About the Author

David Davenport

Dave = Davenport=92s interest in=20 audio was sparked as a teenager, building hi-fi kits and learning = electronics.=20 Upon retiring from a 30-year career as an engineer with IBM, Dave is now = active=20 writing articles and providing audio kits through his business, Raleigh = Audio -=20 www.raleighaudio.com

=20

Audio Component Grounding and Interconnection

David Davenport

,=20

21st April 2010

This article focuses on audio system design for the DIY enthusiast=20 implementing a stereo system for home use. There is a lot of information = available for designing audio circuits, but precious little on = interconnecting=20 those circuits into a total audio system to achieve maximum performance. = Interconnecting circuits to create an audio system involves connecting = signals=20 and grounds , both within a single component as well as between = components. This=20 article has the most benefit for the person constructing his or her own=20 equipment, rather than merely interconnecting finished components. = Although much=20 of the information presented is applicable for professional = installations, those=20 installations face a myriad of problems due to physical size, number of=20 components and different types of components, which are not addressed = here.=20

Although I use electron tube circuits for many examples in this = article,=20 the concepts presented are valid for solid state circuits as well. Most, = if not=20 all digital circuits will use integrated circuits and I cover = interconnecting=20 digital electronics with analog electronics.

Chapter 1 - Ground

The term =93ground=94 is problematic = because it can=20 mean so many different things. Often a designer will lump all of the = meanings=20 together and in an indiscriminating manner, connect everything that = needs to be=20 =93grounded=94 together. Lumping all of the meanings together and = treating them all=20 the same causes problems that wouldn=92t happen if the different uses of = ground=20 were kept separate and each treated in a manner appropriate for its use. = Therefore a good place to start would be to tease all of the meanings = apart so=20 we can address them separately.

The first ground we encountered = was=20 playing in the dirt as a child: we call the place where the grass and = trees grow=20 ground. However, considering the electronics in an airplane or a cell = phone,=20 this ground is not necessary for electronic circuits to function =96 its = only=20 interest in electronics is as a sink for lightning strikes. I=92ll refer = to this=20 ground as =93earth.=94

Recognizing that high voltage is lethal, = the various=20 standards organizations around the globe have instituted safety = standards for=20 mains-attached electrical equipment. The first assumption in these = standards is=20 that a human body in the vicinity is at earth potential. Therefore the = standards=20 require that a metal chassis and any exposed metal parts be at earth = potential=20 by specifying a wire be connected between the chassis and a connection = to earth=20 at the circuit breaker panel. This ensures that any electrical fault = within the=20 equipment will not be a hazard to anyone that comes in contact with the = chassis.=20 A second benefit of this earth connection is that it provides a = low-impedance=20 path to earth for a lightning strike. This minimizes potential circuit = component=20 damage as well as providing protection from electrocution. A third facet = of the=20 standards is the means by which a fuse will be blown or a circuit = breaker=20 tripped in the event of a fault where the mains voltage is connected to = the=20 chassis. As shown in figure 1-1, in the event of a fault, a large = current will=20 flow from the circuit breaker panel, through the electrical wiring to = the=20 equipment, through the fault to the chassis, and then through the safety = wire=20 back to the circuit breaker panel. If there is an appropriately-sized = fuse in=20 the equipment, it will blow; otherwise the circuit breaker will trip. It = is a=20 good idea to provide an appropriately-sized fuse in the equipment = because it=20 will blow a lot sooner than the circuit breaker, thereby quickly = eliminating the=20 large current which may damage expensive components. Note that the = connection to=20 earth at the circuit breaker panel does not come into play for a fuse to = blow or=20 breaker to trip. Also, no current flows in the safety wire when there is = no=20 fault present. I=92ll refer to this ground as =93safety ground.=94 =

Figure 1-1, = Safety=20 ground: showing the current from a fault between the power line and the = chassis.=20

Adhering to the safety standards by providing a safety = ground=20 provides some protection for equipment manufacturers from expensive = liability=20 law suits. While you may not be worried about law suits =96 you are = liable for any=20 damage, injury or death caused by the equipment that you build =96 you = should be=20 very concerned about your personal health and well being as well as that = of your=20 loved ones. Equipment lacking a safety ground has caused fires, injury = and=20 death. Do not under any circumstance, for any reason, disconnect or = disable=20 an existing safety ground, or fail to include a safety ground in any = equipment=20 that you build. But, you say, =93what about my CD player, it has a = two-wire=20 power cord without a safety ground?=94 Consumer component manufactures = have=20 engineers with the skills to design a double-insulated product which is=20 inherently safe without a safety ground. This is not a trivial task and = should=20 not be attempted by amateurs that do not have the necessary skills. = Double=20 insulated products are designed such that any single fault cannot cause = power=20 line voltage to be present on any exposed part, including the chassis. = Special=20 provisions must be made to preclude a winding-to-winding fault within = the power=20 transformer.

The good news is that, in a properly designed audio = system,=20 a safety ground will not hurt and may actually help produce optimal = sound=20 quality. We=92ll see how this works a little later.

Most audio = components=20 will have one or more power supplies. I think of the purpose of a power = supply=20 is to provide the operating environment for the audio circuitry. That is = to say,=20 it provides the necessary voltages and currents to establish an idle = state=20 separate from any function that the circuit is intended to produce. For = example,=20 in a triode vacuum tube amplifier, the power supply supplies plate = current,=20 plate voltage and a bias voltage. It will also supply screen grid = voltage for a=20 pentode amplifier. Current will flow from the positive terminal of the = supply,=20 through the load and back into the negative terminal of the supply. For=20 convenience in making voltage measurements, understanding and using the = power=20 supply, a designer will usually designate some point in the power supply = as=20 =93ground.=94 This is usually the negative terminal of a positive = supply, or the=20 positive terminal of a negative supply. I will refer to this ground as = =93power=20 common.=94

Figure 1-2, = Power Common:=20 In this example the negative terminal of the power supply is chosen to = be power=20 common.

What would happen if some node with voltage from the = power=20 supply were to become shorted to the chassis? The chassis is at earth = potential=20 (through the safety ground) so the shorted node would be at earth = potential =96 so=20 far, so good. We would like to detect the fault and cause the fuse to = blow,=20 which requires a complete path for the fault current to flow from the = power=20 supply, through the fault to the chassis and back to the power supply. = Although=20 it doesn=92t have to be, to avoid confusion most designers choose power = common as=20 the point to connect to the chassis for the return path. Note that under = normal,=20 non-fault, conditions, no current will flow in the connection between = power=20 common and the chassis. The only purpose for this connection is to = direct fault=20 current.

Figure 1-3, = Power Common=20 and a fault

Figure1- 3 shows that a fault causes a large = current to=20 flow in the secondary of the power transformer, which in turn induces a = large=20 current in the primary of the transformer. Note that no current flows in = the=20 safety ground for this fault.

Power common should also be used as = the=20 =93ground=94 reference for any vacuum tube filaments. This may be by a = direct=20 connection or by a bias voltage.

It goes without saying, that in = order=20 for an audio component to be worthwhile, it must process an audio signal = in some=20 manner. For a signal voltage to be meaningful, it must be referenced to=20 something. In some cases, as in a differential amplifier, the signal is=20 presented to the positive input and the reference is the negative input. = In=20 other cases, as in a single-ended amplifier, the signal is presented to = the=20 input and the reference is =93ground.=94 I will call this ground = =93signal reference.=94=20 A triode is a three-terminal device: the input signal is presented to = the grid=20 with the cathode as the reference, and the output is taken from the = plate with=20 the cathode as the reference. The cathode is usually connected to power = common,=20 either directly or through a bias resistor. Thus, signal reference is = connected=20 to power common; however no current flows in the connection between = power common=20 and the signal reference. Each circuit has a signal reference and some = circuits=20 may be analog while other circuits are digital. It is a good idea to = keep the=20 digital references separated from the analog references.

The = last use of=20 the term =93ground=94 that I will address is shields. There are = basically two=20 different shields in an audio system =96 chassis and cable shields. = There are two=20 important characteristics of a shield: the continuity of the enclosure = and the=20 material it is made of. By continuity, I mean that for a shield to be = maximally=20 effective there must be no holes or gaps in the chassis or in the = connection=20 between the cable shield and chassis. This may not be practical in an = audio=20 system, but it is good to keep in mind as a goal. Of course, plastic or = wood=20 will offer no shielding, while copper or aluminum will provide = electrostatic=20 shielding and steel or mu metal will provide both electrostatic and = magnetic=20 shielding.

