
I uploaded both files here - an example file Dave posted on the 3.7 beta thread, which tested the new FDS effects (HXX and JXX, IXX is missing from the example), expanded by me and rendered to NSF. As I re-imported it back to Famitracker, I noticed it records the pitch rather than the modulation, as you can see in "fdsfmimported".
The results, as you can see, are far from accurate. (Though they still sound rather cool.)
EDIT: I just noticed you used 0.3.6 to create the NSF importer. D'OH! Though I actually expected the file to get a lot of FDS instruments with different modulation settings, at least, to compensate for the lack of the new effects.
It's not incredible, it's just a recording of what is happening on hardware each second. There can't be any actual instruments unless the software somehow figures out patterns (or takes data from famitracker-generated nsfs).
Also this "tracking style" is pretty crazy, as in "why would you do it this way??" crazy.
) 
