Power quality and grounding: When using the newer digital-based equipment,one must abandon many older methods of grounding, bonding and shielding.
Sep 1, 1997 12:00 PM, Warren H. Lewis
Times change, and so does our equipment. The analog power supply,electronic circuit and power amplifier world has steadily been giving wayto digital switching and logic-based designs. In the lighting controlworld, this change is nearly complete. Although most of us have at leasttaken passing notice of this progress, what does the change mean inrelation to the reliable operation of our new digital-based equipment whenwe connect it into the typical AC power and grounding, audio and videowiring systems, as we used for the old analog equipment?
To begin with, the old analog audio pre-amplifier and mixer usedpotentiometers to vary directly the level of a signal input to analogcircuit elements used to amplify and combine signals into one signal of anew level. The potentiometers could be rotary or slide types, and eithertype could get noisy over time. The newer digital-based circuits can usethe same types of controls but with a difference - they can now change thebinary count of a digital number applied to an integrated circuit whosechannel gain is step-variable in accordance with the binary value of thenumber produced via A-to-D conversion of the potentiometer's resistance.Noise is filtered out prior to and during the A-to-D conversion process.
Some controls for mixing and gain are no longer even potentiometer-basedtypes of controls at all. Now, they are optically or magnetically coupleddigital encoders whose shaft rotation or slide position is directlyconverted to digital pulses and, hence, binary numbers. Then, of course, wehave the lighting control systems, audio and video mixers and similarequipment controlled via a digital link to a laptop or desktop personalcomputer system. Links are often of the popular parallel, RS-232 or -422type, although fiber-optic links with even better performance are coming onstrong. Wireless systems also exist, and they may also use control andsignal links involving infrared (IR) or radio frequency (RF) for thewireless portion of the path but still wind up using wiring to get thesignal into and out of the equipment to which these links are connected.The wired portion of these devices may use these types of digital-formatsignal protocols or other proprietary types.
Once we get to the actual control, audio and video signals our newerequipment deals with, we find that the signals are also in full digitalformat. No neatly phase-shifted or amplitude-varied AC waveforms orDC-referenced sine or analog-shaped waves can be seen anymore on theoscilloscope used to view these signals. They are now streams ofsquare-wave-shaped, pulse-width modulated (PWM) signals or packets ofencoded binary numbers representing a fully digitized control, sound orvideo signal. Those of us with sound and video cards plugged into our PCsare certainly familiar with this because the only analog signals are wherethe microphone and loudspeakers are connected.
Even looking into our AC-to-DC power supplies, we now find about only twoplaces where DC exists (the input filter capacitor and the logic orutilization voltage output buses); the rest of the circuit handles squarewaves of one sort or another. Frequencies into the tens of kHz and PWMschemes are now encountered inside power supplies. In contrast, analogpower supplies had only 50 Hz or 60 Hz AC on the line input to therectifier, and everywhere else you could pretty much count on findingeither pulsating 120 Hz DC or some "pure" level of DC - stable or changing,but nevertheless DC. The block diagram of an older style of linear powersupply (an analog or linear design) is shown in Figure 1; a newer type ofswitch-mode power supply (SMPS) is shown in Figure 2. The differences aresignificant.
The power amplifier is probably one of the last holdouts from the digitalrevolution. Generally, its output is ultimately fed to an analog transducercalled a loudspeaker. What happens ahead of the point where the analogsignal finally emerges from the amplifier is another story! The digitalcircuits are everywhere ahead of the final output point on the poweramplifier, which may even use a digital-based power amplification schemeand some special filtering to remove high-frequency components (a carrier).The filter then finally helps reconstruct the recovered audio frequencyrange signal as an analog voltage and current.
In the analog equipment world, we typically lusted after the much-discussedsingle-point ground (SPG) system and quiet earth grounding electrode,usually of a mystical 1 V character and not connected to anything elseexcept our equipment. This practice has been and remains a serious NECviolation per Article 250 - Grounding. All earth grounding electrodes onthe premises are required for fire, shock and lightning safety purposes tobe made by means of bonding conductors, electrically common to all otherelectrodes and to the building's electrical system's equipment (safety)grounding conductor (EGC) system consisting of metal conduit-raceway, metalequipment enclosures and the famous greenwire on the power cords.
This situation is now totally changed because the modern recommendedpractice for grounding both analog and digital-based equipment is to use amultipoint grounding (MPG) design, accept connection to the AC powersystem's equipment grounding greenwires, and ignore special connections toearth of all types. For example, where we have rooms full of equipment inracks and where this equipment is interconnected by signal-level cables, asignal reference grid (SRG) is used to ensure that broadband groundingcapability is achieved and common-mode noise is attenuated amonginterconnected units. A diagram of an SRG and related equipment is shown inFigure 3. More will be said about the SRG later.
Accordingly, the new digital equipment is clearly different and also quiteimmune to some of the problems that bothered us when we used analog-basedequipment. However, as some of us have discovered, new problems haveemerged for us to be aware of and to cope with involving the digitalequipment. There are also some old rules of installation that were usedwith analog equipment but now must be thrown out. We must now observe new,important rules peculiar to the newer digital equipment; these rulesgenerally apply to the analog equipment as well. For example, there aresome equipment grounding rules that have undergone some significantrevision. Also, as we will later see, there are some cable shielding rulessimilarly affected. And lest it be overlooked, lightning and electricalsurge protection requirements have also emerged as being poorly understoodand more important than ever when we are using electronic load equipmentand, in particular, digital-based equipment.
With this in mind, let's begin exploring the new world of digital-basedsound and video equipment and its relation to building AC power, groundingand audio or video cabling systems.
Power quality and performance problemsNo practical difference really exists between a logic circuit used in acomputer system to represent a numerical bit value (0 or 1) and that usedin digital-circuit-based sound and video equipment. Therefore, a lot ofinformation that has been successfully developed for use in the computerworld is directly applicable to the equipment with which contractors work.(Don't forget this now involves computer equipment.) Nowhere is this moretrue than when we consider the CBEMA curve.
CBEMA stands for an old main-frame computer manufacturer's organizationcalled the Computer Business Equipment Association, and the acronym ispronounced See-B-Mah. The curve developed by CBEMA's Power InterfaceSubcommittee No. 3 (SC-3) within that group can be used to characterize thegeneral relationship between AC power quality and the reliable performanceof digital-logic-based equipment of almost all types. The curve ispresented in its original form in Figure 4.
As can be seen in Figure 4, the x-axis is constructed in terms of time andthe related number of cycles-per-second of the standard U.S. 60 Hz powerline's frequency. The y-axis is constructed as a +/- reference for voltage.The reference point on the y-axis is set to equal 100% of the chosennominal rms voltage of the AC line being considered. For example, 120 VACis at the 100% line; when the voltage increases, you go up on the curve'sy-axis (marked by double, triple and so on increments of the AC line'snominal voltage as set to 100%). When the line voltage goes down, you godown on the y-axis all the way to zero, as in a complete power loss. Thelogarithmic x-axis determines the duration of the event being tracked. Forexample, if the AC power line went to zero voltage for one minute (inaddition, the y-axis line would be drawn along the bottom of the curve fromthe far-left and at zero V until the one minute point, also 3,600 cycles,was reached on the x-axis), then it would be restored to the 100% line. Ofcourse, this all assumes that the voltage did go back to 120 VAC andtherefore 100%. If it didn't, then you would need to restore the line towhatever level to which it finally returned.