Rule 1. In Morrison [1] (page 39) states, =93An = electrostatic=20 shield enclosure, to be effective, should be connected to the = zero-signal=20 reference potential of any circuitry contained within the shield.=94 =

It is=20 Important to not have a chassis isolated =96 it must have some ground. = Everything=20 has a potential to everything else. Either control it or you have noise=20 potential. Thus a chassis or cable shield must be connected to signal = reference;=20 however, no current flows in the connection between the shield and the = signal=20 reference.

I use the terms =93ground buss=94 and =93star = ground=94 in this=20 article. A ground buss is a piece of wire at ground potential. A star = ground is=20 two or more grounds connected to one physical point. A star of stars is = just=20 that =96 several star grounds in turn all connected to a single point. =

So=20 in an audio system we have earth and possibly multiple safety grounds, = power=20 commons, signal references and shields; all interconnected. I have said = that=20 under normal, non-fault, conditions that no current will flow in these=20 interconnections. This is not entirely true for there is noise in the = system=20 which may flow between the different =93grounds=94 and cause degradation = of the=20 sonic performance of the audio system. Therefore the goal is to minimize = the=20 noise traffic in a system by first reducing the noise and second, by = reducing=20 the propensity for the remaining noise to move around; or alternatively = to=20 control the paths so as to minimize the effect. Let=92s start by taking = a look at=20 noise.

Chapter 2 - Noise

All electrical noise, with the exception = of=20 lightning induced noise, is produced by man-made electrical or = electronic=20 devices, including all of those in your home. Of course we can=92t = exclude the=20 audio system itself from producing noise; each component affecting = itself as=20 well as all of the others in the system. The noise is either radiated = through=20 space or conducted through wiring into the audio system as well as = between the=20 components in the system.

We think of the mains power as a solid = 120V,=20 60 HZ (or 230V, 50 HZ) but in reality it isn=92t very clean. It harbors = noise from=20 DC, to harmonics of the power line frequency, to spurious junk into the=20 megahertz range. All of this noise can get into the audio circuitry and = produce=20 readily identifiable noise in the form of hums and buzzes, or = aberrations in the=20 sound, which while not something you can put your finger on, messes up = the=20 sound.

The noise is running on a two-way street with power line = noise=20 projected into the power supply and power supply noise projected into = the power=20 line. Additionally, noise from both sources is projected into the = chassis, and=20 thus the system of grounds. This is discussed in detail by Eric Juaneda = here and by Bill Whitlock in Jensen AN-004.

Power line noise comes in two = flavors,=20 differential noise and common mode noise. With differential noise, the = noise=20 voltage is impressed between the two power lines =96 hot and neutral. = With common=20 mode noise, there is no noise voltage difference between the two power = lines;=20 rather, the noise voltage is impressed between safety ground and the two = power=20 lines.

So, what can we do to ameliorate the noise from the power = line?

Clean up the power line. Any electrical device in your home = that=20 contains a switching power supply is dumping tons of electrical = garbage into=20 the power and safety ground wires. Of particular interest are TVs and = DVD=20 players that are interconnected with your audio system or plugged into = the=20 same AC power branch circuit. In an AudioAsylum post here, Charles Hansen identifies the problem and = recommends=20 disconnecting and un-plugging all such components. Merely turning them = off=20 doesn=92t cut it =96 even though you think that they are off, their = power supplies=20 are still on and are injecting noise into the power system. The = problem is not=20 limited to TVs and DVD players; rather, devices such as computers and=20 microwaves anywhere in the home, and particularly in the same branch = circuit=20 as the audio system, are also culprits.=20
Clean up the safety ground. Safety ground is really the = flip side=20 of the power line because the power is referenced to the ground. By = cleaning=20 up one, you can=92t help but clean up the other. Even though safety = ground is=20 connected to earth, it is a long wire that is routed around the house, = picking=20 up all sorts of electrical noise because it is an RFI antenna. Some folks have suggested adding a second earth ground = directly=20 from the audio system, while others advise that an additional ground may = actually cause=20 more problems than it solves. A single, dedicated safety ground along = with a=20 dedicated power branch may be a good idea; however this is expensive = because=20 it must be installed by a licensed electrician. Much the same result = can be=20 achieved by dedicating one of your existing branches to your audio = system.=20 This branch should ideally have no other equipment connected to it, or = at=20 most, only what I call =93benevolent loads=94 =96 for example, = ordinary incandescent=20 light bulbs without a dimmer control.A thorough explanation of a = =93Technical=20 Ground=94 is contained in Jim Brown=92s grounding article here.

Class 2 devices side-step the issue = of a noisy=20 safety ground because they, of course, do not have a safety ground. = However,=20 it is important to note that this solves the problem only as long as = all=20 powered devices in the audio system are Class 2 devices. That is, = there can be=20 no connection to earth anywhere in the audio system, or in any = attached=20 component, for example, a computer that is not galvanically isolated. = The=20 problem of interconnecting Class 2 devices (without a safety ground) = with=20 Class 1 devices (with a safety ground) is shown in figure 5 of the = Juaneda article and discussed in detail in the Jensen = AN-004.=20
Eliminate DC on the power line. We normally think of noise = as some=20 sort of alternating current phenomena; however direct current on the = primary=20 of a power transformer is also noise. A thorough discussion of the = problem and=20 a solution is presented in Rod Elliott=92s article.=20
Choose the proper power line polarity on each component. = Noise from=20 the power line in the form of AC leakage current will be transferred = to the=20 core and frame of the power transformer through the parasitic = capacitors as=20 shown in the Juaneda article. Because of the way a power transformer is = constructed, the parasitic capacitors are different values; and = because the=20 neutral power line is connected to the safety ground, the amount of = noise on=20 the two sides of the power line is different. These two factors work = together=20 to cause the amount of leakage current to be different depending on = the=20 polarity of the connection of the power to the transformer. The result = is a=20 voltage, with respect to earth, impressed on the chassis. It is most = unlikely=20 that this impressed voltage will be the same on two different chassis = in the=20 audio system and when those chassis are connected together, noise = voltage will=20 flow in the interconnection. This is discussed in Charles Hansen=92s = AudioAsylum=20 post here and a way to test the polarity is shown here. Remember, the final polarity change should = be done=20 where the power line connects to the power transformer so as to keep = the fuse=20 and power switch on the hot side of the line. Note that this problem = exists in=20 class 2 devices (without a safety ground) as well as class 1 devices = (with a=20 safety ground.) This is discussed in detail in Jensen AN-004.=20
Include a line filter on equipment that you build. The best = kind to=20 use is one that has the filter integrated into the power entry module, = like=20 that shown here. Unfortunately, a power line filter is a = two-edged=20 sword; although it does filter the noise, it dumps common mode noise = into the=20 safety ground (which we want to clean up.)=20
Use a power conditioner. By its very nature a power = conditioner=20 isolates its attached components from the power line. The conditioner = can be=20 used two ways: First, for isolating sensitive audio components from = the noisy=20 power line, and second, for isolating noisy devices that contain = switching=20 power supplies from the power line. Of course, the audio components = and noisy=20 devices should not both be connected to the same power conditioner. = Although=20 the power conditioner does a good job of cleaning up the power line, = the=20 safety ground is passed directly through the power conditioner. = Therefore, the=20 suggestions for cleaning up the safety ground discussed above may = still prove=20 beneficial.=20
Use a shielded power transformer. A shielded transformer = has a=20 shield that eliminates the parasitic capacitors between the primary = and=20 secondary windings, thus eliminating the AC leakage currents between = those=20 windings. It isn=92t often that a shielded power transformer is needed = but it is=20 good to know that it is available for that rare case.

As = well as=20 minimizing the noise coming from the power line, it is important to = minimize the=20 noise that a component injects back into the power line. We will see why = this is=20 so later when we talk about interconnecting components.

Power = Supply=20 noise can be minimized by careful design of the power supply, paying = attention=20 to the choice of rectifiers, snubbers and filter networks. What can we = do to=20 minimize the noise injected back into the power line?

Don=92t include switching power supplies in any equipment that = you=20 build.
Provide a more constant load. Because audio components are = not=20 resistive loads, the manner in which they draw current from the power = line is=20 not constant. This irregular load presented to the power line = contributes to=20 the noise on the power line. Some things you can do here are regulated = supplies, class-A circuits, choke-input power supplies and Minimal=20 Reactance Power Supplies.=20
Use low-noise rectifiers. Different kinds of rectifiers = used in a=20 power supply convey different characteristics to the sound of an audio = circuit. Some sound better for one circuit and worse in another = circuit.=20 Whichever one you choose to use for a particular circuit, low noise is = an=20 important criterion. Tube rectifiers, Schottky diodes and Fast = Recovery=20 Epitaxial Diodes (FRED) are good choices.=20
Use a snubber on the rectifiers if needed.
Include a line filter on equipment that you build. Since = the noise=20 is on a two-way street, the filter will help both ways.=20
Shielded power transformer. Again, rarely needed but good = to have=20 available.

