[ Home | Staff & Contacts | DIY & Tweaks | Listening tests | HiFi Playground | Music & Books ]

Product: big thick wire & a big spike & new sockets
Approximate cost: less than an interconnect
Warning: do not conduct any experiments on your mains wiring.
Do employ a qualified registered professional electrician to make any alterations to your house wiring.
Reviewer: Mark Wheeler - TNT UK
Experimental period: April 1987 to May 2008

In Spring 2005 I began writing the following piece, "Recently several noise problems have plagued my system. There is endless mutation of the system and its component locations due to the regular influx of stuff to review and stuff I'm just experimenting with.

This creates much system upheaval. I take every opportunity to listen to any items others recommend so component changes occur several times every week causing wear & tear to plugs & sockets that compromises the integrity of connections, which is why the pro-sector wouldn't use phono plugs to pierce their cigars.

Colin Yallop of Chevin Audio was visiting in Spring 2005 with the Transcendent Sound Phono Preamp that he has upgraded with Vacuum State Electronics super regulators. This is not a review item as I have never heard the standard product, he just thought I would enjoy hearing it and comparing it to my own modified Naim phonostage and the Concordant Excelsior borrowed from Mark Orr, the proprietor of Something Solid.

As soon as we changed the lp arrangements by moving my butchered Naim the problems started. First we were plagued by high frequency oscillation that sounded like RF intermodulating with the audible band. This happened with all 3 phonostages and we traced it to the Orbe motor power supply. The NC power supply leaves the motor housing unconnected with earth or the system signal ground. My Kimber Silver Streak balanced cable rewire of the Hadcock arm, being unscreened, is not sufficiently immune to RF to avoid interference from the Orbe's new tacho-feedback speed regulator signal. I am using a transformer attenuator to match levels and this had also broken the system ground connections leaving the highest gain stages (the phono pre-amps) no longer referenced to the notional zero volts of the earth spike immediately outside the wall of my 2005 listening room. The Hadcock unipivot arm does not help as the pivot does not make a satisfactory earth (ground) connection, so extra earth links have now been added to various points of the Mitchell chassis and Hadcock arm assembly. These run the shortest possible distance to a common point, using copper wire (lower Z than silver despite gurus trying to rewrite physics) dressed carefully to avoid reducing subchassis suspension movement. That common point is connected by thick low resistance copper wire to the earth screw (a loudspeaker binding post) added to the exterior of the pattress box containing an audio spurs double socket. The binding post is connected to the external 'technical earth' and to the ring-main earth on that circuit. This is OK with the type of UK wiring that I have, and was inspected, tested and certified for 'Part P' by an electrician.

Once the oscillation had been solved we were now faced with an earth loop somewhere as the intensity of hum could be varied by choice of high or low primary combined with equalised settings on the variable secondary of the transformer. Colin had to leave before I had solved the problem. Again a simple solution sufficed. I created a switchable jumper for signal-ground input to metal chassis in the variable transformer case; this allows user selection of ground-lift or fully connected through from input to output. Getting earth connections right is essential. The problems caused by poor earth arrangements are not usually so obvious. Typically the noise-floor of any audio system may be improved by careful dressing of interconnect and mains wires and further improved by inexpensive attention to the mains supply of the system. If there is any credence to widely held belief that 1 metre of mains wire can make a difference, surely improved mains power supply to your system must be an order of magnitude better." That article remained dormant on my hard drive as other reviews took precedence.

Then, in April 2007 I wrote this reply to an enquiry from one of our readers:
"Noise in audio systems is what really separates many of them, and there are many sources of noise and non-linear distortion. One of these is electrical noise caused by the so-called zero-potential reference point of the system (ground or earth in various parts of the Anglophone world) being more or less zero-potential in different parts of the system. When this is wildly different on two sides of an amplification stage we hear the well-known 'ground loop hum', a loud hum at 50Hz or 60Hz (depending on which side of the Atlantic we're suffering the problem) and its various multiples (for you 100Hz, 150Hz, etc).

