But dCS claims that the network stuff in the NB and in the Upsampler is the same…but it seems that your experience does not confirm it…well…I am still hesitant whether I should buy the clock or the upsampler…and it is not easy to find a shop where I could listen to it to spot which combination I like the most…
So Pete, you’re passing through DSD at native rates, correct? And are you finding that imagery satisfactory?
Pete isn’t the first to report that the networking solution in the Vivaldi Upsampler sounds better than that in the Network Bridge - I’ve read on another forum of people who made this exact same comparison, with the Upsampler in Clone Mode, of course, coming to the same conclusion. I wasn’t there, just saying, learning never stops…
Greetings from Switzerland, David.
1 Like
As to passing DSD through, I’d be curious what DSD64 for DSD128 actually does to the signal technically, because it sounds very much like back-and-forth upsampling/remodulation via DXD (PCM) to me, in other words, I prefer using the “DSD Pass Through” function. It also sounds better to me playing back DSD64 than if the Upsampler is set to DSD to DSD (input = output) - doesn’t make sense (or does it)?
Greetings from Switzerland, David.
Slightly off topic however Network bridges have been mentioned.
Is the network bridge hardware in the Rossini the exact same board as what is currently in the new Bartok?
Best
Gregg
There have been a few comments from dCS about this topic within a few subject threads fairly recently ( although I can’t find them this afternoon for some reason).
The most recent one that I recall said that the implementation of the network interface is better on the Vivaldi ( so that isn’t necessarily the networking board per se ) and Andrew has also remarked upon the superiority of the Vivaldi Upsampler’s implementation of network replay compared with the Network Bridge. I have to say that if one has the opportunity of hearing both in the system with which you are most familiar the difference is pretty clear.
However the Network Bridge is terrific in its own right and offers a major percentage of the Vivaldi Upsampler’s network streaming performance at a minor percentage of the cost.
1 Like
Yes, native DSD64 or 128 using the pass through option sound pretty glorious to me. No problems with imagery when I hear Rachel Podger and Brecon Baroque virtually appearing at the end of my room ( DSD128). The illusion is so strong that I almost have to say " excuse me" when I walk between the speakers to go upstairs to the loo 
.
In fact I have duplicated some Pentatone recordings that I own in 24/96 PCM with DSD64 files as Pentatone’s website implies that the original recording was made in the latter format. I find DSD superior in this instance.
2 Likes
Thank you for that however perhaps I was not clear with my question. I have a new Bartok and may be considering a change to a Rossini. I would like to know if the Network interface board in the Bartok is the same as in the Rossini (physical circuit board). I guess then another question could be if the functionality is the same between the two - or perhaps the Rossini has firmware which also changes the Network streaming performance vs. the Bartok (just like 2.0 changes the RingDac mapping and performance even though the FPGA is the same between the two).
Best
Gregg
Bartok, Rossini, Vivaldi Upsampler, and Vivaldi one all use the EXACT same network board. Network bridge uses an ever-so-slightly different one, but same exact CPU architecture.
The software is all a common code base so there is effectively no difference in the software between any of the products. There is not anything in the code which changes performance of the streaming interface depending on the device in which the card is installed.
…but… none of that matters one bit in terms of actual performance. The network interfaces all “sound” the same. The sonic differences between the products come from the differences in digital decoding, power supply, analogue stage design, etc, etc.
Rossini performs better than Bartok because it is a different product that was conceived with different performance targets in mind. The increased performance is realized through a combination of hardware, software, and physical construction.
6 Likes
Andrew
Perfect. Thank you for the detailed explanation.
Perhaps more thread drift however one more question. My understanding is that the Rossini was released in 2016 and based upon the new architecture developed with the Vivaldi. The Bartok followed at the end of 2018/early 2019. Was the Bartok a continuation of a trickle down of the same Vivaldi/Rossini architecture or were there new advancements and developments unique to the Bartok - and which might carry over to the next generation of dCS players?
Best
Gregg
As was the progression established in Scarlatti > Paganini > Debussy so is the progression in Vivaldi > Rossini > Bartok.
The underlying hardware architecture used in Bartok is very similar to that used in Rossini which, in turn, is very similar to that used in Vivaldi. Vivaldi has a much more complex sophisticated processing architecture than the other two and Rossini has a more sophisticated power supply and enclosure than Bartok. The big difference between the three is in software and that’s where Vivaldi is far ahead of Rossini… and Rossini is far ahead of Bartok.
1 Like
Thank you. This was helpful to determine the remaining life span of each unit. If things continue as in the past one would expect there will first be a replacement Vivaldi followed by a replacement Rossini 1-2 years later.
In other words (for my interest) it sounds like the current Rossini will be with us for quite a while longer.
If you look at the interior of the Vivaldi upsampler you’ll see that the streamer is a small green pcb connected to the pcb concerned with upsampling which is connected to a large pcb which is exactly the same as the Vivaldi clock pcb. In fact the Vivaldi dac appears to be a pcb attached to the same pcb found in the Vivaldi clock. It appears dCS built the clock first and incorporated that into the other boxes. This supports my theory that the clock isn’t really a super duper clock design just the same clock with an isolated power supply. Maybe the clock software is better but why wouldn’t they use the same code in the dac version of the board?
