This month Ben Duncan ends the mains supplies saga with a further word on earthing.

Here's a 9-step guide to building a clean earth, for readers who don't live ten floors up.

1. You'll need access to the soil outside. Even in a heavily built up area, there'll usually be gaps in the concrete jungle adjacent to the building but steer well clear of pathways as tripping over a hunk of cold metal in the dark is not much fun! Also look out for existing electricity board or Buzby earthing stakes. You should steer well clear of these, preferably placing your stake on the opposite side of the building, to avoid contamination.

2. A suitable earthing stake can be obtained from an electrical wiring contractor or electrical wholesaler. The type required is a BICC 'Bi-clad,' type BL 70165-01, or equivalent. This model consists of a solid steel core, clad with copper, and comes with various accessories, in particular, a pointed steel tip to make the task of pile-driving easier. It's also extendable, so that when the first 4 ft length has been rammed in up to the hilt, you can add-on a second 4 ft length. It's good to go as far down as possible, but beware of over-enthusiasm because earth stakes are difficult to remove, and a 2 ft length poking out of the lawn (because it won't go down any further) can look a little unsightly. Indeed, in some stony soils, even driving a 4 ft length fully into the ground can prove tiresome. It's also advisable to make an intelligent assessment of the likely course of any pipes lying underground. If you hit something solid about 2 or 3 feet under, it may well be the water main, the sewage pipe or the gas main, and none of these take kindly to being punctured.

3. To drive the stake into the ground, you'll need to borrow a sledge-hammer. A 'tamping bar' may also prove useful. This is a hunking great steel bar, held horizontally with both hands. Meanwhile, the rod itself will need to be fitted with a steel driving cap, otherwise the end will splay-out, when walloped.

4. Don't drive the stake underground; if you leave a couple of inches (excluding the driving cap) above the ground, the connecting wire will be less likely to corrode, easier to inspect, and visible if it falls off! With the stake in place, it's time now to spruce up the copper surface, adjacent to the clamp, with wire wool.

5. Connecting the cable comes next. The wire chosen will need to have a low resistance and be rugged enough to withstand any abuse that's meted out to it over its vulnerable passage to the building. Attacks by lawnmowers, well-aimed spades and local traffic, either human or motorised, may be anticipated. Of course, it helps if you can lead the cable down a deep gap between the concrete or tarmac. Or, for the flower bed or lawn, we can thread the wire through a length of steel water pipe.

6. But what of the actual cable? Use the thickest one you can afford; 80/0.4 (80 strands of 0.4mm diameter wire), rated at 60 amps, is a healthy size for the average technical earth, and costs around 75p per metre. For best results, the wire must have green insulation, (of course)!

7. Now it's time to strip back the wire and clamp it firmly to the stake. Note that the recommended BICC stake comes with a heavy duty clamp, secured with high tensile bolts, instead of the poxy and ill-befitted strip-clamp used on domestic earth stakes. Using a real spanner (please leave the Taiwan adjustable in the workshop) will help you attain the pressure necessary for a gas-tight seal.

The idea is to clamp the wires so tightly that there's no room for a surfeit of oxygen molecules; no oxygen, no corrosion. The end result is that whilst the exposed surface copper may corrode badly, the internal connection will remain good for several years. Be sure, however, to leave a little slack at the stake, so you can cut back the wire and remake a fresh connection should this become necessary in the future.

8. Once inside the building, you'll need to terminate the monster cable, and split out to several smaller feed wires to the various 13 Amp outlets requiring the services of a clean earth. At this stage, we recommend you consult a knowledgeable electrician. The ELCB (Earth Leakage Circuit Breaker) that's often connected in line with the mains earth should be avoided if at all possible. Instead, ask if a 10mA RCCB (Residual Current Circuit-Breaker) can be used, as this will avoid introducing a high resistance into the technical earth. To cut costs, a friendly electrician may be willing to advise you initially on the work to be undertaken, then return later to inspect and test your work, charging you for this alone. In any event, any modifications you make to the earth wiring must be tested to the appropriate IEE regulations (for your own safety) before the power is reinstated.

9. If you didn't manage to push more than 4 ft of earth stake into the ground, your technical earth may fade away in dry weather.