Most earthing, or grounding, is carried out by an earth bus. This is a thick wire or copper bar through a piece of electrical equipment, to which each active stage is connected to achieve a common zero-potential (0 volts with respect to the earth under our feet) point for the whole piece of equipment. Voltage simply measures the difference between two points, and the easiest way to define a universal point is to say that the ground is defined as zero volts. Hence the impedance Z (resistance ohms, capacitance Farads and inductance Henries combined so that the actual resistance at any given frequency can be calculated) between any part of an audio circuit and the defined ground zero needs to be equal at any part of the circuit to minimise noise. If we plug our audio system into different sockets in the room and connect them with long interconnects we have just introduced multiple connections to earth (in the UK the so-called neutral wire is usually bonded to earth near our electricity meter) with different impedance characteristics for each, thus introducing a noise generating circuit. Within the whole system something approximating star-earthing can be achieved by using a multiple socket for the hifi system with a same-make same-length mains cable from each socket to each audio component. Better yet is something like the hydra-headed connector Naim used to take to hifi shows with a clutch of cables and IEC plugs emerging from one 13A plug (Maplin supply an ideal plug for this). If you are qualified and technically competent to rehouse your hifi components or build DIY from scratch, you can arrange the internal wiring such that all zero-potential connections are taken to one central earth point in each case. This can be connected to a central star-earth point (Russ Andrews is so convinced by this arrangement that he makes a product to enable this). Thus all supposedly zero-potential points in your audio system are the same electrical distance from the zero-defined point that is the earth we stand on."

Do not attempt any electrical work unless you are suitably qualified, competent and sober at the time.

Perhaps some definitions are in order, the words earth and ground are both terms used in the English speaking world to mean zero potential in an electrical system. The names imply that the soil or dirt outside where plants grow is chosen as the defined zero volts point. This is confused by the fact that in most countries our domestic power supply system uses the surface of our planet as the return path for the European 230v and American 120v circuit. Americans use the term ground for all these and English people use the term earth for all of them.

The 3 square prongs of UK (British Standard) domestic plugs are therefore not as straightforward as they appear. Just to confuse the situation, our single-ended (unbalanced) hifi connections (RCA phono plug, not amplifier topology) are spoken of as having signal and ground. For this article the various terms are used thus:
Earth or ground - the stuff outside we grow plants in and any electrical connection directly to it, or to the PME or TT house wiring (UK terminology) that makes multiple connections to water pipes, distribution unit, third core of ring main etc and any other theoretical zero point in domestic mains wiring
Signal return - the zero volt defined parts of an audio circuit connected to the outer ring of the input or output phono sockets usually referenced to the chassis ground or power-supply ground.
Neutral - the 'other' half of the 3 phases of UK electrical supply. Unless your house demand includes a kiln or a large welder your house will be supplied by 1 phase, and each of your immediate neighbours will be supplied by one of the other two. This effectively distributes the load on the grid from your street. The land upon which you walk, or dance, provides the means of completing the circuit for all three phases as well as providing the theoretical zero reference point for 'earth' or 'gound'. Hence the potential problems for audio tweakery.

Many articles have been written about mains upgrading and several companies make a living out of it. The importance of correct house wiring and safety cannot be overemphasised. The Twisted Pair at acoustica.org have produced a good summary of UK mains requirements and audio and concisely summarise the benefits of a separate spur, but if you go to read it now, please do not forget to come back to TNT-audio. Descriptions of the various UK mains wiring topologies are at electrical installation work, or the work must be 'Part P' certified upon completion. Usefully, from the audiophile's perspective, "Installing additional earth bonding" is permitted "conditional upon the use of suitable cable and fittings for the application, that the circuit protective measures are unaffected and suitable for protecting the new circuit, and that all work complies with all other appropriate regulations". My advice is to get everything tested by a qualified electrician. That is what I do.

