Power & Grounding for a Post Facility: by Eric Wenocur (This article originally appeared in Broadcast Engineering magazine, June, 1996)
Over the years certain aspects of facility power and grounding have become wellaccepted. In particular, single-point grounding techniques have dominated asthe best way to create a "clean" technical power system. The objective is toreduce hum, buzz and radio frequency (RFI) components on power and signalgrounds, thus reducing noise impressed on audio and video signals.
Recently, however, there have been stirrings around the television and audiorecording industries suggesting other approaches to solving noise issues.There have been articles and seminars dealing with the sources of noise, hownoise gets into signal circuitry, and noise control through the use ofbalanced (symmetrical) AC power. My own experiences with conventionalgrounding techniques have been so unsatisfactory--both in terms of failure tosolve problems and difficulty in implementation--that I have chosen to try someunconventional, but theoretically sound, ideas in recent installations.
Sources of Noise
The noise problems familiar to audio and video engineers range from audiblehum, buzz and interference at various frequencies to visible hum bars and RFmoiré in video signals. Many of these can be traced to coupling ofenvironmental noise into signal cabling, while other noise originates in theincoming power or in equipment itself.
One primary source of ground-related hum is voltage differentials betweendifferent points in the power system. This is the classic "ground loop"phenomenon; two points are at differing potentials, so current flows betweenthem. In the case of broadcast systems, ground currents can arise from severalsources; commonly these include leakage currents from line filters andtransformers in equipment power supplies, or devices which present anunbalanced load--that is, when some of the power delivered to the load leavesvia a path other than the neutral, generally meaning the ground. RecallingOhm's law, a voltage drop occurs when current flows through some resistance.The resistance component is the impedance of the grounding conductorsthemselves.
A second, minor, source of noise is induction of AC hum into signal wiring.Any conductor carrying a current radiates magnetic and electrostatic fields,which can induce current in a nearby conductor. The intensity of the inducedcurrent depends on proximity of the two conductors but even following goodcable dress practices, such as keeping power and signal wires perpendicular toeach other and using steel enclosures, there is enough AC field activity aroundequipment and signal wiring that this can be an issue. Even if your signal iscarried in a shielded cable, a nearby AC field still induces a current inthe shield, thus becoming a source of ground differential voltage.
The typical solution for ground differential hum is to try creating a system inwhich all equipment is grounded via only one path: the power cord third pin.This requires isolating chassis from racks, utilizing iso-ground powerreceptacles and lifting the shields of signal cables at one end. This lastpractice is absurdly difficult to maintain through patch fields and in systems,such as video facilities, with much unbalanced equipment (because the groundconductor is also the signal return). Additionally, it makes the shields intovery appealing antennae for RF signals floating by. In either case--shieldsgrounded on one or both ends--ground-borne hum appears at the "earth ground" ofequipment to which the shield is connected. This is typically the enclosure,or some part of the internal structure, which should be protecting thecircuitry from noise but instead allows the noise to be coupled in with signalsdue to poor design topology (the so-called "pin 1 problem"). This problem isso common it is shameful. (1)
You can try bringing all your shields to some central "earth ground" point inthe hope of bleeding off the ground noise, but the noise does not "flowinto" the earth! This is a common fallacy. Aside from safety, ground rods (orbonds to building steel) help to equalize any voltage potential between thetechnical power and utility power grounds, and reduce static charge buildup onshields, but they do not absorb ground noise. (2) In the end, single-pointgrounding and heavy earth grounds are attempts to treat the symptoms of noise,but not the source problems.
An important point to remember is the value of transporting signals onpairs of conductors because of their inherent ability to cancel unwantednoise; if all equipment used balanced/differential floating inputs and balancedoutputs most ground noise issues would be moot. This applies to power signalsas well; the hot and neutral conductors in a line cord carry equal and oppositecurrents which effectively cancel most radiated fields.
Balanced AC Power
The notion of balanced AC is not that new, it simply has been forgotten,particularly since modern electrical distribution (at 120V) relies on a single"hot" conductor and a grounded neutral. Balanced (symmetrical) power worksmuch like balanced audio; there are two hot conductors each carrying the sameAC voltage but 180 degrees out of phase. In the case of typical electricalsystems, this translates to two conductors at 60V relative to ground. The endresult at the line cord is still 120V but there is no "neutral".
The primary advantage of this approach is that, as with audio balancing,common-mode noise (which is equal and in-phase on both conductors) is nulled atthe ground. Leakage currents in a power supply, particularly those from linefilters (which are balanced by design), are summed and cancelled at the groundconductor. (3) In addition, distortion and noise caused by non-linearcomponents in power supplies are also cancelled. Switching power supplies,ubiquitous in today's equipment, tend to introduce more of this type of noisedue to their lack of a large transformer at the power input acting as alow-pass filter.