Temporary PA system earths benefit from the need for engineers to continually dispense with the by-products of Stella Artois. Rich in salts, it keeps the soil around the stake moist and conductive, but for a permanent location, this technique may lead to some distasteful rumours in your neighbourhood and can lead to unpleasant smells. To appear civilised, we can simply go for the watering can or lawn sprinkler when dry weather threatens. Dissolving some salt in the water will certainly improve ground conductivity, but can dramatically accelerate corrosion, meaning the stake would need replacing after a few years.

Ecology

So far, we've looked at the mains as a source of spikes and RF interference. Granted good suppression and filtration, the absence of these effects may reveal one new irritation; voltage fluctuation. The primary mechanism at work here is wiring resistance. Increasing the load on any circuit with a fixed voltage (ie. 240 volts AC) is to say we've decreased the load resistance across the power lines. Given that all practical lengths of conductor have some small, fixed resistance, the lateral by-product of a smaller load resistance is to emphasise the residual wiring resistance, meaning that a greater proportion of the available power is being burned up in the conductors. This appears as a voltage drop, at our end of the wire.

Fluctuation is a bulk effect: Switching any single appliance can create spiky perturbations, so spikes and nasty RF noises can occur at any time: Who is to say when the umpteen appliances within interfering distance from your studio are turned on or off? However, significant fluctuations are only possible when a certain number of appliances are operating at the same time, so they're most likely to occur at particular times of the day. (Ask yourself what everyone does first thing in the morning and at 6pm.) Leaving the hedonistic replies aside, televisions, kettles and microwaves have a lot to answer for when it comes to lousing up the mains.

Fluctuation FX

If you've followed the suppression and filtering techniques so far, the stepping effect (Figure 1) of a drooping mains voltage isn't directly audible, essentially because the total change in voltage is relatively small (even an outrageous 10% fluctuation is only ± 24 volts, compared with 100V, up to 2000 volts in the case of a spike (see September H&SR), and the dv/dt (the rate-of-change of voltage) is, by definition, relatively sluggish: a matter of volts per millisecond, rather than per nanosecond. So what's the problem? Well, it's all down to how sensitive your equipment is to voltages, (different from the tidy, precise figure printed on the back panel). For ordinary, analogue electronics, much depends on whether the DC supply is regulated or not, and how tolerant the electronic guts are to voltages aside from the normal. As a general rule, power amplifiers don't have regulated supplies, and some are tolerant to astonishingly low voltages. For example, the Rauch DVT 250s will keep working if the mains sags to an all time 100V low, a 58% reduction! Naturally, power output is somewhat diminished, but that's hardly reasonable criticism. Many up-market power amps exhibit a similar tolerance to variation, essentially because the op-amp topology in use has enough current sources sprinkled about to make it work within ± 10% of full performance, over a 10:1 supply voltage range. Similarly, many IC op-amps can operate from rails of ± 4 volts, up to ± 22 volts. On the hand, auxilliary parts, like fans, may drop-out quite early.

Excepting a few historic specimens, desks, effects, crossovers, control units and all digital equipment have stabilised supplies. That's fine until the unregulated voltage drops, and approaches the (fixed) regulated level. Then the regulator ceases to work, and you'll hear the outcome as a rasping buzz. In NYC, they call it a 'brownout,' in Manchester, it comes across as a 'Regg-you-laiter drop-hout.' Either way, the buzz may come and go, for as the fluctuating main voltage ebbs and flows, the unregulated input to the regulation circuit is lifted above and below the regulator's operating threshold. It's ironic that this should happen, because the actual audio circuitry would probably go on working with further diminuation. In effect, modern power supplies 'regulation' is used more to suppress ripple, (the source of the raspy hum), than regulate per se. Bear in mind that more equipment is rendered useless by loss of ripple rejection, than by outright under-voltage failure.

For tape transports, AC motors are invariably synchronous, meaning their speed is primarily defined by the frequency of the mains supply. This is nominally 50Hz and is kept within reasonably tight limits by the CEGB (Central Electricity Generating Board). It is in fact related directly to the generator revs at the power station and typically varies by ± 5%. For any synchronous motor, too low a voltage means loss of torque: when the mains supply is low, speed is most likely to suffer under heavy loading. In other words, the effects of under-voltage will be most apparent when a 10" spool begins its long rewind trip from one end, but you needn't assume that small voltage changes won't affect playback pitch, if the supply droops at all badly. This is a double-edged sword, because major voltage fluctuations usually come in hand-in-hand with frequency 'adjustments', as the generator revs recover.