Notice that the CBEMA curve has an overlaid funnel-shaped set of lines onit. The inside area represents that set of time periods and voltage levelsover which typically well-designed digital-logic-based equipment should beexpected to reliably operate with respect to the quality of the AC line'sinput power. However, if the conditions on the AC power line to theequipment should find themselves outside of the funnel, an increasing levelof trouble will occur with the operation of the equipment. Ultimately, wewill experience a failure of the equipment and even component damage, suchas if a high level of surge voltage were applied to the equipment's ACinput for a sufficient duration.
The area above the funnel's top line represents the area of increased linevoltage for whatever reason. The area below the funnel's lower linerepresents the area of decreased line voltage for whatever reason. Thecross-hatched areas represent relatively small areas of disagreement oruncertainty that were expressed by the members of SC-3 during the creationof the curve and have never been solidly resolved. Note that the AC line'svoltage is shown as reaching zero at the 8.33 ms point. All this means isthat once every half-cycle, the 60 Hz voltage waveform passes through thezero voltage point on its way to the next half-cycle and a polarityreversal. We all hope that our equipment will continue to work while the ACline does this; otherwise, we will have to provide for a DC powerdistribution system.
To the left of the 8.33 ms or half-cycle point is the sub-cycle, impulse ortransient-surge voltage realm, as you prefer. To the right of this point isthe longer-term or rms voltage event realm. For example, we might find avoltage swell or a voltage sag here if we are talking about events lastingonly a few cycles. After that, we might be discussing problems such as along-term, high or low nominal rms line voltage or loss of voltage. Ingeneral, then, impulses are to the left and high or low rms voltages on theAC line are to the right of the 8.33 ms mark. The former lasts for lessthan one-half cycle; the latter lasts for more than one-half cycle and outas far as you might contemplate (minutes, hours, days, weeks, whatever).
The momentary low-voltage or sag conditionOne of the most commonly encountered power quality problems is called asag. The sag has been variously described as being a dip, dive or somethingsimilar, but the term sag is official according to the IEEE's Emerald Book(Std. 1100-1992). An idealized view of a sag is shown in Figure 5, where anominal voltage exists before and after the sag event, which resulted in anoticeable decrease in rms voltage for more than one, and over a period ofseveral cycles.
The typical sag event is often caused by the sudden application to anelectrical system or circuit of loads that have a high momentary startingor inrush current. Such loads are typically represented by wholepanelboards, motors, large rectifiers and AC-DC power supplies that have alarge-value input capacitor in them, which is charged directly across theline via a rectifier. With a big, empty capacitor, a whopping big chargingcurrent can therefore exist on the first half-cycle and progressivelylessening currents on subsequent half-cycles until the capacitor is chargedto near the peak line voltage being applied.
The momentary high-voltage or swell conditionAs might be expected, the swell is the exact opposite of the sag condition.The swell has also been called a surge voltage, but this phrase is notcorrect because the term surge is more properly applied to shorter durationevents involving momentary high voltages, such as those produced bylightning. The term swell is now the IEEE's official description of thedescribed event. An idealized view of a swell is shown in Figure 6, where anominal voltage exists before and after the swell event, which resulted ina noticeable increase in rms voltage for more than one and over a period ofseveral cycles.
The typical swell event is often caused by the removal on an electricalsystem or circuit of large loads that have a high running current. Suchloads are typically represented by panelboards, motors, rectifiers andlarge groups of AC-DC power supplies that can be disconnected at the sametime by one power-off control.
The impulse voltage conditionThe typical impulse event has many names, such as glitch, spike, notch,whisker, zot, transient and, of course, impulse. The event is characterizedas being a sub-cyclic event of either polarity and any amplitude. It isgenerally of a singular nature but can occur in trains or strings ofimpulses that may or may not all be related to a single cause. The impulsemay be synchronous or asynchronous with the amplitude of the AC line'svoltage or current with which it is being compared. The impulse willtypically appear in that portion of the CBEMA curve to the left of the 8.33ms point and in the sub-cyclic area. The impulse may remain fully above orbelow the 100% line, or part of it might be above and part below that line.It may also be of a decaying oscillatory nature.
A fast transition time is also typically associated with an impulse event,but the term fast is not always clear except in comparison to the rate ofchange for the AC power system's fundamental frequency. All in all, theimpulse is generally conceded to have a transition time expressed in termsof less than a millisecond or several microseconds on an AC power circuit.Faster transition times are generally only seen when the source of theimpulse is very close to the point at which the measurement is beingundertaken. This occurs because AC power circuits are lossy transmissionlines at high frequency and tend to attenuate HF signals rapidly withdistance. An example of an idealized impulse event is shown in Figure 7.
The decaying oscillatory voltage eventThe impulse event may also involve a decaying oscillatory current orvoltage waveform, depending upon how it was first generated and how it hasbeen transmitted through the intervening AC power system's wiring. Notethat the involved AC power system does contain reactances and is, afterall, an LC circuit that is even resonant at a fundamental frequency andharmonically related ones. Therefore, oscillatory events are commonly seenon this circuit. The degree of damping of these oscillations is variabledepending upon the Q of the LC circuit involved on the wiring system andthe other losses it exhibits. In general, only a few decaying repetitionsof the higher-frequency impulse will be seen because it will quickly bedamped-out in exponential fashion on the typical AC power circuit.Lower-frequency decaying oscillatory events may take longer to damp out andcan be propagated over longer wiring paths.
The typical decaying oscillatory event on the AC power line generally isfound to the left of the 8.33 ms point on the CBEMA curve and in thesub-cyclic disturbance area. Sometimes, however, this is not the case, andthe 8.33 ms line will be crossed by the tail of the oscillatory waveform.The oscillation generally moves above and below the 100% voltage line, butin some cases, it may be DC-biased and stay mostly above or below this line.
One good example of a troubling, decaying oscillatory event on the AC powersystem is that produced by banks of power factor capacitors being switchedin (sometimes out). These events both have some energy behind them andcontain a lot of low-frequency content. Together, these factors allow theresulting disturbance to be propagated without much attenuation over adistance up to several miles on the electrical supply system beforeentering the facility and the equipment. A typical circuit diagram of aswitched bank of power factor capacitors is shown in Figure 8. A generalrule here is that the closer the capacitor bank is to the affectedelectronic equipment, the worse the potential disturbance will be and viceversa.
Most power factor correction capacitor banks are installed on three-phasedistribution circuits for the electrical supply system itself, within afacility on its main feeder system, or at both locations. In either case,their unwanted effects easily reach down the building's secondary feederand branch circuit system to reach connected electronic load equipment.
