Pete's Place
AC Line Conditioners: Friend or Foe?

by Pete Goudreau

Introduction

Garbage in, garbage out. Isn't that the reason so many have spent so much on power-line conditioners? After all, it seems so reasonable that if the incoming power is "dirty," then the outgoing power must be "dirty" as well. Well, t'ain't necessarily so.

All audio equipment, in one form or fashion, converts the incoming AC power to DC power by means of its power supply. Any noise on the incoming AC line can be filtered out through proper design of the power inlet filter, associated wiring, transformer design, and bulk DC filter design. Seems simple enough but, again, t'ain't necessarily so.

The circuitry within all audio equipment converts the DC power output from the power supply to AC yet again. This AC is the signal, either sourced or processed by the equipment. The equipment is then essentially an AC-to-DC-to-AC converter. It's just that the output AC is a tad different from the simple sine wave of the incoming AC power. In light of this, the power supply should then theoretically act to break the link between the incoming AC power and the outgoing AC power, making the issue of noise on the incoming AC line moot. Or so they say.

PSRR

Power supplies are notoriously difficult beasts to design for extremely quiet operation, electrically speaking, and the circuitry that they power may or may not be adequately capable of rejecting power-supply noise which makes it through, or around, the power supply to the DC rails. This parameter of design is called PSRR, or Power Supply Rejection Ratio, and some circuit topologies have better inherent PSRR characteristics than others.

Unfortunately, PSRR is always frequency-dependent, owing to the nature of the reactive elements and parasitics in any circuit. This frequency-dependence may be adequately broadband so as to remove from the output any noise present on the DC rails, at least over the audio range, but is rarely sufficient to eliminate it entirely.

Should negative feedback be employed in the design, this noise becomes an additive disturbance which the feedback suppresses, as a function of its "excess gain," over its control bandwidth. To be specific, power-rail noise will appear at the output, suppressed by an amount roughly equal to the low-frequency open loop gain, with increasing amplitude, above the open-loop gain's corner frequency, up to the frequency at which the open-loop gain reaches unity. At which point the amplitude of the noise at the output is equal to the amplitude of the noise on the rails. This is the point where PSRR reaches 0dB on the plots that often accompany op-amps.

In circuit topologies that don’t utilize negative feedback, at least not global feedback, the PSRR is the open loop PSRR of the gain stage. This PSRR may or may not be successful in suppressing power-rail noise at the output of the stage. Regardless of the care of the design, however, this open-loop PSRR will still fall off with increasing frequency, which results in increasing noise at the output with increasing frequency.

Differential stages are often used to convert this power-rail noise to a common-mode noise at the two outputs of the stage. This is to say that the noise is equal on both of the differential outputs. Since "differential" implies differencing, the next stage will then subtract like noise voltages by the nature of its design, resulting in zero amplified noise. Another way to say this is that the common-mode gain is unity while the differential gain can be greater than unity. This sounds perfect on paper, but it is not so easily achieved in reality as it's extremely difficult to achieve perfect differencing over a wide frequency range.

The bottom line here is that no matter the topology, no matter the presence or absence of feedback, no matter the care with which a circuit is built, there will always be rail noise on the output, especially at high frequencies. Since the linearity of any circuit decreases with increasing frequency, and much more so with those that employ negative feedback, any high-frequency noise that finds its way from the output of one stage to the input of the next will naturally end up amplified, nonlinearly, by each succeeding stage. This leads irrevocably to intermodulation components (IMD) for any signal that contains anything other than a single pure tone, like music for instance, and these intermod products can and do appear within the audio band. These intermod products are themselves an error. Whether they are audible is another issue and not one to be addressed here.

Power-supply interactions

There is another source of noise that an AC line conditioner can affect and that is the noise generated by the power supply itself in the process of converting AC to DC. In the case of a power supply that uses solid-state rectifiers and capacitor-input bulk filtering, any inductance added by the line conditioner will be reflected through the power supply's transformer to the secondary side where it will add to the leakage inductance of the transformer. This inductance can resonate with the bulk capacitors during the conduction angle of the rectifiers. The resultant effect is one of line distortion and a broadening of the current pulse as well as a reduction of current slew rate.

