why expensive streamers


@soix and others

I am unclear about the effect on sound of streamers (prior to getting to the dac). Audio (even hi-res) has so little information content relative to the mega and giga bit communication and processing speeds (bandwidth, BW) and cheap buffering supported by modern electronics that it seems that any relatively cheap piece of electronics would never lose an audio bit. 

Here is why. Because of the huge amount of BW relative to the BW needs of audio, you can send the same audio chunk 100 times and use a bit checking algorithm (they call this "check sum") to make sure just one of these sets is correct. With this approach you would be assured that the correct bits would be transfered. This high accuracy rate would mean perfect audio bit transfer. 

What am I missing? Why are people spending 1000's on streamers?

thx

 

128x128delmatae

@delmatae - yes, that is the approach used by many DACs (including my Denafrips Terminator Plus), and this helps, but doesn't fully solve the problem. A FIFO buffer can be used to reduce timing errors, but as @andy2 notes, this doesn't really deal with noise issues.

And if a synchronous data interface is used (such as SPDIF, I2S, TosLink, or AES), there are challenges with FIFO buffers. In these cases, the source clock is used to clock data into the FIFO. If the DAC uses its own clock to clock data out, then you risk overflow or underflow conditions. Many Denafrips users reported this problem, particularly those using the lower-end models connected to modest-priced streamers and transports (where the clock accuracy of both devices is not as tight as higher-end models). . 

Many DACs use a phase-locked loop or some other similar mechanism to adjust the output clock frequency to match the input clock, but its significantly more difficult to achieve the timing accuracy with this approach compared to a high quality oscillator. 

There are ways to reduce the overflow/underflow potential, such as resetting the FIFO between songs (when possible), using very deep buffers, adjusting the buffer depth based on the difference in source clock and DAC clock frequency, and using highly accurate clocks in both the DAC and streamer. 

Using deep and/or variable depth FIFOs also has issues though, particularly if the DAC output needs to be synchronized with another media stream (such as video).

Using an asynchronous data connection, such as USB, allows the DAC to control the timing, which eliminates the overflow/underflow situation, but USB is notorious in the amount of noise that is carried with the signal, particularly if it is generated by a noisy computer or cheap streamer.

An optical connection will eliminate noise carried on the ground, but not on the data signals themselves. The optical signal is still an analog signal and will carry whatever noise was on the electrical signal in the source (streamer or transport) before the signal was converted to optical. This noise will still be present when converted back to an electrical signal in the DAC. That said, eliminating the ground noise is still a significant benefit. 

Some DACs have clock outputs which can be used to control the timing of the source, so that a relatively small FIFO can be used inside the DAC to reclock the data without worry of FIFO over/under flow. But this requires non-standard devices, or an additional digital-to-digital converter that uses an asynchronous source connection (e.g. USB) and a synchronous output that is clocked by the DAC clock. 

Removing noise on high-speed digital signals is far from trivial. It's a lot easier (but still challenging) to prevent (or at least minimize) the noise from being generated in the first place. Any circuitry implemented in the DAC to reduce noise and timing errors has a much easier time when the problems are minimized in the first place.

In my system, I use a Denafrips Gaia DDC which accepts clock inputs from the Terminator Plus DAC. The DDC is fed with USB from the streamer, and then uses a synchronous connection to the DAC (I like I2S best). I started out using a fanless NUC with LPS, then switched to using a Sonore Optical Rendu, and finally to a Sonore Signature Rendu SE. Even with my moderately high-end DAC/DDC, I was able to discern improvements in clarity and soundstage width/depth moving up in streamer performance.

I'm now working on building my own DIY streamer which will use multiple levels of data isolation and reclocking, very high quality SC-cut oscillators, super-capacitor power supplies (allowing off-grid operation), and extensive electrical and mechanical isolation, to provide the cleanest possible signal to my DAC. 

The streamer is the first point at which timing (jitter) becomes a thing. It's job is to unpack the packets/frames of data which arrive asynchronously and convert them into a bitstream. An highly accurate clock at this point can have a huge impact on sound quality, as of course can the streamer not adding in unnecessary noise which won't make any difference to the 1s and 0s but if it reaches the analog(ue) parts of the DAC then it very much can.

I still only have a modest Innuos Zen Mini with a souped up Zen Mk3 LPSU but the improvement in sound quality over my Bluesound Node 2i is clearly audible, and that's despite the latter having been "pimped" with a Sean Jacobs DC3 linear PSU and a Mutec MC-3 reclocker.

This whole thread reminds me of the debate about amplification.  There are audiophiles that claim that if 2 amps measure the same then there is no sonic difference between them.

  I’ve tried several streamers in the $500-$4500 range and there are definitely differences. 

I don't claim to be the world's most sophisticated person (let alone audiophile) when it comes to digital audio and playing music from a cloud-based music service, or a hard drive/server.  I do know more than the average person about audio playback systems, acoustics and reverberation time, amplification, cabling etc.
What I have experienced firsthand is the sonic improvements that I have heard when moving source from a high-quality compact disc player, incorporating an outboard DAC (the initial offerings from Theta Digital, Wadia, etc.) and then moving onto a compute solution with an outboard compute-based DAC system. 
First a small Benchmark and then to a Lampizator Baltic 3.  Then USB from compute to the USB input of the Lampizator.  And the differences there were not subtle. 
Then along came a Lumin U2 and WOW! 
Can I technically explain why it sounds so much better?  Maybe a little.
But you know what else adds to the overall sound quality of these more expensive products?  It's still a lot of good quality and attentive detailed engineering.  These things still plug into a wall, have a power supply, have circuitry that are sensitive to resonance and heat, etc.  And there is still that pride of ownership.  Being able to look at something and appreciate what went into designing it, the build quality, chassis finish, the tubes that glow (in some cases) are also all parts of why we purchase what we do.
If that didn't exist then Rolex, Breitling, Phlip Patek etc. wouldn't be providing their watches in the marketplace.  I mean, don't all watches just tell time?  Why own an expensive one?
Same goes for cars; yeah, you can buy a car based on specifications alone, but specifications don't tell you what the leather smells like, how the door shuts so quietly, and how the engine sounds when you turn it on and how fun it feels when you take a corner at 70 miles per hour while still feeling so safe.  (These are just analogies so don't critique my driving habits, please!)
I know I am not helping the discussion specifically, just something I wanted to throw in there!
          

Streamers and dacs work in tandem. I wouldn't get the best out of a DCS Vivaldi DAC  using a Bluesound Node, just as I wouldn't get the best out of a Grimm using a Topping DAC.