How do capacitors cause decaying oscillatory events? The problem istypically caused when a set of discharged capacitors is directly connectedright across the AC line at or near the peak line voltage. Alternately, acharged capacitor can be connected to the AC line at a point where theline's voltage is much lower than the voltage charge in the capacitor.
In the first case, the capacitors take a lot of current from the AC line.Because of the upstream power circuit's impedance, they cause a largemomentary voltage drop to occur, which is seen by downstream loads as anunwanted voltage reduction. In the latter case, they add voltage to the ACline. This voltage then causes a momentary unwanted voltage rise to flow inboth directions in the line conductors from the capacitor's point ofconnection. Although this rise is seen as a momentary voltage rise in bothdirections, it occurs mostly in the direction of the load.
Another problem occurs when the mechanical contacts used to connect thebank of three capacitors do not have properly or well-synchronized contactclosure times. This results in two of the capacitor's being connectedbefore the third and last one is. The immediate result is a definitenormal-mode disturbance on the two involved power systems' conductors. Thisdisturbance is later accompanied by a common-mode disturbance.
In either of these cases, the LC products involving the capacitors and theAC line cause an oscillatory condition to occur until the capacitor'scharging becomes synchronized to the line voltage's frequency (Figure 9).Appearing as a simple reactive load connected across the line, thecapacitors produce the intended power factor correction effects because oftheir reactance. The resulting oscillation is quickly damped out because ofthe losses in the AC line's path and is typically unnoticeable after a fewcycles of oscillation.
The commutation notchVoltage waveforms on the AC line are also sometimes seen to exist with whatis best described as one or more notches having been taken out of thewaveform (Figure 10). These notches may appear anywhere along the time-baseand may move around in both starting time and duration. These are calledcommutation notches and are generally caused by the momentary short-circuitthat a controlled rectifier will place across the AC line during the timethat one rectifier is turning off and another is turning on. With SCRs thatare phase-shift fired, it is easy to see the notches move around on thex-axis. Notches generally stay on the left side of the 8.33 ms line on theCBEMA curve and in the sub-cyclic area.
Multiple zero-crossings on the voltage waveformIn the extreme case of a capacitor's decaying oscillatory event, thezero-voltage line may be multiply crossed by the oscillatory part of theevent or a harmonic voltage (Figure 11). This can also occur if someoscillation exists on a deep notch's edge and crosses the zero-voltageline. In either case, the result is multiple voltage zero crossings, whichcan significantly affect any equipment that depends upon the 60 Hz linevoltage's zero crossing point for timing, SCR commutating or both. SomeSCR-based lighting control systems are seriously affected by this kind ofproblem. Also, some digital clocks and timers that count zero crossings onthe voltage waveform can really speed up if they get more than one voltagezero crossing every 180 degrees on the 60 Hz circuit!
Harmonic voltage waveform distortionWith nonlinear loads, such as rectifier power supplies of all types (linearand SMPS), current is taken from the 120 VAC power line at harmonicallyrelated frequencies to the 60 Hz fundamental. In general, all current istaken in the form of an impulse near 90 degrees and 270 degrees as opposedto being linearly taken all along the applied voltage waveform. A typicalinput current waveform for an SMPS is shown in Figure 12. This waveform caninvolve a very high peak current on the affected circuit with aconcurrently large voltage drop occurring across the AC supply circuit'simpedances, including the wiring, the impedance of the supply transformer,an alternator's windings or any other AC power source's internal impedance.
Of special note and interest, the above impedance can be largely providedby the flexible power cord sets used with temporary AC power distributionsystems. Long branch circuits are another contributor, and if they are usedin conjunction with an extension cord, good luck! Also, placing a powerconditioning device, such as a line voltage regulator, between thenonlinear load and the AC supply can sometimes just make things worsebecause the typical power conditioning device usually has a lot of internalimpedance, which can directly add to the voltage waveform distortionproblem.
The typical nonlinear load, such as a SMPS connected to a 120 VAC line,requires both fundamental frequency current and harmonic currents up toabout the 19th order (19 Hz x 60 Hz or 1,140 Hz). Mostly, the currents aredrawn from odd-ordered harmonics (3, 5, 7, 9...) and from the lower orders(particularly the third, fifth and seventh harmonics).
Once a harmonic current is demanded by the nonlinear load, such as arectifier, from the AC supply circuit, it will produce voltage drops acrossany of the impedances in the series path. Thus, if a third harmonic currentof a given amount produces a 10 V drop in the upstream circuit, thiscondition will be seen as 10 V at 180 Hz algebraically added to the 120 V,60 Hz fundamental voltage. So as shown in Figure 13, the observed voltagewaveform resulting from this condition will exhibit peaks and valleys, orhumps, depending upon how you want to describe the distortion. It is alsoimportant to understand that because of the total amount of inductivereactance (XL = 2pfL) in the current's path, 1 A at 180 Hz (third harmonic)will produce three-times the voltage drop than would be produced per amp atthe fundamental current's frequency of 60 Hz. This relationship also holdstrue for each higher order of harmonic current as well because theinductive reactance increases proportionally with frequency.
Note that a harmonically distorted voltage waveform as shown does not haveas much total area under its curve as does an undistorted sine wave, andits peak voltage may be lower as well. This spells more trouble for aconnected linear power supply than it does for an SMPS design. The formerloses regulation headroom and runs hotter because it sees what itinterprets as a low-voltage condition on its input and reacts accordingly.The SMPS has more headroom available and later on compensates by simplydrawing a higher peak charging current for its main energy storagecapacitor at 90 degrees and 270 degrees. With enough distortion, bothsupplies will eventually lose regulation, but the linear supply will almostalways go first and by quite a bit.
An important bit of advice: If you are attempting to make either a voltageor current measurement on an AC power conductor that is not carrying asinusoidal waveform, you cannot use anything other than a true-rmsinstrument. The typical analog or digital current or V meter is nottrue-rms but is rather an average-actuated, rms-calibrated device. In otherwords, it is a full-wave-rectified DC instrument that has its scalefudge-factored to make it agree with the rms value of 0.707 on a pure sinewave. On a typically harmonically distorted waveform, the area under thecurve is insufficient to allow this kind of instrument to give accuratereadings. The result is typically a reading that may be off as much as 50%.In other words, a 20 A rms current could read around 10 A on such aninstrument and subsequently mislead the user into believing that thecircuit is not heavily loaded. Similar problems occur on the voltagewaveform, where the user is mislead into thinking the AC line voltage istoo low and needs to be raised. Computations involving voltage and currentare, of course, really fouled up by this situation.
The effect of a sag on a typical SMPS and linear power supplyAnother way to look at the CBEMA curve is to identify the area below the100% line on the y-axis and to the left of the 8.33 ms point as being thearea where the lack of good energy storage in the typical SMPS' inputfilter capacitor comes into play. In well-designed SMPS units, this is arelatively large-valued capacitor fed from a full-wave bridge rectifierconnected directly across the input AC line. This capacitor was once fullycharged to nearly the AC line's peak line voltage (169 VDC on a 120 VACline) and can store a lot of charge (Q=CE), which can then be drawn off bythe inverter in the supply prior to its use in the logic voltage regulationcircuits in the SMPS. It is therefore somewhat like a flywheel from whichone can keep the machine going once the motive power is removed.
