Keeping It Quiet
Aug 1, 2001 12:00 PM,
By Bill Whitlock
THAT UMPTEEN-THOUSAND-DOLLAR HOME THEATER system loses some of itsallure when you turn it on and see interference bands moving acrossJulia Roberts’ face and hear teeth-rattling static coming out ofsurround-sound speakers. While professional balanced interfaces largelypreclude noise problems, most audiophile and home theater systemsconsist mostly, if not entirely, of audio and video equipment withunbalanced inputs and outputs, a danger zone for signals. In all butthe smallest systems with very short cables, noise problems are likelyto exist.
The ultimate goal of sound and image reproduction is realism, thekind that creates suspension of disbelief in the listener or viewer.And nothing breaks the spell like background buzz during a quietmusical passage or a suspense-filled moment in a movie. No matter howgood the reproduction technology is, interference can really spoil theexperience.
Because a reasonably quiet home listening room may have a backgroundnoise level of some 20 to 35 dB SPL anyway, many professionals believethat reproduced equipment noise below these levels will be masked andtherefore inaudible. But in fact, on average, a listener can detectequipment noise some 15 to 30 dB below ambient room noise! Ourears and brain are very smart, using learned spectral signatures anddirectional cues to easily identify hum, buzz or hiss, even in thepresence of other ordinary household noises.
Therefore, the dynamic range of a sound will need to be somewherebetween 100 and 125 dB. Recent advances in converters and digitalrecording have already extended the dynamic range of CDs, and it’s asafe bet that program sources will develop even wider dynamic range.Playing an ordinary CD with about 95 dB of dynamic range on a systemwhose residual noise limits dynamic range to 80 dB will bury thequietest passages in noise, painfully reminding the listener that he orshe is listening to a mere recording.
Consumer expectations are high, and the ranks of “goldenears” are growing. For video, dynamic range requirements are abit more relaxed. Even expert viewers cannot detect improvements insignal-to-noise ratio beyond about 50 dB.
WHAT’S MAKING NOISE?
Any signal accumulates noise as it flows through the equipmentchain; and once it is contaminated, no process can remove the noisewithout degrading the original signal. Since the dynamic range of anentire system can be no better than its weakest link, noise must beavoided everywhere along the signal path. In most systems, the worstproblem is not signal processing in the equipment itself but so-calledpickup, or noise coupling, in the interconnect cables.
This noise is most often a mixture of 60Hz harmonics and otherhigh-frequency noises that normally exist on AC power lines. Since itenters the audio or video signal path via ground wiring, this noise isusually referred to as ground noise. It should not be confusedwith random noise, which manifests itself as hiss in an audio system orgranular movement (snow) in a video image. A predictable amount of thiskind of electronic noise is inherent in all electronic devices and mustbe expected. Ground noise produces artifacts such as hum, buzz, clicksor pops in audio systems. In video systems, it generally produceshorizontal bars (light or dark) or bands of specks that slowly moveupward in the image.
The subject of grounding is widely misunderstood and plagued withnonsense rules. As a result, eliminating system noise often becomes along series of experiments that finally end when someone says, “Ican live with that.” At best, this method leaves a systemvulnerable to a reappearance of noise when someone adds a new piece ofgear. At worst, it results in an electrocution hazard and a legalliability when someone disconnects required safety grounds.
WHY IS EQUIPMENT GROUNDED?
Just the term ground itself can cause confusion. For thisdiscussion, we need to talk about three kinds of grounds.
- Earth or Neutral GroundUtility AC power systems require an actual earth connection toprotect people from lightning. Lightning involves millions of volts andtens of thousands of amps in discharges from clouds to earth — itreleases incredible amounts of power. Before modern standards like theNational Electrical Code were developed, a strike to a power line wouldoften just follow the power line directly into a building, killingsomeone or starting a fire. Power companies quickly realized thatconnecting one of the lines solidly to earth was a very good idea. Itgave the lightning an easy and relatively safe path to real earthground, before the lines entered a building. Virtually allmodern electric power is now distributed over lines having anearth-grounded or neutral wire in addition to other live orline wires.Electric companies supply most residential customers with a 3-wireservice. One of these wires, which often is not insulated, is thegrounded neutral conductor. This earth ground, along with those of yourneighbors and other grounds at the power poles, provide lightning aneasy path to earth.
