DKak
05-03-2006, 08:26 AM
Warning...long...
Digital cable measurements – my interpretation, experiment and results.
In my system in my home I hear differences in cables, both analog and digital. It seems fairly intuitive to me why I hear differences in analog cables: The components are affected by the cables’ capacitance, inductance, shielding, connection quality and so forth. Each interaction produces a different result…possibly better to our ears, or maybe not so good. It becomes a matter of synergy – what works best in one system may not in another. Therein lies the basis for the endless debate.
With digital, however, things are much more quantifiable. I have heard differences in digital cables with my system, but I didn’t know quite why. After all, the cable’s job is pretty basic: get all the data from one end to the other, intact and at the correct time. On the surface it looks like a simple job, but upon further investigation, it’s one that’s not so easily accomplished. The bottom line is that it does it right – or it doesn’t.
I sent a variety of digital cables to AK member House de Kris who graciously offered to perform his TDR measurements on each. All are extras and duplicates. My goal in lending these cables for testing was to see if I could learn more and understand why I hear differences. Maybe I could correlate differences I’ve heard to actual measured results. Maybe these results would lead to other conclusions for improving my system’s sound.
So just what does it take in order for a digital transmission line to get this job done right? It appears that some clearly defined factors are required and here are three that are very important:
1) Fast rise time and sufficient bandwidth. If I’ve done my math correctly, about 2.82MHz is necessary for 44.1k/16-bit digital audio. For 192k/24-bit digital audio, we’d need more like 18.4MHz. But that’s not it. The waveform is a square wave. There are infinite harmonics needed to reproduce the square wave perfectly. That’s just not possible. Some accept that 5x the basic frequency gets us a good enough representation of the wave for the task. But being a perfectionist audio type, I’d want at least a 20x bandwidth margin to make it even better, but that’s still “only” 368MHz bandwidth for the 192/24 audio needs. All but one of my cables Kris tested easily surpassed this tough 20x test at the –1dB mark and all did at the –3dB mark.
2) 75-ohm transmission line. Not just in the cable, but within the terminations, connectors and signal path within the connected gear, too. Sounds simple, but Kris’ test bear out that many cables do not meet this spec. How important is this?
3) Great shielding. It’s simple; noise contamination must be kept out. I don’t think Kris tested for this.
Knowing these requirements, what gets in the way and degrades performance in a potentially audible way?
With many thanks to House de Kris, we now have some clear measurement results among a sampling of 75-ohm “digital” cables, specifically sold as S/P-DIF coax cables. The cables I sent Kris were a mix of Canare and Belden wires (all with Canare 75-ohm BNC or Canare 75-ohm RCA connectors) plus a couple of brand name cables, all of them with BNC connectors. I selected these because it is a mix of cables where I did and didn’t hear differences. I wanted to see if those that sound the same to me also measured the same, and hopefully those that measured differently also are those that sounded different. In my system, the differences I hear between digital cables aren’t huge, and possibly in a less revealing system they might be inaudible. But I’ve spent years tweaking and ‘drilling down’ to improve my hi-fi at as many levels as I can. I doubt I’ll ever be done!
To describe the differences I hear, they aren’t so much in the spectral domain, i.e. bright, dull, warm, etc. Rather, I hear them as differences in articulation. Some cables seem to separate the instruments better, letting me hear more in the performance. At first listen you might say that one cable sounded brighter than another, but upon further listening, it becomes clear that one is more ‘smeared’ sounding than another and that it isn’t really a brightness difference. Hmmm, I wonder of the differences really are temporal. Argh...What’s causing it???
Funny how the best sounding cables in my system are those with Belden 1694A or Canare L-xCFB wire and Canare connectors. They certainly were inexpensive, with each costing less than $20. All were constructed by Blue Jeans Cable, HAVE, Inc. or Markertek. The Better Cables Silver Serpent cable is in the same league for sound, but at a higher cost. It has Canare BNC connectors, too. Interesting that these cables all measured the best.
Virtually all the cables tested have enough bandwidth (and then some) and all are designed with very good shielding. So, what’s left for differences in the lesser cables are most likely discontinuities in the transmission line. All the Belden, Canare and Better Cables cables’ impedance measured very smooth and close to 75 ohms – from end to end including the connectors! Could this be why these sounded so similar to me?