However, a poorly made SMPS supply or a good one that is overloaded couldbe susceptible to a lack of stored energy in its main input capacitor oncethe AC line voltage goes below that necessary to provide recharging currentfor the duration of the sag event. In an SMPS this point is simplydescribed as being when the peak AC line voltage out of the full-waverectifier that is used on the SMPS' input is equal to or less than thevoltage level on the capacitor for that half-cycle. Because the linearAC-DC supply operates all of the filter capacitors at low voltage obtainedfrom an input step-down transformer, not much voltage difference existsbetween what the capacitor is charged to and what the linear voltageregulators constantly work against. This difference is called head room,and there is no comparison between the two designs - the SMPS wins thiscontest hands down.
Again, because Q=CE, you can see that for the same size capacitor, theamount of available energy stored at 170 peak volts from a 120 VAC line isa lot more than what might be stored at 40 peak volts from a 25 VACtransformer secondary. Also, just think about how much more energy can bestored in an SMPS that is otherwise almost the same as a 120 VAC input unitexcept that it is operated at 240 VAC input, and where the capacitor isstill directly connected to the rectifier's output! With the linear design,there is no difference because the secondary voltage from the stepdown transformer doesn't change, only the primary voltage does; hence no change in headroom results either.
The idealized effect of a typical line voltage sag on the DC output of ananalog supply and SMPS of similar output rating are compared in Figure 14and Figure 15. A line voltage sag does little to the connected SMPS incomparison to its equivalently power-output-rated, linear type of AC-DCpower supply. The digital difference provides better performance in a lessexpensive, smaller volume, lighter weight SMPS than for the equivalentlyoutput-rated linear supply. The linear supply can compete only if it isallowed to be inefficient, much more costly, much larger and accordinglyweighs much, much more than the item it is supporting!
Not surprisingly, most AC power-quality studies (IBM, Bell, et al) haveidentified the AC line voltage sag event as being the most commonlyexperienced power quality problem by most electronic and computer equipmentusers. Even though the SMPS-based digital equipment type is better thananalog equipment at ignoring AC line power quality problems, such as a sag,there is still some point at which the sag can produce performance problemswith this type of equipment.
Effect of a swell on a typical SMPS and linear power supplyWhen a swell arrives at the input to a typical SMPS, the result ispredictable: It tries to charge the input energy storage capacitor afterthe full-wave rectifier as fast as it can. This is a fairly large amount ofcapacitance in the typical SMPS, so this effort is not as easy as it mightat first seem. An RLC time constant is involved here, and it not onlyinvolves the capacitor, but also the reactances and resistance of the wholeupstream wiring system from which direction the swell is arriving. So, itis hard to get the capacitor filled up and instead it just absorbs what isavailable and stores it for use. If this results in a higher voltage thanthe line's peak, then the capacitor will simply not accept any new chargeon the next half-cycle or until some subsequent half-cycle after the SMPS'inverter load has depleted the capacitor's voltage to below the AC line'speak voltage.
The SMPS power supply's inverter is a high-frequency (usually in the tensof kHz) switched on and off circuit involving alternately cut-off andsaturated devices being supplied from the main input energy storagecapacitor. The inverter transformer's secondary is then full-wave rectifiedand used to charge a secondary energy storage capacitor at a voltage (withheadroom) near that to be used by the connected loads. By pulse-widthmodulating the inverter (controlling its duty-cycle), we can control thecharge level on this capacitor.
The final result is a stable voltage on the inverter-served secondaryenergy storage capacitor even though the main input energy storagecapacitor may have a great deal of voltage variation across its terminals.Note that the inverter may even be temporarily held cut-off by thepulse-width modulation circuit whenever too much voltage begins to appearacross the secondary energy storage capacitor. This is an automaticallyresetting protective action on the part of the SMPS and it works quite wellin keeping input line overvoltage disturbances from reaching the servedloads.
With a linear power supply, the swell often causes the main energy storagecapacitor on the low-voltage secondary of the input transformer to becomeovercharged. This can get to the point that the linear voltage regulatorcircuit it feeds cannot withstand the applied voltage, and it then losesregulation to the served load. The result is typically too much DC voltageon the power supply's output, which can upset or damage the connectedloads. Alternately, the regulator circuits themselves can become damaged. ADC level crowbar circuit is sometimes the only protection from this kind ofproblem, and it is not an elegant solution, especially if it is notautomatically resetting or can be triggered by electromagnetic interference(EMI).
Other effectsBecause impulses and oscillatory events of almost all types containhigher-frequency components than typical sags and swells, they can affectthe power supply that they are impressed upon because of EMI effects.
The typical EMI problem involving AC line propagated electricaldisturbances generally involves a power supply design that providesunwanted coupling between its input (and any associated higher levelcircuits) and its output. This is especially true when higher frequenciesin the tens to hundreds of kHz are involved and where small amounts ofstray reactance can create significant coupling between circuits. The usualproblem within a power supply involves in-fields as opposed to e-fieldsbecause the currents are relatively high and both circuit impedances andvoltages are relatively low. Thus, most problems involve stray magneticfields and the routing of, for example, wiring harnesses and PC boardtraces.
In regard to EMI immunity, the SMPS wins hands down because it is already aprolific generator of EMI in its own circuits. Thus, the SMPS must be welldesigned from an EMI and electromagnetic compatibility (EMC) standpoint, orit will not work and will also generate interference for the equipment withwhich it is being used. Making it immune to its own poison has thedesirable effect of also rendering it pretty much immune to externallyapplied EMI, such as that on the AC power line to which it is connected.This is not the case with the typical linear power supply, which is fairlyquiet by itself but is often not well designed from an EMI immunitystandpoint.
Let me give you an example. Often, when an item of equipment has a linearpower supply in it and it is being affected by AC line EMI problems, anuninterruptible power supply (UPS) that is shown to be compatible with theload is placed between it and the offending AC line. This change cantypically cure the problem, but how does it do it?
Simple. The problem has been fixed by putting an SMPS between the AC lineand the linear power supply in the "victim" equipment. After all, what'sthe real difference between an AC power UPS and the SMPS in equipmentexcept that the UPS is higher power, doesn't have a rectified inverteroutput and supplies 60 Hz instead of DC to its loads? The similaritybetween the UPS' battery and the SMPS' main input energy storage capacitoris pretty obvious, so I won't elaborate.
Mitigating problemsChoosing the AC power source: Proper AC power for the electronic equipmentdefinitely involves more than just plugging it into the nearest availablewall outlet. It also involves avoiding the common mistakes that are made inthe industry when attempting to get special or dedicated AC power for theequipment. The task of obtaining truly proper AC power therefore startswith the AC source itself.
Go upstream: The best advice to follow is to originate the AC power from apoint on or as close upstream to the service equipment (SEQ) for thebuilding as is practical. This means a dedicated feeder into the SEQ andbeing routed to the electronic equipment room, where it can be interfacedto one or more panelboards via either an isolation transformer (IT) oranother suitable power conditioning device.
