Jun 1, 2000 12:00 PM,
Learn to handle hum and buzz with a critical look at system noise from theground up.
For sound and video systems in stallers, voodoo and wives tales aboundabout how to achieve minimal system noise. By noise, I mean hums andbuzzes, which are power- and grounding-related issues. Hiss, on the otherhand, is a gain-structure issue not covered in this article.Troubleshooting noise can eat into profit margins for installers,technicians and end-users, ultimately consuming an unpredictable amount oftime and money. This article is intended to help everyone who has ever beenburned by a problematic system.
The lack of noise immunity in some equipment is an insidious problem – theprimary topic of this article – and the industry’s dirty little secret.Long before software and hardware developers started pointing fingers ateach other, all eyes turned toward the installer when the system powered upfor the first time humming a tune. At least bad analog gear has somewhat ofa voice to express its unhappiness – hums and buzzes for audio and avertically creeping, horizontal bar for video. Digital gear lets us knowonly when it is too late, but it has increased manufacturer awarenessregarding noise immunity, yielding a heightened consciousness that hastrickled down and into some modern analog gear.
The first part of this article will cover systems basics and what shouldhappen. Later in the article, I will address the fixes that have beenattempted to accommodate bad gear and why they do not always work. Some arevalid as insurance policies; others are borne out of paranoid overkill andmostly unnecessary.
I could not have written a single word about system noise withoutexperience in the field. I have admittedly made mistakes and am lucky tohave a tenacious, curious nature. All this makes me knowledgeable, but I amnot an expert. In a parallel universe, brains far more brilliant than ourshave wrestled these thorny issues. You can find more detailed informationfrom Bill Whitlock at Jensen Transformers (www.jensen-transformers.com),including a library of reference books for your suggested reading.
Another authority is Neil Muncy, who chairs the AES Standards Committee(SC-05-05) on the subject. You can find his condensed overview of suggestedpractices at the Rane site (www.rane.com/note110.htm), and honorablemention of Muncy appears at the Philip Giddings site(www.engineeringharmonics.com/papers/pin1.htm) specifically addressing thepin 1 issue (more on that later). At www.josephson.com/audiofaq, there is awonderful FAQ compiled from the rec.audio.pro newsgroup by the late GabeWeiner.
There are four solutions typically applied to circumvent system noiseproblems. It is not my intent to debunk what some may have embraced asreligion, merely to shift the focus to the real problem, which involves thegear itself. The first solution is power distribution, an important part ofevery system’s foundation. Before considering noise insurance options likebalanced power, an isolated ground or star grounding, all of which requireconsiderable effort and cash, it is important to remember how easilyinstallation costs can exceed the cost of the gear. Few customers can claimthat money is no object, and none of the aforementioned items are anabsolute guarantee of long-term silence.
The second common solution for noise problems is the use of balanced gearand avoidance of troublesome and, most likely, unbalanced equipmentwhenever possible. The third fix requires the installer to tame the problemchildren with transformers or an active-balanced interface. Necessity beingthe mother of invention, the fourth, only the most tenacious, uses gearmodifications to drive the evil spirits away. Some of these modificationsare simple; some are not.
Many designers and installers would choose to start from scratch ratherthan perform surgery on a system with a preexisting condition. It may behard to convince a potential customer of this approach, but the experiencedtraveler will avoid the dark road that leads to wasted time, lost profitsand customer dissatisfaction. It is too easy to put the blame and the curseon a previous installation and installer, respectively, so avoid the habitof flaming the competition.
Let us examine the foundation of the problem. Power from a standard walloutlet is an alternating current (AC) at 60 Hz. At minimum, this humradiates from the power cable at the moment a device is plugged in andturned on. Sound converted to electricity is also AC, but batteries producedirect current (DC). Note that most equipment ships with alonger-than-necessary power cable. In cramped spaces, power cable bundlesmake for a crowded, unserviceable installation. Belden makes an 18 inch(457 mm) IEC power cable (part number 17002A-B1-10) available from NewarkElectronics.
Additionally, electronic equipment is typically built on a metal chassisthat is referenced to the earth via the round prong of a three-conductor ACpower plug. If you are barefoot, pregnant and standing on a slab of dampconcrete, all properly grounded equipment is safe to the touch.
