NOISE MAY BE THE MOST MISUNDERSTOOD PROBLEM in any professional audio or video facility. Today’s recording technology offers unsurpassed quality and accuracy, yet grounding system noise still baffles the experts.
Grounding noise is one of the most common complaints from audio engineers, but it's difficult to explain noise and grounding problems without addressing power. Grounding noise is closely linked to AC power under a variety of impedance load conditions. At first glance, the causes of noise can seem mysterious and perplexing. Often, dirty power is cited as the reason. But in fairness to the experts, it is important to remember that conventional power standards were adopted long before we had sensitive electronics. Eighty years ago, no one predicted the effects of industrial loads on power integrity.
More relevant to modern audio and video production is AC noise that is created locally by equipment power supplies and other impedance type loads within the studio. Most people don’t understand that there is a dynamic relationship that exists between the AC voltage phase, an impedance load, harmonic power distortion and objectionable RF program-signal noise. Conventional beliefs state that the causes of noise stem from the way a studio is grounded, however the problem is deeply rooted in the background of the electrical industry and modern electrical power distribution standards.
A brief history
In 1882, Thomas Edison wired the town of Sunbury, Pennsylvania using a shared-common three-wire DC distribution system. The cost of copper wire was an important factor, so Edison's engineers devised a way to distribute two circuits using only three wires. The forerunner to modern power distribution, DC and AC versions of the “Edison Circuit” are still widely used.
Beginning with the Niagara Falls power project initiated by George Westinghouse and supervised by Nicolai Tesla, AC distribution overtook DC and became the primary power system. With mostly lights and electric heaters loaded on the power grid, there was never much concern for voltage phase beyond what was considered to be "distribution convenience" for the utility. From almost the very beginning, 120-volt AC wiring has been conveniently unbalanced -- a "split" off half of a 240-volt single-phase grid.
Eventually, three-phase power was developed as a standard to suit heavy industrial users. Little was known at that time about harmonic distortion or other adverse effects on power systems created by impedance loading. Meanwhile, single-phase took a back seat to bulk three-phase power distribution because of the huge demand created by the efficient three-phase industrial motor. The three-wire Edison circuit was expanded another phase: 120/208- and 277/480-volt three-phase "wye" systems allowed for running three circuits using only four wires.
In electrical parlance, this multiple circuit wiring method is called a "Round Robin." The three-phase wye design enables single-phase fluorescent lighting, three-phase air conditioning and other three-phase motor loads (e.g. elevators) to be fed by one power distribution grid with the load current evenly distributed across the system -- ideal for commercial use.
But this system has one glaring fault. The level of interference created when a three-phase wye system is split up and used as three single-phase circuits is truly something to behold. For example, as much as 20% (or more) of the power used by fluorescent ballasts is reflected back onto the power grid in the form of reactive or harmonic currents -- now that’s a lot of distortion. In the late 80’s, a 40-plus-story office building in Los Angeles actually burst into flames because of these reactive currents. Incredibly, the origin of the fire was determined to be from excessive harmonic distortion in fluorescent lighting circuits which created a high-frequency current overload and literally a meltdown of the electrical wiring system. The First Interstate Bank fire in Los Angeles in May of 1988 was the event dubbed by the media as "The Towering Inferno" a la the Hollywood movie. Codes were adapted to remedy the fire danger, but the noise problem itself was never completely resolved.
Three-phase power nonetheless remains the bulk power of choice for utilities. When a utility furnishes single-phase service to an area, the standard procedure is for the utility to derive single-phase power by using one or two of the distribution grid's three phase elements. So even single-phase power is linked to the distorted three-phase grid. Typically, electrical power furnished by utilities contains 3% to 5% harmonic distortion. Single-phase service itself remains "split" into two 120-volt circuits as per Edison's original wiring design. In one form or another, these standards have been adopted and put into use around the world.
Meanwhile, in the audio industry
The post World War II era brought a revolution in electronics. Among these developments was the balanced circuit. RCA jacks gave way to XLRs, direct outputs gave way to balanced outputs and "L" circuits gave way to "T" circuits. The result was a quieter and more stable signal circuit that was less prone to interference. The technique might have been extended to power applications, but it wasn't. Power was generally regarded as something altogether different from RF or audio signal, perhaps more as a sort of fuel. Bulk power generation and distribution standards were etched in stone. Consequently, serious power problems have manifested with the onset of balanced RF circuitry and ever more sophisticated electronics. Today, data loss and common signal-circuit malfunctions attributable to background noise and power distortion are cited as critical issues in audio and other high-tech electronics.
A century after Edison, one might assume that definitive answers to common problems would be self-evident truths found in any engineering text book. Not so. In facilities with noise problems, there seem to be as many suggestions as engineers. Although volumes of theory have been written on electrical interference and noise problems, there are few standards which genuinely address the issues.