A power supply of this type when driven from an AC source exhibiting very low inductance would exhibit very high peak-current pulses with very sharp rise times at the point where the rectifier begins to conduct. At the point where the AC waveform reaches its peak, the rectifier shuts off hard. Reverse recovery characteristics of the rectifier can lead to ringing through interaction with its capacitance, as well as that of the transformer's, and the transformer's leakage inductance. Depending on the type of rectifier in the circuit, this noise can be very broadband. This is likely one of the reasons that hyper-fast, soft-recovery rectifiers can often yield a slight improvement in sound quality in certain pieces of equipment, although it should be mentioned that adding appropriate lossy snubbers across rectifiers, or the transformer secondary, can have a similar effect.

Installing an AC line conditioner that has any significant inductance ahead of this equipment will then necessarily change the current waveform to one that approximates a narrow half-sine wave. One of the benefits of this effect is that the current through the rectifier at shutoff is reduced dramatically, as is the slew rate of the current. Thus the reverse recovery-induced ringing and concomitant broadband noise generation are ameliorated.

The drawback, however, is that the distortion of the AC-power sine wave, "flat-topping" as it's called, during the conduction angle, will increase the ripple on the unregulated output of the supply as well as reduce the absolute DC voltage. This additive noise, however, is low in frequency, and if the circuitry it powers has a high PSRR at low frequency, there should be little if any audible effect. The loss in absolute DC rail voltage, however, may cause slight shifts in operating points or limit excursion should this not be taken into account by the designer. This may yield a more audible result that would be strongly dependent on the equipment itself.

While these are small effects, and really shouldn't be considered a major reason for choosing a conditioner, they can, however, help explain why some conditioner topologies exhibit subtle differences in effect between different pieces of equipment. If it were easy, we'd all own a Yorx, I suppose. Then again, maybe not.

Noise sources

From all this, one can reasonably conclude that to design a piece of equipment immune to the effects of high-frequency noise present on the AC power line, from whatever source, is an expensive proposition. This is not to say that it can't be done at reasonable cost; it's just that it's enough of a rarity in the real world, in my estimation, that it's not entirely unreasonable to seriously consider the purchase of what is commonly called a line conditioner. At least it's an easy way to eliminate one potential source of noise and possibly mitigate subtle sources of noise generated by the equipment's power supply. Worth a shot anyway.

This conclusion, of course, assumes that there exists a real source, on the AC line, of noise sufficiently high in frequency to be of concern. But where would this high-frequency noise come from? It can be conducted by the AC mains themselves, picked up by the giant antenna that is the distribution network outside your home. It's in the air as radio waves that we're immersed in every day, everywhere. Radio waves, or RFI, can also be locally generated by high-speed digital equipment such as CD transports, DACs, CD and DVD players, VCRs, TVs, tuners, computers, cell phones, cordless phones, etc.

In the days before digital and TV, we had only a turntable, a preamp with an RIAA stage and a few line stages, a power amp, and a couple of speakers. No locally generated RFI there. Since radio waves' intensity falls off so rapidly with distance from the source, with no sources in close proximity, all the above problems essentially disappear. Not to say RFI pickup was never a problem; it was, but it wasn't nearly the problem it is today. A bad solder joint was pretty easy to fix -- hell to find, but easy to fix. Ever seen a piece of poorly designed digital equipment completely muck up the picture on your TV? Not exactly a solder joint to reflow, is it? 'Nuff said.

In addition to these sources of very high-frequency noise, there's the pulse noise that often occurs on AC mains. This is generated by such things as motors starting and stopping and thyristor phase control units operating in a linear mode rather than as zero-crossing burst switches. Air conditioners, blower motors, refrigerators, washers and dryers, ceiling fans with and without speed controls, large lamps and dimmers are all examples of these. You have them and your neighbors have them. The industrial facilities across town have them and they're considerably more powerful. Luckily the transformers and limited conducted bandwidth of the power distribution grid isolates your home from most of the industrial noise but likely not that generated within your home or from that generated by your neighbors.

This noise can be wideband and bursty in nature, which is often the case, and is a bit more difficult to filter than one would normally expect unless inured to this class of problems by experience. With the advent of the increased presence of noise in today's world, coupled with the difficulty in designing equipment to completely reject such noise, one comes irrevocably to the conclusion that some filtering of the AC line must be mandatory.

Filters

Generally, equipment which is likely to be an emitter of RFI is designed so that this is minimized since equipment sold for use in residential environments must meet certain emission limits defined by government agencies. This means the addition of a line filter, internal to the equipment, that brings the conducted emissions within limits set by the agency’s rules. This filter is selected, installed, and wired such that emissions originating within the equipment are not conducted or radiated outside the enclosure by the power cord. This filter also acts to minimize the noise that can be conducted into the enclosure from outside sources since these filters are generally bilateral designs.