We have talked about electrical noise and = what can be=20 done to minimize the noise. All the electrical noise in the world = doesn=92t matter=20 a bit unless it affects the signal. On the other hand, just a tiny bit = of=20 electrical noise can wreak havoc if it shows up in the wrong place. In = order for=20 the noise to affect the signal it must be routed in such a way as to = interact=20 with the signal. Careless mixing and interaction of grounds is the = biggest cause=20 of audible degradation of the music in an audio system, so let=92s take = a look at=20 how this works.

To understand what can go wrong we need to = understand=20 something called Common Impedance Coupling. Simply stated, Common = Impedance=20 Coupling is the way that noise gets mixed in with a signal. Noise can = get mixed=20 with the signal in a couple of ways: first, =93conductive coupling,=94 = when two=20 circuits share the same path, and second, =93radiated coupling,=94 when = noise from=20 one circuit is radiated into another circuit carrying the signal. = [12]

Figure=20 2-1, Two forms of Common Impedance Coupling.

The circuit on = the left=20 of figure 2-1 shows two current loops, one loop with the signal and a = second=20 noisy loop. Impedance Z1 is a sensitive spot in the circuit, perhaps a = signal=20 reference buss, which is common to both loops. The signal current, I1, = and the=20 noise current, I2, both pass through Z1 creating a voltage that is the = product=20 of the impedance and the sum of the two currents. The circuit on the = right of=20 figure 2-1 also shows two current loops, one with the signal and the = other with=20 noise. In this case the sensitive impedance, Z2, is not common to both = circuits.=20 However the two loops are situated such that there is capacitive or = inductive=20 coupling between them and the noise voltage is coupled into the signal = loop. The=20 resultant current is a composite of the signal current and the noise = current=20 which creates a voltage across Z2 that contains a mix of the signal and = noise.=20 There are two loops in each of these examples, but in reality the Common = Impedance Coupling could be the result of several loops or a combination = of=20 conductive coupling and radiated coupling.

An example of a = problem=20 caused by radiated coupling is with the orientation of transformers in a = vacuum=20 tube amplifier. Depending on the orientation of the transformers with = respect to=20 each other, the output transformer may pick up the radiated field from = the power=20 transformer, causing an audible hum. Simply rotating one of the = transformers may=20 be sufficient to remedy the problem. Otherwise, increasing the distance = between=20 them will eliminate the problem.

Careless or poorly chosen = routing of=20 wiring can also cause problems with radiated coupling. It is important = to keep=20 filament wiring tightly twisted and laid close to the chassis. Low-level = signal=20 wiring should be kept as far as possible from power and filament wiring. = It is a=20 good idea to use a twisted pair for signal wiring, the signal wire = twisted with=20 its signal reference. When signal wiring must cross power or filament = wiring,=20 they should do so at right angles.

Chapter 3 - Interconnecting Components

There are two aspects = to an=20 audio connection: the signal and the ground. While both are equally = important,=20 designers have focused on the signal and left the grounding to chance. = Thus=20 poorly chosen grounding is the largest cause of audible hum and buzzes = in an=20 audio system. We will cover both aspects of an audio = connection.

Audio=20 components have either balanced or unbalanced interfaces. Those = interfaces are=20 similar in that they both have two wires =96 a signal and a reference. = The=20 difference is: on a balanced interface the impedance to ground is the = same for=20 the signal and reference; while on an unbalanced interface the impedance = to=20 ground for the signal is different than the impedance to ground for the=20 reference. Usually the reference of an unbalanced interface is connected = to=20 ground and the interface is called =93single-ended.=94 This term is = unfortunate=20 because it can obscure the fact that it is still a two-wire interface = and lead=20 designers to indiscriminately choose any convenient ground as a = reference. Note=20 that the kind of connecter used for the interface has nothing to do with = if it=20 is balanced or unbalanced; however most balanced interfaces use an XLR = connector=20 and most single-ended interfaces use an RCA connector. Why would you = want to use=20 one over the other? The ubiquitous single-ended interface is cheaper to = build;=20 while the balanced interface gives better noise performance. That being = said, if=20 you are using commercial components, it really comes down to what those=20 components hand you.

Let=92s start by looking at the ground = aspect of=20 interconnections.

3.1 - Balanced Interconnections: Grounding

The good news is, = compared to a single-ended connection, it is relatively easy to get a = balanced=20 interconnection right; the bad news is, it is also easy to get it wrong. = In fact=20 many, if not most, of the balanced interfaces available on both = commercial and=20 professional audio equipment got it wrong. [12]

To see what = could go=20 wrong let=92s take a side trip into power supplies. It goes without = saying that=20 the largest contributor of noise within an audio component is its power = supply.=20 I won=92t go into power supply design here but I will cover one = important area =96=20 ripple noise on an internal power supply ground buss. Figure 3.1-1 shows = a=20 schematic diagram of a simple power supply with an internal ground buss. = Ripple=20 current flows through this buss from the filter capacitors to the center = tap of=20 the power transformer. There is more ripple current in the capacitor = closest to=20 the transformer, with progressively less as the filter progresses. Even = though=20 the buss may be a short wire or printed circuit trace, it has finite = resistance=20 so a noise voltage is developed by the ripple current flowing through = the buss.=20 If signal reference current is routed through the power supply buss, = this noise=20 voltage will be impressed onto the signal by Common Impedance Coupling. =

Figure=20 3.1-1, A typical power supply.

Now let=92s see what happens = when we=20 combine this power supply with a poorly chosen grounding system, which = is=20 employed in many components. Figure 3.1-2 shows an example of a piece of = equipment that has a severe ground loop problem. Knowing that the center = tap of=20 the power transformer is the noisiest point in the device, the circuit = designer=20 connects this point to the chassis at the same point that the safety = ground=20 connects to the chassis. He thinks that somehow this will drain all of = the noise=20 to earth. Unfortunately, the earth is not a huge electron vacuum cleaner = =96 in=20 order for current to flow anywhere, there must be a return path and a = voltage=20 difference between the input and return of the loop. Of course the power = supply=20 must be connected to the audio circuit so the designer dutifully = connects the=20 quietest point on the power supply ground buss to the audio circuit = ground. Come=20 time to connect the input XLR connector to the circuit, the signal on = pin 2 and=20 the reference on pin 3 is connected to the + and =96 inputs of the = differential=20 amplifier. What to do with pin 1 on the connector? Well, pin 1 is the = shield and=20 the shield should be grounded, and the shield is on the same cable with = the=20 signal, so why not connect pin 1 to ground of the circuit where pins 2 = and 3 are=20 connected? Okay, done. The designer should have been thinking outside of = the box=20 (his own box) when he designed this grounding arraignment. If he had, he = would=20 have seen that he had created a ground loop with the attached component. = The=20 attached component connects the safety ground to the cable shield, = either=20 directly by the chassis (as it should be), or worse, indirectly through = a lame=20 grounding arrangement like the one in the first device. The loop = contains the=20 power supply ground buss and the signal reference. The noise current on = the=20 ground buss together with the finite resistance of the buss, provides a = voltage=20 across the buss which will drive noise current through the loop. The = noise=20 current through the loop together with the finite resistance of the = signal=20 reference will develop a noise voltage in the signal reference. Now that = the=20 signal reference is dirty there is no hope in achieving a clean signal. =

Figure=20 3.1-2, The Pin 1 Problem.

Neil Muncy identified this = problem and=20 dubbed it the =93Pin 1 Problem=94 in his AESJ article [12] because pin 1 = of the XLR=20 connector was connected to the wrong place. The shield and pin 1 should = have=20 been connected to the chassis as shown in figure 3.1-3 rather than the = signal=20 reference.

Figure=20 3.1-3, The Pin 1 Problem Fix.

By connecting the shield to = the=20 chassis rather than the signal reference, there is no longer a loop for = noise=20 current to flow in - thus the signal reference is clean. The internal = ground=20 structure is still not right and I=92ll address that when I cover = unbalanced=20 interconnections.

