I assume that the interconnections between the two components were via unbalanced (RCA) cables, since the preamp does not provide balanced outputs. I see that the amp has both unbalanced and pseudo-balanced inputs. I presume that there were no jumpers or other connections on the XLR inputs.
This is just a shot in the dark, for which I can't formulate a precise explanation, but perhaps the problem would not have occurred if the amp's inverting input on XLR pin 3 had been grounded (e.g., with a jumper connecting that pin to XLR pin 1).
With the original 100K input impedance, the unconnected inverting XLR input would have more closely approached a "floating" condition, compared to when the input impedance was subsequently changed to 47K. That would perhaps have made it more likely that the amp would behave in an uncontrolled manner. Without detailed design information on the amp it is probably not possible to say what that behavior might have been, but an oscillation of some sort seems conceivable to me.
The different (and probably lower) output impedance of the NAD that you temporarily connected would have changed (and probably reduced) the impedance between the non-inverting input signal path of the amp and ground. Again, without having specific knowledge of the design of the amp's input circuit it's hard to explain precisely how that might have prevented the oscillatory condition, but it seems conceivable to me that the changed circuit conditions might have had that result.
Regards,
-- Al