Hum and buzz are often attributed to the elusive ground loop. Thisphenomenon might more accurately be described as ground current in that notwo pieces of gear can be exactly at the desired zero-voltage reference(the earth). When connected by audio or video cables, an extremely smallamount of AC current flows in the shield and ground wiring. Inside eachelectronic device, leakage currents from power transformers induce anelectrical charge on the chassis. Even though the unit is grounded via thepower plug, the length of all power cabling (back to the breaker box)amounts to enough resistance so that no chassis can be held to exactly thesame voltage as la terra firma. Because this difference in potential (ameasurable voltage between any two chassis) exists in nearly all instances,you might think we face insurmountable odds, but in reality, ground currentaffects only flawed gear. The typical external fixes, such as runningground wires and flying the shields at one end (!of audio cables), solves the immediate problem; ground wires lower the impedance of the path to ground,reducing shield current. Disconnecting the shield at one end eliminatescurrent flow, but neither is the long-term solution.
Properly designed balanced gear should tolerate shield current. Unbalancedgear, on the other hand, cannot. By design, a balanced input circuit isimmune to most cabling noises by way of its common mode rejection ratio(CMRR). At minimum, CMRR sees noises common to both signal wires asout-of-phase and as such, cancels the unwanted intruders. Not all balancedoutputs have signal on pin 2 and pin 3, but the source impedance at eachpin must be the same for CMRR to be effective.
Another source of common-mode noise is a wall wart. In close proximity to abalanced audio cable, any transformer will induce hum into both the signalwires and the shield. For wiring that feeds unbalanced gear, pay strictattention to cable dress and run wires away from power transformers bothobvious (wall warts) and hidden (in gear). Balanced gear can tolerate bothcommon-mode (induced) noises as well as shield current noise.
Internal product grounding will be referred to as the pin 1 issue not onlyas it relates to XLR pin 1 but also other connectors ( inch, RCA, BNC). Thecorrect approach is easily illustrated by observing the many inchconnectors on a Mackie mixer, all of which are metal jacks on a metalchassis. This approach creates a low-impedance firewall that protects theinternal, high-gain circuitry from external influences.
It was not so long ago that manufacturers chose connectors that wereplastic-insulated from the chassis. Such a choice may be cost effective oreven streamline the manufacturing process, but it undermines the internalground scheme, decreasing the noise immunity of even good product designsmaking the resulting gear inherently vulnerable to any power- ortransmission-induced (RF/TV) interference. The reason is simple. Printedcircuit board (PCB) ground traces do eventually make contact with thechassis. If shields do not go directly to the chassis, the pin 1 groundcurrent (and its noises) will infect the PCB ground and all amps referencedto it. Remember that a mic preamp can have 60 dB of gain, more than enoughto amplify even the slightest ground noise. Higher ground and noisecurrents can also raise the copper’s temperature, increasing its alreadyhigh resistance and decreasing ground integrity within the unit.
I once received an e-mail regarding a mic preamp that picked up televisioninterference (TVI) with a certain condenser mic and not another.Incidentally, TVI is often heard as a 59.94 Hz buzz, video’s vertical syncrate, with harmonic variations that correspond to the transmitted videoimage. This was a clear case of pin 1 not being connected to the chassis atthe point of entry. Resolving the problem initially seemed complex – how tofix the problem via e-mail without the customer opening the preamp to makeany modifications.
Because condenser mics require phantom power, there is DC current flowingin the shield in addition to whatever external noise currents the shieldacquires while trying to do its job. The phantom power requirements ofcondenser mics may vary, but all should be within the industryspecification. The higher current requirements of this particular mictugged on the ground enough to cause the preamp to detect the TV signal. Inessence, piggybacked on to the DC current in the shield was the ACtransmission of a local TV station, the most obnoxious of which was thevertical sync frequency.
To fix the immediate problem, I simply had the customer open the male XLRof the mic cable and wire pin 1 to a lug on the XLR connector designed toroute the shield to chassis. (See Figure 1.) The customer was now happy,and I felt rewarded for the number of times I have attempted to improve aproduct’s noise immunity on the bench. Time spent on equipmentmodifications may not always be profitable, but it sure is enlightening.
Perhaps you can now understand why, when customers ask about the relativemerits of balanced power or any external ground scheme, I have difficultygiving a straight answer. I believe that if all gear were balanced withwell-executed internal grounding, then no deviation from standardelectrical power distribution would be necessary. That said, customers withenough cash to pay for noise insurance will reduce potential system noiseswith the aforementioned items. On the other hand, as systems grow in size,a multi-room or multi-floor facility with great cable distances, it becomesmore difficult to implement and distribute a power and ground scheme. Moreimportant is to make good equipment choices. To installers who learn thehard way that certain gear is consistently problematic, please forward thatinformation to the manufacturer.
There should not be any voodoo required to achieve low system noise, butmere knowledge of the noise devil does not solve the problem. In order tocomplete the circle of designer, installer, end-user and technician, wemust all provide feedback to encourage manufacturers to take the lead.
