The only thing I needed to do to the cable was straighten out the twist.
Hum... I wonder if you just 'got lucky' with the Chinon FZ-357 (the additional jumpers you mention are not available on most drives, and may be important).
I haven't tried to 'untwist' my cable, but my cable has 4 connectors, so there's 3.5" and 5.25" before the twist, and again after the twist.
Should amount to the same thing.
Also, I wonder if untwisting the cable would have much of an effect, aside from swapping the meaning of the drive select jumpers.
That seems to be pretty standard, so I don't think Commodore deviated from that. If they did, then the drive select wouldn't have worked.
The idea with the twist is that all drives are configured as 'drive B:', and the drive that is connected after the twist is actually drive A: in practice, because the lines are reversed by the twist.
As far as I got with my experiments, hooking up drives before and after the twist, changing drive select jumpers etc, everything seemed to work as expected.
The TTL/C-MOS jumper in particular caught my attention: that seems to indicate different voltage levels for signaling.
Could it be that the Commodore controller uses a nonstandard voltage, which the drive logic does not respond to? Except for selected drives... apparently Chinons are what you often find in these Commodores. I'll have to check my Amiga drives again, I think at least some of them are Chinon, but I don't think they are FZ-357s. They are from A600s/A1200, so that is newer than the PC10/20-III and also the A500/2000 models.
Perhaps Commodore had already moved to a more standardized drive interface by that time...?
But who knows, some of my drives may have a TTL/C-MOS jumper, or at least some other jumpers I can experiment with.
Well, some areas I can explore