If you do a google search on XLR-to-RCA adapter, every pin-out diagram I have seen shows Pin 2 to signal pin on RCA and Pins 1 and 3 to the GND shield on the RCA.Yes, most of them are (incorrectly) designed that way. My understanding is that Cardas adapters are one exception, which correctly leave pin 3 unconnected on adapters that have an XLR-female connector, and therefore would be used to adapt XLR outputs to RCA inputs. Shorting an output signal on pin 3 to ground, as I see it, accomplishes nothing whatsoever (apart from in certain situations involving transformer-coupled outputs; I presume that's why it's done), while creating the potential for adverse sonic effects on some output stages (as illustrated in the thread I linked to).
As I indicated previously, though, pin 3 SHOULD be shorted to ground on adapters that have an XLR-male connector, so that pin 3 on the XLR input they would presumably be connected to is not left floating.
The way I connected it up originally was just using the standard RCA outputs since my Simaudio integrated amp has only RCA inputs (so just used RCA cables).I would point out, therefore, that the change you implemented actually changed three things at once:
1)You corrected the absolute polarity of the system (which per my earlier comment won't necessarily be better for many recordings).
2)You changed the output of the DLIII that is being used from its RCA output to its XLR output, which may have even called into play a different output stage.
3)You changed the basic design and internal physical configuration of the interconnect cable that is being used.
If the improvements you have noted can be heard consistently across a broad range of recordings, my suspicion is that most and perhaps nearly all of the improvement is attributable to some combination of nos. 2 and 3.
In any event, I'm glad it worked out well.
Regards,
-- Al