Posted by Al Sekela on March 17, 2003 at 19:19:20:
In Reply to: Re: Re: Re: Field report on InnerSound ESL amplifier posted by Wally Benson on March 17, 2003 at 11:34:14:
It is easy to fall into the trap of thinking a low driving impedance is all that is necessary to make an interconnect immune to external noise. I did so for a long time, and the Jensen transformer people still think this is true. However, consider the topology.
A source of some impedance is connected to two wires, which are fixed together in the interconnect cable, either as hot and ground in the case of a single-ended system, or as positive and negative phase, in the case of a balanced pair (there will be a third wire as ground in the balanced case, but it can be ignored for this discussion). These two wires are connected to a load resistor at the receiver. The voltage developed across the load resistor is the signal, as far as the receiver is concerned. It includes the original signal from the source as well as any noise induced on the wires.
Now, visualize what happens if lines of time-varying magnetic flux get between the two wires. An emf is developed that is dropped across the source and load impedances. If the source impedance is zero, then all the emf is developed across the load resistance. In this case, a source with zero output impedance does nothing more than shift all the magnetically-coupled noise to the load.
The noise emf comes from a source with a high effective impedance because the number of flux lines that can penetrate a tightly-twisted pair of wires is small. This is why making the load resistance be 600 ohms is effective in reducing the magnitude of the noise voltage at the receiver. As long as the source is capable of driving 600 ohms, the signal-to-noise ratio is increased.
This is also why interconnect cables are typically twisted or braided: lines of magnetic flux that originate from a distant source penetrate twists almost equally. The twists make a series of loops and the aggregate emf from flux lines that go through the loops with one orientation (say, positive phase on top and negative phase on bottom) almost cancels the emf from lines that go through the loops with the opposite orientation. Almost, but not quite.
Most audiophile interconnect cables are not shielded these days because about ten years ago people found they got better sound, and little or no hum, from unshielded cables. Shielding concentrates the signal's electric fields in the wire insulation, which is not as good a dielectric as air for the sound. Teflon is better than most synthetic polymers, but some people think it still has a deleterious effect on sound. Many cable designs attempt to minimize the amount of any insulation next to the wires.
However, the noise from digital sources is increasing as time goes on. It is no longer automatically true that shielding causes more harm than good. Foil or thin braid shields are ineffective at low frequencies, but that is simply a limitation imposed by cable designs for RF transmission. I have personal experience with very heavily shielded interconnect cables made by Joe Cohen. They deliver the best sound I've heard from my system, even without 600 ohm termination. Adding 600 ohm termination to unshielded cables gave some, but not all, of the benefit of Joe's shielded design. By the way, the same shielded cables make stunningly good digital cables, perhaps because they minimize noise leakage.