The rationale for this advice is simple. No matter where you get the ACpower in the building, whatever affects the SEQ will similarly affect alllevels of power distribution in the building. For example, a momentary lowvoltage (sag) at the SEQ will also be seen at all electrical outlets in thebuilding. However, if one is obtaining power from an outlet very fardownstream in the distribution system, then anything affecting the feedersor panelboards between the selected outlet and the SEQ can affect the powerquality at the selected outlet. Statistically, power quality is best at theSEQ and becomes progressively worse as one moves down through thedistribution system, where the building's own loads can create problems.
Get a low-impedance power source: All power sources have an internalimpedance. This is what limits the short-circuit current available acrossthe terminals of a battery or a transformer, for example. Internalimpedance is therefore both unavoidable and beneficial unless you fail toaccount for it and get in trouble from its predictable effects.
Typical electronic load equipment power supplies involve rectifier inputs.These kinds of inputs require non-sinusoidal and high peak currents on eachhalf-cycle and will cause a significant voltage drop across the internalimpedance of a power source. This then causes a harmonic distortion of theavailable voltage waveform from the source. Harmonically distorted voltagewaveforms thus created cause numerous problems with the basic operation ofboth normal and electronic types of load equipment.
Harmonics on power circuits can be strong and therefore can usually getinto both telephone and electronic equipment's signal level circuits andcause a lot of mischief. For example, simple voltage waveform harmonics onthe AC power system are serious because they also cause the involved ACpower wiring to propagate wide-band audio frequency interference intonearby control or signal cables. For example, the interference from simpleAC-DC power supplies may range from 60 Hz to around 2 kHz. Where higherfrequency notching is present on the voltage waveform, the availableharmonics can extend upward into the tens and hundreds of kHz, where theycan additionally affect either video or digital switching processes.
Harmonic voltages are reduced to reasonable levels by ensuring that thechosen AC power source is of sufficiently low internal impedance. For thetypical service transformer (ST), this correction is not a problem, but itbecomes one where dry-type transformers or other types of powerconditioning equipment are subsequently installed within a building andfrom which equipment is being powered. Recommended practice is to employbuilding AC power sources of larger kVA capability than the load requiresand with internal impedances in the range of 2.5% to 5% to avoid most ofthese problems.
Note that most voltage regulators and UPS types of power sources cannotmeet the foregoing internal impedance recommendations, but this compromiseis sometimes still acceptable. The trick is to make certain that the chosenregulator or UPS has an internal output voltage feedback circuit that looksat the output voltage waveform and sends an error correction signal backinto the control circuit. This signal will keep the output waveformdistortion and rms voltage level under simultaneous control. Thus, anactive method can be used to compensate for the high internal impedance.
If the power conditioning equipment does not have good output voltagewaveform control, then it is common for the distortion of the outputvoltage waveform to be significantly worse than when it is being supplieddirectly from the building's power system (Figure 16). Sometimes thisproblem gets so bad that the electronic load equipment won't even workproperly when connected to a power conditioning unit but will work whenconnected directly to building power. In the trade, this is known as anexpensive lesson.
NEC wiring methods for electronic loadsHistorically, the NEC did not allow much flexibility in wiring the AC powerbranch circuit supplying the electronic load equipment. Only two ways wereallowed: solid grounding (SG) and isolated or insulated grounding (IG). TheAC system supplying the equipment typically could only be a solidlygrounded system. (This means bonding the supply transformer's neutral tothe equipment grounding conductor system and to the nearest NEC-acceptableearth grounding electrode to the transformer for AC systems of 2,150 VAC toground. See Sections 250-5 and 250-26.) Anything else was not permitted tobe used on the branch circuit, even if it might be a good idea from anequipment performance standpoint, such as decreasing hum and other noiseproblems on typical installed equipment.
Since the product safety testing and listing services (such as UL) werecompletely aware of this point, it should be apparent that all of theequipment that they tested and the standard for safety to which it wastested had to meet the NEC03 requirements as they were written, not assomeone would like them to have been written. Together, the NEC and UL madeit impossible to try anything else until the NEC was changed. This was donefor the 1996 edition of the NEC.
AC system and branch circuit groundingThe two traditional ways of grounding an AC system and its branch circuitsthat were to be used to support typical installed electronic load equipmentsubject to noise and hum problems are shown in Figure 17 for the SG designand in Figure 18 for the IG design. Until the 1996 edition of the NEC,these were the only two permitted designs.
What's important about the two methods of grounding shown is two-fold.First, on a typical 120 Vrms branch circuit, the hot wire is at a potentialof nearly 170 peak volts to anything grounded or to ground itself. Second,because of common- mode noise current flow in the grounding conductorsystem, an unequal common-mode (CM) noise current can occur on the neutraland hot conductors of the branch circuit. This unequal current subsequentlygets partially converted to normal-mode current on the AC power wires tothe load (the hot and neutral conductors).
The first situation means the electric field (e-field) is quite highnearest to the hot conductors and in relation to anything grounded, such asa shield on an audio cable installed near power wires. This high e-fieldmaximizes the electrostatic coupling between the AC wires and the groundedshield and which then introduces hum and other noise into the audio cable.This point is particularly important for high- impedance circuits becausethey are voltage-driven. Note that if the voltage to ground available onthe AC wiring were lowered, so would be the amount of coupled interferencefrom the e-field into nearby signal-level conductors. This is importantbecause the e-field is a near-field phenomenon that is both proportional tothe voltage and tends to fall off rapidly with increasing distance betweenthe involved conductors.
The second situation means that the normal-mode converted common- mode (CM)noise current becomes algebraically additive to the expected power currenton the circuit and, after interaction with the circuit's impedances,becomes a normal-mode noise voltage impressed upon the AC voltage waveform.Thus, the load equipment must have an AC input circuit immunity to noisevoltage across the entire audio frequency range and above, not just at 60Hz and 120 Hz. This implies special design considerations for the filteringsystem following the rectification process, such as the use of very lowseries-resistance leakage value electrolytic capacitors and low-leakageinductors.
For example, a linear power supply might pass this noise voltage acrossitself and onto the DC output of the supply, where it could affect theability of the regulation circuits to perform. It might even pass throughor around the regulation circuits and gain access to the final DC busstructure directly connected to the electronic circuits. This is especiallyvexing when these are low-level mixers or amplifiers that are analog innature and have a band-pass that ranges across the audio range.
Note that with the typical SMPS, this is not much of a problem because theinterstage inverter transformer provides a sufficient amount of attenuationbetween stages and keeps the noise out of circuits such as the regulatorcircuits.
The new wayThe newest way to provide AC power system wiring and grounding is to use a120 VAC centertapped AC system for a single-phase AC power source and toestablish the branch circuit system with all of the conductorssymmetrically arranged to ground. The nominal AC voltage to ground must beone-half of what it was before, for example, 60 Vrms instead of 120 Vrms.This type of AC system and the subsequent wiring and grounding requirementsare called out in NEC Article 530 - Motion Picture and Television Studiosand Similar Locations, Part G; Separately Derived Systems with 60 V toGround. A typical wiring design using the solid grounding (SG) method withthe new AC system grounding method is shown in Figure 19; the design forthe insulated-isolated grounding (IG) method is shown in Figure 20.