- Safety GroundAny AC-powered device can electrocute a user if it develops defects.Most electric devices have exposed metal parts. The transformers,switches, motors and other internal parts use insulation to keepdangerous currents from reaching external parts. But insulation can anddoes fail for various reasons, and often fails in a way thatelectrically connects the hot power line to the exposed parts. Forexample, if the insulation inside the electric motor of a washingmachine failed, someone could be electrocuted if they touched themachine and a water faucet (grounded) at the same time. To prevent sucha tragedy, most devices have a third wire connecting, or safetygrounding, any exposed conductive parts to the third prong of theoutlet.The outlet safety ground is routed, through either a green wire ormetallic conduit, to the main breaker panel where it connects toneutral. The circuit from the black or hot wire, through the defectiveequipment, returning to neutral via the green wire or conduit, iscalled the fault current path. With the safety ground inplace, the potentially dangerous equipment failure simply causes a highfault current that trips the circuit breaker, safely and quicklyremoving power from that branch circuit.Electrical safety is an extremely important issue. While nobodywants to intentionally create a lethal shock hazard, ignorance of howthe safety system works can kill.Never Use Devices Such as 3-Prong-to-2-Prong Adapters to Solve aSystem Noise Problem. These are sometimes even referred to asground lifters. Remember that cables connecting equipment canalso carry lethal voltages throughout the system if one lifted devicefails. The judge in a liability lawsuit won’t care about your humproblem.Many appliances and electronic devices are supplied with 2-prong ACcords. Sometimes called double-insulated, these devices mustbe specially designed and certified to guarantee safety even whenoverloaded or experiencing an internal component failure. Such designs,common in consumer electronics, are generally advised only inrelatively low-power equipment. Because they’re not grounded, theparasitic capacitances of their power transformers form a voltagedivider causing their chassis voltage to be a substantial fraction ofthe AC line voltage with respect to ground.If their chassis, or the shield of their RCA output connector, isconnected by a wire to safety ground, a leakage current up to about 0.5mA will flow. When this current flows through a person, it causes avery mild, harmless tingling sensation. In any AC-powered system, theexistence of these noise currents must be accepted as a fact of life.High-frequency power-line noise is created by power supplies inelectronic equipment, fluorescent or dimmer controlled lights, andintermittent or sparking loads such as switches, relays or brush-typemotors. Since the noise current is coupled through small capacitancesin each piece of equipment, high-frequency power-line noise is coupledmuch more efficiently than the pure 60 Hz.
- Earth ConnectionAs I’ve described, a power engineer or electrician definesground in terms of electrocution and fire hazard and earthconnection. To an electronics engineer, ground is simply acommon reference point that serves as a return path for various circuitcurrents inside a piece of equipment. It has nothing to do with earthground! Sound systems in airplanes can operate quite well: Whether theyare quiet or noisy depends on how the system is designed and wired, notwhether it has a good earth connection.
NOISE CURRENT COUPLING
There’s a common misconception that noise is something airborne,picked up by cables, which can, therefore, be cured with more cableshielding. In real-world AC-powered systems, small leakage or groundnoise currents will always flow in any wire connecting two devices.This tiny voltage drop is actually what causes 99% of consumer systemnoise!
As shownin Figure 1, when two pieces of equipment are connected via anunbalanced interface, the noise current flows in the shield conductorof the cable. Because the shield has impedance, a small noise voltagedrop appears across the length of the cable, according to Ohm’s law.Since the cable shield is also part of the signal circuit, the noisevoltage will be directly added to the signal at the receive end of thecable, which is the sum of all the voltages in the loop from point A topoint C. Because the shield impedance is part of two circuits, noisecurrent and signal current, this mechanism is calledcommon-impedance coupling.