This is especially true for hi-fi turntables, but even DC motors, fed from a regulated supply with servo-control will go out of order, if the voltage falls below their critical regulation threshold.

At some point below the regulation threshold, audio circuitry will begin to malfunction. Amplifiers and the filters in EQ and effects units are apt to turn into wild oscillators. Sometimes the oscillation is loud, raucous and extremely rude. At other times, it's inaudible, the ultrasonic sort. We need only wait for the aroma of cooked power amplifier guts or speaker voice-coil drifts across the room to identify this one.

At a still lower voltage, the equipment will eventually give up altogether. The scale of these events is very variable, and well designed equipment is designed to cope on 188 volts, which is the -6% low on the lowest European voltage: 200 volts. On budget equipment, the tolerance is often on the tight side, with a big protest groan coming form the equipment if the supply drops below about 225 volts.

If the equipment designer has observed real mains and it's erratic behaviour, this is usually reflected in the specification, which will state limits (eg. '215 volts to 245 volts') rather than a nominal, spot voltage (eg. '240 volts'), with the subtle insinuation that you shouldn't ask any awkward questions. Here we must be careful to distinguish between the rated voltage and the limits. For like a reserve tank, the low and high voltages aren't for joy-riding, rather they're for your comfort and security, providing leg and headroom.

Voltage dropsy — a dissection

Whenever the mains supply droops, there are three possible causes, each of them cumulative. But before spelling them out, we need to determine a reference point. To begin with, we can state unequivocally that the nationwide supply is a uniform 240 volts. True, odd places across the kingdom have had 200 volts DC, 215 volts AC, and a host of other unusual supply voltages prior to nationalisation. Beginning with the introduction of the grid system in 1924, these idiosyncrasies have been ironed out, although for some of the remotest parts, it wasn't until the early 60s that the grid arrived.

The CEGB and local boards are jointly responsible for maintaining the 240 volts at every premises in the country within ±6% by law. ±6% on 240 volts represents a minimum of 225 volts, and a maximum of 254 volts. This voltage is measured at the point where the board's feeders enter your premises, usually at the meter (or meters).

1. Of the three causes of voltage drop, the first is also responsible for potential uprisings in the supply, above 240 volts. The CEGB switch their generator plants and lines all the time, in order to maintain the voltage and frequency within the limits, regardless of how the national load varies. Clearly, any minute delay in switching plant in and out, and adjusting generator revs after a sudden change in load (20 million people might turn on their kettles at 8am!) can result in short term deviations. If the load falls off abruptly, the voltage will overshoot (254 volts) as an unloaded genny is switched in, and vice versa.

If the load is very heavy, as might be the case at peak time on the coldest night of the year, the CEGB will reduce their voltage as far as they legally can, to 225 volts. Ultimately, the shortfall represents a potential overload on the generators, thus beyond this point, power gets lopped off to an unfortunate few million, for a few hours at a time, so the majority can get by on a supply that's hovering close to the legal minimum.

In case you didn't know, the CEGB is the sole manufacturer and wholesaler of grid juice. Sold in bulk packs of 275kV under the 'pylon' brand, a -6% reduction represents short measure, a mere 260 kilovolts. This has no effect on the cost of electricity, because meters read the true power consumption, and if the supply goes low, power consumption drops pro rata. However, it does mean that the 240 volt (retail!) supply is already tight, without considering more local voltage sags.

2. The local electricity board owns the poles and underground cables that bring juice to every home in the land. It's instructive to estimate the cost of millions of kilometres of chunky copper, steel and aluminium cables. It's certainly a massive capital investment, so they're apt to make the most of it. As with the ring main, this means applying the principle of diversity. In other words, working on the assumption that everyone won't draw their full rated current at the same time. This saves a valuable resource, namely world copper. Nevertheless, there are bound to be times when peak demand synchronises down the road or street. For instance, my own house is fed by a 120 Amp feeder, shared with my neighbour. The rated current for both houses is 60 Amps, defined by the incoming supply fuse. Of course, currents of 90Amps or more can be drawn without blowing it. But let's assume 60 Amps is the maximum. This is fine from a current rating angle: 2 x 60A is 120A, so the feeder cable is within its rating. However, it's also 300 metres back to the substation transformer. This spells a resistance of (0.0006 x 300 x 2) ohms, remembering to add the resistance of the neutral return. This equals 0.36 ohms, which would result in a whopping 43 volt drop at the rated current. Even at a more reasonable current level, the everyday 20 Amps between two houses, the 120 Amp feeder drops 7 volts, out of sheer length.

