ND555 - RAAT vs UPnP

Roon RAAT versus UPnP

Is there a difference in terms of sound quality between RAAT and UPnP?

I just attached to my home network this little server running RooUPnP.

I can now stream to the ND555 from two sources :

  • from a dedicated server running Roon Rock (like a Nucleus)
  • from a Raspberry Pi running RooUPnP

The Roon Server feeds the ND555 using the RAAT protocol

The Raspberry Pi feeds the ND555 using the UPnP protocol

Like so :


[Roon Server]—[CISCO3560]–[EtherRegeg]—[ND555]

UPnP :

[Roon Server]
[Raspberry Pi]


My Soulution 760 DAC internal network adapter, which is minimalistic, supports the UPnP protocol only.

I’ll use the Raspberry Pi as a bridge: RAAT to UPnP.

Will I lose something in terms of sound quality?

So far, no difference.

It makes no difference through my ND555.

But I only listened to a few of my reference tracks.

I’ll continue testing during the week. My best half seems interested, which is somewhat unusual, so she’ll be part of the game.


yes, Raspberry Pi is the perfect front end for Soulution amps.

Neither the Raspberry Pi nor the Roon Core will front the Soulution or is fronting the ND555.

The EtherRegen is.

I tried to do this earlier in the week, albeit using a ultraRendu as a UPnP bridge and ran into problems. Just found a setup guide in a past post so will give it another try.

Was the UltraRendu directly connected to your streamer/DAC ?

If so, then it is a very different setup.

The only device directly connected to my ND555 is the EtherRegen (+optical fiber).

1 Like

No it’s connected to the same Ethernet switch as my Roon core and then via the same network route to my NDX2

1 Like

Perfect! :+1:
We are then two making the same little experience :smiley:

Let us know your findings!

1 Like

Another question lying behind this little experience is: does the quality of the server, when connected to the home network, matter at all?

I have already compared my pimped little NUC server + LPS against my main extremely noisy PC. No difference at all.

Now the El-Cheapo Raspberry Pi with its even more El-Cheapo switch mode power supply acting as a RAAT_to_UPnP bridge.

I suspect that what makes a difference is what is directly connected to the streamer/DAC : the switch.

But I’ll leave it for a week or two.


Will do :+1:


Hi @Thomas, I had used both the RooUPnP and the Sonore UPnP bridge (currently using the Sonore) to stream from Roon into my NDS. This is the way I roonified my NDS since it doesnt support the RAAT protocol. They both sound the same.
For compering whether I was loosing some SQ. I did a subscription to TIDAL (I am only using Qobuz now) to bypass the bridge entirely since NDS supports Tidal. All sound the same. No difference.
My conclusion is that all the bridge is doing is switching the protocol from RAAT to UPnP without touching the audio signal. I am happy with it. As long as the NDS is roonified for using Qobuz no rush to buy a ND555.
The creator of RooUPnP is very active in the Roon forum. You can chat with him for technical details.

1 Like

I thought this for a while until I switched from my regular Roon server to using a NuC i use to run Plex as I was doing some manitenance on the usual server. The normal server is nothing fancy just a fanless PC from Tranquil PC in a custom milled case of their design, its not a nuc just some other motherboard, no audio tweaks just a ZeroZone LPS. I noticed the NuC sounded a bit harsher. It wasnt night and day but I noticed it as soon as I played a song I was very familiar with. Now the nuc lives in my data cupboard which is full of noisy equipment sound wise and electrically. Modem, router, main switch that feeds all the other rooms and switches, electricity circuit breaker, my Roon server lives in my dining room and at the time fed a PI that was connectd to my RME DAC for headphone listening. It was on this system I noticed the difference, never got to use my Naim so cant say if it affected that.

1 Like

These are my questions too and I come to it with a preconceived notion the same as yours, that the server hardware located several switches and dozens and dozens of feet of ethernet cable away from the player matters little or none. And that what’s directly connected to the player (a decent hi fi switch) does matter the most.