The catchily titled 'Amendment No 2, 2004' to BS 7671:2001, Requirements for Electrical Installations (the 'IEE Wiring Regulations'), changes the cable core colours of mains wiring. The new (harmonised) colour cables may be used on site from 31 March 2004. If I have understood correctly, and it IS YOUR RESPONSIBILITY TO CHECK ALL FACTS BEFORE embarking ON ANY WIRING CHANGES, new installations or alterations to existing installations may use either new or old colours, but not both, from 31 March 2004 until 31 March 2006. Only the new colours may be used after 31 March 2006. This matters because TNT-audio readers in England and Wales might discover that it will be very difficult to sell their house if they have made uncertified changes to the wiring, which will be easily identifiable by the core insulation colour changes. You have been warned.

Then in October 2005, 10 months after the regulations changed, I moved house to a 200 year-old cottage that has been extensively extended at various times since 1963. It looks like the house had been partially rewired in 1963 (double insulation and steel earth) and partially rewired again (pvc double insulation and copper earth) about 20 years later, and the wiring modified yet again with more building work in 1996. At some time the circuits had provided for night storage heaters (on the same ring) which had later been disconnected, but their extra cable-breaks and screw terminals remain. This is a nightmare of potential noise sources, high impedances, intermittent or semi-conducting connections, for domestic mains supply wiring.

My listening room, being an extension to the original house does not share wiring with any other part of the house. Despite this isolation, the audible haze typical of noisy high-Z wiring seems constantly present, regardless of time of day. The room was probably wired in the 60s and the terminals of the sockets look very discoloured, and the circuit has three, now disused, additional connection points for night storage heaters. This room could really benefit from a dedicated audio circuit with its own low-impedance connection to its own local earth.

My 1998 audio spur did benefit from 1.5metre copper rod driven into the ground in a low, damp, North facing corner outside. This position is as important as the spike itself. It has been demonstrated, by Ben Duncan, among others, that the audio system mains earth must be as low impedance as possible, especially at radio frequencies in order to maximise the attenuation of interference effected by the system screens that are connected to this earth reference. A short spike in a dry position will be less effective than the usual earth system relying on connection to the mains service cable and the water pipes. I have met enthusiasts who, having located their audio system earth in a warm sunny spot just outside their south facing hi-fi room, report that during dry weather, their system sounds better if they water the earth spike!

While this seems like the wildest loony excess of audiophool activity, it is actually grounded (groan) in simple physics and basic common sense. I have not tried it myself.

In order to make some sense of my current (haha) house wiring I asked a qualified electrician, Phil, to visit my house and take a look. I chose Phil carefully as he's an electrician who not only holds the correct NIC EIC certification and Part P registration, he also holds B.Eng. (Hons) and MIEE, making him perfectly qualified to consider the knotty problem of audio mains supply. His first reaction is to ask why any equipment with adequate PSRR (power supply rejection ratio) should be affected by the mains supply at all. He notes that mains is a filthy signal of fluctuating amplitude, but also that no properly designed equipment should be much affected by this, in theory, and his curiosity is piqued by the whole project (sufficiently to travel well outside his normal radius to my house).

My conversation with Phil raises more questions than it answers: here is a qualified electronics engineer, experienced in high-power amplification design, who is surprised by the idea that an extra earth could be audible, but interested to know whether it can be, and we both want to know what mechanisms are at work. One hypothesis might be that it is the close proximity of the extra spike minimising any RF pick-up by the shorter wiring, but this hypothesis ought to be refuted by the irrelevance of RF on the mains to an amplifier with good PSRR.