These filters are generally not sufficient to completely eliminate noise, but they go a long way toward it. They are designed explicitly to control emissions in the radio bands generally above 1MHz, but there are some that have some attenuation down to 100KHz or lower. These latter filters are quite large and quite expensive, more expensive than one can tolerate in anything other than exceptionally expensive equipment.

Just adding a filter of this type is rarely adequate, as anyone who has done any EMC work knows, since the mechanical nature of the installation and its associated wiring, chassis bypassing, etc., are just as critical. Done wrong, the filter's presence can be nearly completely undone. It's safe to say that just the presence of an EMI/RFI filter on a piece of equipment does not guarantee that the equipment will be immune to the noise on the AC line. They are important to have though, and if done right will allow the equipment to at least not create any obvious problems in the presence of other equipment.

Due to the presence of "Y" capacitors in these types of filters though, there exists leakage currents when powered from an unbalanced AC source. These leakage currents can and do generate interchassis potentials, which were discussed in the first article on the nature of interconnect designs. Many designers avoid this sort of filter purely for this reason. Alternately, neutral and line can be completely isolated from chassis or there may simply be a high-value resistor, or parallel RC network, between neutral and chassis.

What this means is that even though some equipment these days has what are essentially tiny little AC line conditioners built right in, some don't. The former, by nature of the design of these filters, works only against the highest-frequency noise and generates interchassis potentials, while the latter avoids generating these potentials but offers no incoming filtering at all.

While an external AC line conditioner may act to isolate all of the attached equipment from the incoming AC line, they generally cannot eliminate leakage current induced noise. Nor can they easily eliminate noise coupling between pieces of equipment. There are ways to do this, however, and some of the conditioners on the market address this very point. Specific solutions to these particular problems are discussed later in this article.

Adding to the complexity of the overall problem, any external conditioner that contains a line filter poses the possibility of creating a resonant interaction with one or all of the AC inlet filters present in the connected equipment if the filter responses are similar. Resonances can lead to no effective noise filtering improvement or possibly even a degradation. This particular problem then places a limit on the types of conditioners that may successfully be used with equipment containing AC inlet filters. Since it's virtually impossible to predict whether this will occur, the only way to find out is to try it and see if the results are sonically pleasing. Then again, you could invest in a couple hundred thousand dollars of equipment and measure the bejeezus out of the system. One of these methods is more fun.

Surge suppression

There's another issue to dispense with, and that issue is surge suppression. Surges are essentially another form of noise -- a really big noise sometimes, but noise nonetheless. Given the investment of time and money in your system, it would seem wise to make an attempt to protect it against damaging or destructive transients that may appear on the AC lines from time to time.

We hear a lot about this as a side issue to power conditioning, and while this is often dispensed with in some designs with an MOV, or two, there are reasons to examine the problem and its solutions in greater depth. MOVs (Metal Oxide Varistors) are nonlinear, voltage-dependent resistors which abruptly become low resistance above a threshold voltage and are rated by the amount of energy they can absorb before self-destructing. By the way, self-destructing MOVs are a lot of fun to watch in the lab but a tad disconcerting to watch in one's living room.

MOVs are typically placed across line-to-earth, neutral-to-earth, and/or line-to-neutral and act to clamp a surge voltage, whether differential or common mode, to safe levels. They are, however, sacrificial elements. That means they wear out and in fact can fail catastrophically in some rare instances. They also don't sharply limit voltage since they depend on an impedance in series with the surge source to limit the peak surge current and thus, through their resistance, the resultant peak surge let-through voltage. This impedance is very often only the line itself or some small inductance in an associated line filter.

They'll still protect connected equipment from damage, to a degree that is largely dependent upon the specific circuit design, but it is possible for their energy-dissipation rating to degrade over time if subjected to a large number of lower-level surges. It is often wise to replace this type of suppressor at regular intervals or after a power failure or two for this exact reason. Of course, if they didn't work as advertised, a company that offers a guarantee against equipment damage wouldn't stay in business long, would it?

Brickwall

There haven't been any alternatives to MOVs, other than motor-generator sets, until the Brickwall by Price Wheeler Corporation devices came along (see their website at www.brickwall.com). It's a clever design that removes the sacrificial element and the potential for wearing out, or that of catastrophic failure, while providing significantly better surge protection than MOVs. Pretty clever actually -- wish I'd thought of it. Basically, it's a "simple" LC filter that rapidly lowers its natural frequency in response to an input peak voltage above a designed threshold.