Installing a balanced technical power system is only moderately different fromconventional power. The most difficult part is getting electrical contractorsto think differently: "balanced" does NOT refer to balancing the loadson a three-phase service, and there IS NO NEUTRAL. Electricians are soaccustomed to neutral and ground being effectively the same that there is notelling what might get connected wrong if they are unclear on the concept.More prosaically, the standard AC line tester with three lamps will not worknormally.
Balanced power is created through the use of line transformers. Whateverincoming service you have, the end result is typically 120V at some amperagerequirement (though it can be another voltage). Therefore the necessarytransformer(s) should have a primary designed for the available service,perhaps 120V or 240V, and a capacity (stated in Volt-Amps, which is essentiallywatts into a purely resistive load) sufficient for your technical powerneeds.
The first step is to take a power inventory of your facility to determine thetotal tech power draw. Most equipment will state somewhere, on the rear panelor in the manual, either current or watts drawn. In general, it is wise to useVAs, rather than watts, when determining power consumption for a device. Thisis because watts can be misleading if the device's power factor is not known.Using VAs, based on a device's actual current draw, will ensure adequate powersystem size even for equipment with heavily reactive load characteristics (suchas big motors). Add it all up, add in your expansion needs and throw in someextra for good measure. The reality is that most equipment draws less currentthan rated, most of the time, but startup surges (such as from 1" VTRs or filmdubbers) must be considered, as well as changes in efficiency due totemperature. Plus, those big digital boxes, like DVEs and switchers, are realjuice hogs and might not be fully loaded when first purchased.
The tech power can be derived from a single transformer or several, dependingon the service. For instance, at Fast Cuts (Fig. 2 -- click below to see the full diagram) there was a 120/208V/100Athree-phase service available, which can provide 3 100A supplies. Threetransformers were chosen which could be strapped for 208V primaries and 120Vcenter-tapped secondaries (thus creating the balanced 120V). Theterminology of these transformers can be tricky, so be sure the supplier orcontractor understands what you want. Purchasing transformers from a companysuch as Equi=Tech eliminates this problem and also provides precision-woundsecondaries which are claimed to give better common-mode cancellation.However, precision transformers are significantly more expensive than commonindustrial models; the 8 kVA unit in this system was about $1500.
For the Fast Cuts facility, Equi=Tech supplied toroidally wound transformers,which have higher power capacity for their size than conventionallaminated-core designs (and well-contained magnetic fields), but had twopeculiarities which caused trouble during installation. First off, if thetransformer is mounted vertically, such as on a wall, support can be providedusing the hole in the plastic core but there must not be acompleted turn of electrically conductive material through the core (Fig. 1).The introduction of a closed electrical circuit through the core acts like ashorted winding turn and can damage the transformer.
Secondly, though the system in Fig. 2 was installed with a 90A three-phase mainbreaker ahead of the transformers, the breaker would not hold when powering upeven one of the transformers. The transformer manufacturer quotedshort-duration inrush currents as high as 5000 amps for the 5 kVA units! Thiswas probably due to the transformer cores being somewhat undersized andreaching saturation too quickly at startup. Since it was not possible to get acircuit breaker with extremely high inrush specs a 90A fused disconnect wassubstituted. This brings up a good point: do not hesitate to contact thetransformer manufacturer for technical help with installation, particularly inthe areas of lead identification, mounting and cooling.
To see a diagram of the entire installation (Fig. 2), click here.
The Code and Other Issues
Balanced power is first addressed in the National Electrical Code with the 1996revision (thanks to work by Equi=Tech's Martin Glasband). This legitimizes thepractice, at least for audio and video facilities, and provides some specificdirectives. For the most part these directives are common-sense anyway, thoughsometimes tedious. Since the NEC's sole purpose is to ensure human safety itis wise to stick with the letter unless you can foresee the consequences ofstraying. In the end, a given electrical inspector may make unreasonabletrouble about a balanced installation, even with support of the code, or maylet things slide if he understands what is being done.
One aspect of the code requirements for balanced power which is problematic isthat each branch circuit must be supplied with a Ground Fault CircuitInterrupter (GFCI or GFI). The reason for this is to protect against shock inthe event that someone touches ground while also touching a device whose caseis connected to the "neutral" line cord lead. The code addressed this problemin conventional power systems with the polarized two-prong plug, but withbalanced power both sides are at 60V so the polarized plug does nothing. Inpractice, however, this scenario is very unlikely since most professionalequipment has a three-wire cord or an isolated chassis.