It is important for the shield to be connected = to the=20 chassis at both ends for several reasons: improved shielding, improved = headroom=20 and maximum CMRR as is discussed here. Galvanic isolation provides the possibility of = eliminating a Pin 1 Problem, it does not guarantee it. Notice that the = cable=20 shield provides a second path (with the safety ground) between the = grounds in=20 the two chassis. If either component has poorly chosen grounding, a Pin = 1=20 Problem will surface. There is controversy in the pro-audio community = whether=20 the shield of a balanced connection should be connected at both ends or = only at=20 the driver end. There are pros and cons to each with the biggest factor = being=20 the complexity of interconnecting a large professional system. This, = together=20 with the large number of devices suffering from a Pin 1 Problem has led = many=20 installers to favor the driver-end-only solution. The technical aspects = of the=20 question are discussed in a side bar in Bill Whitlock=92s article on audio interfaces. A home stereo system is = a lot=20 simpler than a pro instillation and the cables are a lot shorter thereby = greatly=20 reducing the common mode noise problem. Assuming that you can eliminate = any Pin=20 1 Problems, the advantages of connecting the shield at both ends win out = in a=20 home stereo system. However, if you cannot eliminate a Pin 1 Problem in = an=20 attached component, the driver-end-only is an easy solution. Note that=20 off-the-shelf cables have the shield connected to pin 1 at both ends, so = you=20 will need to modify the cable to implement the driver-end-only = connection. I=20 prefer modifying the cable rather than the wiring of a component because = at some=20 later date you may want to attach a different component and don=92t want = to have=20 to remember to re-wire the interface. It is a lot easier to keep a pair = of=20 driver-end-only cables in your stash of cables =96 but remember to label = them as=20 such.

Before I go into single-ended interconnections I want to = clarify=20 one thing about safety ground in the figures. The figures show safety = ground=20 directly connecting the two chassis. Of course this is not accurate; = rather both=20 chassis are connected to safety ground at the power outlet. I am = assuming that=20 both devices are plugged into the same power outlet, thus their safety = ground=20 wires are directly connected in the outlet. I have dropped the power = outlet from=20 the pictures to simplify them. The absolute value of the voltage with = respect to=20 earth of the safety ground at the power outlet is not germane to the = discussion=20 because it is common mode to both devices; remember =96 earth is not a = vacuum=20 cleaner.

3.2 - Single-ended Interconnections: Grounding

As shown in = figure=20 3.2-1, a single-ended connection can suffer from a noisy ground loop. = Even=20 though there is no XLR connector to have a pin 1 in a single-ended = connection,=20 the problem is still called the Pin 1 Problem because it is the same = structure=20 as that in a balanced connection.

Figure=20 3.2-1, The Pin 1 Problem in a single-ended connection.

The = problem=20 is worse in the single-ended case than the balanced case because the = shield is=20 also the signal reference. The noise current through the shield develops = a=20 voltage across the finite resistance of the shield and the resultant = noise=20 voltage on the reference is impressed on the signal in the amplifier.=20 Consequently, the problem is harder to fix in the single-ended case than = for the=20 balanced case. However since connecting the shield to the chassis solved = the=20 problem for the balanced connection; let=92s take a look at that. =

Figure=20 3.2-2, Connecting the shield to the chassis.

Connecting the = shield=20 to the chassis may help a little because it shorts out the loop that = includes=20 the cable shield. However it does not solve the problem because it = causes=20 another noisy loop within the device, that is: chassis -> signal = reference,=20 -> noisy power supply buss, -> chassis. Well how about if we were = to=20 connect the shield just to the chassis and not to the signal reference? = Nope,=20 there is still a problem; now the signal reference of the amplifier is = connected=20 to the signal reference of the cable through the noisy power supply. = Hmmm=85 okay,=20 is there something that we can do to reduce the noise current in the = loop? Sure,=20 we can add some resistance to the loop like shown in figure = 3.2-3.

Figure=20 3.2-3, Safety Loop Breaker Circuit.

The Safety Loop Breaker = Circuit, as explained in Rod Elliott=92s article, is a=20 clever circuit that inserts a high impedance in the loop for low = voltages, and a=20 low impedance in the loop for high (fault) voltages. It is effective; = however it=20 attempts to ameliorate the symptom rather than eliminate the problem. = The=20 problem is twofold =96 the loop and the noise generator; so let=92s = address them.=20 Figure 3.2-4 shows a better power supply solution.

Figure=20 3.2-4, A clean power supply.

In figure 3.2-4, the internal = ground=20 buss is collapsed into a point, forming a local star ground. Bringing = everything=20 to a point forces us to make a connection to that point =96 no more = multiple-point=20 connection over which a noise voltage could form. Before, we had two=20 connections: a high noise one connected to safety ground and a low noise = one=20 connected to signal reference. Note that now the high noise point is = directly=20 connected to the power common where it can be connected to safety ground = to=20 drain the AC leakage current, and the low noise point is directly = connected to=20 the power common where it can be connected to signal reference. =

A=20 power supply is a two-terminal output device =96 a voltage and power=20 common. Do not make any external connections to internal points = in the=20 power supply. Of course, if the power supply produces both a positive = and=20 negative voltage output with a shared power common then it is a = three-terminal=20 output device.

Figure=20 3.2-5, A clean power supply attached to the loop.

We have = solved=20 one source of Common Impedance Coupling but there is still another = lurking. The=20 next noise generator is the AC leakage current from the power = transformer=20 thorough the power supply to power common. In this case the Safety Loop = Breaker=20 Circuit may be detrimental. The Safety Loop Breaker Circuit will inhibit = this=20 current from reaching safety ground, and the current will take the path = of least=20 resistance through the signal reference and shield to find safety ground = through=20 the attaching device. Another problem is that the signal reference is = not=20 directly attached to the chassis so the chassis is not as an effective = shield as=20 it could be. I=92ll come back to the appropriate use of a Safety Loop = Breaker=20 Circuit later but for now let=92s get rid of it and attach the signal = reference=20 directly to the chassis.

Up until now we have looked at the Pin = 1=20 Problem only on an input connection. Let=92s now look at the problem = from an=20 output connection perspective.

Figure=20 3.2-6, Pin 1 Problem on an output.

A Pin 1 Problem on an = output=20 really isn=92t any different than a Pin 1 Problem on an input. As we are = getting=20 close to a final solution on the input side, let=92s apply what we have = found to=20 the output side as well.

Figure=20 3.2-7, A clean power supply in both components.

Okay, the = loop is=20 shrinking and there is no longer any reason for the AC leakage current = to choose=20 the shield rather than the Safety ground. We have ameliorated the = conductive=20 coupling problem but we still have a radiated coupling problem so = let=92s take a=20 look at that next. We still have the shield current going to the chassis = through=20 the signal reference. Figure 3.2-8 shows all of the different grounds = connected=20 together in a star of stars configuration.

Figure=20 3.2-8, Star of stars.

It is now clear that the shield and = safety=20 ground no longer form a problematic ground loop and are now merely = parallel=20 paths.

Loops aren=92t bad =96 it depends on what is on the=20 loop. Unless there is a voltage generator to drive a current = around the=20 loop, or radiated current into the loop, it is merely a parallel path. = Consider=20 the parallel shields of a left and right channel stereo cable. However, = parallel=20 paths do form a loop antenna and can pick up RFI by radiated coupling. = Therefore=20 minimize the use of parallel paths to only where necessary and then = minimize the=20 area of the loop.

Speaking of RFI radiated coupling, it is = possible for=20 audio cabling and input circuitry to pick up RF noise. In extreme cases, = shielding won=92t resolve the problem and more aggressive techniques = like RF=20 filters on the audio inputs must be employed. A = Ham=92s Guide to=20 RFI, Ferrites, Baluns, and Audio Interfacing by Jim Brown is a = particularly=20 good coverage of filtering RFI.

There is still one thing left to = address:=20 Since it is certain that AC leakage current will flow through safety = ground and=20 the shield from both components, the two chassis and thus the two signal = references will be at different AC potentials. This result s in signal = noise:=20 much less than we had with the Pin 1 Problem, but still some noise. We = can=20 reduce the voltage difference between the two chassis by reducing the = impedance=20 between them. First we use a larger safety ground wire in the power cord = =96 the=20 larger, the better. And of course you could even go to a hefty = silver-wire power=20 cord. 3D""=20

=20 Second, use a shielded twisted pair for the interconnect cable, with one = of the=20 wires in the pair (as well as the shield) being the signal reference. If = all of=20 this is not enough, you could consider a Parallel Earth Conductor (PEC). = A PEC=20 is simply a heavy wire connecting the two chassis. Jim Brown calls this = "local=20 bonding" in this article.

Figure 3.2-8 shows the star = grounds on=20 the chassis where the safety ground comes in. The star does not have to = be there=20 and it may be more convenient to move the star onto a PC board. A couple = of=20 examples of this are shown in figure 3.2-9.

Figure=20 3.2-9, Some options for star grounding.

Let=92s move on to = the=20 signal aspect of Interconnections.