The Brickwall units aren’t designed explicitly to act as line filters, but owing to their inherent LC filter topology, they do provide this function free of charge. There is, however, no common-mode filtering present in the basic design, but since all typical home installations share a common earth point, common-mode filtering of the AC line is of no concern. Since they are an LC filter though, there does exist the potential for interaction with AC inlet filters in connected equipment. Luckily the normal operating point of the Brickwall units provides a filter corner frequency of about 10kHz, well below the typical corner frequency of power inlet filters, thus making it very unlikely that any interactions will arise.

Since this type of filter contains an inductor in series with the AC line, questioning whether interactions of the power-supply's current draw with the driving point impedance of the line conditioner creates audible artifacts is a natural reaction. There are two effects that this condition may create though and they each need to be evaluated in turn. One is the modulation of leakage currents due to the distortion of the waveform being placed across the "Y" capacitor between line and earth with a concomitant signal-correlated modulation of the interchassis potential. Another is the previously mentioned low-frequency noise generation in the power-supply output as well as the associated reduction in absolute DC rail voltage.

The first of these effects, modulation of leakage currents, can only be eliminated by eliminating the "Y" capacitors. This is often not possible, owing to the requisite nature of common-mode filtering to meet emission regulations. Balanced AC power, while capable of eliminating the problems associated with the dreaded "Y" capacitor, is not an acceptable solution if any of the connected equipment is of the "dirty chassis" type. The use of true balanced interconnects throughout the entire system may be the only optimal solution.

The latter effect, AC line distortion, may or may not be a real problem, depending on the nature of the equipment's power-supply design, as discussed previously. However, many instinctively seek to avoid any conditioner that has a series inductance as the common myth of "choking off" the sound is the unavoidable end result. This effect is most often associated with power amplifiers. Since power amplifiers, especially solid-state designs, exhibit the highest "crest factor" (ratio of peak current to RMS current) current draw from the line, the effects on the operation of the power supply within the power amplifier will likely be the most audible of any piece of equipment in the system. Again, only listening will provide a reliable answer in most cases.

To summarize, since surge suppression is often a desirable function of a line conditioner and since surge suppression can be achieved with varying degrees of success and varying degrees of risk through the use of MOVs, or like components such as gas tubes, or through the use of active filters such as the Brickwall components, the balance of the choice rests on in-system performance with cost as an additional parameter. The former choice is considerably less expensive than the latter, but performance of the latter is demonstrably superior. In-system performance, however, is something that can only be addressed through audition. I would suggest contacting the folks at Brickwall and discussing your needs prior to acquiring a unit for audition as every system is different, as is every listener's perception of the resultant music.

Conclusions

The foregoing analysis yields several possible overall solutions that are dependent on the type of AC inlet design of the connected equipment. These solutions attempt to simultaneously eliminate the threat of surge-related damage, interchassis-potential noise, intercomponent noise coupling and interaction, and AC line noise isolation. They break down into three broad categories, one using unbalanced power, one using balanced power, and one using a mix of technologies.

Unbalanced power

The first category, the use of unbalanced power (the power that appears at your wall outlet) first and foremost, is optimal only when used in conjunction with components that provide true balanced interconnections in order to eliminate the problems associated with interchassis potential noise. If, however, all of the equipment is designed without EMI/RFI filters, there will exist no "Y" capacitors, and therefore the requirement to use all balanced interconnects is eliminated, allowing the use of single-ended interconnects.

A multiple-output Brickwall unit may be used in this instance both as a line filter as well as a surge suppressor if this forgoing caveat is followed since the unavoidable leakage current modulation it may induce is no longer a concern. However, the design of the equipments' power supplies may lead to audible artifacts due to their interaction with the unit's driving point impedance, and only a listening test will provide a reliable answer, as is true with any line conditioner. Naturally, the option exists to connect all source equipment to the unit while connecting the power amplifier directly to the wall outlet. This is a common method used by many audiophiles and it works well, but there will be no surge protection for the amplifier. Tradeoffs are everywhere.

Isolation between connected equipment is not explicitly addressed in this configuration, but then this is an often misunderstood problem. Multiple components will "interact" if and only if their respective current draws develop an additive voltage on the AC line through summation at the output of the conditioner. Thus one's signal-correlated current draw may create a noise component on the DC output of another's power supply. Again, this effect can only be determined to be a problem through audition in one's own system.

Common-mode isolation of interconnected equipment is a different problem, however, in that it is likely not to be of a signal-correlated nature, but rather that of high-frequency noise contamination of the common power-feed point due to the lack of EMI filtering in the associated equipment. This problem can, however, be solved with the use of an isolation transformer.

