Hi All,
I've been reading through a number of threads regarding cable choice for unbalanced analogue interconnects. I have found it stated that the correct choice of cable for this purpose is coax and not shielded twisted pair (which would be the correct choice, according to the threads, for balanced interconnects). What I have been unable to find in the threads or elsewhere is the technical reason as to why. The tone of the threads leads me to think that using STP in an unbalanced interconnect would be detrimental to the signal. However, other threads on related topics such as wiring a phono pre-amp inside its case, led me to think that STP is ideal i.e. there should be a signal wire, a return wire and, if necessary, a shield tied back to signal ground at one end only (thus a shield is left to shield and not carry signal). This initially led me to think that STP would be a good idea for unbalanced interconnects but after reading through the interconnect specific threads I am now not sure. Any assistance in helping me understand why coax is the correct choice for unbalanced interconnects is greatly appreciated.
Many thanks to all.
I've been reading through a number of threads regarding cable choice for unbalanced analogue interconnects. I have found it stated that the correct choice of cable for this purpose is coax and not shielded twisted pair (which would be the correct choice, according to the threads, for balanced interconnects). What I have been unable to find in the threads or elsewhere is the technical reason as to why. The tone of the threads leads me to think that using STP in an unbalanced interconnect would be detrimental to the signal. However, other threads on related topics such as wiring a phono pre-amp inside its case, led me to think that STP is ideal i.e. there should be a signal wire, a return wire and, if necessary, a shield tied back to signal ground at one end only (thus a shield is left to shield and not carry signal). This initially led me to think that STP would be a good idea for unbalanced interconnects but after reading through the interconnect specific threads I am now not sure. Any assistance in helping me understand why coax is the correct choice for unbalanced interconnects is greatly appreciated.
Many thanks to all.
A general rule is currents should not flow in a shield. Maybe not true in the RF world, but true for audio. That leaves you with the shielded twisted pair as the best solution. In one of the Ott books on noise and shielding he has an example of about every possible arrangement of connection between two devices and the expected signal to noise ratio. Need to find that as he's pretty much "the guy" when it comes to this stuff.
In theory, the outer conductor of a coaxial cable is (on average) coincident with the inner, so the two experience exactly the same magnetic hum field, which ought to cancel at the receiving end. In practice, the cross-section of a decent screen is much greater than the inner, so the resistance is lower, so the developed voltages do not cancel. As mentioned in the previous post, you don't want to return audio currents via the screen because they then have hum loop currents added to them. At RF, you have to use a transmission line and terminate it correctly at each end to avoid reflections, and coaxial cable is the easiest way to do it. Ott seconded.
A coaxial or shielded twisted pair cable does not protect against magnetic hum. The outer shield will protect against RF pickup. If your source has very low impedance, you don't need any shield.
Grabbed my Ott- it's Noise Reduction Techniques in Electronic Systems, 2nd ed., C1988. Starting on page 90 he covers the various cable and ground connection schemes. Ott has several books but I find this one most useful for audio. He covers everything you might want to know about grounding, bypassing and low noise techniques. Tends to be expensive unless you find a good used copy, or try 
<link removed at poster's request>
<link removed at poster's request>
Last edited by a moderator:
Sadly Henry Ott is no longer with us. His web pages disappeared shortly after his death.
His 850 page 2009 book "Electromagnet Compatibility Engineering" is rather expensive, but should be a first source for anyone interested in EMI/RFI problems. His older 1988 might be available used for $20.
His 850 page 2009 book "Electromagnet Compatibility Engineering" is rather expensive, but should be a first source for anyone interested in EMI/RFI problems. His older 1988 might be available used for $20.
You are not kidding about "rather expensive". However, EMC compliance testing typically runs at £100/hr and takes two days, so (professionally) anything that avoids retests is worthwhile. The most useful things I learned from going through compliance testing are:
- Even tiny gaps (0.2mm clearance between two parts) let 600MHz in and cause radiated susceptibility failures
- Put copper tape over gaps and ferrites on cables (buy copper tape as "slug tape" for gardeners - much cheaper than from electronics suppliers)
- Think of cables as watertight extrusions between conductive enclosures, and if their extruded screen has a hole in it (pig tails etc), it will leak and fail
- Think low inductance - ground planes and short-as-possible chassis bonds inside the enclosure
- Zeners are slow and do not protect against the mains spike test, but TVS diodes are fast and do
Always felt uneasy about using the shield as a signal wire, yet some rather prominent audio manufacturers do it. Dartzeel is one. Some of their equipment uses 50ohm coax with BNCs and with proper 50ohm source and termination. Subjective response is very favourable compared to more conventional interconnects.