Cost no object dac/streamer

[tourneychamp23]

Choosing a Streamer & DAC has never been easier then now, looks & various preferences excluded.

Choose a streamer from a well known supplier supporting the Application protocols you need eg Roon's RAAT, they're all computers in various disguises:

View attachment 18682

Then pick a DAC which has been measured to be audibly transparent fulfilling what other preferences yoy may have in terms of looks and functionality.

How hard can it be? ... ;-)

Haha can you give any examples?

 

[Thomas savage]

 

[Deleted member 65]

Haha can you give any examples?

Starting with the Streamer, look up the Silent PC I've built or Amir's build documented here in ASR. Go buy a similar ready made product if building is not your thing.

As for DAC, wade through Amir's measurements and you'll find some audibly transparent ones, it then comes down to your prefs ...

 

[tourneychamp23]

Starting with the Streamer, look up the Silent PC I've built or Amir's build documented here in ASR. Go buy a similar ready made product if building is not your thing.

As for DAC, wade through Amir's measurements and you'll find some audibly transparent ones, it then comes down to your prefs ...

So all audibly transparent dacs sound the same and any increase in money spent beyond that point is wasted?

 

[Deleted member 65]

So all audibly transparent dacs sound the same and any increase in money spent beyond that point is wasted?

Not wasting any more time on this thread ... Good Luck!

 

[tourneychamp23]

Not wasting any more time on this thread ... Good Luck!

Not trying to stir the pot honestly looking to understand?

 

[jacobacci]

Why was the DCS chosen as a reference? It was the one you wanted to buy.

Actually not quite. The initial impulse to compare at all had been: "What the heck, lets see how a 20 year old dCS Purcell / Delius holds up against some of these miracle DACs we keep reading about. And while we're at it, let's see how my Weiss does against the dCS". I was not predisposed to buy the dCS. In fact if the listening sessions had confirmed that I liked my Weiss the best, I would have happily kept it and saved myself a significant amount of money.

For the tests I only had access to the dCS Purcell / Delius combo (which is 20 years old) of my friend's and my Weiss for an extended period of time and had to loan the other DACs one by one from dealers. So to make any meaningful comparison, we used the dCS as a reference (sorry not reference as per your definition) to match the other DACs against. As it turned out, we all

• were able to reliably (around 90%) distinguish the dCS combo from the other DACs
• did not prefer the sound of any of the other DACs (including my Weiss) to the dCS combo. Why is not relevant here. Simply personal preference

During the comparison session with my Weiss, I preferred the dCS by a substantial margin. Again, why is not of issue here. Had I liked any of the other DACs better than the Purcell / Delius, we would have compared the DACs I preferred (to the dCS) against each other. As that was not the case, we were able to leave out that step (which would have meant getting several DACs again from the dealers.
The Purcell / Delius is not manufactured and not serviced any more, so as a final round we compared the current Rossini to the Purcell / Delius combo and we preferred the Rossini (no miracle, as it is three generations further along the family tree. Had the Bartok been available at that time (half of the Rossini's cost), we would have probably chosen the Bartok for the final comparison.
The fact that none of the new DACs sounded better than the 20 year old dCS combo was a huge surprise to all of us.
But I take your point, we could have taken along my Weiss as reference. I do not feel that would have changed the result, as I clearly preferred the Purcell / Delius to the Weiss in the direct comparison.
Call me biased. I will not lose any sleep over it. For the purpose of making a choice the process fulfilled its purpose. I am sure it could have been improved, if the intention had been to conduct a scientific experiment fit for peer review.

 

[jacobacci]

Not trying to stir the pot honestly looking to understand?

I am afraid no amount of asking will spare you the effort of actually doing a listening comparison to find out whether YOU can hear a difference or not. See my post above.
If you can't hear a difference, buy the cheapest one that measures well. If you can, get the one you like best and can afford.
Just make sure the test setup is suitable to the task, otherwise you are cheating yourself.

 

[andreasmaaan]

Not trying to stir the pot honestly looking to understand?

Fair enough If the DAC doesn't produce any audible noise or distortion higher in level than -80dB (0.01%) under normal operating conditions, it's likely to be transparent. -90dB (0.003%) and I'd say extremely likely. -100dB (0.001%) and I'd say there's next to no chance it won't be transparent, assuming nothing surprising is going on that Amir's thorough but not all-encompassing measurement procedure doesn't catch.

Many DACs fulfil these requirements, which is why people are suggesting you look more at the features and price, and just use the measurements to filter out any units that are obviously poorly designed.

Off the top of my head (and without checking - sorry) popular units here that fulfil the most stringent requirement above include in the budget category most of the Topping units, the Schiit Modi 3, the Khadas tone board, and various JDS units, and for more money you have e.g. most Benchmark and RME products, and no doubt piles of units that aren't measured here. Stereophile is another good source for quite thorough measurements, although they focus more on mid-high priced products.

 

[andreasmaaan]

@jacobacci sorry if you've already mentioned it, but would you mind outlining also the setup you did the comparisons on? Thx

 

[jacobacci]

Most of what I read on here suggests 90% of what I am going to spend should be on speakers because there isn't a huge delta difference on electronics

Rule of tumb: 50% for speakers, 50% for electronics (Streamer, DAC, Amp). If your speakers / amp are not resolving enough, no point spending more than a few hundred on a DAC / Streamer. You simply will not notice a difference.

