AES/EBU Cables - what to use?

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Hrodulf, Jan 18, 2021.

  1. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    It's all about wavelength versus conductor length. Transmission line effects start to exhibit themselves once signal wavelength is shorter than the conductor. There are also special cases for half, quarter and eighth-wavelength transmission lines. For electricity at audio frequencies wavelengths are in kilometers, so the wave comfortably "fits" into the cable and it "just" conducts. For transmission lines to comfortably carry the signal you need impedance matching, so the conductor presents "no obstacles" for the waves traveling in the conductor. For example, for SPDIF you need 75Ohm coax cable and matched 75Ohm BNC connectors, so the signal "sees" no obstacles. Now, if there are impedance mismatches, a portion of the wave will be reflected back to the transmitter where it is "caught". A worse scenario is when you have an incorrectly terminated cable at both ends and the reflection can bounce multiple times until it dissipates.

    The question is, does it matter? I haven't got a definite answer. I'm sometimes anal about doing thing the right way, so having incorrectly terminated HF lines would bug me. At the same time Schiit and Hypex, both companies headed by proficient engineers, have no qualms about using RCA as SPDIF connectors. I would imagine that for RF and video this would matter more.
     
    Last edited: Jan 19, 2021
  2. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    Our testing confirms this difference. But it also confirms it is very, very slight (less than1/4 of a picosecond).

    PI2AES XLR into an AP 2722:
    aes_jitter_xlr.PNG

    BNC:
    aes_jitter_bnc.PNG

    And OPTO for comparison:
    aes_jitter_opto.PNG
     
  3. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
  4. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:

    What is the vertical scale on these pictures? As important as the pulse reflections are, if they do not change the sample point on the signal then they are not affecting the downstream clock recovery. A better picture if they had one would be what's called an eye diagram. That will let you see whether any signal disruptions impact the signal integrity. the reflections may look ugly but they may or may not actually be relevant.

    But, really good find and thank you for bringing it to our attention. That's why this forum is so much fun!
     
  5. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    Sorry, that's all there ever was. Again, as you say - this all might be more objectionable to our eyes than to our ears.
     
  6. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,467
    Trophy Points:
    113
    Location:
    Winnipeg
    Wavelength is only the partial answer. Even at digital signal frequencies, your wavelengths are still going to be much longer than the conductor length in a typical home setting. However, signal shape is affected. Your digital signals are usually square waves. While mentally we recognize this as an on/off signal, mathematically it is far more complex with many many higher order odd harmonics. It's when these harmonics start getting rolled off or shifted or whatever due to impedance mismatch/filtering/etc that you lose your square save. You get things like jaggedy or rounded or triangular waves instead.

    Now think of that poor receiver chip at the end of the line. He ain't too bright and hasn't had his morning coffee yet. His incredibly dull job is to push a button when he sees you raise your hand. You on the other hand are drunkenly dancing flailing YMCA at double time while jumping in a bouncy castle.

    http://www.dmx512-online.com/dmx512_electrics.html
    https://en.wikipedia.org/wiki/Square_wave
     
  7. OJneg

    OJneg The Most Insufferable

    Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,923
    Trophy Points:
    113
    Location:
    Grand Rapids, MI
    Thank you for demonstrating, and putting things in perspective.
     
  8. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    You are welcome. One thing I should add though is that the cables I used were inexpensive Amazon brand cables, but they were only 3 to 4 ft in length each. For longer cables it could very well be worth a little bit more expensive.
     
  9. Andre Y

    Andre Y Friend

    Pyrate
    Joined:
    Oct 4, 2015
    Likes Received:
    220
    Trophy Points:
    43
    Location:
    Southern California
    The other thing besides mismatched impedance that's not great about XLR termination for high speed signals is that the shield does not form a 360-degree full circumference cover around the signal wires, unlike a crimped BNC or one of those crimped RCAs. Instead a pigtail is made from the shield to connect it to the XLR pins, and not only does that break coverage, but now also forms an antenna that can both receive and transmit noise. Neutrik a long time ago made an XLR connector that you could crimp around the full circumference of the shield, but I don't think it was in production for very long or very popular.

    If you have to use RCAs, the Canare RCAPs that Blue Jeans uses are as close to 75 Ohm as you're going to get an RCA connector, and they have the full shield crimp too. But if you're going to be be OCD, use BNCs on both ends with a good cable.
     
  10. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,467
    Trophy Points:
    113
    Location:
    Winnipeg
  11. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    Here are eye patterns I ran today on a few cable types.
    Setup:
    dScope digital output to <cable under test> to dScope digital input

    AES XLR Belden 1800F
    01 Digitral Loopback XLR Belden 1800F Eye.png
    Difference between AES and spdif BNC is slight at best. Better noise immunity is likely for 6Vpp balanced system vs 1Vpp unbalanced shown below.

    spdif BNC Canare LV-61S
    02 Digitral Loopback BNC Canare LV-61S Eye.png

    spdif RCA Audioblast HO-1
    03 Digitral Loopback RCA Audioblast HO-1 Eye.png
    Slope at zero crossing is slower than BNC cable setup. Jitter in the second crossing is wider than BNC cable.

    toslink Opt generic plastic
    04 Digitral Loopback Opt generic plastic Eye.png
    This last one has an offset for the zero crossing which implies duty cycle distortion in the toslink connection.