It’ll not be answered other than by our own ears, as the reports run the entire spectrum. If the owner of my local hi fi shop uses a Roon Nucleus and Roon (but not wi fi; it’s all wired) to feed the $750,000 system in his main demo room, it’s because he doesn’t hear any issues. He’s told me this.

Hi Bart,

I spent the last couple of hours listening to my favourites tracks.

It is as good as it is with my fancy NUC.

My subjective perception is there isn’t the slightest difference.

The networked device (Raspberry Pi or NUC) doesn’t make a difference, nor does the protocol (RAAT or UPnP).

And this makes sense!

RAAT lies on top of UPnP (see it as a pimped version of UPnP).

Both RAAT and UPnP rely on TCP, which guarantees data integrity.

Anyway, I’ll have fun with my wife at comparing Raspberry Pi vs NUC.

I’m not stubborn; if she hears a difference. I’ll investigate a little further.

I even renamed the devices, so she doesn’t know from which device she is streaming. :grinning_face_with_smiling_eyes:

1 Like

But one question remains.

Would a 30K hifi server, connected directly to the DAC, sound better than the 760 internal network adapter.

At some point, I’ll do the test.

I discussed the matter with Cyrill Hammer. The answer wasn’t clear cut as seem to believe some folks on other forums.

But when I mentioned the fact I intended to use the 760 internal network adapter, I was asked about my switch.

The answer is no, as they are or can be actually effectively the same down on the wire, however there may be, and indeed I can detect very subtle differences between different servers, whether they be UPnP media transfer or RAAT media transfer. Also I find again subtle sonic differences can vary on different clients.

So really the comparison can only be effectively evaluated with server, transport and client combinations.

1 Like

there are differences but like others I cant blame just one specific cause and blame just UPnP, HQplayer/NAA or RAAT. just use the one you enjoy.

Is there a difference in the quality of the clocks in the R-Pi vs the Dedicated Server?

Would this make an audible difference if they were the last component before the ND555?

But in your system there are 3 more components where the signal is potentially checked and re-clocked.

Thus any difference between the quality of the clocks in the R-Pi vs the Dedicated Server may be eradicated by the signal processing of the 2 switches and the streamer/DAC itself.

FWIW in my system in almost all cases the switch was the last in the link to the (Ethernet connected) streamer, an ND555 in my case, and differences between different servers have always been audible, sometimes dramatically so.

When comparing the Asset UPnP music server vs. Roon I cannot recall having heard significant differences between UPnP vs. RAAT with Roon on the same QNAP NAS, although I was more focused on what Roon brought to the party in other ways.

I did, however, find the Nucleus+ sound quality more to my liking - even more so when swapping the bundled SMPS for a Sean Jacobs DC3+. As for the Innuos Statement that made a much bigger difference.

I also found the switch makes a difference - relatively subtle with GS105 and medical power supply, a bit more with the Cisco, and quite z bit more with the EtherRegen, with which a good LPS again notches things up further. The PhoenixNet is more impressive still, but I wouldn’t say it brings more to the party than the Statement downstream.

So, in short, @Thomas , I don’t remember hearing any difference between UPNP and RAAT. In fact I heard more of a difference between music server software (Asset, Minim and Twonky) but I have found significant differences in the sound quality of music server hardware, from NAS to NUC/Nucleus and Statement.

The main difference is that Thomas uses optical cable - I think it’s from Cisco to ER, iirc.

So that cuts upstream electrical noise, but presumably adds artefacts connected with the conversion to and from fibre mode.

So the only electrical noise coming into the ND555 is what the output of the ER and the ethernet cable themselves generate.

Plus any potential clocking issues that may occur anywhere along the chain, unless these are corrected.

So one has to choose his poison - fibre conversion effects or electrical noise management.

That’s a good point. I’d actually intended to mention I had no experience of optical but then forgot to. Thanks. It will be interesting when someone makes a fuller comparison.