These types of transformers have an electrostatic shield between the primary and secondary windings, which is connected to an earthed chassis. In order to make use of this technique, however, all the equipment has to be connected directly to the output of the isolation transformer; thus this method doesn't provide any surge suppression or differential line filtering unless a high-current Brickwall unit is cascaded ahead of it. All of the foregoing caveats apply, of course, when this configuration is used. Once again, only an audition in one's own system will reveal whether this is the ideal solution.

Balanced power

The second category of solutions, the use of balanced power, requires the connected equipment to have symmetrical impedances from line to chassis and from neutral to chassis. There is also the requirement that the chassis be earthed through the power cord. These requirements, in most cases, are satisfied when the connected equipment has modern AC input EMI/RFI filters installed. The presence of these filters strongly attenuates any of the internally generated common-mode currents that might be kicked back at the common AC line source, but it is not uncommon to find equipment with no impedances between chassis and both line and neutral, in which case there exists the possibility of these currents being present.

While the problem with the power-supply's current draw modulating interchassis potential, in response to the interaction with the conditioner's driving-point impedance, doesn't exist when balanced power is used, as discussed in the first article on interconnect design, the problem with waveform-distortion-induced DC-power-rail noise does. Since balanced power converters use a transformer, it's unavoidable that some series inductance will be introduced. This is why balanced converters use a toroidal transformer as they have a much lower leakage inductance than EI core transformers. The problem here is identical in nature to that just discussed with respect to the inductance present in the Brickwall surge suppressors and naturally requires the same listening tests to determine its suitability in any given installation. Again, tradeoffs are everywhere.

By adding an electrostatic shield to the balanced power converter's transformer, the converter is then imbued with the additional function of an isolation transformer at negligible added cost. This provides excellent common-mode isolation between connected equipment by shunting to ground any common-mode currents kicked back to the common AC source, as discussed earlier. Since all of the connected equipment is supplied AC power from the output of the converter, the requirements outlined earlier are met simply and effectively without the addition of another piece of equipment.

This category of solution does not, however, provide intrinsic surge suppression, just as an isolation transformer alone does not. However, it is possible to cascade a high-current Brickwall unit ahead of a balanced power converter, in the same way that one may be cascaded ahead of an isolation transformer, thus adding differential-mode line filtering and superior surge suppression to the advantages of the balanced power converter.

The "interaction" problem discussed in the first instance is then exacerbated in this particular instance and is an unavoidable consequence of adding more inductance to the output of a line filter. Again, the only way to know if it's a problem is to audition the equipment in one's own system. The only redeeming grace is that the transformer's additive inductance is likely to be smaller, in absolute terms, than that present in the upstream line filter.

This category of solution provides the most comprehensive coverage of the problems associated with AC power distribution in that it addresses all the issues outlined initially. It does, however, mandate certain restrictions on the connected equipment, which may or may not be a showstopping problem for some users as it is often rather difficult to ascertain the exact nature of the AC input topology in any given piece of equipment. This then limits the application of balanced power to systems where this information is a known quantity.

The API concept

A third solution category involves the use of a mix of balanced and unbalanced power. The Power Wedge Ultra product line from Audio Power Industries (see the API website at www.audiopower.com)  is an excellent solution for system applications where a mix of equipment types is desired. The original Power Wedge supplied individual outlets from a transformer secondary without providing a connection of the neutral conductor to earth. This naturally isolates individual connected equipment in such a manner that leakage current couldn't flow or at least reduced it to a negligible magnitude. The Ultra designs provide for selection of each outlet to have either this connection (non-polarized), a standard connection where neutral is tied to earth (polarized), or a third option of balanced power through the addition of a switch at each outlet. Pretty clever if you ask me.

This particular technique provides the best flexibility in that each individual piece of connected equipment can be powered in the manner that allows the greatest relative performance. For instance, a piece of equipment that has no EMI filter can be connected to a filtered piece of equipment via an unbalanced audio interconnect simply by setting the Power Wedge output of the former to float, or to earth the neutral line, while setting the latter to balanced power. This completely eliminates any source of leakage current that would lead to interchassis potential.

However, there is one caveat. There is no transformer isolation for high-current loads such as power amps. This means that in any given system connection there is going to exist one source of leakage current. The use of a power amp without any "Y" capacitors is therefore strongly recommended if the user wants to have an ideal setup. The use of a power amp with "Y" capacitors on the other hand must be considered a compromise configuration.

