From the above graphs I can see that the microrendu drops the "noise floor" by 10 - 15dB & with far less perturbations than the iFi DAC directly connected to USB - I would consider that a substantial improvement, wouldn't you?
Actually no. As I mentioned we are dealing with two completely different audio stacks in the measurements that I provided. Specifically, the default Windows player is going through the Windows audio stack with its dither component in place.
I re-tested, this time with Foobar2000 and Asio driver for the ifi DAC. Here are the results:
Green is ifi DAC connected to USB as before, using Windows audio stack.
Yellow again is the same old data with the data path now being through microRendu with my lab power supply.
The new graph in teal is using foobar 2000 using ASIO interface to talk to ifi DAC. The noise floor is now dropped by the same amount because of the removal of dither in Windows audio stack. And performance is now identical to using microRendu which likewise doesn't dither (it is a pass through device).
So in summary, microRendu has not improved anything when using it with my lab power supply. The output of the DAC is the same with or without microRendu. Using ifi Power supplied as provided by Sonore, substantially degraded performance however, as I showed before.
Objectively then, based on this data, microRendu has only potential to do harm. Again, I welcome other comments and data to the contrary.