3. Finally, the wiring between the fusebox and the studio adds its contribution. Here, the peak load is under more control, but in the archetypal home studio, beware of Sunday mornings: Dad's had a bath, so the immersion is on, as is the cooker, the microwave, and the drier. Total current, about 35 Amps.

Taking one ring circuit alone, this can legally cover up to 1000ft². Over a rectangular area, this implies around 46m of 2.5mm² wiring cable. The source resistance is highest in the centre of the ring which means that either way, the current has to travel 23m. The two 23m paths are paralleled, but of course, the current also has to return via the neutral conductors. The source resistance of a full size ring is 0.15 ohms, measured at the sockets in the midpoint of the rings circumference. If the contacts or wires are grimy or tarnished, the exact figure could be considerably more: 0.2 or 0.25 ohms. Even in a ring of smaller nominal area, the actual cable length may be higher than expected, when vertical drops are brought into the calculation. Anyhow, multiplying the calculated 0.15 ohm wiring resistance by the current drawn gives us the anticipated voltage drop. For example, 0.15 ohms x 30 Amps (the design maximum for a ring-main) will cause a 4½ volt drop, so the regular 240 volts becomes 235½ volts. If the studio's current draw is more reasonable, at say 6 amps, the droop will be less of a nuisance, at 0.9 volts. Nevertheless, the cumulative action makes any voltage drop triply inauspicious: if the grid, the local feed cable and the ring all drop 4.8 volts each, then you're right on the -6% limit.

Surges

So far, we've concentrated on voltage reductions, because upward fluctuations are more avoidable and controllable, and so less frequent or severe. The exceptions relate to thunderstorms, and three phase distribution.

Often when there's a storm, protective breakers on the grid are blown out. A few seconds or minutes later, power is restored by switching through another part of the network. With whole sections of the grid being turned abruptly off, then on, and being cross-switched, there's a lot of scope for manic over-voltages. Added to this is the likelihood of having lightning coming up your supply feeder.

As with spikes, voltage surge is a statistical beast. Some people suffer the effects again and again, while others are sure it'll never happen. One thing is certain: it's doubly risky to leave any sort of equipment plugged in during a storm. Don't wait for the power to go off, or bother to switch off: just pull the plug on the whole lot, or reach for the isolating switch.

Second, three phase power distribution can go 'high' if the three (live) phases are unbalanced, and the neutral wire goes adrift. If, for instance, an overhead power line is damaged by high winds. Voltage from any phase to neutral (≈ground) is 240 volts, but between phases, it's 415 volts. If you're really unlucky, a neutral fault can send the supply shooting up to this level. The effect of a steady 415 volts is assuredly most spectacular, and expensive: the Electricity boards don't publicise the fact, but if they shrink from compensating you, you can sue for negligence on three counts:

1. The Electricity Lighting Act 1937. The board has to make full compensation for injuries caused to a third party.

2. The Electricity supply Regulations of 1937. The board has to protect the customer's installation against excessive energy. This is defined as:

3. 'anything outside the prescribed 240 volts, ±6%'.

Investigating fluctuations

The pragmatic approach is to monitor the mains (Figure 2). A digital meter is best, but for real accuracy, it needs to be the sort that measures true RMS vales. Others, there'll be errors if there's any waveform distortion. Data gathering is similar to the RF interference procedure: simply make a note of any readings below about 235 volts, the time, and the date. The investigation can be passive; just jot down readings when you have a moment to notice them. Over a period, you'll be able to build up a picture of how the voltage fluctuates throughout the day, and the days of the week.

If you notice any regular fluctuations below the limit (225 volts), you can ask for matters to be sorted out by the board, who are responsible for any voltage shortfall up to the fusebox. But before doing so, it's a good idea to check that the voltage droop isn't predominantly the outcome of shoddy DIY wiring on your home ground. Given a sound DIY knowledge of electrical safety, you can test for this by contrasting the change in voltage (^dV/_V) at the fusebox, with the ^dV/_V in the control room, as a heavy load is switched on. A portable 2kW fan heater is handy for this test. If ^dV/_V at the fusebox is less than 10% of the control room figure, then we can say that voltage drop in the wires between the two are 90% responsible. In this instance, the electricity board might feel themselves let off the hook, and might then want to charge you for calling them out. But if it's not predominantly your own wiring's fault, there's no charge. Of course, if your studio wiring's been professionally installed, but shows an excessive sag, the installer is ostensibly responsible for putting matters right.