Kris’ measurements really struck me, not because of how my favorites measured, but because of my second favorite: the PS Audio cable, which definitely has measured problems. While the cable itself seems to be good, the connectors and quality of the connections to them eradicate its performance capability.
Maybe I wasn’t hearing ALL the potential of any of the cables? How so? Maybe the components’ connectors and signal paths have discontinuities, too, so differences in the cables are minimized? Hmmm.
Now that I know I have some cables with virtually dead-on performance for the task, I wondered just how good my digital signal path could get. The first step was to get some really good 75 ohm BNC receptacles to replace those in my components. I’d like to say I could improve the 75 ohm signal path traces on the components’ boards, too, but I haven’t plunged that far overboard yet! I installed Canare receptacles in my Assemblage D2D-1 sample rate converter and Vampire PCBNC receptacles in my Audio Alchemy DTI*Pro32. What’s left to upgrade are the receptacles in my DAC, a highly upgraded Counterpoint DA-10. All these components have good chassis and pc board damping, and the Pro32 has a highly upgraded power supply. Transport is an Audio Alchemy DDS*Pro with an upgraded power supply and it feeds the Pro32 via I2S. All four of these components get their AC power from a PS Audio P300 (with MultiWave 2+ upgrade). All power cords are my own design and construction.
Without even having all the new BNC receptacles in yet, the improvement is of a similar magnitude as replacing the cables themselves. The difference is similar, too: better musical articulation. It’s much like cleaning the cloudiness from a dirty window. The view is the same, but you can see things more clearly. This is all heard using two 3 foot Belden 1694A/Canare BNC cables between the Pro32, D2D-1 and DA-10. I just received some much shorter 1694A/Canare BNC cables from Blue Jeans Cable and will try them next. One is 18” and the other 24”. Nearly half the total cable length!
This is still a new change to my system, and once I replace the BNCs in the DAC I’ll try the other cables that didn’t measure so well. I want to see if the connector changes minimize the differences or make them even more noticeable.
In any event, this proved to be most interesting. For a small amount of money I was able to improve my digital signal path substantially. And from the sound so far, the inexpensive cables are something else, too! I highly recommend this experiment for anyone looking deeper into ways to improve their system’s performance.
Digital cable measurements – my interpretation, experiment and results.
In my system in my home I hear differences in cables, both analog and digital. It seems fairly intuitive to me why I hear differences in analog cables: The components are affected by the cables’ capacitance, inductance, shielding, connection quality and so forth. Each interaction produces a different result…possibly better to our ears, or maybe not so good. It becomes a matter of synergy – what works best in one system may not in another. Therein lies the basis for the endless debate.
With digital, however, things are much more quantifiable. I have heard differences in digital cables with my system, but I didn’t know quite why. After all, the cable’s job is pretty basic: get all the data from one end to the other, intact and at the correct time. On the surface it looks like a simple job, but upon further investigation, it’s one that’s not so easily accomplished. The bottom line is that it does it right – or it doesn’t.
I sent a variety of digital cables to AK member House de Kris who graciously offered to perform his TDR measurements on each. All are extras and duplicates. My goal in lending these cables for testing was to see if I could learn more and understand why I hear differences. Maybe I could correlate differences I’ve heard to actual measured results. Maybe these results would lead to other conclusions for improving my system’s sound.
So just what does it take in order for a digital transmission line to get this job done right? It appears that some clearly defined factors are required and here are three that are very important:
1) Fast rise time and sufficient bandwidth. If I’ve done my math correctly, about 2.82MHz is necessary for 44.1k/16-bit digital audio. For 192k/24-bit digital audio, we’d need more like 18.4MHz. But that’s not it. The waveform is a square wave. There are infinite harmonics needed to reproduce the square wave perfectly. That’s just not possible. Some accept that 5x the basic frequency gets us a good enough representation of the wave for the task. But being a perfectionist audio type, I’d want at least a 20x bandwidth margin to make it even better, but that’s still “only” 368MHz bandwidth for the 192/24 audio needs. All but one of my cables Kris tested easily surpassed this tough 20x test at the –1dB mark and all did at the –3dB mark.
2) 75-ohm transmission line. Not just in the cable, but within the terminations, connectors and signal path within the connected gear, too. Sounds simple, but Kris’ test bear out that many cables do not meet this spec. How important is this?