Consider this typical scenario. Two devices are connected by an RCAaudio cable. Both devices have 2-prong power cords, and theirpower-line capacitances cause a 300uA, 60Hz noise curent to flowbetween them through a typical 25-foot unbalanced cable. The cable hasa foil shield and a 26-gauge drain wire, making its resistance, or 60Hzimpedance, about 1 ohm. Using Ohm’s law, we calculate the resultingvoltage drop to be 300 uV. With respect to the nominal consumer signalreference level of 300 mV, this noise level is only 60 dB. And thehigh-frequency noise may be even worse and more audible as buzz.
The magnitude of the noise current can be much higher if bothdevices are safety grounded (i.e., having 3-prong power cords).Referring to Figure 1, consider that leakage currents from all deviceson a branch circuit cumulatively flow in the safety ground wiring.Since leakage currents for safety-grounded devices are not limited tothe 0.5 mA allowed for 2-prong equipment, it is not uncommon to haveleakage currents of 100 mA flowing in portions of the safety-groundwiring. Because of the impedance of the safety-ground wiring, thiscould result in well over 10 mV of noise voltage between the safetyground pins of two different receptacles.
And this can get much higher between two outlets on different branchcircuits. For two safety grounded devices, this voltage will beimpressed across the length of the signal cable shield and be directlyadded to the signal. For 10 mV of noise voltage, the noise would beonly 30 dB with respect to reference level.
In a standard video interface, reference level is 1 voltpeak-to-peak, including sync. The active range from reference black toreference white spans about 600 mV peak-to-peak. Since our previousnoise voltage of 10 mV (rms) equals about 30 mV peak-to-peak, oursignal-to-60Hz hum ratio would be 26 dB. This would cause a visible humbar in a video display. Even higher voltage differences can result ifone of the devices is connected to an outside ground point such as aseparate earth ground or a CATV drop.
Weeding out power-line noises such as hum and buzz from an audio orvideo system can be frustrating and time-consuming. The followingmethod is easy to understand and perform, and requires no testequipment other than ears. It points the way to solutions that are bothsafe and effective.
The success of any troubleshooting has a lot to do with how youthink about the problem. First, don’t think that, because you’ve donesomething a certain way many times before, it can’t be the problem.Remember, things that can go wrong do! Sometimes, just gatheringinformation will reveal the problem. Get as many clues as possiblebefore you try to solve a problem. And write everything down. Imperfectrecall wastes a lot of time.
Ask lots of questions. Here are a few good ones to start with:
- Did it ever work right?
- What are the symptoms that tell you it’s not working right?
- When did it start working badly or stop working?
- What other symptoms showed up just before, just after, or at thesame time as the failure?
Tinker a Bit
Use the equipment’s own controls, with some logic, to provideadditional clues. For example, if the noise is unaffected by thesetting of a volume control or selector, then it must be entering thesignal path after that control. If the noise can be eliminated byturning the volume down or selecting another input, it must be enteringthe signal path before that control.
Use a Visual Aid
Sketch a block diagram of the system. Show all interconnectingcables and their approximate length. Mark any balanced inputs oroutputs. Generally, stereo pairs can be indicated with a single line.Note any equipment grounded via a 3-prong AC plug, and any othergrounds such as cable TV or DSS dishes.
Ground Dummy Tests
The term dummy isn’t intended to demean the personperforming the test. It refers to special adapters that don’t passsignal. Dummies allow the system to test itself and pinpoint the exactentry point of noise or interference. By temporarily placing the dummyat strategic locations in the interface, precise information about thenature of the problem is revealed. The tests can specifically identifycommon-impedance coupling in the cable, magnetic or electric fieldcoupling to the cable, or common-impedance coupling insideequipment.
Dummiescan be made from standard parts and wired as shown in Figure2. Since a dummy doesn’t pass signal, it should be clearly markedso it doesn’t accidentally find its way into a system. Testing with adummy is described in the sidebar on page 42.
Equipment manufacturers are seldom a good resource for solvingsystem noise problems. Their technical advisors will usually blame badgrounding for hum and buzz problems. Some are so uninformed or carelessthat they’ll actually recommend disconnecting safety grounds as asolution! When a system contains two ormore ground connections, whether through power cords to safety groundor other grounded signal cables such as the CATV connection inFigure 3, a ground loop is formed, which provides acomplete circuit path for ground noise current.
