Do I still need a DDC in between?

Yes, @soix, the U18 has galvanic isolation on its USB input, and so does the Micro DAC. As I mentioned earlier, the Micro DAC also allows the user to select either the I²S or the local clock, and I slightly prefer using the local clock.

The reason I'm asking now is because the return deadline for the U18 will have passed by the time the Rivo+ arrives.

Anyway, thank you for your response.

The harmony (micro) Dac is the only Dac I know so far that allows one to choose which clock to go with, I2s vs local (dac) clock.  I try both and it seems local clock option yield slight better SQ but I won’t swear to it.  Both units (U18 & micro Dac) might have similar grade of clock.  I use Ifi Ipower X for the micro Dac and I use 0.5m DHL hdmi 2.1 cable as shown earlier.

With that said, @sns, what you have said below confuses me a bit and please provide some clarification.

By the way, I have Harmony with Denafrips Gaia DDC, sync’d to Gaia via Tubulus Ximius I2S cable. 

What clock option you have chosen in Harmony Dac to sync., local (dac) or i2s clock?

This superior to optimized usb vis Sonore OpticalRendu via Audioquest Diamond usb.

This makes sense.  The U18/micro DAC via I2s also outperforms the usb (galvanized isolated) input on micro Dac BUT that is because of the weaker source (Wiim Ultra; but still very good when combined with DDC).  That is why I am thinking about the upgrade.

@audiotroy  I think Passion for Sound addresses your question.  The i2s sounds comparably, if not better, to usb which is perceived better than AES.  Did the homework.  But nobody as far as I know does the comparison with and without DDC.  Anyone? 

https://www.youtube.com/watch?v=D-fEVEdFBnE&t=317s

It is good idea to try a good ethernet cable.  Thx.  Do I also need a good switch box and / or ethernet-optical-ethernet converter there, @audphile1, you think?

I’m fine with returning the U18, but I’m fully satisfied with the Harmony Micro DAC and the Rivo Plus/LPS I’m considering. If I ever need a DDC again, I’ll just get a better unit like the SU-6 or the Micro DDC. I’m not sure the folks making that statement have ever actually auditioned these units. I know exactly where the point of diminishing returns is for my system.

By the way both DDC and dac use I2S clock, I2S is native pathway in every dac...

No.  Don’t assume that.  Go back to read your Harmony manual again regarding I2s Clock (Sec. 4.2).  It clearly says "This feature enables you to select between the local DAC clock or the I2S clock when utilizing the I2S input."  The ’local DAC clock’ option is manifest by itself.  If you opt for ’I2S clock’, the DAC will be slaved to DDC’s clock and use the I2S clock unless the I2S clock becomes unstable or incompatible. 

This is an unique feature offered by Harmony DAC and µDAC. 

Keep it simple!  I like that answer as long as it is effective.  Any suggestion?  DH Labs or other vendors that won't break my bank?

@sns  I am not trying to be argumentative but I respectfully disagree with the generalization you have made regarding:

I2S is the native digital pathway in every dac, this is the DEFAULT clock used regardless of input chosen.

"local clock is in FACT a I2S clock", "just a matter of semantics"

That directly conflicts with the statement in the manual: 'When users switch to other digital inputs, the DAC seamlessly transitions back to using the local clock.' When other inputs are selected, there is no such thing as an 'I²S clock' in that timing information is not pipelined separately from the signal data. Please be precise with the terminology. You may want to double-check with Harmony.

@mattw73  Thank you for the recommendation. I like that the cable is affordable, and Cat 6a is more than sufficient for audio applications. I’ll gladly take a $15 cable as long as it works. For the same length, a DH Labs Ethernet cable would cost ten times more (around $300). I’ll definitely give it a try.

@moonwatcher You are welcome.  It is interesting that you are curious about the Rivo+ jitter performance.  There is no measurement yet since it is just released but I found the measurement for Motivo, which is basically Rivio+ plus the DAC and headphone amp.  Their streaming circuit and performance is neck to neck according to iiwi but the Rivo+ has a slight edge.  See below.

The average (Lt / Rt channel combined) RMS (or so-called 1-sigma) jitter is around 8.2ps (I am not sure which output this measurement is taken from though). See yellow highlighted.  Quite impressive.  

@moonwatcher The 38fs is the jitter rms level in the femto CO (crystal oscillator).  The jitter level at outputs can not be that low.  Here are some measurements taken from coaxial output of various DDCs by Goldensound for their RMS electrical noise and jitter noise at 48khz.

                                  Jitter(rms)   Electrical(rms)   Jitter(peak)  Electrical(peak)

Holo Red                     73ps           2.97mV                213ps           12.9mV

Singxer SU-6              75ps           3.3mV                  225ps            13.7mV

Singxer SU-2 KTE      85ps           3.2mV                247ps            13mV

Gustard U18             453ps           5.2mV                1133ps            22mV

You could see U18 jitter and electrical noise performance is lagging.  That is why I mentioned I could pick up better DDC like SU-6 or micro DDC later on if needed.