The operative section within the NEC for a 60 VAC to ground system is530-70, General, where it is stated that the "use of a separately derived120 V, single-phase, three-wire system with 60 V on each of two ungroundedconductors to a grounded neutral conductor shall be permitted for thepurpose of reducing objectionable noise in A-V production or other similarsensitive electronic equipment locations provided that its use isrestricted to electronic equipment only and that all of the requirements inSections 530-71 through 530-73 are met." The relevant NEC sections'complete wording is provided in the sidebar, "NEC Codes for Contractors."The new requirements call for the addition of a ground-fault interrupt(GFI) on either the receptacle or the branch circuit breaker used on thenew type of AC system grounding.
Three important changes occur when the new 120/60 VAC system is used.First, the voltage to ground on the hot conductors is one-half of what itwould be if it were a standard form of circuit, which drops the e-field toground level proportionally. Second, the grounding system is fullysymmetrical, which limits the ability of a common-mode noise current in thegrounding system to be converted to normal-mode current and hencenormal-mode noise voltage on the circuit. This change reduces the abilityof the noise to get through the equipment's power supply and to thesubsequent circuits. Finally, the requirements of the NEC provide for the120/60 VAC system not to support any equipment other than electronicequipment that is equipped with the special connector configuration thatthe article requires. This prevents a great deal of unwanted interferencewith the electronic equipment from other types of equipment that mightotherwise share the same AC system and branch circuit wiring, for example.
Don't forget that when an AC system that is nearly completely separatedfrom the normal AC system in a building is in use, it also becomesgenerally more reliable from a power-loss standpoint. This simply occursbecause loads such as lighting or motors, which can cause nuisance trippingof branch circuit and feeder breakers under starting inrush and faultconditions, are simply not on the same transformer secondary, feeder andbranch circuit system as the contractors installed equipment that is now onthe dedicated 120/60 VAC system.
What to do?For best results, install a completely separate transformer, secondaryovercurrent and disconnect device, feeder and branch circuit all at the new120/60 VAC level for the installed equipment, which meets the requirementsof the NEC for use on such a system. This is a separate AC system from thebuilding's other systems, but it will still need to be AC system groundedvia a grounding electrode conductor (GEC) to an NEC-acceptable earthgrounding electrode that is electrically common to the building's normalearth grounding electrode system used by the service equipment and otherseparately derived AC systems.
Please note that one small glitch exists in the whole process of using the120/60 VAC system arrangement, and it involves locations that do not haverestricted access to qualified persons only. Per the NEC, when the 120/60VAC system is used in areas of general access, it will need to be uniquelyconfigured and identified for use with this system.
The problem involving this unique configuration requirement is two-fold:First, you have to obtain the special receptacles and plugs, which are notnecessarily in-stock at all suppliers; second, you must modify alreadyproduct safety listed and labeled equipment by cutting off the existing 120VAC plug on its line cord and replacing it with one of the new plugs. Thislatter action makes for another problem: You generate another NEC situationbecause of the combined effects of Sections 110-2, Approval and 110-3,Examination, Identification, Installation, and Use of Equipment; Paragraph(b), Installation and Use. For example, one cannot modify listed or labeledequipment in any way without invalidating its listing or label process.This then causes the equipment to no longer be in listed or labeledcondition, which then creates a problem with the two sections justidentified. The only out on this is to get the electrical safety inspectionauthority having jurisdiction at the location to provide a written waiverthat permits the installation and operation of such modified equipment. Ifthis is not done and a fire or shock occurs, the lawyers will have a fieldday in assigning liability to those involved in the modification process atall levels. A change in the NEC is needed in this area to permit equipmentalready listed and labeled for use on standard NEMA-5-15 or-20 types ofreceptacles to be so modified without losing either the listing, labelingor any other NEC compliance benefits.
Equipment grounding for hum and noise controlWhen equipment is installed together in a room, such as for audio or videoediting, there are typically several racks of equipment involved with anumber of signal-level cables routed among them. This type of installationis often involved in hum and other noise problems, all typicallyoriginating in the common-mode (CM).
Common-mode currents and voltages on the equipment grounding system cancause operational problems with digitally based equipment when the impulsesinterfere with clocking and related set-reset operations of the logicelements. Once converted into normal-mode current or voltage by thecircuit's impedance imbalances, it mimics the desired signals and can sodirectly affect control, audio or video signal processes.
Common-mode currents and voltages classed as noise are problem areasusually addressed by marginally and randomly effective or hazardouspractices, which generally involve creating electrical safety problems viaequipment grounding methods that violate the NEC or employinggrounding-bonding schemes that are based in unreliable art as opposed topredictable engineering practices. A good example of this is the use of thesingle-point grounding method, which is marginal and unpredictable inperformance for analog-based equipment, useless for digital-basedequipment, a real lightning damage-prone grounding design for the attachedequipment, and an NEC equipment and AC supply system grounding violationwhen implemented in its classic form.
However, a really effective methodology does exist to deal with all ofthese problems and also to keep the AC supply system and equipmentgrounding safe. The method is called a signal reference grid (SRG), and itinvolves AC power and signal-level surge protectors being employed alongwith some special rules for the termination of cable shields. Taking theseitems in order, let's start with the SRG.
The signal reference gridIt is generally agreed that if all the electronic equipment is grounded toan underlying ground-plane (a flat, wide-area solid sheet of copper), thenthe best form of grounding available across the broadest range of frequencycan be obtained. The problem is that a ground-plane is typically difficultto implement and somewhat costly in practical forms. However, if oneinstalls a grid instead of a plane, all the benefits of the ground-planecan be had except that the upper effective operating frequency limit willbe lower. Remember, topologically speaking, a grid is just a plane withsome openings made in it. For the typical installation of a grid, however,having openings in the plane is not a practical problem because therecommended practice designs for SRGs are very effective from DC toapproximately 25 MHz to 30 MHz. (See the IEEE Emerald Book titled"Recommended Practice for Powering and Grounding of Sensitive ElectronicEquipment" and Federal Information Processing Standard Publication No. 94[i.e.; Fips-PUB-94], "Guideline on Electrical Power for ADP Installations,"U.S. Department of Commerce, National Institute of Standards and Technology[NIST], National Technical Information Service, U.S. Department ofCommerce, Springfield, VA 22161.) This is good broad-band grounding that iseffective across the entire range of frequency needed by both analog anddigital-logic-based commercial audio and video equipment of all types.