3.3 =96 Balanced Interconnections: Signal

A signal interface = is=20 comprises of a signal and an associated reference. By definition, on a = balanced=20 interface the impedance to ground is the same for the signal and = reference. Most=20 balanced interfaces these days are implemented with electronic circuits, = usually=20 op-amps or similar integrated circuits. Classical balanced interfaces = were=20 implemented with audio transformers, and a few audio components still = use=20 transformers on their interfaces. Notice that the two lines on the = interface are=20 called =93signal=94 and =93reference.=94 These are different from the = =93signal reference=94=20 ground within the component.

Figure=20 3.3-1, Balance interconnection with active circuits.

A = cursory=20 search will provide many available active chips for a balanced = interface. Bill=20 Whitlock goes into some of the circuits in his article here.

There is one other issue to be aware of = with a=20 balanced interface when both ends have electronic circuits rather than a = transformer on one or both ends. Even though a balanced input takes the = signal=20 between the two inputs without reference to a ground, it is important = that the=20 two inputs have a reasonably close potential to ground, otherwise the = CMRR will=20 suffer. Consider for a moment a tube differential amplifier with both = its inputs=20 to the grids at plus or minus 50 volts. In this extreme case the tubes = would be=20 either saturated or cut off. Thus, it is important to have an = established ground=20 reference between the sending and receiving components.

Audio Transformer

A transformer? Yes, a transformer. = Transformers=20 have received a bad rap for use in high-quality consumer audio = equipment. They=20 are said to be large and heavy, exhibit poor frequency response and = distortion,=20 and are expensive. As to the size and weight, we are talking line input = and=20 output transformers here, not tube amplifier output transformers. The = Lundahl LL1690 line=20 input transformer is PC board mounted and is lighter and smaller than = many=20 high-quality film and foil capacitors that I use for coupling. As for = frequency=20 response and distortion, take a look at the Jensen JT-10KB-D which is down -3dB at 0.5Hz (less than 1Hz = folks!)=20 and 180kHz, with less than 0.001% THD at 1kHz. These are just a couple = of=20 examples; both Lundahl and Jensen have several line transformers = available to=20 meet your specific needs. High-quality transformers are expensive; there = is no=20 getting around that. You get what you pay for and I suspect that most of = the bad=20 reputation has been generated by the use of cheap transformers. Bill = Whitlock=20 wrote the audio transformer chapter of the Handbook for Sound = Engineers=20 that will give you an opportunity to understand audio transformers. =

Figure=20 3.3-2, Balanced interconnection with transformers.

Okay, so = what do=20 you get with a transformer? Galvanic isolation and an excellent = common-mode=20 rejection ratio (CMRR) are the main things, but you can also get free = gain or=20 attenuation and along with that an opportunity for a lower output = impedance. The=20 primary advantage of a balanced transformer input versus a balanced = active=20 circuit input is a vastly improved CMRR. This is explained in section = 3.2 of=20 Bill Whitlock=92s Understanding, Finding, & Eliminating Ground Loops = in Audio=20 & Video Systems.

A good introduction to the advantages of = balanced circuits and transformer interfaces is presented in So=20 You Thought Your Amplifier Was Balanced? by Andy Grove and Peter=20 Qvortrup.

Figure=20 3.3-3

Figure 3.3-3, It doesn=92t make any difference if one = end of=20 the connection has a transformer while the other end has an electronic = circuit.=20 The interconnection is the same for all varieties of balanced=20 interfaces.

3.4 Single-ended Interconnections: Signal

The ubiquitous=20 single-ended circuit is available on most audio components.

Figure=20 3.4-1 A single-ended interconnection using electronic circuits.=20

Most, if not all, single-ended output circuits will = exhibit a DC=20 voltage offset, meaning that the quiescent interface will be at a = voltage level=20 different from the reference, or shield. This voltage offset will = produce a loud=20 audible =93thump=94 when the component is powered on. At best, the thump = is=20 startling and at worst it may damage a speaker. Also, some (not all) = volume=20 controls can be damaged by DC current over a long period. Therefore the = designer=20 of the output circuit will include a capacitor to block DC offset on the = output.=20 Because there are no standards addressing the interface, not trusting = that the=20 designer of the output circuit included a capacitor, the designer of the = input=20 circuit will also include a capacitor there. Being in series, the = effective=20 value of the combination of the two capacitors will be less than either = alone.=20 Therefore, unless the designers greatly oversized the capacitors, bass = response=20 may suffer. Also, the capacitors will interact with other impedances in = the=20 circuit, creating a low-pass filter; thereby affecting the = high-frequency=20 response. And of course, unless the capacitors are of the best quality, = they=20 will degrade the quality of the audio signal. There is a way to = eliminate the=20 capacitors.

Figure=20 3.4-2, A single-ended interface using transformers.

A = transformer=20 is inherently a balanced device; however it can be used in a = single-ended=20 circuit by just grounding one side of a winding.

Figure=20 3.4-3, Single-ended interconnection with transformer = input.

Figure=20 3.4-3 shows two components interconnected with a single-ended interface. = The=20 component on the right has a transformer input and notice that the = shield, or=20 reference, is connected to only the transformer and specifically is not=20 connected to ground in that component. The only connection between the = grounds=20 in the two components is through the safety ground. The signal and = shield that=20 are connected to the primary winding of the transformer are connected to = the=20 driver and ground in the left hand component. Thus, even though the = transformer=20 is physically located in the right hand component, its primary winding = and the=20 cable are part of the output circuit of the output circuit in the left = hand=20 component. The signal connection between the components resides in the = flux of=20 the transformer and not in any wires. This is what is meant by = =93galvanic=20 isolation.=94 Since there is only a single connection between the = grounds in the=20 two components (by the safety ground,) galvanic isolation precludes the=20 possibility of any ground loops between the components. This completely=20 eliminates the Pin 1 Problem. The capacitor on the output of the = component on=20 the left in the figure is still needed because, even though the = transformer will=20 block any DC offset, a DC offset may saturate a small input transformer. =

You may provide an RF connection between the cable shield and = the=20 chassis on the right hand component by connecting a 10nF ceramic = capacitor=20 between the chassis and the shell of the RCA jack. Keep the leads as = short as=20 possible.

This same interconnection scheme could be used if both = components had transformers on their interfaces; however, in that case = it would=20 make more sense to implement a balanced interface.

Figure=20 3.4-4, Single-ended interconnection with transformer output. =

The=20 transformer output shown in figure 3.4-4 is the mirror image of the = single-ended=20 transformer input shown in figure 3.4-3. However there is one important=20 difference between the two configurations: the output amplifier in = figure 3.4-3=20 has a low output impedance, while the input amplifier in figure 3.4-4 = has a high=20 input impedance. With a high input impedance, the attached cable and = transformer=20 primary winding is an antenna which picks up RFI noise. Grounding the = shield at=20 the source provides a low impedance and eliminates the problem. = Therefore in=20 this case, both ends of the interface must be grounded. Depending on the = bias=20 requirement of the input circuit, the input capacitor may not be needed. =

3.5 - Mixed Interconnections

Sometimes you will have two = components,=20 one with a balanced interface and the other with a single-ended = interface; and=20 you wonder if there is a way you could interconnect them. Yes there is = and we=92ll=20 go into that now.

Single-ended to Balanced Interconnection

The single-ended to = balanced interconnection is pretty straight-forward, requiring that you = only=20 make a special cable. I=92ll skip the cases where both components have=20 transformers on their interfaces because in these cases it would be best = to use=20 a balanced interconnection.

Figure=20 3.5-1, Single-ended to balanced interconnections.

In both = of these=20 cases the component on the left has a single-ended interface and the = component=20 on the right has a balanced interface. The only difference is that with = a=20 transformer in the right-hand component, a blocking capacitor is needed = in the=20 left-hand component. The capacitor is optional in the case of both = components=20 having electronic circuits on their interfaces. Even though the = component on the=20 right has a balanced input, the interconnection is unbalanced because it = is=20 referenced to ground.

Figure=20 3.5-2 Single-ended to balanced interconnection cable.

You = will need=20 to make a special cable from a shielded twisted pair with an XLR = connector on=20 one end and an RCA connector on the other end. Pin 2 of the XLR = connector is=20 connected to the center pin of the RCA connector, and both pins 1 and 3 = of the=20 XLR connector is connected to the shell of the RCA connector. This is = important;=20 if pins 1 and 3 are connected together at the XLR connector end instead = of at=20 the RCA end, the noise rejection will be poorer.

Balanced to Single-ended Interconnection

The balanced to=20 single-ended interconnection is more complex than the single-ended to = balanced=20 interconnection.

Figure=20 3.5-3, Balanced to single-ended interconnection with transformer=20 output.

