DAC-free streamer; modular approach

The DAC could do this anbsolutely .. But think how the DAC internally would send the data from the buffer to the internal DAC / DSP. Will that not use a clock?

That clock (and other electronics) however is susceptible to modulation from powerline, ground plane or EMI from nearby electronics which may be sharing common resources. This is jitter caused by cross talk. But yes theoretically if you achieved 100% effective decoupling, or to the point that any distortion iwas below the noise floor .. then you will have produced perfect jitter distortion  rejection, and all digital sources will sound the same.

Simon

Sure, and I would expect the accuracy of that clock and all the downstream processes and conversion technologies to make the difference in sound quality!

I understand that the steps that have to take place after the data has been transferred and stored locally are by no means trivial. And this is the kind of technlology I expect companies like Naim or Chord to deliver and which I would be happy to pay for.

But I would expect these processing steps and technologies to be largely independent of the data transfer mechanisms and protocols used to bring the data to the dac. And I tend to consider designs that exhibit such dependency immature.

This is the reason why I was initially a bit disappointed by the nDAC. That said, I have nothing to regret: at current s/h prices, I consider the nDAC (and, of course, more recent devices in that price class: hugo, mirus, etc.) to be very viable options.

This is very sad, isn't it ? Investing in streamerless and dacless amplifiers and buying second hand dacs is perhaps not the worst strategy, after all.

Interesting points Simon and one of the reasons you can never fully decouple the output data stream in a RAM buffer from the input when running in near real time playback (assuming a small delay introduced by the buffer)

Taking S/PDIF and assuming a bit perfect source. If you were able to have a large enough buffer to be able to store, say, a single track, could turn off the input clock (and clock recovery circuitry) and disconnect the input cable from the source, then the resulting playback from the buffer should be completely source agnostic (although from a usability point of view a complete PITA !)

Why just a single track? It could be a whole album or playlist. I am not sure usability drawbacks would be a necessary implication. The dac could still support elementary control of the replay process. I can imagine these options would be easily customizable: users aiming at the best possible sound quality would default to stop data transfer and replay control options before starting replay. Users wishing to trade sound quality for usability could opt to enable replay control and simultaneous data transfer during replay. I see no problems in principle.

Yes why not, the single track was just mentioned to illustrate my point. 

James

Cool discussion.

These choices, for data rate, buffering, clocking, etc., are design choices rather than theoretical constraints. You can imagine many different schemes, but implementation of non-standard solutions generally requires more local engineering and, thus cost...perhaps with better quality as the result. What's being discussed here is analogous to the design decision in CD playback to save money/effort clocking variable data rates (with a constant speed motor) by paying money/effort for a variable speed motor that senses read-head radius and runs accordingly (to keep the data rate steady). They could have made other decisions - including changing data density at the time discs were burned to have both constant data rate and constant rotational speed, as in LP turntables - but they didn't and everything else had to follow. There is more flexibility now, since everything is faster and buffer memory is cheaper. Streamed video replay seems to make different choices to exploit this progress: with quick start times highly prized, lower quality is given for the first few seconds of, say, Netflix, while the buffer fills...then higher quality is offered once things settle down. A different trade off, with correspondingly different design choices.

When I was in school, especially in the lab, we had a saying: In theory, there is no difference between theory and practice; in practice there is!

Regards alan 

The technicalities of the discussion are beyond me. But, trying to distill some practical lessons for the committed end user (ie me) I think the following points have emerged:

1. Using a separate dac-free renderer from a dedicated hifi manufacturer into an external DAC is possible in today's market, but not (yet) with Naim. Some forum users might be interested in a Naim dac-free renderer - others would not. For myself the idea arose with the arrival of the Chord Hugo, but the V1 and nDAC raise the same point;

2. Using a separate renderer and external dac may introduce noise, depending on the quality of the equipment and material used. [For myself I am still unsure whether there is an advantage in using a USB connection into an external dac];

3. Using a separate (non-Naim) renderer may allow greater flexibility, depending on the services (internet radio; lossless streaming etc) available and on the access software [another issue which would not have arisen for me but for my recent woes with the Naim app];

4. The greatest flexibility comes from a computer - at this point my brain waves a little white flag, given that the possibilities for self-build and pre-build are almost endless. There doesn't seem to be any clear opinion as to whether the sound from a computer (pre-built; self-buit) into an external DAC can match a Naim streamer into a DAC. But you can use any software, access all the streaming services you like and generally play around more. And if you add video to the computer, well...

For some general interest, I've come across some pretty cool-looking stuff in the last week. I hope Richard will allow me to mention the following (no links obviously, but search around) because their existence suggests there is a market for such products which is not (yet) met by Naim:

Auralic Aries; Simaudio Moon 180- MiND; Krell Connect; Bryston BDP-2; bel canto REFStream

Yes - but are you not missing the point? Essentially is this not what a USB memory stick is? But hey do memory sticks all sounds the same? And on the NDAC does the rear USB sound different to the front one? - and whilst we are at it does FLAC sound the same as WAV?

There are many possible sources of cross talk - other than decoding and reclocking SPDIF.

Agreed - but i suspect a player designed just to read data from a single USB stick (or Memory card etc) in one format can be optmised for that single role more successfully than a player designed for multiple inputs and formats. 

Yep - that sounds reasonable

S

Or instead of a USB stick, why not use an optical disc? 

Ah yes - the simple solution 

Sure there are. I am not arguing buffering would eliminate all jitter problems, of course. But it could eliminate some, it seems to me. I would also expect that, if data were transcoded and buffered prior replay, all USB sticks and file formats would sound the same (or, at least, that the differences in sound quality would decrease). I hope that the next generation of dac devices will implement a cleaner separation between data transfer concerns and digital to analog conversion concerns and, as a result, be more source agnostic than the current generation. Best, nbpf

nbpf - I agree with your conclusions