When they arrive, the Electricity board will take some meter measurements. If serious fluctuations are alleged, they'll set up recording equipment, ie. an analyser with a printout. Once the evidence is established, you can expect to see some amendments to the supply, such as heavier cables, or changes to the substation transformer. After paying out £1000s for electricity, this is a deeply gratifying experience.

If you can't wait for the board to arrive, or you want to pursue your own investigation, the answer is to hire a mains disturbance analyser. The classic mode is the Dranetz 626, which is a micro-mainframe, into which various power line analysers can be plugged.

Figure 3 highlights some spot readings on my own domestic supply. The data summary records the maximum and minimum voltage, frequency and impulse level. Impulse is Dranetz's word for a spike or transient. The next printout gives data on these, in this instance showing a series of impulses caused by contact bounce on a thermostat: no excessive voltages here because it's been well suppressed, but over a period of a week, you'd expect to catch the odd monster voltage. It might even happen in the next millisecond.

The third printout shows the fluctuation in action, with the voltage going between 240 and 248 volts over a few milliseconds. Note the readout is from the bottom upwards, because of the way the machine's printer is set. With this amount of detail, it's possible to sketch a detailed graph of the disturbance, which can help the detective work if the source of disturbance is a mystery.

Peak Currents

Big monitoring systems and heavyweight digital (or micro) power supplies can pull surprisingly high peak currents. The same applies to televisions and VDUs. Essentially, the massive capacitative reservoir in top-notch power amplifiers, and the switch mode action common to digital and video gear's supplies tend to concentrate their current demand into a short 'time window'. For example, instead of drawing a steady 4 Amps (for instance), as implied by the rated power consumption, the unit will take very little current for most of the waveform, but when the voltage hits the peak of the waveform, 10, 20, even 50 or 60 Amps is suddenly sucked out of the system. This behaviour creates waveform distortion (Figure 4), and naturally exacerbates any existing voltage drop. For this reason, a cost-no-object studio would incorporate several radial supplies to feed pulsating sources individually. This is an extension of the 'technical supply' concept.

Line conditioning

The 'Variac' (Variable AC) is a continuously adjustable transformer: turning a knob adjusts the supply output from zero volts, up to a slight boost (in the region of 250 volts). In conjunction with a digital meter, it's invaluable for evaluating the behaviour of studio equipment at voltages either side of 240V. It can also be used to make routine, manual adjustments to the supply, but only if the fluctuations are relatively stable and long-lived. Otherwise, it's boring.

More esoteric, the 'Servo-variac' is a piece of electronic archeology that can be picked up cheap at equipment auctions. As the mains fluctuates, the motor tirelessly winds the variac up and down to compensate. The snag is that there's a limit to this reaction speed, so some under and over-shoot must be expected on rapid fluctuations. It also makes a lot of acoustic noise.

Principally because a host of heavy military and industrial digital electronics has also suffered a hard time from the mains, there's a big market today for sophisticated electronic conditioners. Mostly originating from the USA, these are worth considering if your studio suffers intolerable mains hassles, so much that some equipment gets declared unworkable at bad times. However, price per watt is high, so outright conditioning of a 30 Amp ring is usually out of the question. More often, the conditioner serves the studio's digital side only. Products range from basic double-wound isolating transformers with near-zero interwinding capacitance (for effective RFI filtering), through to completely regulated AC mains packs, with programmable output voltage and frequency (so we can select 60Hz or 110 volts, or 220 volts, as necessary, for foreign gear). Uninterruptible supplies take this a step further: by combining a 12V DC to AC invertor, the 240 volt output is derived from batteries whenever there's a power cut, or even a brief glitch in the supply. This level of performance will become increasingly sought after as more studio equipment goes micro and the quality and reliability of the mains supply becomes increasingly intolerable.

To end this series, I'll leave you with three addresses.

The Dranetz 606 line analyser mainframe can be hired from Microlease, (Contact Details).

Sources for mains conditioners include:
Calex Electronics, (Contact Details).
Powerline Electronics, (Contact Details).