With the interconnect scheme shown in figure 3.5-3 = we have=20 a balanced output on the left, a single-ended input on the right with = the two=20 interconnected. Even though the component on the left has a balanced = interface,=20 the interconnection is unbalanced because it is referenced to ground. =

In=20 this case, with a transformer output on the balanced interface, = you can=20 use an interconnect cable similar to that shown in figure 3.5-2. This = cable is=20 wired the same but the gender of the XLR connector is opposite of the = cable=20 shown in figure 3.5-2.

Most equipment that provide balanced = outputs do=20 so with differential amplifiers rather than transformers. The = differential=20 amplifier is most often two single-ended amplifiers, one with its signal = inverted. When a single-ended output is desired, the negative output is = not=20 used, thereby loosing one-half of the signal, resulting in a 6dB = difference in=20 the output level between single-ended versus balanced operation. A = transformer=20 output does not exhibit this 6dB difference in signal level because the = whole=20 output signal is used for both balanced and single-ended = operation.

Now=20 let=92s consider the case where both components have active circuits on = their=20 interfaces. This is where the complexity lies. A good explanation of the = interconnection for balanced and single-ended components is given in = Jensen=92s AN-003, and=20 I will present a summary here.

I have purposely shied away from = showing=20 specific circuits in order to simplify the illustrations and also to = suggest=20 that the examples are generic. However it is important to know the type = of=20 circuit used for a balanced output driver if you want to connect it to a = single-ended input. Some circuits want to have the unused output = grounded while=20 other circuits want the unused output left floating. Some circuits want = the=20 unused output grounded at the driver end while others want the unused = output=20 grounded at the far end. The wrong choice can degrade the sound or even = damage=20 the circuit. Rather than sort out all of the possibilities, I am going = to=20 side-step the issue and give you the best-quality solution. We know that = a=20 transformer input will accept any balanced output circuit and we know = that a=20 transformer can be used for a single-ended output, so let=92s put those = two=20 together and use a transformer for balanced to single-ended = conversion.

Figure 3.5-4, = Balanced=20 to single-ended interconnection.

An off-the-shelf version = of this=20 solution is the Jensen ISO-MAX=20 PC-2XR. This unit is specifically designed to interface between = professional=20 and consumer equipment. Professional equipment usually operates at a = 12dB higher=20 level than does consumer equipment, so the PC-2XR provides 12dB of = attenuation=20 to better match the devices.

We=92ll come back to this, and = similar=20 devices for some solutions to the Pin 1 Problem.

Composite Interfaces

We can see that there is not a whole = lot of=20 difference between the configurations of a single-ended transformer = input versus=20 a balanced transformer input; likewise for a single-ended transformer = output=20 versus a balanced transformer output. Therefore, for the cost of an = additional=20 connector and a switch you can have both.

Figure=20 3.5-5, Balanced connection with single-ended = connection.

The input=20 transformer is connected to both an RCA jack and an XLR connector in = parallel=20 and the signal is provided through one or the other connector. If the = input is=20 single-ended, the ground for the shield is provided by the attaching = component.=20 The output transformer is also connected in parallel to both an XLR = connector=20 and RCA jack, providing the signal through one or the other connector. = If a=20 single-ended output is desired, a cable is connected to the RCA jack and = a=20 ground provided for its shield by closing the RCA/XLR switch. The switch = is left=20 open if the XLR connector is used to provide a balanced connection. =

3.6 - Pin 1 Problem Remedy

Even though you employ all of the = proper=20 grounding practices in audio components that you build, you can still = have a hum=20 if you attach to a piece of commercial equipment that has a Pin 1 = Problem. You=20 have some choices in handling the problem.

First, you could = modify the=20 offending device so that it has a proper grounding scheme. Sometimes = this is not=20 practical because of the way the device is constructed so the next best = thing=20 would be to add a Safety Loop Breaker Circuit to the problem device. =

But=20 what if you have a piece of vintage classic equipment that you don=92t = want to=20 modify in any way so as not to impact its resale value? The solution is = simple=20 if the equipment has a balanced interface =96 merely open the shield at = the=20 receive end of the interconnect cable. As shown on page 27 of this article, it is important to open the shield at the = receiving=20 end rather than the sending end to maximize CMRR. You might want to = experiment=20 with placing a 10nF capacitor from the shield to the chassis. Keep the = leads as=20 short as possible.

If the offending equipment has a single-ended=20 interface you can turn to a transformer to solve the problem. As = discussed here,=20 Jensen has a complete line of isolators similar to the ISO-MAX PC-2XR = that can=20 interface any type of connection to any type of connection while = breaking the=20 problematic loop.

It is always better to solve the problem in the = offending device (by modifying it) or on its interface (with transformer = isolation) rather than to compensate for the problem in the equipment = that the=20 offending device is attached to. We=92ll see why this is when we cover=20 interconnecting equipment that use a Safety Loop Breaker circuit for = isolation=20 and also in the section on ground isolation.

3.7 - Effective Interconnection Schemes

In this section I = will=20 present four interconnection schemes that should cover just about every=20 situation. I=92ll be using some terms rather loosely so I=92ll define my = usage=20 here:

Class 1 device =96 This is an analog component that has its = chassis=20 connected to safety ground and does not have a Safety Loop Breaker = Circuit or=20 a Pin 1 Problem. It does have single-ended non-galvanically isolated=20 interfaces.=20
Class 2 device =96 This is a commercial analog component = that is=20 designed to be safe with no connection to safety ground. It has = single-ended=20 non-galvanically isolated interfaces. An example of a Class 2 device = is any=20 component that has a two-prong power plug, or is battery powered, like = a=20 laptop computer.=20
SLB device =96 This is an analog component that has its = chassis=20 connected to safety ground and has its system star ground isolated = from the=20 safety ground by a Safety Loop Breaker Circuit.=20
Pin 1 Problem device =96 This is an analog component that = has its=20 chassis connected to safety ground and does have a Pin 1 Problem. That = is, it=20 has a poorly chosen grounding system that injects noise into the = ground=20 system. Most, if not all, tower computer (not laptop) sound cards have = a Pin 1=20 Problem.=20
Computer =96 This is a catch-all for any kind of digital = audio device=20 that has digital power supplies dumping a lot of noise into its ground = system.=20 An example is an SPDIF interface from a computer or a satellite TV = system.=20

The first effective grounding scheme is a system with = all Class 1=20 components shown in figure 3.7-1. This is the grounding scheme that we = developed=20 in section 3.2. This interconnection scheme has a single-level ground = system=20 with all of the signal shields and safety grounds connected to the = system star=20 grounds.

This scheme is the only one of the four that allows = noise from=20 the power line in the form of AC leakage current to flow in the signal = ground=20 system. That noise will be transferred from the primary of the power = transformer=20 through the parasitic capacitors to both the core and frame of the power = transformer as well as power common as shown in the Juaneda article and also Jensen AN004.=20 However, since the whole ground system has a low impedance, the level of = the=20 noise will be low and should not be a problem. If it is a problem, the = impedance=20 can be reduced using Parallel Earth Conductors (PEC) or =93local = bonding=94 as=20 described by Jim Brown here.

Figure=20 3.7-1, Simple system of Class 1 components.

The second = effective=20 grounding scheme is a system with all Class 2 components as shown in = figure=20 3.7-2. This interconnection scheme has a single-level ground system with = all of=20 the signal shields connected to the system star grounds.

Figure=20 3.7-2, Symple system of Class 2 components.

This scheme = does not=20 have a problem of noise from the power line in the form of AC leakage = current=20 because the ground system is not connected to safety ground and thus = there is no=20 return path for the leakage current. Without a path, there can be no = current.=20 And of course, because there is only one connection between the = components =96 the=20 shields, there can be no ground loops.

The third effective = grounding=20 scheme is a system with all the components having Safety = Loop Breaker=20 Circuits as shown in figure 3.7-3. This interconnection scheme has a = two-level ground system with all of the signal shields connected to the = system=20 star grounds forming the first level. All of the safety grounds are = connected to=20 the chassis forming the second level. The two ground levels are isolated = from=20 each other by Safety Loop Breaker Circuits.

Figure=20 3.7-3, Simple system of SLB components.

This scheme does = not have a=20 problem of noise from the power line in the form of AC leakage current = because=20 the first-level ground system is isolated from safety ground and thus = there is=20 no return path for the leakage current. Without a path, there can be no = current.=20 You can think of the first-level ground system as the same as the ground = system=20 in the Class 2 ground scheme. However, AC leakage current that has been = injected=20 onto the chassis from the transformer does flow in the safety ground = circuit.=20

Of course, no grounding scheme is very useful if it is = restricted to=20 having only one type of component in the system. So let=92s see how we = can mix=20 different types of components in a ground system. We=92ll start simply = with just=20 two interconnected components and then expand the grounding to cover = several=20 components.

