Esoteric K-03X Runin RCA & XLR Outputs Separately?

Yes, definitely. You should break-in the XLR output circuitry using the same combination of upsampling and filters (or none to your example) to ensure the XLR circuitry has the same amount of time on it as the single-ended you wish to A/B. Esoteric gear sounds its best (IMHO) with its fully balanced circuitry in use; the break-in is vital to hearing this unit at its best. Also, no-up/no-filter may not be the best you can do with this unit. See many comments on other threads by many of us who own the upper Eso units (P/D and K); I think you'll find that after you break-in the unit with XLR, you'll also want to check out S_DLY2+176 upsampling and several other combinations.

Guido's feedback is dead-on, those are the best filter settings. More warmth, try S_dly1...

Having heard the results of all available modes, I prefer 4X Upconversion and S_DLY2. 8X (8Fs) is quite good but for overall natural, musical presentation, imaging, etc...I prefer 4Fs (4X).

I do like it but for some reason the imaging is not as precise as with 4X.....will give it another try though. Same with upconversion to DSD (meaning giving it another try).

As I mentioned above, I do like the overall sound quality when the transport is set to output to an ESL3-capable DAC using Dual AES/EBU cables with UPCONV=>8Fs (352.8) (also, NoFilter on the P-02 and S_DLY2 on the D-02, Master CLK sync "MCK-IN" at 22.576 Mhaz between the two units), however, I just tested a few key tracks again using both UPCONV=>4Fs (176.4) as well as 8Fs (352.8) and found that again, I prefer the 4Fs due to my hearing alot more ambient information and the overall depth and 'thickness' of the soundstage being very real and very much "better" in every way at least to my ears.

There is a very tell-tale thing that happens when switching from 4Fs to 8Fs UPCONV, at least on the P-02/D-02, namely that on the face of hte D-02 you see it switch from "ESL3 176.4" to "DUAL 352.8". This is the DAC's way of saying the communication channel bit-depth dropped from 48-BIT(!) to the legacy 24-BIT depth AND it is no longer an ESL-linked communicaton channel. I believe the additional bit-depth is what is causing the increase in ambient information, soundstage depth, palpable thickness, etc...that I definitely prefer.