AES/EBU cable shootout

@rudyb Anyone who thinks their cables make better zeros and ones is unfamiliar or disingenuous about the nature of digital communications. Cables live at Layer One of the model, so copper, fiber, radio are all interchangeable. Errors in digital communications happen all the the time and are accounted for by the protocol used. For example, an Ethernet frame is preceded by a preamble and start frame delimiter (SFD), which are both part of the Ethernet packet at the physical layer. Each Ethernet frame starts with an Ethernet header, which contains destination and source MAC addresses as its first two fields. The middle section of the frame is payload data including any headers for other protocols (for example, Internet Protocol) carried in the frame. The frame ends with a frame check sequence (FCS), which is a 32-bit cyclic redundancy check used to detect any in-transit corruption of data. (Wikipedia). Dropped packets are retransmitted and reassembled in their correct order by the receiving device. Remember, this is all happening at at Gigabit speeds, and audio only requires Megabits, a thousand times slower. The other issues with digital, things like quantization errors (when a sampled signal is between the voltages recognized as a zero (<300 mv) and a one (> ~3V) have nothing to do with cables  and are all about the digital filter design in the DAC.  Noise, in the analog sense, doesn't exist in digital. Spurious voltage spikes that might be sensed as a 'one' are trapped by the error correction. Voltages below 300mV are treated as zeros, and ignored. Everything a listener interpret as 'definition' or 'space' or whatever is occuring in the DAC filters or in the analog circuits, not in the digital cable. And that's by design. Do you think your ISP is using Nordost cables between your streaming service, across the internet, through lord knows how many intermediate routers, switches, and buffers? No. I can assure you it's the same Belden  Cat 6 or fiber you get off the shelf at any electronic supply house. I was trained by AT&T on both copper and fiber, and am a certified fiber splicer (which takes a very cool $20,000 tool). Digital cables either work or they don't. Binary, zero or one. If they have ANY impact on performance, they are defective.

@ghdprentice I'm still waiting for someone to explain to me how the last 1 or 2 meters in the digital communications chain can make a "huge difference" and the previous one or two thousand miles doesn't make any difference. Digital is not analog. I totally get how inductance, capacitance, source, line, and destination impedance interact to cause subtle differences in SQ, I have no issue with that. But digital is fundamentally different and the things you attribute to digital cables simply have no basis in digital signal transmission. Attributing a difference in what you are experiencing to a digital bitstream that is identical at the source and the destination simply defies credulity. Whatever you are experiencing is happening in the analog realm, or it is a heuristic error or a cognitive bias. No shame there, we need those to survive. But our heuristics and cognitive biases are mental shortcuts and demonstrably imperfect. Dismissing them out of hand is never a good plan.

@ghdprentice I was Principal Architect for a Fortune 10 company for 10 years before I retired after a career in IT. System resilience and availability were my responsibility. Whatever was built and went into production was either my architecture, or had to have my review. So yeah, I have a pretty solid grip on theory and practice, and the business implications of failure when a system responsible for $100M a DAY breaks.

And yes, I still had to know how to build cables and splice fiber.

And I am still waiting for an explanation on how a Layer One passive digital component can have any impact on analog sound quality. 

@ghasley You presuppose I have not, and that is an error. I will be happy to test any cabling beyond what I have done, just as soon as someone can provide a plausible explanation for how a digital cable can impact analog sound quality. The cable, not the DAC, not the power supply, not some oscillator induced phase noise, just the cable.

You say"Now, imagine streaming music. It absolutely requires proper buffering. FULL STOP. You cant have the data stream begin and then “error check” on the fly. Jittery at best, drop outs at worst. SO, how is the buffering performed, how much of the data is able to be buffered into memory, what and how is the data streamed from the buffer and what/where/method is the clocking/reclocking performed? There was alot more to the conversation and todays streamers/dacs do a wonderful job with the challenge. Some do it better than others and some do it completely differently." And I couldn't agree more.

Then you continue, " We are still in the process of understanding why certain materials, lengths of cable, shielding or lack thereof along with which method of transmission provides the best results and which do not. But it is audible. Ive tried to buy equipment from companies that seem to have a grasp of the importance of timing, lower jitter and clocking." None of this has anything to do with cables, but rather how the interfacing equipment performs. In your own words," Ive tried to buy equipment from companies that seem to have a grasp of the importance of timing, lower jitter and clocking." Those are attributes of active components, not cables. If an active component is so marginally engineered that it cannot perform properly with an in-spec cable, and has some special cable attribute as a requirement, I suggest the manufacturer needs to either document that or re-engineer the device.

@svenjosh I missed the part where I said there is no noise in digital, of course there is. Crank up the ASA speed on a digital camera and you'll see plenty of it. One of the biggest differences between your cellphone camera and a serious digital camera is the bit depth of the sensor and thus dynamic range. This makes cellphone cameras much more sensitive to noise, especially quantization noise, as they have many more opportunities for values between bit levels. But none of this has anything to do with cables.

Within the digital domain any change in sound from a digital source must be accompanied by a corresponding change in the digital bitstream describing that sound. However, if that were introduced by a cable, that would change the checksum and/or parity of the bitstream and would be sensed as an error, and corrected whether by CRC or parity error on the receiving end. This is true for copper, fiber, or wireless Ethernet, AES/EBU, or USB. 

Changes in sound - and their corresponding changes in bitstream are prevalent before and after transport at Layer One. Buffering, latency, jitter, all the side effects of various digital filter algorithms, introduction of noise from all kinds of sources, phase noise, quantum and thermal noise from electronics, noise carried in from imperfect grounding schemes, RF noise; in fiber there is quantum (shot or 1/f) noise, thermal noise. My personal favorite, is dark current noise (eery sounding innit?) But i digress, that's a camera sensor thing. None of these are caused by an in spec interconnect.

The point is any change be being attributed to the cable is being done so contrary to the basic operations of frame or packet based digital signal transmission. Those audible changes may be occurring, just not during transmission of the in-band bitstream.

@jerrybj Physics is physics, acoustics is a branch of applied physics as is digital communications, an area I've dedicated several decades of my life studying and implementing. So I am comfortable saying I have a lot more experience in that area than most. I have been involved with audio, both personally and professionally, even longer.

You are trying to attribute a perceived change in subjective sound quality to a part of the digital signal chain that is demonstrably unchanged. That leaves only three options: (A) the actual cause is not what you think it is. (B) you are falling prey to any number of forms of observer bias. (C) you want to propose that the digital bitstream, from which all content is derived, is mysteriously influenced by some unidentified external force capable of manipulating bits beyond the ability of error correction protocols, and that can only be mitigated by the most exotic cables, alloyed, extruded, woven, braided and terminated according to the incantations of the elders.

So I go with Option A.