PITFALLS IN audiophile systems
Feb 1, 2000 12:00 PM, Bill Whitlock
For most designers, technicians and users, system interfacing and groundingpractices are considered a black art. These columns are intended toincrease your understanding of these interfaces and offer guidelines forpreventing and solving the inevitable real-world problems. Experience tellsus that equipment interconnections are an all-important factor inassembling trouble-free A-V systems of all kinds, but especially importantin high-performance audiophile sound systems.
What is high performance?
The ultimate goal of sound reproduction is realism, the kind that invokessuspension of disbelief in the listener. To do this, a reproduction systemmust do two things. First, it must be capable of reproducing naturally loudconcert levels. Just how loud, of course, depends on what kind of music orprogram it is. Classical music may have peak levels ranging from 103 dB SPL(sound pressure level) for solo piano up to 122 dB SPL for percussivematerial. For amplified country, jazz, or rock, peak levels can easilyreach 125 dB SPL to 130 dB SPL.
Second, equipment noise must be inaudible when listening at these naturallevels. Nothing breaks the spell like a background buzz or hum during aquiet musical passage or a suspense-filled moment in a movie. Because areasonably quiet home listening room may have a background noise level of20 dB SPL to 35 dB SPL, until recently it was widely accepted that anyequipment noise below these levels would be masked and, therefore,inaudible. The fact is that, on average, a listener can detect equipmentnoise some 15 dB to 30 dB below ambient room noise. Our ears and brain,using learned spectral signatures and directional cues, are smart and caneasily identify hum, buzz or hiss even in the presence of ordinaryhousehold noises.
The dynamic range of a sound reproduction system is defined as thedifference between its loudest undistorted output (hopefully concert level)and its residual noise floor (expressed in dB). Therefore, to satisfy ourrequirements for realism, we need a dynamic range somewhere between 100 dBand 125 dB. There is an excellent paper on this subject [ref 1]. Somemodern CD recordings use special digital techniques to extend their dynamicrange to around 115 dB, and it is a safe bet that most program sources,especially digital ones, will develop even wider dynamic ranges in the nearfuture. The point is that even for an ordinary CD having 90 dB to 95 dB ofdynamic range, realism can be totally lost if it is played on a systemwhose dynamic range is only 80 dB. The quietest passages would be buried innoise, painfully reminding the listener that he is listening to a mererecording. Consumer expectations are high, and there are more golden ears(real and self-appointed) out there than you might imagine.
Signals accumulate noise as they flow through the equipment chain and oncecontaminated, it is essentially impossible to remove the noise withoutdegrading the original signal. Because the dynamic range of an entiresystem can be no better than its weakest link, noise and interference mustbe prevented all along the signal path. In the vast majority of audiophilesystems, the worst loss of dynamic range is not caused by the equipment'sinternal signal processing but by the coupling of power line noise at thesignal interfaces.
Most real-world systems consist of at least two pieces of equipment thatare also connected to and powered by the AC line. These power-lineconnections unavoidably cause a significant voltage to exist between thechassis or local ground of any two pieces of equipment, whether safetygrounded or not. Because line voltage normally consists of a broad spectrumof undesirable harmonics and noise along with the pure 60 Hz sine wave, theinter-chassis voltage will be noisy, too. We must accept this as a fact oflife. This inter-chassis voltage is what causes hum, buzz, pops, and clicksin most systems.
The price alone of high-end audiophile equipment implies state-of-the-artdesign, and makers often tout impressive measurements of performance.Because these measurements are made in a laboratory setting, however, theyreveal little, if anything, about how the equipment will perform in areal-world system. Unfortunately, most of this expensive audiophile gearstill uses an audio interface system introduced more than 50 years ago. Theubiquitous RCA cable and connector form an unbalanced interface, which isextremely susceptible to noise coupling from the power line, making itnearly impossible to assemble a noise-free system.
As shown in Figure 1, when two pieces of equipment are connected via an RCAcable (or any unbalanced interface), the noisy voltage between the twochassis grounds causes current flow in the shield conductor of the cable.This causes a small but significant noise voltage to appear across thelength of the cable (Ohm's Law). Because the interface is unbalanced, thisnoise voltage will be directly added to the signal at the receive end ofthe cable [ref 2]. Note that the noise is not "picked up from the air" asis so widely believed.
Effective, safe solutions
Because interfaces are the danger zone for signals, systems with fewerinterfaces will obviously have fewer noise problems. Of course, selectingequipment with balanced inputs and outputs (the only kind worthy of beingcalled "professional" in my opinion) will preclude most noise problems, butmost audiophile and home theater systems will consist mostly, if notentirely, of equipment with unbalanced inputs and outputs. In all but thesmallest systems with very short cables, noise problems are highly likely.
As shown in Figure 2, when a transformer ground isolator is inserted in thesignal path, the connection between devices through the cable's shield isbroken. Because power-line noise current cannot flow acrossthe transformerwindings or through the cable's shield, no noise is coupled to the signal.A ground isolator is not a magic filter that can remove hum and buzz nomatter where it is placed to eliminate noise; it must be installed at theinterface responsible for the noise coupling. This can be determined easilywith a simple test adapter and procedure [ref 3].
High-quality ground isolators, such as Jensen's ISO-MAX series, not onlybreak the ground loop while transparently coupling the signal, but theyalso provide inherent suppression of ultrasonic and RF interference.Isolators using poorly designed or undersized transformers, however, cancause loss of deep bass, bass distortion and poor transient response.Beware of cheap products with scanty or non-existent specs; they often use$2 transformers.
If you ask equipment manufacturers about a hum and buzz problem, most willdeny that their product has anything to do with it and blame bad grounding,whatever that is. Some are so ignorant or reckless that they actuallyrecommend the use of ground-lifting AC adapters to break the ground loop.
If a short develops in a "ground lifted" piece of equipment, the audio orvideo cables that interconnect equipment will carry lethal voltagesthroughout the system and/or start a fire. Never defeat the function of thethird prong on any equipment's AC plug. It is both illegal and verydangerous.
Beware of marketing hype. There is nothing unexplainable about audibledifferences among cables. For example, the physical design of a cableaffects its coupling of ultrasonic power-line interference. Even low levelsof this interference will cause audible spectral contamination indownstream amps. The real solution is to prevent the coupling in the firstplace with a ground isolator, not agonize over which expensive designercable makes the most pleasingly subtle improvement. Expensive and exoticcables, whether double or triple shielded, made of 100% pure unobtainium orhand-braided by Peruvian virgins, will have no significant effect on humand buzz problems.