The "Wire we Here?" Blog and Forum
A place for SPDIF digital cable discussions.
Ethernet BER is 1E10^10, and that's a REALLY small number. Seldom do we lose bits in serial digital or Ethernet systems. Errors are a result of the DA and AD decisions that impact ringing and other errors but the digital stream once converted by the filters on the cable are super, super reliable. It is the cable ends where the problems lie.
The SPDIF cable's are designed for pure digital, no analog. The copper or silver plated copper signal wires are heavily shielded (better than 100 dB SEED tested noise isolation) and have exceptionally low RL. RL is Return Loss into a FIXED 75-ohm termination load. The loss from RL can be viewed as a form of attenuation as the signal isn't applied across the load and requires a perfectly matched cable impedance to the resistive load and across frequency. RF level SPDIF coaxial cables should not ever be compared to low frequency designs, they share nothing in common except maybe both being a coaxial design. Don't use the SPDIF cables for your analog.
I would like to start by leaving feedback for the spdif coax cable I purchased. Afterwards I have a question that I hope someone can answer.
Of the 3 cables that I purchased I started with this cable first. I already had a spdif coax cable that I had previously purchased from BJC. This was the 1694A Belden with canare connections. Honestly I didn't think it would make much difference between the two. Once again I was wrong. In fact the difference was quite noticable. I used the same media at the same volume to compare and with the BAV iconoclast spdif the music opened up at a higher performance level.
Now here is the question I have. While doing some research on spdif coax cables I ran across a topic on the length of these cables. I don't know if this is a myth or something from the early days of the spdif cables. Anyways the claim was that spdif coax cables should always be no shorter in length than 1.5m/5ft or you would get reflections between your source and your DAC. Does anyone know if this is true? If it is true it certainly doesn't seem to be audible with my system. The BJC 1694A Belden cable I was using was only 2 ft in length. I had to special order that length and the guys at BJC never mentioned any of this and it seemed to work great. The new BAV iconoclast spdif I purchased is 3 ft in length and it seems to be flawless as far as any audible noise or jitter from reflections.
Any truth to this claim of 1.5m/5ft minimum length with spdif coax cables?
@edmund - Yes, there is some validity to the claim. In an ideal world, you'd want the LONGEST wavelength (lowest frequency) harmonic to match the cable length for one full wavelength, what ever that length is and is frequency dependent. This lowers the RL reflections off the 75-ohm resistive load a SPDIF cable is terminated into. That makes it easier on the error correction system. l is a form of attenuation and noise, and that impact the ACR, Attenuation to Crosstalk Ratio margin.
In the real world, good quality 75-ohm precision coaxial cable like 1694R and 4694P will mimimize RL relection such that most lengths will not trigger errors but length to match the longest wavelength is "technically" better, true. That is where the minimum length comes from, the SPDIF frequency in a length form factor recommendation.
The minimum length is more critical on fiber optics as the detector can get swamped with too much light. An in-line attenuator is needed as there is a maximum light budget a detectro can properly encode. It can easly be too good, or high level signal. Most SPDIF inputs anticipate shorter lengths so it is adjusted to not impact the I/O.
As strange as it may seem, SM singlemode fiber is natorious for too strong a signal at the receive end and you need to make sure the signal is padded to match the detector. The signal has very little attenuation and even though it is a small optical power, can swamp the sensitive receiver.
I just tried a BAV audio cable between my cd transport and DAC and am getting surprisingly good performance.
Would you expect better performance from the 4694P?
If you are using the BAV RCA for a digital output, it is a ~ 110 ohm cable designed for analog. The 4694P is a true 75-ohm RF serial digital cable and is better matched to the expected load that is a 75-ohm resistor. The cable to load has lower RL reflections than the BAV. Technically yes, the 4694P is a better cable for DIGITAL where the BAV is the better cable for ANALOG. We don’t try to mix the two applications but optimize to the expected load.
The proper resistive load for RF is matchiung to the cable impedance =SQT(L/C). There is no phase angle at RF as it cancels out. Thus the cable is a pure resistive vector at RF. We can use a high tolerance resistor matching the cable impedance. Analog isn’t like this…read on.
On shorter digital runs, where the RL reflections can be high, the best resistive load match is important. Longer improper cable impedance (too low 50-ohm or too high 110 ohm) cable lengths attenuate the RL reflections more for moderate distance runs but as you get longer, those RL reflections are a form of attenuation and will limit the maximum reach (less signal voltage) to a lesser distance. On a shorter run with improper cable (110-ohm analog), RL reflections are pretty severe and can swamp the error correction.
Analog cable does not need the extreme geometry as digital and we can use AIR TUBE design (the BAV) to lower capacitance as analog IC cable is a voltage transfer function into a high impedance 47-kohm load (not a 75-ohm resistor like RF, big difference) with analog wavelengths that are far, far in excess of the cable length so the RL reflections aren’t a thing for analog cable.
Using the proper 75-ohm cable will allow optimum use from the shortest to the longest runs because the RL reflection losses are removed and the signal strength is maintained over longer runs. Shorter 75-ohm runs also see far less RL signal bounce and the error correction can be less impacted.