The typical SRG consists of a network of bare copper conductors with bondedintersections of somewhere between one and two feet, all of whichcompletely covers the floor area where the equipment is installed. The SRGis typically installed on the floor and all of the equipment is jumpered toand from it via grounding-bonding straps. All electrical conduits, theequipment ground, and the isolation transformer (IT) serving the equipmentroom are grounded or bonded to the SRG to make it the commonly sharedground reference for everything. Yes, everything, because the idea ofisolation is long gone; it has too many problems associated with it,including electrical safety ones. (The neutral terminal on the secondary,the metal case or enclosure and the greenwires are all tied together withinthe transformer. The junction point for these items is thengrounding-bonding jumpered to the SRG. The isolation transformer istypically placed directly atop the SRG and is not remotely installed.)
Typical constructionGenerally, the construction of the SRG involves the use of a cellularraised floor (computer room style flooring) under which the SRG, electricalpower and signal wiring can be routed. The underfloor volume is also oftenused for an air-handling space for the supply of HVAC-process cooling airinto the room above.
The cellular raised floor type of installation allows the SRG to beinstalled directly atop the structural sub-flooring or suspended just belowthe floor's pedestal post's top cap by SRG wire-holding and bonding clamps.Either way, it works quite well. The materials used generally involve barecopper wire AWG#36 or a copper strap about 0.1 inch (2.54 mm) thick and 2.0inches (50.8 mm) in width. The crossover points where the junctions aremade are typically at 2x2 feet (61 x 61 cm) but are sometimes made a littlelarger or smaller, the differing effects of which are not too noticeable.
For carpeted or uncovered floor surfaces, a flat copper foil is generallyused to construct the SRG from. The foil is generally about 0.030 inches(0.762 mm) thick and 2 inches (61 x 61 cm) in width, with all of thecrossovers soldered together. This type of SRG can be directly applied to afloor's surface and then either walked directly on or covered with ananti-static electricity carpet. (You do use this kind of carpet inequipment rooms, don't you?) T-shaped access slots cut into the carpetallow grounding straps or jumpers to be passed through and then soldered tothe underlying SRG foil, after which the carpet's edges are folded backdown.
Surge protectionThe biggest threat to electronic equipment from a physical damagestandpoint is from lightning-related transient voltages, commonly calledsurges. Although somewhat related to the geographical location of theequipment, surges are a real threat almost everywhere in the United States.How much of a threat can be estimated by consulting ANSI/NFPA-780, thenational Lightning Protection Code 1992.
A direct strike to the building or the incoming AC supply conductors is notnecessary to create a damaging surge current and voltage for the equipmentinstalled within the building. A cloud-to-cloud overhead or even a nearbystrike is often sufficient to do the trick.
Because many typical installations will involve the routing of cables allover a facility, there is ample opportunity for damaging lightning currentsto be near-field coupled into them. This is both an e-field(electric-field, capacitive coupled) and in-field (magnetic-field,inductive) set of phenomena that is unavoidable. However, they can bemitigated by proper grounding, bonding, shielding and surge protectiontechniques being employed.
The surge is coupled into the building wiring system and the variouscontrol or signal processcables of the affected electronic load equipment.The amount of surge being coupled is proportional to the amount of areaenclosed by the affected wiring. Big areas mean larger surge currents andvoltages are developed from the lightning discharge.
>From the victim equipment's standpoint, the surge threat arrives from oneof two ports: the AC power input and the signal or control cableconnectors. Therefore, a surge current arriving from one port's conductorsis passed into the victim equipment and out the other port and into itsconductors. The victim equipment is therefore seen to be in the center of aloop into which a surge impulse has been coupled. This explains why addingsurge protection to only one of the two ports does not do much to protectthe equipment from lightning-coupled damage.
Don't forget that with the increased use of computers, the control orsignal-level port is generally configured to use one of the computerindustry's standard protocols, and this is fortunate, as we will see. Notethat this is true unless problems arise because of some special interfaceplug-in card having been installed, which may use a nonstandard signalingprotocol or other configuration.
Proper surge protection needs to be installed on the equipment itself andon both the AC power and control or signal cable ports using performancecoordinated surge protective devices. Typically, this is a metal-oxidevaristor-based form of protection on the AC power input port. The signallevel port may use a combination of Tranzorbs, gas- tubes, seriesresistances or impedances and common-mode chokes in a specially designedprotection circuit particular to the signal port's characteristics andcoordinated with the performance of the associated AC power port'sprotection.
For example, an RS-232 signal port for digital use needs a surge protectorthat is specifically designed for the RS-232 protocol. It must be used withan AC power port surge protector with which it has been performance testedand rated to work. Anything else will most likely fail to do the job, andthe signal port will most likely be the one damaged. The best protection isprovided when both the AC power and signal protectors are mounted onto andgrounded-bonded to the protected equipment's metal frame or enclosure. Inmost cases, using a rack's metal framework is adequate for this purpose,and this practice is necessary for larger groupings of equipment in anycase.
A surge protector unit is generally available for any industry-standardsignaling protocol but not for proprietary signal protocols. These need tobe specially engineered for the specific application.
The building's AC power system also must be surge protected so that thelevel of the surge arriving at the electronic load equipment on the ACpower circuit is as low as possible. The protection provided at theelectronic load equipment is not usually capable of doing the wholeprotection job involving high-energy surges but is rated to do a good jobwith lower-level surges. By today's means, one cannot provide top-notchperformance and high energy level handling in the same package as installedat the victim equipment's level.
The recommended practice is to start with the building's SEQ and to installa device called a secondary lightning protector (or arrestor) on theservice conductors. The metal frame or enclosure of the SEQ must be usedfor the surge current's reference point. Generally recommended practice isto parallel this protector with an AC capacitor connected from each line toequipment ground. This is often properly referred to as a wavefrontmodification capacitor. It is sometimes built in as a part of the overallprotector chosen to do the job. Most of these protectors are rated for2,600 VAC systems and are available for single- and three-phase services.
Next, recommended practice encourages the application of surge protectorssimilar to the one used at the SEQ on each level of switchboard andpanelboard that exists between the SEQ and the victim electronic loadequipment on the branch circuit. In each case, the protector is applied inshunt between each line and equipment ground, such as the metal frame orenclosure of the switchboard or panelboard.
The final level of surge protection ahead of the equipment's portprotection is applied at the end of the branch circuit itself. Recommendedpractice is to install a surge-protected receptacle or to plug in aprotector to the normal receptacle into which the victim electronic loadequipment is then plugged.
Taken together, all of these measures serially shunt and progressivelyattenuate the level of surge that arrives at the SEQ from the powerdistribution system to a lesser level at each point at which thecoordinated protection has been applied. What is left is mopped up by thedevice-specific and coordinated protection, which is applied at the ACpower port for the electronic load equipment and is coordinated with thatequipment's signal port's protection.
Cable shield terminationsWith analog circuits and lower-fre-quency signal processes, the golden rulewas always to ground the shield at one end only. However, this is not arecommended practice with digital signal circuits because of thehigher-frequency nature of the circuits. It is not recommended practice forthe analog circuits either if one is concerned with having the shield doanything to attenuate in-field coupled interference, such as that fromlightning.
