5 ms·

One thing I can say about copper is that it is noisy. We are running FPGAs (with NICs) and were getting lots of packet errors over the copper cable SFP cables.

One thing I can say about copper is that it is noisy. We are running FPGAs (with NICs) and were getting lots of packet errors over the copper cable SFP cables. Turns out, we either needed to write out noise cleanup logic into the FPGA, or add in a noise cleanup module which would add an extra 50ns to packet processing to the FPGA. Simple solution was to convert to fiber and the noise and errors all went away. I'd say if audiophiles truly want unmolested electricity, they ought to use optical fiber for as much of the run as possible.

lmilcin 8y ago
Digital transmission/storage ensures the signal is not affected no matter how much noise is on the line (well, as long as it can be transmitted at all...) Now, what I think you mean is that the copper in the cable picks up noise either from the device on the other end or from the surroundings and that has the possibility of affecting the DAC. Even moderately well designed receiver should be able to galvanically isolate itself from outside sources of noise without having to resort to fiberoptics or wireless. It's just couple of dollars max for extra few components on the board.
- TylerE 8y ago
  > Digital transmission/storage ensures the signal is not affected no matter how much noise is on the line (well, as long as it can be transmitted at all...) There is no such thing as a digital transmission. All signals are analog.
  - mrob 8y ago
    Digital transmission refers to the whole transceiver system, not just the transmission medium. If I can put a "1" in and reliably get a "1" out, it doesn't matter that it was only a 0.9 part way down the wire.
    - TylerE 8y ago
      But that's the question, isn't it? If 0.0001% of the time you get 0.45v instead of 0.9v on the far end, now what?
      - lmilcin 8y ago
        I suppose you need to read up on digital protocols, it is all well handled. If it wasn't there could be no Internet or digital media or even processors or motherboards.
        TylerE 8y ago
        That is a higher level, though. I'm just so tired of all the bullshit where it "Must be a 1 or a 0". That is patently not true. HDMI video, to name one example very relevant to this thread, has no error correction, and bit errors will indeed crop up as the "sparkling" artifacts GGGGGP reported.
- gnufx 8y ago
  For an example possibly familiar in computing circles, Infiniband has easily-accessible counters to record transmission errors. In a properly-built system with hundreds of communication-busy nodes connected with copper, you see remarkably low error rates, even in comparison with the spec (maybe a few a week in my experience).
  - bayindirh 8y ago
    Infiniband is using extremely isolated and well balanced cables for passive/copper transmission. Also the cable lengths are very short for EDR and beyond (3m-5m is max usable length). For anything longer you need to use fiber cables with embedded converters. We have a cluster with ~1000 nodes and all generations of Infiniband (DDR/QDR/EDR) equipment, and routing all these copper cables without damaging them is not easy. The quality of digital tranmsission over copper cables is dependent on the size and clarity of "eye", which is the pattern which two sinusodial signals when run through and oscilloscope. Lower quality cables (HDMI/DP/Coax/etc.) has blurry and small eyes, which increases chances of artifacts and drives the error correction modules harder, while better cables are easier on the both sender and receiver, since they produce lower number of errors to correct. Trivia: Amphenol / Gore has high speed digital interconnect cables called Eye Opener Plus [0], which also used some of the Infiniband cables that we use. [0]: http://www.spectra-strip.com/Eye_Opener_Plus_Cable.cfm?Nav=Eye%20Opener%20Plus%20Cable http://www.spectra-strip.com/Eye_Opener_Plus_Cable.cfm?Nav=E...
mbruce 8y ago
Sorry, copper is not ‘noisy’! Any medium has limits and methods to properly use it. E.g. shielding / grounding. Have to look at whole system. Agree fibre might be more practical in some situations.

lmilcin 8y ago

Digital transmission/storage ensures the signal is not affected no matter how much noise is on the line (well, as long as it can be transmitted at all...) Now, what I think you mean is that the copper in the cable picks up noise either from the device on the other end or from the surroundings and that has the possibility of affecting the DAC. Even moderately well designed receiver should be able to galvanically isolate itself from outside sources of noise without having to resort to fiberoptics or wireless. It's just couple of dollars max for extra few components on the board.

TylerE 8y ago
> Digital transmission/storage ensures the signal is not affected no matter how much noise is on the line (well, as long as it can be transmitted at all...) There is no such thing as a digital transmission. All signals are analog.
- mrob 8y ago
  Digital transmission refers to the whole transceiver system, not just the transmission medium. If I can put a "1" in and reliably get a "1" out, it doesn't matter that it was only a 0.9 part way down the wire.
  - TylerE 8y ago
    But that's the question, isn't it? If 0.0001% of the time you get 0.45v instead of 0.9v on the far end, now what?
    - lmilcin 8y ago
      I suppose you need to read up on digital protocols, it is all well handled. If it wasn't there could be no Internet or digital media or even processors or motherboards.
      - TylerE 8y ago
        That is a higher level, though. I'm just so tired of all the bullshit where it "Must be a 1 or a 0". That is patently not true. HDMI video, to name one example very relevant to this thread, has no error correction, and bit errors will indeed crop up as the "sparkling" artifacts GGGGGP reported.
gnufx 8y ago
For an example possibly familiar in computing circles, Infiniband has easily-accessible counters to record transmission errors. In a properly-built system with hundreds of communication-busy nodes connected with copper, you see remarkably low error rates, even in comparison with the spec (maybe a few a week in my experience).
- bayindirh 8y ago
  Infiniband is using extremely isolated and well balanced cables for passive/copper transmission. Also the cable lengths are very short for EDR and beyond (3m-5m is max usable length). For anything longer you need to use fiber cables with embedded converters. We have a cluster with ~1000 nodes and all generations of Infiniband (DDR/QDR/EDR) equipment, and routing all these copper cables without damaging them is not easy. The quality of digital tranmsission over copper cables is dependent on the size and clarity of "eye", which is the pattern which two sinusodial signals when run through and oscilloscope. Lower quality cables (HDMI/DP/Coax/etc.) has blurry and small eyes, which increases chances of artifacts and drives the error correction modules harder, while better cables are easier on the both sender and receiver, since they produce lower number of errors to correct. Trivia: Amphenol / Gore has high speed digital interconnect cables called Eye Opener Plus [0], which also used some of the Infiniband cables that we use. [0]: http://www.spectra-strip.com/Eye_Opener_Plus_Cable.cfm?Nav=Eye%20Opener%20Plus%20Cable http://www.spectra-strip.com/Eye_Opener_Plus_Cable.cfm?Nav=E...

mbruce 8y ago

Sorry, copper is not ‘noisy’! Any medium has limits and methods to properly use it. E.g. shielding / grounding. Have to look at whole system. Agree fibre might be more practical in some situations.