Ground Loops: How They Work and How to Deal with Them
A ground loop is created whenever two or more pieces of mains-powered equipment are connected together so that mains-derived AC flows through shields and ground conductors, degrading the noise floor of the system. The effect is the worst when two or more units are connected through mains ground as well as audio cabling, and this situation is what is normally meant by the term “ground loop.” However, ground currents can also flow in systems that are not grounded galvanically; they are of lower magnitude but can still degrade the noise floor, so this scenario is also considered here.
Ground currents may either be inherent in the mains supply wiring (see Section 13.3.1) or generated by one or more of the pieces of equipment that make up the audio system (see Sections 13.3.2 and 13.3.3).
Once flowing in the ground wiring, these currents will give rise to voltage drops that introduce hum and buzzing noises. This may occur either in the audio interconnects or inside the equipment itself if it is not well designed.
Here I have used the word “ground” for conductors and so on, whereas “earth” is reserved for the damp crumbly stuff into which copper rods are thrust.
Hum Injection by Mains Grounding Currents
Figure 13.2 shows what happens when a so-called “technical ground,” such as a buried copper rod, is attached to a grounding system that is already connected to “mains ground” at the power distribution board. The latter is mandatory both legally and technically, so one might as well accept this and denote as the reference ground. In many cases this “mains ground” is actually the neutral conductor, which is only grounded at the remote transformer substation. AB is the cable from substation to consumer, which serves many houses from connections tapped off along its length. There is substantial current flowing down the N+E conductor, so point B is often 1-V rms or more above earth. From B onward, in the internal house wiring, neutral and ground are always separate (in the United Kingdom anyway).
Two pieces of audio equipment are connected to this mains wiring at C and D and are joined to each other through an unbalanced cable F–G. Then an ill-advised connection is made to earth at D; the 1-V rms is now impressed on the path B–C–D, and substantial
current is likely to flow through it, depending on the total resistance of this path. There will be a voltage drop from C to D, with its magnitude depending on what fraction of the total BCDE resistance is made up by the section C–D. The earth wire C–D will be of at least 1.5 mm2 cross section, and so the extra connection FG down the audio cable is unlikely to reduce the interfering voltage much.
To get a feel for the magnitudes involved, take a plausible ground current of 1 A. The 1.5-mm2 ground conductor will have a resistance of 0.012 Ω/m, so if the mains sockets at C and D are 1 m apart, the voltage C–D will be 12 mV rms. Almost all of this will appear between F and G and will be indistinguishable from wanted signal to the input stage of unit 2, so the hum will be severe, probably only 30 dB below the nominal signal level.
The best way to solve this problem is not to create it in the first place. If some ground current is unavoidable, then the use of balanced inputs (or ground-cancel outputs—it is not necessary to use both) should give at least 40 dB of rejection at audio frequencies.
Figure 13.2 also shows a third earthing point, which fortunately does not complicate the situation. Metal water pipes are bonded to the incoming mains ground for safety reasons, and because they are usually connected electrically to an incoming water supply, current
flows through B–W in the same way as it does through the copper rod link D–E. This waterpipe current does not, however, flow through C–D and cannot cause a ground-loop problem. It may, however, cause the pipes to generate an AC magnetic field, which is picked up by other wiring.
Hum Injection by Transformer Stray Magnetic Fields
Figure 13.3 shows a thoroughly bad piece of physical layout that will cause ground currents to flow even if the system is grounded correctly to just one point.
Here unit 1 has an external DC power supply; this makes it possible to use an inexpensive frame-type transformer, which will have a large stray field. However, note that the wire
in the PSU that connects mains ground to the outgoing 0 V takes a half-turn around the transformer, and significant current will be induced into it, which will flow round the loop C–F–G–D, and give an unwanted voltage drop between F and G. In this case, reinforcing the ground of the audio interconnection is likely to be of some help, as it directly reduces the fraction of the total loop voltage that is dropped between F and G.
It is difficult to put any magnitudes to this effect because it depends on many imponderables, such as the build quality of the transformer and the exact physical arrangement of the ground cable in the PSU. If this cable is rerouted to the dotted position in the diagram, the transformer is no longer enclosed in a half-turn, and the effect will be much smaller.
Hum Injection by Transformer Stray Capacitance
It seems at first sight that the adoption of Class II (double-insulated) equipment throughout an audio system will give inherent immunity to ground-loop problems. Life is not so simple, although it has to be said that when such problems do occur they are likely to be much less severe. This problem afflicts all Class II equipment to a certain extent.