More importantly, the typical GFI is designed to trip at only 5 mA of leakagecurrent. Unfortunately, in any collection of average professional equipment afew devices will exhibit this much leakage just from shunt currents in theirpower supply filters or leaky capacitors. There is nothing functionally wrongwith the equipment, but the GFI will trip the instant power is applied. Onesolution is to install GFI circuit breakers, with adjustable tripcurrent, in the tech power panels, but this is very expensive. Anotherapproach is to examine the offending equipment to determine and fix whatever iscausing the leakage current. In the end, the actual danger is so minimal thatone might consider a more obvious solution.
The code also specifies double-pole breakers for the branch circuits, which isimportant. Though it increases the cost marginally, without double-polebreakers a branch circuit is never truly OFF! From an overload orshort-circuit standpoint the protection is the same as with conventional power;excessive current draw on either leg will trip the breaker. In addition, thepanel box must have both power buses isolated from the box itself, such as for220V distribution. Isolating the ground bus is not really necessary (see finalsection). Other code requirements, such as labeling circuits and receptacles,are actually a good idea anyway in order to maintain the integrity of thetechnical power system. That is, all equipment with system signal connectionsmust be powered from tech power, and all non-production equipment (suchas copiers) must be kept off tech power.
Clean Tech Power
A few issues about the power and grounding system remain. As far as lineconditioning, regulation and backup power go, these must be addressed on anindividual basis depending on budget. In the facility described here noattempt was made to provide voltage regulation. Most decent equipment alreadyhas internal regulation which is sufficient to handle minor fluctuations inline level. I have also known large tap-switching regulators to go haywire andwreak havoc with the power. Similarly, no backup system was provided, thoughindividual UPS units may be added for specific vulnerable devices such asgraphics computers and editors.
In terms of spurious noise and spike control, since balanced power is createdusing transformers there is the inherent benefit of an isolation transformerbetween your tech power and the power company. The transformer acts as a steeplow-pass filter which reduces high frequency noise. Surge protection for FastCuts was handled by installing some very large MOVs (metal oxide varistors) atthe breaker panels. These should be rated for the nominal RMS voltage (butalso sized to handle potential long-duration power line fluctuations whichwould damage a MOV).
Another potential source of equipment damage is when power browns or blacks outthen flickers on and off before returning. To alleviate this we installed somerelays which monitor the incoming power phases. If any of the three phasesdrops out a giant master relay (a 90A industrial "contactor") disconnects allthree phases. This relay must be reset manually, by pushing a button, whichallows facility personnel to determine when the incoming power hasstabilized.
Finally, what about grounding? As previously mentioned, I have lost faith instar grounds, ground rods, telescoping shields and the like. They arecumbersome to implement and solve little. In the facility described here a"brute force" approach was taken: all cable shields connected at both ends,power system grounds connected to racks via third pins and power strip cases,all racks bonded together with bolts and star washers as well as heavy groundstraps to outlying racks, extra large ground conductors installed betweentransformer center-taps, breaker panels and distant sub-panels.
The idea was to create the lowest-impedance ground possible. This makes iteasier for ground-borne noise to return to its source, and reduces thepossibility of voltage differential problems by bringing all parts of thesystem closer to the same electrical potential. In practice, the use ofbalanced power and good layout and shielding practices are meant to prevent theadmission of noise into the system, but a solid low-impedance ground is thefoundation. The transformer center taps and electrical grounds are bonded tothe building (power company ground) via a large conductor coming in with thethree-phase from the electrical room (the three-phase neutral is not used).
The presumption here is that shields can always be lifted, or ferrite beadsinstalled, if noise is found but so far this has not been necessary. Thefacility has been in operation since January of 1996, with technical areascovering about 2500 square feet, and there has been no sign of video hum oraudible noise (apart from the occasional bad equipment). While there areplenty of balanced audio and video devices, there are also numerous unbalanceddevices, some with two-wire line cords. It just doesn't matter. Admittedly,implementing this type of power and ground system in a very large facilitywould be more difficult, but I believe it is worth considering these"unconventional" approaches. One day they may be considered conventional.
Bibliography and Further Reading
1. Neil Muncy, "Noise Susceptibility in Analog and Digital Signal ProcessingSystems," Journal of the Audio Engineering Society, June 1995
2. Ralph Morrison and Warren H. Lewis, "Grounding and Shielding in Facilities,"John Wiley & Sons, 1990
3. Martin Glasband, "Lifting the Grounding Enigma," MIX, November 1994
Eric Wenocur owns Lab Tech Systems, specializing in audio and video systemdesign, installation and troubleshooting in the Washington, DC area. Specialthanks to John Frey for additional consulting on this article.