Hi All, just joined the club after moving on from a Dave + Mscaler + USB conditioner + Streamer with fewer LPS and the spaghetti I had to deal with and loving the sound and simplicity now. Would like to take the sound up another notch and I have a question…
From most resources I read…ethernet seem to be the best input. I have the chance to build a top notch music server from scratch based on SOtM’s offering from the motherboard to nic card to usb card and obviously clocking them from a single master reference clock. I am of the opinion its not going to benefit me if such a build will function only as a roon core since and sending files to the Bartok via ethernet.
The question is what if you have a really good music server serving local files + streaming from Tidal like the one I am thinking of building and connecting it into the USB input of the Bartok? Will this provide a better audio experience compared to ethernet on the Bartok? Frank comments appreciated. If I’m dreaming tell me lol.
Currently using the ethernet input of the Bartok and have my roon core installed on an off the shelf Intel NUC powered by an LPS.
Disclaimer - I’m really happy with the sound now and the question is really in the interest of experimenting and learning from others here.
Hi dstubked and welcome to the forum. I hope that you will enjoy it here.
As you seem to have discovered just about all experienced users of dCS processors will advise that the best sound quality is found via network connection.
History of dCS products shows that Network Bridge was originally intended to include a USB output. However dCS engineers found the fundamental engineering of USB did not allow the level of quality that they required. Time was spend considering re-engineering the USB device but, in the end, it was decided that the time and expense of so doing would be such that it was uneconomic and so the promised USB output was abandoned. Bear this historic fact in mind when making your decision. There is or was an FAQ on this in the archives here but I am currently unable to locate it.
I commenced using a USB based streaming system with dedicated computer, USB interface conditioners etc. but the change to ethernet streaming and playing local files was very significant in terms of improvement that I would not revert. Incidentally I stream Qobuz rather then Tidal.
If you can get to hear the difference between the interfaces then I would recommend this so that you will not have any doubts. If you can use a good network streamer/NAS ( I use Melco) for ocal files then you will probably get the best performance from the dCS Bartok. Bartok’s internal streaming of Tidal is already round about as good as you can get at that level.USB will be “OK” but , perhaps, less than you may ideally hope.
Thanks Pete, makes alot of sense and glad you shared your comments.
In the case of your Melco server, you are just streaming files off the network and not connecting it via USB?
Also, I am wondering if it will be an overkill if I build a custom server and connect it directly to the ethernet of the Bartok via a direct connection without a network switch and clock the ethernet output of the music server with the Bartok’s ethernet via the same master reference clock. Any thoughts on this one? Won’t touch the USB in this case.
Hi Phil, yes, direct without a switch and of course will make sure the music server comes with dhcp to provide the IP address…
What I am trying to do is to feed a high quality stream to the ethernet port of the Bartok since this is its best input as I understood from reading in this forum. I am trying to understand if making sure the NIC of the music server and the Bartok (2 x Mutec Mc3+ clocked from a 10mhz master clock) will provide sonic benefits if they are being clocked by the same master reference clock.
I also read that no matter where the data stream is coming from, whether its USB or Ethernet, once it gets received by the Bartok, the signal will get reclocked anyways. I am not sure if my understanding is correct but assuming this is the right understanding then it is really meaningless to do what I wrote above and I should just focus on getting data to the Bartok via Ethernet and let the Bartok do its magic…
Unfortunately living in a house with a garage is a luxury in Singapore for only a lucky few so I got
to have my NUC located next to my DAC over here
Anyways, are you saying ethernet is just ethernet and it doesn’t matter how it will be received by the Bartok and I should probably put the NUC as far as possible so it does not pollute the signal chain?
Remember that it is very rare for music to actually be streamed in real time across a network … even with Roon the audio is sent in asynchronous bursts (with accompanying timing data) rather than as a synchronous stream of data.
Applying an external clock to a servers network port would certainly seem to be an awful lot of work for no practical purpose that I can envisage. Similarly, although it is possible to link both the server and Bartok directly, I would strongly dissuade you from doing so and just stick to well established and reliable networking practices … I can’t imagine how many times I’ve said “You can’t tweak yourself into a better network but you can easily tweak yourself into an unreliable one”…
Transferring audio data across a network is an absolutely trivial load for any network that you are likely to set up today unless you are totally cheaping out. Totally reliable gigabit NICs, switches and network cables are the norm and have been for some years - chanting shaman and ritual sacrifices have absolutely no place in corporate networking where even a single error in data could cost thousands or millions of pounds.
Streaming even typical uncompressed high-res audio uses less than 1% of the bandwidth of any half decent modern network and has technically been possible using “current tech” networking since 1983 (we’ve come a long way since then).
There is also absolutely no advantage (from an audio streaming perspective) to running 2.5, 5 or 10gig networks and just makes life more difficult.
That would be my way of looking at it…
What I did with my NUC was put it in a silent Akasa “Plato” enclosure so that there was absolutely no possibility of fan noise but it is still in the same stack as my audio and video kit.
Thanks Phil, this helps alot and definitely makes alot of sense!
Last question - what are your views around a high end music server like those from Aurender (N20, W20SE) with a wordclock input and having them plugged directly into the USB input of the Bartok while having the Aurender sync its clock from the Bartok’s word clock output? Will such a setup offer more benefits sonically than the Bartok’s ethernet input?
Unlike S/PDIF, USB and Ethernet are an ASYNCHRONOUS packetised data flow (as opposed to a SYNCHRONOUS bitstream) so clocking the USB source (in this case the Aurender) to the Bartoks Word clock out wouldn’t be of any (logical) use.
However, I’m sure that there will be many out there that claim that it does make a (night and day?) difference so please do feel free to try it if you wish and decide for yourself…
…but…
…the way you have described it there would also be the wrong way round too.
Even if you were running S/PDIF (synchronous audio) from the Aurender to the Bartok you wouldn’t want to sync the Aurenders word clock input to the Bartok word clock output because that would possibly/likely/definitely/maybe cause synchronisation headaches any time there was a change in sample rate between a 44.1 and 48kHz based track being played from the Aurender because the Aurdender would be being fed an incorrect Wordclock by the Bartok (which it would be trying to use as it’s clock because that is the external sync and the external sync is supposed to be its “reference master”) til the Bartok saw that the sample rate had changed, re-sync’d to the clock of the new track and then adjusted its Wordclock output to match …
Imagine tripping up a marching solder, watching him stumble a bit, gather himself and then try to get himself back in time with his squad only to find that the squad has changed time.