The NWB obviously doesn’t have the Vivaldi clock Board or the same quality power supply. The power supply in the NWB is likely the same found in the Rossini clock.
The upsampler has to be overkill for just streaming.
In a sense yes but the decision for Vivaldi owners is buy the Upsampler or have no streaming at all ( which for many means no music at all).
The common PCB that you see in all of the Vivaldi stack components is dCS’ Control Board (the “DCS400540”, left in the image below).
If I’m not mistaken, it’s an updated design based on the control board in the Scarlatti, and has a lineage that dates back to the earliest days of dCS platforms; I/O processing, RS232, Clock generation and PLLs, and connections into the DAC processing board etc. It wasn’t originally designed just for the Clock.
All of boxes in the Vivaldi stack have this same common Control Board in them, with appropriate power, I/O, and software tweaks for each.
3 Likes
There was another thread about the upsampler. Someone suggested combining streaming in the dac might produce suboptimal results. Spent some time thinking about this, admittedly with the expertise of a naive five year old, and question if the benefits of having the external streamer are negated by the use of aes cables into the dac. My working theory is that it would make little difference and streamer was included in upsampler because it wasn’t considered a “must have” interface during design. Don’t have sufficient expertise to examine the question of why the streamer was tied to upsampler while the dac has a USB port.
The existence of the NWB supports my theory that upsampling & streaming aren’t intimately connected processes. If this is true then the upcoming Ligeti dac will definitely include streaming.
Sorry… I don’t have much to do these days.
This is partly because one has the opportunity to buy all 4 boxes of Vivaldi simultaneously or one or more at a time . So I guess that dCS decided that someone who bought only the DAC may want to have a USB input. That is a different proposition to placing a complete network streaming board there. In fact my initial Vivaldi configuration used that USB input on the DAC ( I had no Upsampler) for PC based streaming. It was , however, significantly inferior in SQ compared to the dCS streaming solution - I wish I had known sooner.
Historically no dCS DAC had a USB input until the current third generation. USB was not a consideration for the first generation. It didn’t exist at the time. The second generation initially had the USB input on the upsampler ( you can see the lineage there). However the USB input was moved to the clock for some reason ( not to the DAC). That move resulted on the ability to process a wider range of input resolutions.
Indeed upsampling and streaming are not necessarily linked. NWB was, of course, introduced to provide appropriate connectivity and functionality to owners of legacy dCS equipment who otherwise did not have the ability to play formats developed later ( hence NWB downsampled not upsampled) including streaming options.
The forthcoming Britten stack may not follow the configuration of its predecessors for all we know. One speculation I have read here is that it could follow a sort of dirty box /clean box concept with a control box containing streaming /upsampling/user control functions including volume plus two mono only DACs in separate cases. It was , however, just guesswork on the part of the poster. We await what it really will be in due course. However I am pretty sure that I won’t be able to afford it anyway and will have to admire it from a distance 
.
The picture of the internals of the clock shows it only contains the control board. I sort of expected to see something unique to the design. However, after thinking it through, in a “cost no object“ design, whatever they did to the external clock they would also do for the internal clock. Unfortunately I’ve tried to understand how/why the external clock improves sq but it’s way too technical considering my background. The external clock would benefit from an isolated power supply and a lower noise environment. That is enough to explain why it would likely be a more reliable signal. Whatever complicated maths involved with integrating the external and internal clocks becomes irrelevant. The problem becomes more interesting when the external clock signal has to flow through 1 meter of wire with electrical properties unknown to the designers. Is it really simple in that regardless of the properties of the bnc cable, the signal will be consistent and that’s what matters?
The reason I’m interested is that I was earlier considering a MSB reference dac. It’s fascinating that two companies can have completely different design philosophies yet produce competing products that are extraordinary. MSB design has an edge in the “yeah I get it” area. Putting the best possible clock 1/2 inch from the dac is pretty intuitive. Again I can’t prove this design is optimal. It’s also possible dCS engineering is much more sophisticated and creative.
The point of a system clock as used in the Vivaldi stack is not just to be more accurate than the clocks in the individual components but also to synchronise those three components. Otherwise each will be running its individual version of a 44.1 or 48 kHz one second long cycle. So, as an exaggerated example, if Transport starts its 44.1 kHz cycle now, DAC might not start for another 10th of a second. So by the time transport has completed its one second cycle of 44.1K “pulses”, DAC is still only part way through its. This means that each individual cycle becomes further and further out of synchronisation with the other as time progresses leading , for example, to audible glitches . The system clock provides a single clock signal distributed to the other components. This alone would be reason for having it even if the clock accuracy per se was no better than the individual component clocks. However in fact it is which is the second reason for an external clock which is pertinent to Rossini ( used without Transport) and Bartok .
Which is, as I understand it, what is still happening in the stack ( I think I have this right). The individual clocks are not switched out but are slaves of the master system clock. So they are positioned close to whatever processor they serve but are all working identically to the same and more accurate master reference.
2 Likes
There are a number of Clock related threads elsewhere with posts that address some of your questions (including BNC cable related ones IIRC).
Maybe start with this post from James;