With digital circuits, the requirement is to ground the cable's shield atboth ends to obtain in-field protection and to preserve the integrity ofthe high-frequency signal on the cable. Worries about cable shield currentcombining with the digital signal on the cable are minimized because of thefact that, at the involved frequencies, the two currents are separated byflowing on the inner and outer surfaces of the shield with little or nooverlap. This is both an e-field and a skin-effect function. In fact, mostdigital signal cables, such as coaxial cables, can have the shield groundedat multiple points along the way with no ill effects and much improvedlightning protection.
The telephone company has always grounded the cable shield at both endsbetween the subscriber loop station and the subscriber's premises. This wasdone for lightning safety reasons and is required under NEC regulations.
"Power Quality"A twisted-pair arrangement is used inside of the telephone cable's shield,and each of the contained pairs is protected by a surge protective device(SPD) connected from line-to-line and line-to-ground or chassis. Unwantedcable shield currents caused by conducted common-mode currents aretypically dealt with by isolating one end of the shield from ground via acapacitor of a few microfarads. This blocks DC and most audio frequencies(including AC power system harmonics) but lets high-frequency surge typecurrents flow with ease. As a result, the shield is able to function as ameans of attenuating in-field-coupled noise and surge currents but isrelatively unaffected by DC and AC power system-related common-modecurrents resulting from ground potential offset between the two ends of thecircuit.
Shield current problems can also be dealt with via a number of otherrecognized techniques, such using Tranzorbs between the shield and ground,opto-couplers at the cable's ends, common-mode chokes on the cable, andtransformer isolation of the signal.
Finally, if you simply must have the cable's shield grounded at one endonly, you can still have the benefits of grounding the shield at both ends(for in-field protection) by pulling the cable into a grounded-at-both-endsmetal conduit-raceway. This practice provides a two-level shielding system.The design also improves the e-field shielding capability of the circuit.The best types of conduit or raceway to use in this role are electricalmetallic tubing (EMT), intermediate metal conduit (IMC) and rigid metalconduit (RMC) in ascending order of effectiveness. Note that galvanizedsteel conduit is markedly better than aluminum conduit in attenuatingcommon-mode noise currents and that tightly made joints and terminationsare necessary for the best effect.
Digital logic and SMPS-based equipment is replacing or has replaced theanalog circuit based equipment of yesterday. This newer design of equipmenthas great advantages but is still somewhat susceptible to electrical noisearriving on the AC power input, the grounding system and the attachedsignal level cables used to interconnect items into a system. Equipment ofthe new and old design types are both still susceptible to the damagingeffects of lightning-induced surge currents and voltages that are impressedupon the equipment's AC power input wiring, the signal cables and thegrounding system being used.
If you understand how AC power problems affect electronic load equipment,you will be able to determine what kind of power conditioning equipment youneed. The CBEMA curve was developed to aid in this process and is usefulwith both the older analog design equipment and the newer digitallogic-based designs.
When using the newer digital-based equipment, one must abandon many of theolder methods of grounding, bonding and shielding if the equipment is towork reliably. This means strictly following the NEC; not using isolatedearth grounding electrode connections; eliminating single-point groundingsystems because broad-band signal reference grid (SRG) designs now must beused; avoiding the grounding of signal cable shields at one end onlybecause both ends are the new requirement for digital signals; and payingstrict attention to providing proper surge protective devices on both theAC power and signal circuits for the newer equipment if there is a risk oflightning damage at the location. Following these guidelines will ensure aground system that is safe for the electronic equipment you install and forthe people who will use it.
530-70. General:Use of a separately derived 120 V, single-phase, three-wire system with 60V on each of two ungrounded conductors to a grounded neutral conductorshall be permitted for the purpose of reducing objectionable noise in A-Vproduction or other similar sensitive electronic equipment locationsprovided that its use is restricted to electronic equipment only and thatall of the requirements in Sections 530-71 through 530-73 are met.
530-71. Wiring Methods:(a) Panelboards and Overcurrent Protection. Use of standard single-phasepanelboards and distribution equipment with a higher voltage rating shallbe permitted. The system shall be clearly marked on the face of the panelor on the inside of the panel doors. Common-trip, two-pole circuit breakersthat are identified for operation at the system voltage shall be providedfor both ungrounded conductors in all feeders and branch circuits.
(b) Junction Boxes. All junction box covers shall be clearly marked toindicate the distribution panel and the system voltage.
(c) Color Coding. All feeders and branch-circuit conductors installed underthis section shall be identified as to system at all splices andterminations by color, marking, tagging, or equally effective means. Themeans of identification shall be posted at each branch-circuit panelboardand at the disconnecting means for the building.
(d) Voltage Drop. The voltage drop on any branch circuit shall not exceed1.5%. The combined voltage drop of feeder and branch circuit conductorsshall not exceed 2.5%.
530-72. Grounding:(a) General. The system shall be grounded as provided in Section 250-26 asa separately derived single-phase three-wire system.
(b) Grounding Conductors Required. Permanently wired utilization equipmentand receptacles shall be grounded by means of an equipment groundingconductor run with the circuit conductors to an equipment grounding busprominently marked "Technical Equipment Ground" in the originatingbranch-circuit panelboard. The grounding bus shall be connected to thegrounded conductor on the line side of the separately derived system'sdisconnecting means. The grounding conductor shall not be smaller than thatspecified in Table 250-95 and run with the feeder conductors. The technicalequipment grounding bus need not be bonded to the panelboard enclosure.
Exception: Other grounding methods authorized elsewhere in this Code shallbe permitted where the impedance of the grounding return path does notexceed the impedance of equipment grounding conductors sized and installedin accordance with Part G of this article.
(FPN No. 1): See Section 250-95 for equipment grounding conductor sizingrequirements where circuit conductors are adjusted in size to compensatefor voltage drop.
(FPN No. 2): These requirements limit the impedance of the ground faultpath where only 60 V applies to a fault condition instead of the usual 120V.
530-73. Receptacles:(a) General. Where receptacles are used as a means of connecting equipment,the following conditions shall be met:
(1) All 15 A and 20 A receptacle outlets shall be ground-faultcircuit-interrupter protected.
(2) All outlet strips, adapters, receptacle covers, and faceplates shall bemarked as follows:
WARNING - TECHNICAL POWER.Do not connect to lighting equipment.For electronic equipment use only.60/120 V 1 F ac.GFCI protected.
(3) A 125 V, single-phase, 15 A or 20 A rated receptacle having one of itscurrent-carrying poles connected to a grounded circuit conductor shall belocated within 6 feet (1.83 m) of all permanently installed 15 A or 20 Arated 60 V to 120 V technical power-system receptacles.
(4) All 125 V receptacles used for 60 V to 120 V technical power shall beuniquely configured and identified for use with this class of system.
Exception: 125 V, single-phase, 15 A or 20 A rated receptacle outlets andattachment plugs that are identified for use with grounded circuitconductors shall be permitted in machine rooms, control rooms, equipmentrooms, equipment racks and other similar locations that are restricted touse by qualified personnel.
(b) Isolated Ground Receptacles. Isolated ground receptacles shall bepermitted as described in Section 250-74, Exception No. 4, however, thebranch-circuit equipment grounding conductor shall be terminated asrequired in Section 530-72(b).