The EMI Project – Part 2 Point Source Field (Magnetic Field)
Point Source Field (Magnetic Field) – Dale Shirk addresses how magnetic interference gets into our audio circuits and how to keep it out.
We’ve all learned about electromagnets and magnetism generating current in a wire in grade school. But we may not have given much thought to how magnetic interference gets into our audio circuits and how to keep it out.
Figure 1 – Magnetic Field source orientation to conductors.
In Figure 1 the magnet is an alternating magnetic source and the victim wire below is a twisted pair, plus and minus marked for identification, although it’s also an AC circuit. In part “A” the positive wire is closer to the source and cuts through more magnetic lines than the negative wire. In “B” both wires are equally energized by the magnetic field. Although this may generate a common mode signal the differential amp at the receiving end will subtract it out, if the impedances are balanced. In “C” the negative wire gets more interference. Two conclusions should be drawn from this.
- 1. If the plus and minus wires have equal exposure to the interference source, there should be equal interference signal on both and it will cancel out. This will more likely be the case with greater source distance and/or with tighter twist rates.
- 2. The pickup of interference is sensitive to the location along the length of certain types of wires. Location “B” should give a deep null. Figure 1 D shows a diagram of star-quad wire. There are 4 conductors. The two pluses get tied together as do the two minuses. This bundle is also twisted. This type of construction is less susceptible to magnetic interference from close distances. It also has less variation in susceptibility as the source is moved along it’s length.
A series of tests were conducted using a Han-D-Mag brand tape head demagnetizer as a source. Several types of wire were tested at various distances (Fig. 2). The tests confirmed all of the above postulates.
The tests consisted of driving the demagnetizer at a constant 120 Vrms, at frequencies from 40 to 10 kHz. Due to it’s high inductance it would have been impossible to drive it with a constant current source.
This resulted in the magnetic field strength (Fig. 3) falling as frequency goes up, but since the magnetic coupling is controlled by the rate of flux change, and the flux changes faster at higher frequencies, the result is flat frequency pickup by the victim wires. Since there were no significant frequency variations (all were straight, horizontal lines), the average of all frequency data points was taken for the summary graph.
The summary graph (Fig. 4) shows the results with 6 different wires at incremental distances. The wires used are:
The noise floor for all measurements is about -116 dBV.
The results are clear. Don’t use zip cord for audio. The rest of the cables had interference levels near or below the noise floor of a normal, line level connection. If you scale everything by 40-60 dB for mic level things get interesting. Star quad may be beneficial on stage where the wire may be exposed to very close magnetic sources, such as line lump power transformers, but the benefit rapidly diminishes with increasing distance. In the wall or in the conduit careful routing should keep the wire away from magnetic sources and “ordinary” twisted pair is fine. Last, it’s apparent why CAT5 seems to work just fine in many situations, especially where shielding for RF is not needed. Overall diameter is also a major factor, and is one reason why CAT5 does so well. Smaller diameters have more closely spaced conductors. Twisting rules! ds