Figure=20 3.7-4, Interconnection of a Class 2 device with an SLB = device.

As=20 mentioned above, the first-level ground of an SLB device is the same as = the=20 ground in a Class 2 device so they interconnect without any problem. =

Figure=20 3.7-5, interconnection of a Class 1 device with either a Class 2 device = or an=20 SLB device.

There is no ground loop in either of these = cases;=20 however AC leakage current from the power transformers in the Class 2 = and SLB=20 devices is directed to earth through the shield and safety ground in the = Class 1=20 device. If this is a problem, the only recourse is to isolate the safety = ground=20 in the Class 1 device from the class 2 or SLB device. This may be = achieved by=20 either adding an SLB circuit to the Class 1 device (thereby making it an = SLB=20 device) or galvanically isolating the interconnection with a = transformer.=20

Figure=20 3.7-6, Interconnection of a Pin 1 Problem device with either a Class 2 = device or=20 an SLB device.

There is no ground loop in either of these = cases so=20 the Pin 1 Problem is remedied; however, just as with the Class 1 case = above, AC=20 leakage current will flow through the shield. This may present more of a = problem=20 with a Pin 1 Problem device than with a Class 1 device because the AC = leakage=20 current will flow through the signal reference before being directed to = the=20 safety ground. The solution is the same as that of the Class 1 device = above.=20

Figure=20 3.7-7, Interconnection of a Pin 1 Problem device with either a Class 1 = device or=20 another Pin 1 Problem device.

This is the classic case of a = Pin 1=20 Problem that was discussed in detail in chapter 3.2. The solution is to = either:=20 modify the Pin 1 Problem devices so that they no longer exhibit a Pin 1 = Problem,=20 or add an SLB circuit to the Pin 1 Problem device, or galvanically = isolate the=20 interface with a transformer. This is illustrated in figure = 3.7-8.

Figure=20 3.7-8, Solution for the Pin 1 Problem device = interconnection.

Since=20 the Pin 1 Problem injects noise into the ground system, the most = effective way=20 to include such a component is to galvanically isolate it from the rest = of the=20 ground system. The same solution effectively allows a computer sound = card to be=20 attached to an audio system. The galvanic isolation for these devices = can be=20 incorporated into the components or provided as separate, external = devices, such=20 as the Jensen ISO-MAX.

In the case of the computer, since = a computer=20 sound card is not in the realm of the highest quality audio devices, = there is no=20 need to spend the money on the highest quality line transformer to = provide=20 galvanic isolation. EDCOR has a line of very good line input and output=20 transformers that will serve well in this application.

Attaching = a=20 computer sound card to the SLB grounding scheme presents the same = problems as in=20 attaching a computer sound card to the Class 1 grounding scheme. The = best and=20 total solution is galvanic isolation. However, for a very simple system = as shown=20 in figure 3.7-9, directly attaching the computer sound card to the SLB = isolated=20 amplifier may provide an acceptable solution. The SLB will inhibit the = ground=20 loop. You will still have the problem of AC leakage current from the = amplifier=20 flowing through the computer but, since the computer sound card as a = poorly=20 chosen ground structure, it may not matter.

Figure=20 3.7-9 A simple SLB system.

However, if the computer is a = laptop=20 then there is no need for the galvanic isolation because the laptop does = not=20 have a connection to the mains power and therefore no ground = loop.

There=20 is one additional problem to take care of here =96 the dirty safety = ground on the=20 computer. We do not want to mix this dirty safety ground with the = relatively=20 clean safety ground of the audio system so provide a separate AC mains = branch=20 circuit for the computer.

Okay, let=92s summarize where we are = with the=20 interconnections of two devices:

There is no problem interconnecting devices of the same type: = Class 1,=20 Class 2, or SLB.=20
There is no problem interconnecting a Class 2 device with an SLB = device.=20
There is a minor problem of AC leakage current when = interconnecting a=20 Class 1 device with either a Class 2 device or SLB device.=20
There is a minor problem of AC leakage current when = interconnecting a Pin=20 1 Problem device with either a Class 2 device or SLB device. The major = =93Pin 1=20 Problem=94 is remedied.=20
There is a major =93Pin 1 Problem=94 when interconnecting a Pin 1 = Problem=20 device with either a Class 1 device or another Pin 1 Problem device.=20
A Safety Loop Breaker circuit helps a lot and may provide an = adequate=20 solution.=20
A transformer will provide isolation, solving the Pin 1 Problem = and the AC=20 leakage problem.

Let=92s move on to interconnecting = three or more=20 devices. Again, as long as all of the devices are of the same type: = Class 1,=20 Class 2 or SLB, three or more can be interconnected without a problem. A = mix of=20 Class 2 and SLB devices may be interconnected without a problem. = Problems arise=20 when Class 1 or Pin 1 Problem devices are introduced into a Class 2 or = SLB=20 grounding system. When there is only a single Class 1 or Pin 1 Problem = device in=20 the system, the situation is the same as for interconnecting two devices = as=20 discussed above. It gets interesting when two or more Class 1 or Pin 1 = Problem=20 devices are introduced into the Class 2 or SLB grounding = system.

When we=20 directly attach a Class 1 device to a Class 2 ground system we cause the = whole=20 ground system to be connected to safety ground, thereby loosing the = inherit=20 ground isolation in a Class 2 system. When we directly attach a Class 1 = device=20 to the SLB first-level ground system, we cause the first-level ground = system to=20 be directly connected to the second-level ground system. This nullifies = the=20 benefit of all of the Safety Loop Breaker Circuits in the system. We now = have=20 what amounts to a Class 1 grounding system. We don=92t have an immediate = problem=20 because we don=92t have a noise generator in any loop. We do have a = problem=20 waiting to happen when a Pin 1 Problem device is added to the system. = This is=20 shown in figure 3.7-10.

Figure=20 3.7-10, A ground loop bypassing the SLB circuit.

The = ground loop=20 flows between the Pin 1 Problem device and the Class 1 device, bypassing = the SLB=20 circuit. The noise voltage is impressed on both the shield from the Pin = 1=20 Problem device as well as the shield to the Class 1 device. Things can = get=20 pretty complex and this figure doesn=92t even hint at the complexity. = The SLB=20 device can actually be several interconnected SLB devices and the Class = 1 device=20 can be anywhere in the system =96 an input or an output. The ground loop = problem=20 can be hard to diagnose. For example, say you had a system comprised of = all SLB=20 devices except for a single Pin 1 Problem device. There is no problem = because=20 the SLB isolation is intact and the ground loop is broken. Now, you buy = a new=20 component that happens to be a Class 1 device and install it in the = system. Now=20 you have hum that you didn=92t have before. You take the new component = out of the=20 system and the hum goes away. You would have reason to suspect that = something=20 was wrong with your new component, while the problem is really caused by = that=20 Pin 1 Problem device that had been in your system all along. Bill = Whitlock=92s=20 article Understanding, Finding, & Eliminating Ground Loops = in Audio=20 & Video Systems contains some great techniques that you will = need to=20 isolate the problem device.

Thinking that if one Safety Loop = Grounding=20 circuit is good, two would be better, some designers separately isolate = the=20 component input and output, each with its own SLB circuit. This does = isolate the=20 input from the output but unfortunately comes with the cost of = increasing the=20 ground noise within the component. We=92ll see how this happens in the = section on=20 Ground Isolation.

Often the preamplifier is the central component = in the=20 audio system, with a single output and many inputs. If both the = preamplifier and=20 amplifier are SLB devices, you can ensure that there can be only one = Class 1 or=20 Pin 1 Problem device in the system at any time by switching the input = grounds=20 along with their associated signals.

Other than that, we are = back to=20 galvanically isolating the problem component with a transformer. =

I know=20 that you have been wondering about the forth scheme and thinking that = there are=20 three kinds of people =96 those that can count and those that can=92t = count. Well=20 there really is a forth system and it is the ultimate solution for = system=20 grounding. This solution, which is shown in figure 3.7-11, is centered = on a=20 preamplifier that has all of its input and outputs galvanically = isolated. Since=20 everything is galvanically isolated, it doesn=92t matter what is = connected to the=20 preamplifier. You can mix Class 1, Class 2, SLB, Pin 1 Problems, and = computers =96=20 none of the grounds are interconnected. The preamplifier may be a Class = 1 or a=20 Class 2 or an SLB device, although it really doesn=92t make any sense to = make it=20 an SLB device because there is no reason to break the safety ground = circuit.=20 Notice that the computer is plugged into a separate branch of the mains = power to=20 reduce the earth noise in the system.

Figure = 3.7-11, A=20 grounding scheme centered on an isolated preamplifier.