    Understanding Eye pattern diagrams
    https://rintintin.colorado.edu/~gifford/5830-AWL/Anritsu Eye Diagram.pdf

    *edit* 20210120 added jitter examples

    spdif BNC Canare LV-61S with 10 nS jitter wide-band noise added to carrier
    05 Digitral Loopback BNC Canare LV-61S Eye + 10nS jitter wide band noise.png

    spdif BNC Canare LV-61S with 50 nS jitter wide-band noise added to carrier
    06 Digitral Loopback BNC Canare LV-61S Eye + 50nS jitter wide band noise.png
     
    Last edited: Jan 20, 2021
  12. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    I personally believe the benefits of the balanced XLR cable (noise immunity mainly) outweigh the impedances mismatches that occur.
     
  13. JayNYC

    JayNYC Friend

    Pyrate Contributor
    Joined:
    Jun 10, 2016
    Likes Received:
    39
    Trophy Points:
    18
    Location:
    Miami Beach, FL
    This thread is awesome. Ok, next nervosa class question after the connector impedance sub topic: Is there a correct or optimal length for an AES cable?

    I feel like over the years a bit of lore has developed that 1.5M is the optimal length for a SPDIF cable terminated with RCAs. Curious if AES cable is same or different. Also curious if the esteemed readers of this thread have an opinion on 1.5M being actually better than 1M or .5M for RCA spdif.
     
  14. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,467
    Trophy Points:
    113
    Location:
    Winnipeg
    My understanding is that shorter is better for cables of all types.

    1.5m is probably close enough to 5ft that it is just a coincidental pair of number that work for both measurement systems and is of a length that's appropriate for most home setups. You don't want it too short or it might put strain on your connectors. For the DIY folks it's also an easy arm span when you're doing quick estimates off spools of cable.
     
  15. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    Like all things audio, there is no hard stop for any item. That is no length at which it all works until you go just beyond it! No impedance mismatch that stops working all the time. Everything is a continuum and signals degrade more or less linearly.

    I can however, say with the certainty that any output of the PI2AES can drive 3 meters (10ft) with good quality cables. In fact, balanced AES via the XLR is designed to drive 44.1K/48K (~6.2Mhz bit clock) up to 100 meters per the spec! It has a higher voltage output of 4V versus BNC of 1V and RCA of 0.6V. This allows for much greater length. Eventually it's the capacitance of the cable slowing the edges down, not voltage drop, that will start to cause failures. But that is well beyond 10 meters.
     
  16. JayNYC

    JayNYC Friend

    Pyrate Contributor
    Joined:
    Jun 10, 2016
    Likes Received:
    39
    Trophy Points:
    18
    Location:
    Miami Beach, FL
  17. Baten

    Baten Friend

    Pyrate
    Joined:
    Mar 18, 2018
    Likes Received:
    1,130
    Trophy Points:
    93
    Location:
    EU
  18. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    added jitter examples to post #34 above

    Oscilloscope displays for spdif BNC cables of various lengths and types at 44 KHz with 75R termination at oscilloscope input.

    spdif 44 KHz BNC Canare LV-61S 50cm
    07 spdif 44 KHz BNC Canare LV-61S 50cm oscilloscope.png

    spdif 44 KHz BNC Custom Cable Hi Def RG6 30cm
    08 spdif 44 KHz BNC Custom Cable Hi Def RG6 30cm oscilloscope.png

    spdif 44 KHz BNC Custom Cable Hi Def RG6 100cm
    09 spdif 44 KHz BNC Custom Cable Hi Def RG6 100cm oscilloscope.png

    44 KHz RCA Audioblast HO-1 65cm
    10 spdif 44 KHz RCA Audioblast HO-1 65cm oscilloscope.png

    From a time domain perspective the differences appear very minor.
     
  19. Andre Y

    Andre Y Friend

    Pyrate
    Joined:
    Oct 4, 2015
    Likes Received:
    220
    Trophy Points:
    43
    Location:
    Southern California
    Nice. I had no idea they were this readily available now. Does anyone know any cable makers who use them?

    One cool thing to note in the FXCC part are the little spring knuckles on the inner barrel that's used to make sure the shell makes good electrical contact with the receiving XLR port.

    In theory, that's definitely an advantage but as with so many things, it comes down the quality of implementation. For example, to maintain balance up to the multi-MHz region that digital audio transmission uses, the physical relationship of each balanced leg to each other and ground has to be maintained very carefully, which means the cable should be using something like Belden's bonded twisted pair where the two inner wires are basically glued to each other to maintain high common-mode impedance for high frequencies.

    But I think at the end of the day, a lot of this stuff is probably moot especially for people like me who use RPi streamers, because no matter how well shielded your cable is, the Pi itself is completely open and potentially has tons of noise ingress and egress.
     
  20. lehmanhill

    lehmanhill Almost "Made"

    Contributor
    Joined:
    May 3, 2018
    Likes Received:
    546
    Trophy Points:
    93
    Location:
    Home of Jiffy Mix
    And there is something new that is said to be an improvement on the FXCC. That is the Neutrik FXX-EMC. As mentioned earlier, if you are in the US, Redco Audio will build a custom cable to your length, your choice of cable, and with the the EMC XLR or other connectors.
     

Share This Page