FIPS 94, for example, is a federal government publication which attempts to answer grounding noise problems in data processing facilities. Having limited options, electrical engineers often apply this engineering material to on-line video recording studios. Thousands of dollars worth of copper bussing and expensive labor notwithstanding, an active and very expensive power conditioning system is often prescribed to compensate for a signal reference grid that somehow failed to sink hum noise as expected.
FIPS 94 is hailed as the authoritative source in electrical engineering circles, but it fails to adequately cover all of the ways (right and wrong) to phase and reference AC power for an intended load application. It is a mistake to provide a power system that is not properly configured to operate sensitive electronics. Industrial motors required a new and unique three-phase power design some 80 years ago. Now is an appropriate time for the electrical industry to do the same for high-tech electronics.
Focusing on the Problem
It all boils down to eliminating noise problems. What, for example, needs to be done to completely rid a production facility of video hum bars. Can the problem of AC noise common in Class-A tube amps be addressed? How about noisy guitar amps? Many other examples of stubborn noise problems that have endured the test of time have failed to yield substantial solutions.
Are RF engineers right in saying that the best thing to do with an AC ground is to avoid it? From an RF perspective, it certainly makes good sense. But how safe is applying separate RF and AC grounding systems -- with a high amplitude potential -- just a matter of inches apart and potentially anywhere within reach? What happens when the AC ground fails to provide a fault current path if there's a short circuit? What happens if a power supply burns up? Will circuit breakers still work? Will rubber gloves and boots work? The practice may be an effective way to reduce noise, but it is also extremely dangerous.
Could it be that the answer to the problem is in the power itself? Certainly there is room for power to be cleaned up. Conventionally, this is done in two ways: improving its level of signal purity, or improving the grounding system's linearity and lowering its impedance. Unfortunately, even the cleanest power signal available and the most linear and low impedance grounding system still yield AC noise in audio and video circuits. How could noise still affect a video production facility with top-of-the-line active power conditioning and a professionally engineered linear signal reference? The reason is that power conditioning systems don't address all aspects of the noise problem. For that matter, the most damaging form of noise (reactive current in the ground) is left untouched. Low impedance grounding has little effect on reactive current so the problem persists. Clearly, a different approach is needed.
The answer is preventing reactive currents from propagating in the grounding system. This can be accomplished by re-phasing the power source. To get a clearer picture of this theory, one needs to look at the load, the ground reference and the voltage phase in the power wires. For some reason, the simple truth has been overlooked for years.
Back to basics
First, let's look at some basic electrical theory that has somehow escaped the view of the majority of the engineering community. To understand the solutions to noise problems presented here, it must first be understood how the power phase can be both referenced to ground and applied to a load. This area of electrical theory is poorly taught, and narrowly applied. Except for filtering, a few multi-phase industrial uses and some critical safety applications, the importance of voltage phase orientation to a reference source is generally ignored. Here are four examples of what is meant by the term "Mode."
Figure 1 shows an alternating current signal in Direct Mode (sometimes called Normal Mode.) One wire carries a voltage potential, the other wire is a ground conductor into which modulating current flows when a load (or signal pick-up) is connected to complete the circuit. This mode is often used in low power, unbalanced or high impedance applications such as antennae, video feeds, data networks and test equipment circuits.
Figure 2 is similar to Direct Mode, but a third wire is added to provide a safety ground that is not part of the circuit's normal current path. This is a much safer way to apply a high amplitude signal such as feeding 120-volt power to a refrigerator or a washing machine. When an appliance is grounded in this manner, the chassis of the appliance isn't part of the normal current path and has no potential unless there’s an internal short circuit. This is the standard 120-volt power circuit configuration used in the United States. The two AC power conductors have inversely phased (differential) current and an unequal voltage potential to ground. One wire is hot, the other is neutral. This power configuration is called "Differential Mode."
If one were to omit a ground reference altogether, and simply apply a 120-volt signal differential to two wires from a source system such as a single-phase transformer or generator, the only voltage potential or current flow is across the two circuit wires (Fig. 3). One conductor is referenced to the other and vice-versa. No part of the power circuit is referenced to ground. There is no current flow or significant voltage potential anywhere other than between the two circuit wires. This configuration of applied signal is called "Transverse Mode." The signal traverses two circuit wires without a grounding reference source. Electro-cardiograph equipment in hospital operating rooms operating on 120-volts AC is one example of a Transverse Mode power application.
A fourth variation is Common Mode. The voltage potential traversing the circuit is referenced to its own zero crossing point. (Fig. 4) The power resembles a balanced audio circuit or an XLR input from an unbalanced to balanced audio transformer. In this configuration, a center tap (ground reference) is added to the transformer output winding which divides the output voltage into two 60-volt-to-ground potentials. The potential across the system's two main current-carrying wires remains the usual 120 Volts. From the center tap reference point to the transformer output terminals, the measured voltages are identical but inversely phased. The power output is symmetrical -- each side of the system is a mirror image of the other.
