Streaming Data Correction

The internet protocols grantee perfect transmission of data. This post will be read anywhere in the world bit exactly as I wrote. Same is true with images and videos I email or download, they will arrive with the exact same bits I sent. But is this true for streaming audio? While the internet bandwidth is sufficient for the DAC and switches to assure all bad packets are corrected resulting in the DAC receiving the same bits sent by Tidal, Qobuz or my ROON server, does this actually happen? There is a blog on Mojo Audio where the writer implies that some DACs will do proper packet inspection and fix bad packets, or request resends if needed. But the writer also states that not all DACs do so and those DACs play back what they receive, good or bad. Does the DCS streamers perform proper packet checking to assure the bits are error free?

The writer should also explain exactly how this can happen in the context of the protocol. Subsequently, the above writer should explain why a streaming DAC producer would choose to give up on a bit-perfect capable technology at a negligible cost for an uncertain one.
Maybe the blog entry was trying to explain why the new Krikkit III Ethernet cables by Golgafrincham Inc. are so important and why you should part from your money to purchase them at some exorbitant price?
Or perhaps the are preparing to launch on the market the Audiophool (whoops) Audiophile-ready Internet connection on a dedicated cable from Qobuz /Tidal to your house?


The blog was on Mojo Audio and his motive was, I assume, to imply their products are better. Here is a link:

Every Streamer/DAC which has an Ethernet port (and gets an IP address) operates with a standard networking stack; Ethernet at the Physical and Data link layer (layers 1 & 2 respectively), TCP/IP at Network and Transport layers (3 & 4), and the Application layer above that.

This standard networking stack has error detection and/or correction in combination at the various layers. There are no Ethernet/IP capable Streamers/DACs that do not have error detection and/or correction built-in. Regardless of whether the Streamer/DAC uses UPnP, or RAAT (Roon), or NAA (HQPlayer), or HTTP directly (like Spotify Connect), they all operate over this standard networking stack.

In the case of dCS, the Rossini and BartĂłk have this networking stack built-in on the streaming boards within the chassis, and in the case of the Vivaldi, the networking stack is within the streaming board on the Upsampler.

Hope that answers your question :grin:

1 Like

Thank you for the link. If the blog owner is so sure that there is a lossy version of some ethernet protocol, all he needs to do is measure it: take the “unchecked” stream from the ethernet card (there must be a backdoor somewhere I am unaware of) and send it to a file on a hard disk. At the end, compare the stored data with the original data and compute some statistics on the errors you find. Rinse and repeat for a series of different cables, power units and all that.

By the way: if you find errors, then explain why you should not use the regular bit-perfect protocols over the Ethernet.

He says that you do not have time for corrections and resending because you are streaming music. Let’s do some maths. CD quality means 44,100 x 16 x 2 = 1,411,200 bit/s, which is less than 1.5 Mb/s . A 100 Mbit/s home Ethernet connection will carry more than 60 times this amount of bits. A 10$ cable will carry these data for 10 or 50 meters without errors, but if an error appears, there is plenty of time to correct it via the 32 bit CRC code embedded in the Ethernet frame, or to ask for a resend of the offending frame. A 3 minutes song will be transferred in less than 3 seconds. At that point you have a different problem: not to saturate the input buffer.

Asking for a frame resend is virtually real-time: an Ethernet frame under the 802.3 standard carries from 368 to 12,000 bits with 336 bits overhead, communication between network cards in a LAN is a matter of milliseconds, one frame takes less than that time to be sent again if found unrecoverable.

All this does not imply that cables are useless, that power units don’t matter and so on. All I am saying is that packet loss or unrecoverable packet deterioration can be measured, so I will believe it when I see it. Cables and other stuff may be important to avoid injecting noise, and that can be tricky to measure; but if they alter the data packets they are carrying, that’s easy to see.



Thanks for the detailed reply. I agree with you 100%. I also agree that if the device is made correctly the ethernet cables and power supply should not matter. Only grossly poor products and cables should degrade streaming performance. Which makes me wonder, are all the people that claim ethernet cables, switches, and all the assorted peripherals used for streaming are improving their sound quality, or are they experiencing a placebo effect? And as you said, the companies that design the DAC know with 100% certainty if the are creating data to replace bas packets or using the streamed bits with no errors. Which is verified when the DCS staffer posted that ROON vs Mosiac are identical.

I can’t answer the question about cables, switches, regens and all that.
That not all cables are the same is confirmed by dCS when they say that you should use an unshielded cable conforming with the standard. This means that if you use a shielded cable you may be injecting noise into your system. Again: no bit loss, noise injection. The same goes with USB: the RJ45 input is clearly superior and that’s not a mystery. Maybe there are other ways in which noise can make it into our system via an ethernet cable? Who knows? There is galvanic isolation at one point, but noise is subtle and inventive, so never say never. I surely want to hear a HUGE difference before I pay big money for an Ethernet cable, that’s for sure.

Meanwhile, I found out that the Geistnote clock cables (less than 70$ each including taxes and transport from the USA to Italy) make my Rossini + clock sound much better than some other 75ohm cables in the same price bracket I was using. Much better is the key here. Not just a bit better, sometimes, with some kind of music. No. MUCH better. So yes, I believe in some cables, have an open mind, but I am not going to believe any marketing gibberish.