Figure 13.4 shows two Class II units connected together by an unbalanced audio cable. The two mains transformers in the units have stray capacitance from both live and neutral to the secondary. If these capacitances were all identical, no current would flow, but in practice they are not, so 50-Hz currents are injected into the internal 0-V rail and flow through the resistance of F–G, adding hum to the signal. A balanced input or ground- canceling output will remove or render negligible the ill effects.
Reducing the resistance of the interconnect ground path is also useful—more so than with other types of ground loop, because the ground current is essentially fixed by the small stray capacitances, and so halving the resistance F–G will dependably halve the interfering voltage. There are limits to how far you can take this; while a simple balanced input will give 40 dB of rejection at low cost, increasing the cross-sectional area of copper in the ground of an audio cable by a factor of 100 times is not going to be either easy or inexpensive. Figure 13.4 shows equipment with metal chassis connected to the 0 V (this is quite acceptable for safety approvals—what counts is the isolation between mains and everything else, not between low-voltage circuitry and touchable metalwork); note that the
chassis connection, however, has no relevance to the basic effect, which would still occur even if the equipment enclosure was completely nonconducting.
The magnitude of ground current varies with the details of transformer construction and increases as the size of the transformer grows. Therefore, the more power a unit draws, the larger the ground current it can sustain. This is why many systems are subjectively hum free until the connection of a powered subwoofer, which is likely to have a larger transformer than other components of the system.
Ground Currents Inside Equipment
Once ground currents have been set flowing, they can degrade system performance in two locations: outside the system units, by flowing in the interconnect grounds, or inside the units, by flowing through internal PCB tracks, etc. The first problem can be dealt with effectively by the use of balanced inputs, but the internal effects of ground currents can be much more severe if the equipment is poorly designed.
C–F–G–D as before. Now, however, the internal design of unit 2 is such that the ground current flowing through F–G also flows through G–G’ before it encounters the ground wire going to point D. G–G’ is almost certain to be a PCB track with higher resistance than any of the cabling and so the voltage drop across it can be relatively large and the hum performance correspondingly poor. Exactly similar effects can occur at signal outputs; in this case the ground current is flowing through F–F’.
Balanced inputs will have no effect on this; they can cancel out the voltage drop along F–G, but if internal hum is introduced further down the internal signal path, there is nothing they can do about it.
The correct method of handling this is shown in Figure 13.6. The connection to mains ground is made right where the signal grounds leave and enter the units and are made as solidly as possible. The ground current no longer flows through the internal circuitry. It does, however, still flow through the interconnection at F–G, so either a balanced input or a ground-canceling output will be required to deal with this.
Balanced Mains Power
There has been speculation in recent times as to whether a balanced mains supply is a good idea. This means that instead of live and neutral (230 and 0 V) you have live and the other live (115 V–0–115 V) created by a center-tapped transformer with the tap connected to neutral (see Figure 13.7).
It has been suggested that balanced mains has miraculous effects on sound quality, makes the sound stage ten dimensional, etc. This is obviously nonsense. If a piece of gear is that fussy about its mains (and I do not believe any such gear exists) then dispose of it.
If there is severe radio frequency interference (RFI) on the mains, an extra transformer in the path may tend to filter it out. However, a proper mains RFI filter will almost certainly be more effective—it is designed for the job, after all—and will definitely be less expensive.
Where you might gain a real benefit is in a Class II (i.e., double-insulated) system with very feeble ground connections. Balanced mains would tend to cancel out the ground currents caused by transformer capacitance (see Figure 13.4 and previous discussion for more details on this) and so reduce hum. The effectiveness of this will depend on C1 being equal to C2 in Figure 13.7, which is determined by the details of transformer construction in the unit being powered. I think that the effect would be small with well- designed equipment and reasonably heavy ground conductors in interconnects. Balanced audio connections are a much less expensive and better way of handling this problem, but if none of the equipment has them then beefing up the ground conductors should give an improvement. If the results are not good enough, then, as a last resort, balanced mains may be worth considering.
Finally, bear in mind that any transformer you add must be able to handle the maximum power drawn by the audio system at full throttle. This can mean a large and expensive component.
I would not be certain about the whole of Europe, but to the best of my knowledge it is the same as the United Kingdom, that is, not balanced. The neutral line is at earth potential, give or take a volt, and the live is 230 V above this. The three-phase 11-kV distribution to substations is often described as “balanced” but this just means that power delivered by each phase is kept as near equal as possible for the most efficient use of the cables.
It has often occurred to me that balanced mains 115 V–0–115 V would be a lot safer. Since I am one of those people that put their hands inside live equipment a lot, I do have a kind of personal interest here.