The culprit is exposed
Let's look at some of the problems created by standard (Differential Mode) 120-volt circuits when one applies power to a typical impedance load.
Figure 5 demonstrates how noise invades a grounding system through the AC neutral. Here, a load is applied transversely to a Differential Mode power source, the usual case. Note how the grounding reference of the system loads up with harmonic currents and voltage potentials. These are commonly manifest as a sort of "voltage signature" in the grounding system, different with every piece of audio equipment. As more gear is turned on, more of these signatures appear. This is perhaps one of the trickiest of all noise problems to handle from both a power and grounding perspective.
If one opts to use a low impedance transformer as a power source, some noise will be attenuated. However the trade-off here is an increased danger of power spikes and voltage variations. In a low impedance system, voltage variations and transient surges can be considerably greater than normal. This means that the possibility of damage to electronics is also considerably greater than normal. Protecting one's equipment from this exaggerated danger condition requires installation of an $800 facility transient voltage surge suppressor and a $2000 line reactor -- a lot to pay for 2dB to 3dB of background silence. Radial or star grounding may somewhat lessen the effect, but as with most known remedies, it's no better than a patch.
Before continuing on, let's look at one more problem that is also very disruptive to an audio ground. This problem manifests when a Differential Mode (unbalanced) power signal is applied to a balanced circuit (again, the usual case) -- for example, an RF filter. Figure 6 shows how RF filter capacitors leak current into the grounding system. This is common hum, a highly reactive 1/2 pulse width leading current that is one of the most engineer-resistant of all known types of AC interference. Switching power supplies in digital gear and similarly reactive, nonlinear loads (for example: SCR dimmers) also shunt current to ground in like fashion. Class-A tube amplifiers are particularly vulnerable to this capacitive noise phenomenon, as are balanced microphone preamp inputs and other high-gain inductive circuits where low-level signals are critically referenced to ground.
Together, Figs. 5 and 6 illustrate the source and the creation of almost all grounding noise (with the exception of interference originating elsewhere in a building's power or grounding grid, unmatched or defective components in audio circuits, sub-standard audio grounding schemes and an occasional magnetic aberration). A few chassis power supplies can also be included under aberrations. But most of these are rare, and when they do appear, they are easy to identify. The important thing to recognize is that the source of most grounding noise is directly linked to unbalanced power.
Let’s apply a balanced (Common Mode) 120-volt power signal to the same RF filter and to the same impedance load.
Figure 7 and Figure 8 are illustrations of a balanced power application. In both cases, inversely phased power elements meet at the common ground. The effect on the grounding system is also observable in the power grid with a near-total absence of locally generated noise under the balanced power/load condition. Note that when balanced power is applied, peak and inverse peak voltages are equally present with respect to ground. This means that reactive currents generated by the impedance load are also perfectly balanced. (In the electrical industry, the term "Counter EMF" is sometimes used to describe these power artifacts.) Counter power elements which are balanced will null (sum to zero) in a balanced system at the ground or center tap, an obvious benefit in sensitive audio and video production facilities where overall noise floor and dynamic headroom are directly affected by grounding noise levels.
Today's facilities are fed typically with three-phase service. Common three-phase power systems lack an inverse phase relationship between power wires. Power wires are phased 120 degrees apart so there is never a point in time where the system is completely off (common zero crossing voltage) as in single-phase systems. Consequently, reactive currents instead of nulling begin to "stack up" in manner of speaking, end to end at very well-ordered intervals on both the neutral and ground wires. After each phase peaks, a pulse of reactive current is released onto the common wire or neutral. This occurs at a rate of 60 times per second per phase. All three phases sum together in this way on the neutral equaling 180Hz, the third harmonic. On line video facilities powered by a three-phase (120/208V) grid are typically beset with noise problems. When three-phase based, 120-volt power is used directly for audio or video electronics, one can be assured of more interference than with all other types of 120-volt systems.
One issue remains. What is wrong with grounding? The answer is simple. As a rule, the grounding circuit is not meant to carry any type of current except for rare occasions when there is a short circuit, and then only momentarily. However, Underwriters' Laboratories standards are somewhat lax in this respect. “Objectionable ground current" is loosely defined in every area except for hospital operating rooms. Outside of hospitals, “objectionable ground current” is gauged more closely to shock hazard levels (about 3.5 milliamps). It seems that this standard has backfired. Data corruption, disk crashes and simple hum noise cost time and money -- sometimes a lot. Clearly, UL standards that define "objectionable ground current" are inadequate by today's standards. On the other hand, UL is not responsible for re-engineering audio equipment and AC systems. Normally, performance standards are left up to manufacturers and the marketplace.