When I describe to Phil the rewiring I undertook for my audio system at my two previous houses (in 1987 and 1997) he is surprised at the improvements I report hearing. He notes that the usual UK earth arrangement, at the point of electrical service entry to a property, typically exhibits an impedance of approximately 0.4ohms, while an external earth rod may be as high as 20ohms. Thus the net improvement of the extra local earth could be as low as 0.01ohms, or 1.8%. The levels of voltage and current, described by various authors, between house ring-main earth and a separate earth spike, up to a couple of volts and maybe 100mA (one tenth of one Amp) are bad enough in a typically 20ohm earth rod, but imagine how far that current would rise if the the earth rod were maybe 0.2ohms, half that of the ring and one hundredth of the typical rod! Obviously at that voltage the current increases by a factor of 100 to that of an electric heater.

Phil suggests the only reliable way an improvement may be gained by external spikes would be multiple spikes local to the system. Back in the late 80s I did hear of this in a Naim 6-pack system, but I had thought it over-the-top at the time. Phil also believes that a multiple radial wiring arrangement would probably be best, and this has been reported by many audio authors too. However, my previous experiments in this area contradict this, so Phil immediately proposes a solution that combines the best of all possible worlds. Now Phil leaves me with the task of trying to identify which sockets and lights are on which circuits, before we plan the new audio circuit.

My first experiment in 1987 arose after many conversations with audio manufacturers, audio dealers and, fellow audiophiles. When installing the 1998 audio mains-circuit I tried various permutations and combinations of audio spur. One comparison is of two adjacent 20amp cable runs to adjacent sockets connected to the same fuse in the distribution unit. One socket has the usual earth connection arrangements, but not via the RCD (residual current detector), which would add a second variable. The other socket is earthed via the spike outside through a 2metre cable.

Trying the system connected to one or the other socket via a distribution unit was a quick and easy comparison. NO contest. The spike connected system sounds more transparent. Much more transparent. Much, much more transparent. Like going from a cheap 150€ amplifier with a weedy power supply to a good 1000€ amplifier with a power supply big enough for a welder. The soundstage is wider and deeper, individual instruments are more clearly distinguished, bass is tighter and more tuneful, and all for a reel of ordinary house wiring cable, some 12mm earth cable and an earth rod.

At a previous house there had not been a spare circuit in the distribution unit, and rather than buy a whole larger fuse-box, in 1987 I had a second 3 fuse box installed just for the audio system. The idea was to star connect the system back to this distribution unit, which was earthed independently via a spike (again in a damp north facing corner). Each spur terminated in a unswitched Crabtree double socket (noted for their strong terminal springs at the time), with one socket pair for sources (vinyl and FM tuner), one socket pair for pre-amp and active-crossover supplies (both Naim SNAPS), and one socket pair for the 2 stereo power amps (both Naim NAP110). All the mains connections were cleaned and sealed with solder over the screw terminals, to minimise oxidisation. Plug pins were polished. This was going to be state-of-the-art.

When it was all connected I played a last lp side using the house ring-main supply through two 3-way adapters. I plugged the audio component into the new wiring. It was dreadful. It was intolerable to listen to it. This is not hyperbole. It really was that bad. All the bass had gone, as if a pole of the RIAA equalisation network had been disconnected. It shrieked and sounded distorted too. I reconnected to the house ring main via the adapters, and immediately normal service was resumed.

I returned to the new mains sockets and the awfulness returned. My (then) wife came from another room to demand what I had done to wreck the hifi, she could hear its deterioration through 2 Victorian doors. I plugged the system back into the ring main, via the adapters, and bliss returned...then I had an idea...

I replugged into just one of the new outlets via the two 3-way adapters. Glorious sound emerged from my Decca London Ribbon & Focal 10N501 speakers. I tried moving one of the adapters to another of the new sockets and the sound deteriorated again, but not as much as before. The problem seemed to be the length of loop from component through spur to fuse-box through spur to component. Given that Naim systems earth through the interconnect braids, this was especially surprising. I tried an alternative system of Doug Dunlop modified Quad valve amps through passive speakers, and the effect was far less dramatic. The new spurs now showed a mixture of benefits and drawbacks. Using a the tuner (Denon) instead of the turntable (Linn+Valhalla) with the Naims also produced a less comparative deterioration with the new spurs.