 

[jacobacci]

but would you mind outlining also the setup you did the comparisons on?

Streamer: Intel S1200RPM, XEON E3-1265LV3, ECC RAM, Windows Server 2012R2, jRiver
Interface: Weiss Int202 FW to SPDIF (Comparison DACs went directly via USB from Streamer)
Baseline DAC: dCS Purcell / Delius
Goldpoint XLR Switch and Attenuator
Amps: Brinkmann Monoblocks
Speakers: Carat First with Stella Extremus Supertweeters

 

[andreasmaaan]

Also curious @jacobacci as to whether you've any theories as to the cause of differences you were hearing? My best stab (other than the possibility of controller induced bias) would be to do with input/output impedances between the DAC and the passive attenuator. The DCS units have very low output impedances compared to the other units (or those that publish this spec at least), while the amp has an unusually low input impedance of 1000 600 Ohm.

Given the use of a passive attenuator and a low input impedance amp, it seems plausible to me that this factor could have audibly come into play. Do you know what the resistor values are of the passive attenuator you were using?

Also interesting that the differences were discernable despite the use of an amp that is not particularly low distortion (spec'd at 0.1% at half power according to the manufacturer), i.e. orders of magnitude higher than all the DACs tested.

Btw, do you have a link for the speakers? I've never heard of them and can't find them online...

Anyway I do appreciate the effort you've gone to do describe what you did here

Correction: Brinkmann Monoblocks' input impedance on the balanced inputs is 600 Ohm (1000 Ohm was the figure for the unbalanced inputs).

 

[Blumlein 88]

So what length cable between attenuator and amps?

 

[garbulky]

Streamer: Intel S1200RPM, XEON E3-1265LV3, ECC RAM, Windows Server 2012R2, jRiver
Interface: Weiss Int202 FW to SPDIF (Comparison DACs went directly via USB from Streamer)
Baseline DAC: dCS Purcell / Delius
Goldpoint XLR Switch and Attenuator
Amps: Brinkmann Monoblocks
Speakers: Carat First with Stella Extremus Supertweeters

Do you have a link to the Carat first speakers? Which brinkman model?

 

[andreasmaaan]

Do you have a link to the Carat first speakers? Which brinkman model?

I'm not sure, but I believe there is only one Brinkmann monoblock.

 

[Guermantes]

@tourneychamp23 The benefits of going down a more modular approach, as Lars has suggested, is that by separating the streamer from the DAC you have:

a. a wider range of products to choose from;
b. better upgrade options, especially with a DIY streamer such as an RPi, silent PC, MAC etc. This makes the setup more future-proof as protocols change, etc.;
c. you can focus the money on what is important to you -- sound, appearance, integration, manufacturer, etc.

Also be aware that some of the players in the all-in-one streamer market are using the same software that is available for free from developers for platforms such as the RPi (i.e. Music Player Daemon) and charging a premium for it, probably in violation of the GPL license.

 

[DonH56]

@DonH56 , you wrote:

«That said, Amir's measurements certainly show a lot of DACs are pretty comparable, though perhaps more time-domain results would be worthwhile».

Could you elaborate om time domain? With examples?

Frequency response and time domain response are two ways of expressing the same thing but transient time events are not always completely captured by steady-state frequency measurements. I would add step and pulse testing. However, things like the J-test and multitone testing also provide some insight into time-domain behavior. Doublets and other things in amplifier compensation that lead to long settlers are not always seen in a purely steady-state frequency plot. And stability issues are load- and signal-dependent though would usually show up on a frequency plot as a peak (and ringing i the step or impulse response).

HTH - Don

 

[svart-hvitt]

Frequency response and time domain response are two ways of expressing the same thing but transient time events are not always completely captured by steady-state frequency measurements. I would add step and pulse testing. However, things like the J-test and multitone testing also provides some insight into time-domain behavior. Doublets and other things in amplifier compensation that lead to long settlers are not always seen in a purely steady-state frequency plot. And stability issues are load- and signal-dependent though would usually show up on a frequency plot as a peak (and ringing i the step or impulse response).

HTH - Don

Interesting. Should one read this as a suggestion for @amirm to broaden his set of measurements?

 

[andreasmaaan]

Further to my previous post, here are the balanced output impedances I was able to find for the DACs tested (some from spec sheets, others measured by Stereophile). If the amp presents a load of 600 Ohm to the passive attenuator, we can assume that the amp + passive attenuator + cables present a significantly lower load than that to the DAC. Hence my suggestion that herein lies the most plausible source of audible differences between the units, particularly if the impedances of either DAC or amp vary significantly with frequency (which is for the most part admittedly unknown):

• Chord Dave: 66 Ohm
• Merlot: (unknown)
• Weiss DAC202: 44 Ohm
• DAC8 DSD: 114 Ohm
• Aune S16: (unknown)
• DCS Rossini: 2 Ohm
• DCS Delius: 1 Ohm

As to any audible difference between the Rossini and Delius, I have no plausible theory...