The basic problem is that when loaded, all differential mode 120-volt AC circuits create noise in the ground reference. Noise problems occur in audio signal circuits due to an unclean ground. The old Edison circuit (where a 240-volt circuit is split into two 120-volt circuits) is still used today as a standard means of 120-volt power distribution. Standard power is unbalanced. Even when a high quality 120-volt isolation transformer is installed, one side is grounded (made neutral) which is really not much different at all. The way the AC voltage phase is referenced and carried by the circuit has everything to do with electrical interference in a grounding system. If any aspect of the circuit is applied or loaded in an unbalanced manner, noise will appear in the ground.
Success, at last
Understandably, such a simple explanation of noise problems can invoke a kind of knee jerk skepticism or denial. But, requiring proof is not an unreasonable demand. In locations using balanced 120-volt power systems, the results speak for themselves. When balanced power is applied systemwide, the results are often quite dramatic. On the average, 16dB improvement in background noise has been noted. Where audio and video components are properly installed and maintained, in no known case has balanced AC failed to substantially lower the noise floor. In high-end systems utilizing 24-bit digital equipment, peripheral gear needs balanced power to approach the noise floor capabilities of the digital system.
Not long ago, a well known audio dealer was checking some class-A tube gear before shipping the exhibit off to the Audio Engineering Society Convention. With several amps on the bench, he was resigned to spending six hours changing capacitors to clean up some hum problems. As fate would have it, a delivery arrived at that moment with two prototype 120-volt symmetrical power systems he had ordered for the show. Predictably, and nonetheless miraculously, the balanced 120-volt power systems rendered the hum inaudible saving the fellow some capacitors and hours of work.
Few if any modifications to electronic equipment are ever needed. However, in cases where a device is supplied with a two prong AC line cord, it has been found that retrofitting the unit with a standard three-wire grounded cord further reduces noise. With rare exceptions (extremely sub-standard internal grounding schemes) ground lifters should not be used. Notably, audio wiring is also much easier to configure, as is the case when connecting unbalanced outputs to balanced inputs. Audio isolation transformers are seldom required to eliminate noise. Another reason for grounding is to shunt RF away from grounded shields and chassis. A clean ground reference for attaching RF shields eliminates many problems. Such pleasantries are too numerous to mention. When balanced power is used, the general rule is: "If in doubt, ground everything and lift nothing." Grounding essentially works the way it was taught in school.
When symmetrical AC power is used, grounding also tends to be more forgiving. Recently, a studio owner needed some consultation and assistance installing a 120-volt symmetrical power system for his home studio, a 24-bit digital facility that also included a fair amount of class-A tube gear. Despite careful planning, there was a miscommunication over some of the particulars regarding radial (star) grounding for the outlets. Consequently, the electrician, unfamiliar with studio AC wiring systems, looped entire strings of outlets together using only one #12 gauge ground wire. In spite of the mishap, the system performed perfectly -- quietly. This episode would indicate that unbalanced power is the true cause of noise, not a poor AC grounding design. .
The system was built full of ground loops, but grounding currents and chassis potentials were nowhere to be found. This is how it should be. Star grounding and linear signal reference grids are nothing more than "Band-aids" that, at best, can only marginally mask over some of the noise. To be free of noise problems, grounding circuits must remain clean -- that's all there is to it. If they're not, a hundred ground rods won't help.
The symmetrical 120-volt system is unique in that it deals specifically with balancing all power and load elements with respect to a single point grounding reference. This is the only prescription for maintaining a clean ground regardless of how big the facility or what sort of gear is used. A specially wound isolation transformer with a center tapped 120-volt output shown in Figure 4 and again in Figure 7 and Figure 8 is basically the heart of the system. Both the load and the power signal are balanced with respect to the common output terminal (center tap) on the transformer. Here is the true single point grounding reference for a recording studio. The Earth ground now functions only as a reference for electrical safety and for shields as it should be, not as an ineffective sink-hole for reactive current.
Every professional audio/video production facility that has tried a symmetrical power system has demonstrated a significantly lowered noise floor. Particularly, in cases where meticulous attention has been paid to selecting high-quality balanced audio equipment and well designed audio wiring, the difference can be astonishing.
Interestingly, the most common "complaint" about symmetrical power comes from guitar players, who interpret hum as a sign of reassurance that the amp is turned on and up to speed. It seems that the silence is an annoying distraction to them. That's too bad. They'll get used to it.
Martin Glasband is the President of Equi=Tech Corporation in Selma, Oregon. Formerly as an electrical contractor and consultant has designed electrical systems for KCET-TV, The Post Complex, Warner Brothers, Oz Audio and Post, ABC Radio Network, Baby'O Recorders and New World Pictures.
Technical Articles Index ...... Equi=Tech Index......MIX Magazine.