A critical look
To sum up thus far, I have detailed the causes of hums and buzzes in audioand video systems, placing the blame on gear that falls somewhat short ofbeing professional in terms of its inability to reject noise. Gearmodifications might be the short-term answer to what ultimately shouldhappen as part of the manufacturing process. Further, all of us have takendetours in an attempt to accommodate unhappy pieces of gear. As atechnician, the worst that I have seen are user-installed systems wherehome-brew fixes include everything from the benign – rubber chassisisolators and plastic rack screws – to the illegal lifting of the third pinAC ground. All are band aids that make for an idiosyncratic installation. Iwill now focus on why various external fixes do not consistently solvenoise problems. Hopefully, after finishing this article, you will haveenough background information to assist those in the trenches who aremoving too fast to understand or investigate the underlying! problems fully.
Regarding the implementation of electrical power, the rules and regulationsof the electrical code vary regionally. I am no expert. Finding aknowledgeable local electrician who is sensitive to the needs of amultimedia system is key. Using a ground adapter as a ground lifter is, bycode, illegal. No matter whether the gear, the signal wiring or the powerdistribution is at fault, every temporary fix will eventually become anintermittent noise problem while also creating a potentiallylife-threatening situation.
The outlet-ground connection is simply to protect humans from the risk ofshock (see Figure 2), and despite all the fuss about the dreaded groundloop, consider how may potential loops are created as soon as a piece ofgear is rack mounted, plugged in, then connected to another piece of gear.It would seem as if we are doomed at the start and that meeting code andachieving a quiet system are disparate goals, but that is not the case.
I will start with ground contamination. Standard 120 VAC Power is deliveredto the outlet from the breaker box as three wires – hot, neutral and ground- all traveling through a common jacket. Cable types include plastic Romex,flexible metal jacket BX and conduit (pipe). Romex has a dedicated groundwire. Metal-jacketed BX cable has a somewhat less-substantial, non-insulated ground wire in addition to using the jacket as a conductor. Metal conduit isalso used to distribute ground. In the latter two cases, the groundconnection is made via clamps to the metal jacket. Following the groundfrom breaker box to a common outlet, the connections along the way are notalways as positive as we would prefer. Only when the outlet is screwed intoa metal mounting box is the ground connection made. This meets code forsafety, but its effect on the system noise is a different matter.
A loose clamp leaves room for oxidation to build up, increasing theresistance and therefore increasing the possibility that intermittentnoises will be injected into the wiring and into susceptible gear asappliances are turned on and off. Once conduit is buried under layers ofsheet rock, you cannot go back and tighten every joint, so a dedicatedground wire provides long-term piece of mind by minimizing some of thevariables.
As power demand changes over time, the neutral wire can also become noisybecause up to three phases (120 V legs) use it as a return line, ultimatelyto ground. When power distribution is suspect, check all connections withthe power off and tighten when loose.
There is also the matter of star worship. A star ground scheme involvesrunning a separate ground wire from each piece of gear, typically to acopper spike (or plate) penetrating deep into damp earth. This is no easytask. Standard rack mounting provides an electromechanical link. Dependingupon your perspective, this either defeats the star or serves as a goodstart toward ground distribution. For example, connecting a massive groundwire to each rack rail rather than to each piece of gear. Remember, I amonly documenting here, not suggesting.
Star fundamentalists feel the rack rail is just another potential loopconnection and would be happier with a wooden rail, leaving space betweeneach chassis so they do not touch, attaching a ground wire to each chassis,flying shields at either input or output connectors and adding groundlifters to all the gear. Ultimately, this is both time consuming andillegal. Again, the solution starts with the gear – balanced inputs andoutputs with pin 1 going directly to chassis.
The problem with attempting to distribute an isolated ground is that theground wire is not so isolated; it is in the same jacket as the powercables. The more current that flows, the more the power cables induce humand noise into what is no longer a clean ground. Guess what? All of yourhard work and money went down the tubes, especially when the pin1-to-chassis rule has not been followed.
To prove that the ground wire had become contaminated, my theory was testedin an installation where a video monitor showed hum bars when connected toa ground-isolated outlet. By simply routing the ground wire independent ofthe power wires, the visual hum was eliminated. To implement this solutionit would first be necessary to find a code-legal way to distribute groundwires through a dedicated ground conduit. Because code varies from state tostate, discuss the matter with a licensed electrician.
Remember that standard outlets make ground connection when screwed into ajunction box, which is mounted to either a wood or metal stud. Metal wallstuds can become a path to some other ground; plastic outlet boxes caninsulate a standard outlet from a metal stud. Hospital-grade orange outletsare often used because the ground is isolated from the mounting hardware.