The "interaction" problem is inherently solved in this design since each output is fed from an independent isolation transformer. This then makes it impossible for several pieces of connected equipment to create an additive noise voltage on the AC line feed through summation of their line currents at a common point that might exhibit a driving point impedance higher than that of the AC line itself. The problem of isolating conducted common-mode noise between connected units is also simultaneously solved through the use of the previously mentioned isolation transfomers.

The use of isolation transformers does, however, introduce the same problems mentioned in the other solution categories -- ie, waveform-distortion-induced noise on the DC rails. Again, as in every other instance, determining whether this is a problem requires individual audition. I should point out, however, that these products are mainly aimed at use with source components where the crest factor of the current draw is typically low, thus minimizing this as a source of problems.

This line of products uses EMI filters at each output, and while this is a good thing when connecting equipment without internal filtering, it can pose a problem with connected equipment that already has one (Please note that API has indicated that they use proprietary techniques in the Ultra series to overcome the problems Pete mentions.  Pete is now researching this area further.). Since these EMI filters are likely to be similar in attenuation response to those present in some equipment, there exists the problem that resonances will occur leading to either a reduction in cascaded performance or no essential improvement. At the risk of sounding like a broken record, try it and see. It's the only way.

Surge suppression in this line of products is addressed with the use of MOVs. While this is generally adequate for most users, those in high-lightning-strike areas or those in urban environments where poor quality power is a constant companion might want to consider the addition of a high-current Brickwall surge suppressor ahead of the unit. Adding a Brickwall unit ahead of an Ultra will, however, add yet more inductance to the driving-point impedance seen by the connected equipment and all the foregoing caveats apply. Audition, audition, audition.

Recommendations

An optimal setup for an all-digital and solid-state system, in my limited estimation, would be to cascade a 20A Brickwall unit ahead of a balanced power converter with an electrostatic shield in its transformer. This allows all connected equipment to contain internal EMI/RFI filters, with their attendant "Y" caps so that conducted and radiated EMI is minimized in the system, which is an important consideration when using digital equipment. The limitation on using only balanced interconnects is also removed since the problems associated with leakage current and modulation thereof are no longer a concern, thus allowing the use of lower-cost equipment. And finally, the electrostatic shield provides the requisite common-mode noise isolation between connected equipment. The Brickwall unit provides the desired ideal surge-suppression characteristics while simultaneously providing differential-mode filtering. Sizing the balanced power converter so that the power amplifier can be run on balanced power is a cost driver though, and the balance in performance versus cost has to be determined by, you guessed it, auditioning the alternatives in one's own system. Click...click...click.…

Cascading a high-current Brickwall unit with a transformer-isolated-type conditioner, such as the API Ultra series, would seem to be an ideal setup for use with "dirty chassis" equipment or systems that have a mix of "dirty chassis" and balanced-power-compatible AC power inputs. It is also an ideal setup where isolation of the power amp is either not necessary or cost effective. As an alternative, a high-current Brickwall unit could be placed ahead of an industrial isolation transformer if all the equipment in the system were, say, tube designs with no "Y" capacitors in any of the chassis. The simplest setup might well be a multiple output Brickwall unit that provides only surge-suppression and line-filtering functions. This, of course, doesn't do anything to minimize leakage current problems should any of the connected equipment have "Y" caps installed, but if used with equipment that has balanced interconnects, the issue is moot.

Given these alternative arrangements, I can't say that using one or the other would be absolutely ideal although there are system configurations where one or the other would obviously be preferable. The choices you make are going to depend strongly on the equipment set and the electrical environment. I would have to say though that protecting one's investment against surge damage is a priority if you live in a high-lightning-strike area, as I do, or if your electrical utility has a nasty habit of making a voltmeter needle bounce around like a ball bearing in a concrete room. It's your call, of course, but having purchased two Brickwall units prior to thinking about writing this article, based entirely on design and qualified test reports, I can recommend them based on their technical merits.

Their effect on sound is another topic though. Bet you never thought we'd get there, eh? Can't say I blame you -- it's been another annoying slog through technogeekland. But if you've followed the discussion this far, maybe it was worth learning a bit about noise, surge suppression, and line-conditioning setups. Coming up in the second part of this interminable article is the actual implementation of balanced power and Brickwall units, both individually and cascaded, the resultant impact on sound quality, and problems encountered along the way. Stay tuned.

...Pete Goudreau
pete@soundstage.com

• Audio Online Main Page