I modified the 3 spurs into a double ring and achieved the best sound yet. This was long before fancy bits of wire became fashionable between wall socket and IEC plug. The audio system was now connected to a low-impedance ring with its own earth (ground) and fuse-box separate from the other lights and sockets of the house. All 3 spurs were now connected to to the 3 double socket outlets, each being connected directly to the other 2 sockets. Naturally this double ring (effectively a figure of 8) was now protected by a single fuse instead of a separate fuse for each spur. My experience of ring vs multiple spurs is the opposite of some experiences published on the web so remember that YMMV. This circuit produced the lowest impedance between sockets, and the lowest dc resistance to earth (ground), and it was checked by an electrician.

The Naims sang like never before when connected to this new close coupled ring. Compared with the two 3-way adapters on the ground-floor ring-main (even with every other appliance disconnected) the effort was finally worth it. Rhythms were tighter, more coherent and more explicit. The soundstage was much deeper (really obvious using the Chesky drums-in-a-basement recording) extending way behind the back wall of the listening room (I used to imagine where in the garden an instrument was being played). The nuances of playing of each individual instrument became easier to detect in a complex passage.

That the Naim active setup was so upset by the wrong mains wiring arrangement seemed very curious and I borrowed an assortment of gear to explore further. My friends & I found that good 80s solid-state component were much more sensitive to the mains changes than classic, or 80s, valve (tube) equipment. My conclusions that a ring main is superior to three separate spurs was reached by extensive listening through various brands and topologies of equipment, especially amplifiers. I do know that there are others (like Russ Andrews) who have reached the same conclusion after extensive experiments.

However, I am also aware of several other audio experimenters with established track-records and impeccable reputations, who vehemently disagree with the 'ring main is best' conclusion. They too have tried ring main vs spur and concluded that spur is markedly superior over ring. I have tried the single spur vs dedicated audio ring in that first house. I tried to make the two arrangements as equal as possible in terms of total impeder back to the distribution unit (in the adjacent garage - hence the ease of such comparisons with fairly short wire lengths, less than 8 metres). The outcome was that the single spur was much more dependent on the distribution block or Hydra, but that generally the ring was more likely to be superior. There were times and particular arrangements of brands (Naim passive crossover being a case in point) or types of IEC mains lead that rendered the single spur (using cooker cable dressed alongside the tripled ring) obviously superior, but most times with most equipment the ring wins, either subtly or dramatically.

When a house move became imminent, I knew an early priority would be a dedicated audio ring and a full house rewire.

The dedicate earth should have very little effect, theoretically. A dedicated earth rod alone as the sole earth point for an audio system should be obviously inferior to the house ring main earth, because its impedance is so much higher. However, experiments have proved that such an individual higher Z earth not tied to the house earth also sounds better than the house earth. All theory suggests that the opposite would be true. My own experience is that adding a 'technical earth to an audio system wrings a dramatic improvement to it.

The choice of 10mm^2 earth cable for my earth spike is technically superior to the existing twin and earth ring main gauge, but it is connected to a higher impedance termination than the point at which one's mains power enters the house. This 10mm cable will have sufficient inductance to render its advantage negligible at RF. Perhaps the advantages are in the 20Hz-20kHz passband providing a more stable defined zero point, but again this argument is undermined by the simple fact that even at 1kHz the house wiring is likely to have lower resistance to ground. Both my earlier experiments were in houses where the technical earth connection required less than 3m of earth cable to a damp spot in the garden. My new home needs over 15m of zig-zagging earth cable to reach a damp cellar floor as the south-west facing windswept hillside outside my house is permanently parched, even during rainfall. Future installments of this treatise will explore how to overcome these local problems.

Conclusion

© Copyright 2008 Mark Wheeler - www.tnt-audio.com