Note that balanced power does not induce hum into the ground wire. Balancedpower starts by installing a power transformer with a pair of 60 Vsecondary windings – the two 60 V legs are 180 degrees out of phase, justlike a balanced audio signal. The junction where the two windings arejoined is called the center tap, the null point or 0-reference that is tiedto ground. Now, the hot and formerly neutral connections are each 60 V(60-0-60) with respect to ground and 120 V with respect to each other.Closely matching the two secondary windings minimizes the noise radiation.
Finally, there is the matter of the other wires. The preferred power outletsecures its wires under a screw or by a screw-tightened clamp. Theundesirable outlet secures its wires by spring pressure. Either way isconsidered to code, but the latter can create problems down the road.
It is quite common (and code) for electricians to daisy chain multipleoutlets, looping the power through two or more outlet boxes. Thecombination of potentially poor physical connections and high current willgenerate heat, further degrading the connections. Expansion and contractionover time increases resistance at every junction, thereby making each hot,neutral and ground connection a potential noisemaker.
Power wiring in such a state will become especially vulnerable to devices,such as air conditioners and elevators, with momentary high-currentdemands. In addition, some power supplies kick back noise into the powerlines. As each device snaps online, the weak links will exaggerate bothtransient as well as continuous noises. These will be particularly hard totroubleshoot in a multi-tenant commercial building.
The local solution is to turn off the power at the breaker box and inspectall connections from the breaker box to the outlets. Tighten screws andrewire any outlets not using the screw or screw-clamp connections. Wiresshould be under screw pressure. A global inspection should include thepoint at which the electrical service enters the building.
Running a dedicated ground wire from a central point to each outlet may noteliminate system noise or fix bad gear, but at the very least, it providesa bit more confidence in the long-term integrity of the ground. Of course,as your system grows so will its problems. Start with a good foundation,and remember that power and ground distribution must also be inspected andmaintained.
Ever notice that when flying at 20,000 feet (6,100 m), you have nosensation of motion? I do not mean turbulence or steering but the actualspeed at which you are traveling. To a certain extent, this analogy can beapplied to ground-related noises. So long as every device is flying on thesame ground (no matter how noisy), no device sees the noise. You might saythat this is the ground plane. You are allowed to laugh or groan.
Here is another analogy. With one foot on a building floor and another onan elevator floor it is obvious that the elevator is not as steady as thebuilding. This correlates to what happens when a direct box inadvertentlylinks the system ground (via the recording console) with a bass amp pluggedinto a utility outlet that is not connected to the officially sanctionedground. The point here is to translate what we already know – the sound ofhum and buzz – into what actually happens when two grounds are not at thesame potential or cleanliness, relative to the earth or to each other.
Tie yourself to two horses facing opposite directions and the aboveanalogies will start to make sense. The idea is that differences betweentwo devices puts stress on their connective as well as internal wiring. Thedifferences come from voltage and noise variations. The problems come fromgear with poor internal grounding and from wall warts placed too close tounbalanced wiring. Even a small audio system can have problems that startwhen multiple power strips are haphazardly plugged into different walloutlets. Distribute power from a single outlet, and chances are most of thenoises will go away. This test is safe assuming that you know the systemwill not exceed 75% of a 20 amp breaker (15 amps) and that nothing else ison that circuit.
Connect the first two power strips to a standard wall outlet, and to thosepower strips, connect the other power strips in the most symmetrical waypossible. You can also use a two-to-six outlet adapter that screws into astandard wall outlet; all strips then plug into that. Remove allground-lift adapters. The emphasis here is to use a single outlet.
Power up slowly so as not to overload the breaker, and, if possible, use acurrent probe at the breaker box to monitor demand. The system should nowbe significantly quieter. If not, find all the wall warts and make surenone of the audio cables go near them. This is only a test. If the treesolves the noise problem, and the breaker is not overloaded, you have atemporary fix. Finding the problem gear is not quite as easy, but the clueis to look for plastic-insulated jacks and products with unbalanced inputs.
An awfully big build up for what seems like a rather simple but temporarysolution, right? When a customer asked for this noise reduction tip, it didnot help because there was one piece of gear that had to be plugged into aconvenient outlet rather than follow the discipline of using a singleoutlet. So much for science.
Most of the noise was a combination of using unbalanced sources and cabling(MIDI modules) to a mixer that hummed with so many cables connected andfaders down. Taking advantage of balanced inputs and cabling would havehelped tremendously, even with unbalanced sources. The downside toaffordable technology is that it makes a professional installation seemrather expensive.