Because a voltage difference that is often substantial existsbetween the ground for CATV and the safety ground at the sub-woofer, anoise current will flow in the shield of all signal cables that arepart of the loop. Common-impedance coupling then adds noise to thesignal in these cables. In general, the noise added is directlyproportional to the cable’s length. This sample system would probablyexhibit a loud hum regardless of the input selected or the setting ofthe volume control because of ground noise current flow in the 20-footcable. The hum might be slightly louder if the TV input were selectedand the volume were turned up because the same ground noise currentalso flows in the 3-foot cable. Because the ground loop is a seriescircuit, current flow can be interrupted by opening the circuit at anypoint. You might be tempted to open the loop with a 3-to-2 prongadapter at the sub-woofer AC plug, but that creates an electrocutionand fire hazard for which you would be legally liable.
There are two basic ways to reduce common-impedance coupling inunbalanced interfaces:
The shield circuit is the most common impedance. You reduce it byfollowing these steps:
- Keep cables as short as possible. Longer cables increase thecoupling impedance. Even short cables can produce severe coupling ifground currents are high. Never coil excess cable length.
- Use cables with heavy, braided copper shields. Cables with shieldsof foil and thin-gauge drain wires increase coupling impedance.
- Maintain good connections. Contact resistance is part of the commonimpedance. Connectors left undisturbed for long periods can develophigh contact resistance. Hum, or other noise that changes when theconnector is wiggled, indicates a poor contact. Use a good commercialcontact fluid and/or gold-plated connectors.
Reduce the Circulating Ground Current
Some tips for doing this:
- Don’t add unnecessary ground connections. With rare exception,additional grounds increase ground noise currents. Of course, don’tdisconnect required grounds to reduce noise current either.
- Use ground isolators at problem interfaces. Commercial isolators areavailable for audio, video and CATV signal paths as well as for digitalinterfaces.
Figure4 shows how a ground isolator breaks a ground loop. Since nocurrent can flow between the insulated transformer windings, noisecurrent can no longer add noise to the signal by flowing in the shield.A transformer uses magnetic coupling to transfer signals from primaryto secondary windings with no electrical connection between them. Aground isolator cannot remove hum and buzz if it’s placed randomly inthe system: It must be installed at the interface where the noise iscoupling. This is easily determined by the testing outlined above.
High-performance ground isolators not only provide true audiophilesignal quality but use internal shields to suppress ultrasonic and RFinterference. Be aware that most audio isolators or hum eliminatorsdon’t use proper shielding, and many use tiny telephone-gradetransformers that lose deep bass, cause distortion and create poortransient response. Beware of cheap products with scanty or missingspecifications — signal quality is at stake!
An audio isolator is a safe solution for the ground loop of Figure3. The isolator could be installed in the audio signal path eitherbetween TV and preamp or between preamp and sub-woofer. Highperformance should always be installed at the receive end of aninterconnect cable.
Another safe solution is to break the ground loop by installing aCATV isolator at the antenna input of the TV. Although less expensivecapacitive isolators are available, the ISO-MAX unit uses a wide-bandRF transformer covering 50MHz to 1GHz to reduce shield current flow tolevels by a factor of about 100.
If the TV in Figure 3 were driving a video projector having a3-prong AC plug, the ground loop might cause hum bars in the display,especially if the video cable is long. Because the signal is basebandvideo (composite, component or S-video), different types of isolatorsare required.
COPING WITH FIELDS
Magnetic or electric fields can sometimes induce noise in cables.Electric fields are generated around wiring or devices operating athigh AC voltages. Their coupling is prevented by a conductive outershield, which completely surrounds and covers the inner signalconductor in cables. Braided shields provide 80% to 95% coverage, whichis entirely adequate for all but extreme cases because electric fieldsare rarely a problem in audio or video systems.
Magnetic fields are likewise generated around wiring or devicesoperating at high AC currents. Building wiring, power transformers,electric motors and CRT displays are a few sources of strong ACmagnetic fields. Increasing distance between signal cables andoffending fields is the best cure for either electric or magnetic fieldproblems. Ordinary cable shielding, whether copper braid or aluminumfoil, has virtually no effect on audio magnetic fields.