3) Great shielding. It’s simple; noise contamination must be kept out. I don’t think Kris tested for this.
Knowing these requirements, what gets in the way and degrades performance in a potentially audible way?
With many thanks to House de Kris, we now have some clear measurement results among a sampling of 75-ohm “digital” cables, specifically sold as S/P-DIF coax cables. The cables I sent Kris were a mix of Canare and Belden wires (all with Canare 75-ohm BNC or Canare 75-ohm RCA connectors) plus a couple of brand name cables, all of them with BNC connectors. I selected these because it is a mix of cables where I did and didn’t hear differences. I wanted to see if those that sound the same to me also measured the same, and hopefully those that measured differently also are those that sounded different. In my system, the differences I hear between digital cables aren’t huge, and possibly in a less revealing system they might be inaudible. But I’ve spent years tweaking and ‘drilling down’ to improve my hi-fi at as many levels as I can. I doubt I’ll ever be done!
To describe the differences I hear, they aren’t so much in the spectral domain, i.e. bright, dull, warm, etc. Rather, I hear them as differences in articulation. Some cables seem to separate the instruments better, letting me hear more in the performance. At first listen you might say that one cable sounded brighter than another, but upon further listening, it becomes clear that one is more ‘smeared’ sounding than another and that it isn’t really a brightness difference. Hmmm, I wonder of the differences really are temporal. Argh...What’s causing it???
Funny how the best sounding cables in my system are those with Belden 1694A or Canare L-xCFB wire and Canare connectors. They certainly were inexpensive, with each costing less than $20. All were constructed by Blue Jeans Cable, HAVE, Inc. or Markertek. The Better Cables Silver Serpent cable is in the same league for sound, but at a higher cost. It has Canare BNC connectors, too. Interesting that these cables all measured the best.
Virtually all the cables tested have enough bandwidth (and then some) and all are designed with very good shielding. So, what’s left for differences in the lesser cables are most likely discontinuities in the transmission line. All the Belden, Canare and Better Cables cables’ impedance measured very smooth and close to 75 ohms – from end to end including the connectors! Could this be why these sounded so similar to me?
Kris’ measurements really struck me, not because of how my favorites measured, but because of my second favorite: the PS Audio cable, which definitely has measured problems. While the cable itself seems to be good, the connectors and quality of the connections to them eradicate its performance capability.
Maybe I wasn’t hearing ALL the potential of any of the cables? How so? Maybe the components’ connectors and signal paths have discontinuities, too, so differences in the cables are minimized? Hmmm.
Now that I know I have some cables with virtually dead-on performance for the task, I wondered just how good my digital signal path could get. The first step was to get some really good 75 ohm BNC receptacles to replace those in my components. I’d like to say I could improve the 75 ohm signal path traces on the components’ boards, too, but I haven’t plunged that far overboard yet! I installed Canare receptacles in my Assemblage D2D-1 sample rate converter and Vampire PCBNC receptacles in my Audio Alchemy DTI*Pro32. What’s left to upgrade are the receptacles in my DAC, a highly upgraded Counterpoint DA-10. All these components have good chassis and pc board damping, and the Pro32 has a highly upgraded power supply. Transport is an Audio Alchemy DDS*Pro with an upgraded power supply and it feeds the Pro32 via I2S. All four of these components get their AC power from a PS Audio P300 (with MultiWave 2+ upgrade). All power cords are my own design and construction.
Without even having all the new BNC receptacles in yet, the improvement is of a similar magnitude as replacing the cables themselves. The difference is similar, too: better musical articulation. It’s much like cleaning the cloudiness from a dirty window. The view is the same, but you can see things more clearly. This is all heard using two 3 foot Belden 1694A/Canare BNC cables between the Pro32, D2D-1 and DA-10. I just received some much shorter 1694A/Canare BNC cables from Blue Jeans Cable and will try them next. One is 18” and the other 24”. Nearly half the total cable length!
This is still a new change to my system, and once I replace the BNCs in the DAC I’ll try the other cables that didn’t measure so well. I want to see if the connector changes minimize the differences or make them even more noticeable.
In any event, this proved to be most interesting. For a small amount of money I was able to improve my digital signal path substantially. And from the sound so far, the inexpensive cables are something else, too! I highly recommend this experiment for anyone looking deeper into ways to improve their system’s performance.