Beware of Marketing Hype!
The only property of cable that significantly effects noise couplingis shield resistance. Coupling of even very low levels of ultrasonicpower line noise can cause subtle signal spectral contamination indownstream amplifiers. Rather than agonize over which exotic cablemakes the most pleasing small improvement, prevent the coupling with aground isolator. Expensive and exotic cables, whether double or tripleshielded, made of 100% pure unobtainium, or hand-braided by Peruvianvirgins, have no significant effect on hum and buzz problems.
Bill Whitlock is president of Jensen Transformers, Inc. He hasdesigned audio and video circuits and systems for 30 years. He lives inOxnard, California, and can be contacted by e-mail firstname.lastname@example.org.
R=1k•, 5%, 1/4 W Resistor
Audio 2C Phone
Same as RCA version, except use Switchcraft 336A and 345Aadapters
For Video RCA
Same as Audio RCA, except R=75•, 5%, 1/4 W Resistor
For Video BNC
Same as Video RCA, except use MilesTek
10-01015 and 10-01016 adapters
FIGURE 2: Dummy construction.
TESTING THE SYSTEM
Each signal interface is tested using a 4-step procedure. Ifpreliminary tests haven’t narrowed the problem down to a specificinterface or portion of the system, always start at the inputs to thepower amplifiers (for audio systems) or the input to the monitor (forvideo systems) and work back toward the signal sources. Be very carefulwhen performing the tests not to damage speakers or ears! The surestway to avoid possible damage is to turn off the power amplifier(s)before reconfiguring cables for each test step.
Unplug the cable from the input of Box B, and plug in only the dummyas shown above.
Theory: This test prevents any noise current that mightotherwise flow in the cable’s shield from entering Box B. It alsoeffectively shorts the input. Therefore, any noise output mustoriginate in either Box B itself or somewhere downstream.
Observe: Is system output quiet?
No: The problem is either in Box B or further downstream.Reconnect the cable and perform this test on the next downstreaminterface.
Yes: Go to next step.
Leaving the dummy in place at the input of Box B, plug the cableinto the dummy.
Theory: This test retains the shorted input while allowingany noise current flowing in the cable’s shield to enter Box B.Therefore, any noise output must be due to common-impedance couplinginside Box B itself or somewhere downstream.
Observe: Is system output quiet?
No: The problem is common-impedance coupling (a“pin-1 problem”) inside Box B or a device fartherdownstream. The hummer test can be performed on Box B to determinethis. If the problem is inside Box B, have the unit repaired, modifiedor replaced with a functional — and functioning —equivalent. If the problem is not in Box B, reconnect the cable andbegin the test procedure on the next downstream interface.
Yes: Go to next step.
Remove the dummy and plug the cable directly into the input of BoxB. Unplug the other end of the cable from the output of Box A and plugit into the dummy. Do not plug the dummy into Box A or let it touchanything conductive.
Theory: By effectively shorting the far end of the cable,this test uses the cable itself as a sensor of magnetic or electricfields. The far end of the cable is left electrically floating toprevent any other currents from flowing in its shield.
Observe: Is system output quiet?
No: The noise is coupling into to the cable by induction.This is most often caused by a strong AC magnetic field near the cable.Sources of potent magnetic fields include high-current AC power wiring,power transformers, and TV or computer CRT displays. Electrostaticcoupling is also possible, though rare, in cables that have a groundedouter shield. Re-route signal cables to avoid such strong fields.
Yes: Go to next step.
Leaving the dummy in place on the cable, plug the dummy into theoutput of Box A.
Theory: This test prevents the output of Box A from drivingBox B while allowing noise currents to flow from Box A to Box B throughthe cable’s shield. Box B is effectively listening to the noise voltageat the far end of the cable caused by noise current flow in theshield.
Observe: Is system output quiet?
No: Noisy ground currents are being coupled by thecommon-impedance of the cable shield. Install an appropriate audio orvideo ground isolator at the input of Box B.
Yes: The noise is coming from (or through) the output ofBox A. Perform these same tests on the cable(s) which connect Box A toupstream devices.