Did = I say=20 that there were four schemes? Well there is really a fifth. That is a = scheme=20 where the isolated preamplifier scheme is turned inside-out with all of = the=20 attaching components having galvanically isolated interfaces. In this = case, the=20 preamplifier does not need to have isolated interfaces. The attaching = devices=20 may be Class 1, Class 2, or computers. They cannot have a Pin 1 Problem = because=20 of their galvanic isolation. Again, it doesn=92t make any sense to have = a=20 galvanically isolated device further isolated with a Safety Loop Breaker = Circuit. I currently use this grounding scheme in my personal audio = system. This=20 scheme is shown in figure 3.7-12.

Figure = 3.7-12, A=20 grounding scheme involving isolated devices.

3.8 Cables

Cables are a science of themselves and to get an = idea of=20 some of the complexity involved, you can read Pin 1 = Revisited by Jim=20 Brown, SCIN, also=20 by Jim Brown and chapter 9 of Morrison [1]. However the good news is = that in the=20 relatively benign environment of the home and with short cables used in = home=20 audio installations, cables do not need to be a critical factor. I will = touch on=20 a few important factors though. Cables both radiate energy and are = affected by=20 EMI from other devices. EMI has two components, an electric field and a = magnetic=20 field, and there are different aspects of a cable that will effectively=20 counteract both of these fields. Electric fields are blocked by the = cable shield=20 and a good shield for that is the woven copper type - the tighter the = weave, the=20 better. A twisted pair of wires is relatively immune from a magnetic = field, so=20 you will want to use shielded twisted pair for all of your cables. This = is even=20 good for the power cable, although you will want heaver wire for this=20 application. Another important factor is the DC resistance, or = low-frequency=20 impedance of the ground wires =96 we want that as low as possible. We = saw why this=20 is important back in chapter 3.2 when we interconnected the components. =

3.9 - Other Interconnections

My mother used to say, = gesturing at=20 something I had found, =93Don=92t touch that, you don=92t know where it = has been!=94=20 Digital audio signals are kind of like that =96 you don=92t know where = they have=20 been. Often the source of the signal will have a switching power supply = that is=20 dumping tons of noise into the ground system. Examples would be a = computer or=20 video system. For best results, the digital system should be isolated = from the=20 audio system as much as possible for both power and signal. Each should = be=20 powered from separate power line branches. The isolation for the signal = will=20 depend on what the interface is. As poor as a Toslink is for many = reasons, it=20 does provide galvanic isolation. An SPDIF or AES interface should be = isolated=20 with a pulse transformer. A USB interface is a little tougher. If you = are=20 designing your own interface, you can use opto-isolators or GMR = isolators, or=20 convert the interface to SPDIF and use a pulse transformer. I haven=92t = used any,=20 but there are plenty of external USB isolation devices available. Thomas = Kugelstadt=92s article, =93When good grounds turn bad!=94 is a good overview of = the problem=20 of interfacing with digital devices and what you can do to alleviate the = problem. Although the article uses an RS-485 data link for an example, = the=20 concepts are valid for any digital interconnection scheme.

Even = the=20 analog signals have problems: according to Jim Brown in his RFI = article=20 =93Virtually all computer sound cards have Pin 1 Problems.=94 =

Cable TV is=20 another problem area for causing hum in an audio system. Almost all = cable=20 grounds are at a different level than the mains safety ground and need = to be=20 isolated with an RF isolator. Satellite TV systems can also cause a = problem and=20 need to be isolated, however these need a different type of isolator = from the=20 CATV isolator. A search of the internet will turn up lots of = options.

3.10 - Interconnection Summary

Provide a dedicated branch power line, or at least a benevolently = loaded=20 one, for audio components.=20
Plug all audio components into the same power strip or power = outlet.=20
Provide a separate branch power line for computers, TVs or any = other=20 devices having switching power supplies.=20
Provide isolated interfaces for connections between:
a. The = audio=20 system and other devices, like computers or TVs.
b. Pin 1 Problem = devices=20 and any other audio component.
c. Class 1 component and a class 2=20 component. Isolation may not be needed for this case.
d. = Class 1=20 component and an SLB isolated component. Isolation may not be = needed=20 for this case.=20
Loops aren=92t bad =96 it depends on what is on the loop.=20
Use shielded heavy gauge twisted pair interconnect cables.=20
Make a map of your system grounds.

Understanding, Finding, & Eliminating Ground Loops = in Audio=20 & Video Systems by Bill Whitlock has a lot of great information = on=20 solving interconnection problems in audio systems.

Chapter 4 =96 Ground Structure within a Component

So far, = for=20 simplicity I have included only a single signal reference and power = supply in=20 each device. However, most components are comprised of several circuit = boards,=20 each with its own grounding scheme incorporating busses or stars on = them. These=20 are then interconnected, together with one or more power supplies and = the=20 chassis. This can get quite complex and it is a good idea to make a map = of the=20 ground structure and power structure when designing a piece of audio = equipment.=20 Individual grounds and power lines should be routed as carefully and = purposely=20 as the audio signals.

In a nutshell, a problem can occur when = two=20 grounds share the same conductor (perhaps by necessity) or something is=20 connected to the wrong ground. System grounding may be established using = a star=20 structure, a star-of-stars, a buss arraignment or a combination of = those. Many=20 problems are eliminated when every ground is connected to a single point = =96 star=20 grounding. However, this is often impracticable: The best we can hope = for is a=20 star-of-stars approach with individual stars connected.

It=92s = all about=20 controlling the paths =96 ground current (of any kind) should go only = where it is=20 needed. Likewise, power current should go only where it is needed. =

4.1 - Grounding Rules

Here are some rules to help you plan = your=20 grounding structure. The first four rules are from what we learned about = interconnecting equipment.

Rule 1: Each of the = following=20 must be connected to the system star ground by one and only one=20 route.

All signal references=20
All power commons=20
Shields of non-galvanically isolated single-ended inputs and = outputs=20
Safety ground and chassis. The safety ground and chassis should be = thought=20 of as a single entity.=20
The connection may be direct, or indirect through a = star-of-stars=20 or buss. This is expanded upon below.=20
The safety ground and chassis may be connected to the = system star=20 ground through a Safety Loop Breaker Circuit.=20
The =93one and only one=94 part of this rule precludes ground = loops. There is=20 no excuse for a ground loop within a single component. =

Rule 2:=20 The shield of a balanced input or output (XLR pin 1) must = be=20 connected to the chassis at or as close as is possible to the=20 connector.

Rule 3: The shield of a single-ended = input or=20 output that is not galvanically isolated must be directly = connected to=20 the system star ground.

The shield is the signal reference = in the=20 cable.

Rule 4: Any circuit associated with an input or = output=20 that is not galvanically isolated must have its signal reference=20 directly connected to the system star ground.

Rule = 5:=20 The mains safety ground must be directly connected to the = chassis.=20

From IEC 60950, =93The wire is terminated with a closed = loop=20 connector which is fixed to the earthing stud or screw with a star or = lock=20 washer and a nut. Other parts of the product that need to be earthed are = connected by closed loop connectors to the same stud and locked with an=20 additional nut. It is important that the earth wire from the power = supply cord=20 is located at the bottom of the stud and locked with its own nut. The = earthing=20 stud must not be used for any purpose other than earthing. It cannot be = used,=20 for example, for the mechanical fixing of parts other than the earth = conductors.=20 Its mechanical structure must also be such that it cannot be loosened = from=20 outside the device. For example, it cannot be a post fixed with a screw = from=20 outside the product.=94

Rule 6: Each signal reference = must=20 be directly connected to its power reference.

That is, no = circuit=20 may have its signal reference connected to its power common through = another=20 circuit=92s signal reference or power common. This rule allows for a = star-of-stars=20 with the signal reference and power common directly connected together = in a star=20 and that star connected to the system star (either directly or through a = buss).

Rule 7: Circuits may be grouped together = with=20 their signal references forming a buss.

The order of the = grouping is=20 not arbitrary. Just as the signal is routed along, stage to stage, the=20 associated signal reference can be routed with the signal between = stages. Keep=20 the signal and its associated signal reference electrically close = together; they=20 should be treated as a pair. This minimizes the risk of noise being = injected=20 into the signal reference.

One end of the buss should be = connected to the=20 system star ground, either directly or by a star of stars.

4.2 - Grounding Examples

Now let=92s take a look at some = examples.

Figure=20 4.2-1, An example of proper ground routing with a ground=20 buss.

Figure 4.2-1 shows a Digital to Analog Converter = (DAC) with=20 an SPDIF input and single-ended analog output. Each of the power = supplies has a=20 single power common output and each of the circuits has a single signal=20 reference output; all connected to a ground buss. Thus we have a buss of = stars=20 which in turn is connected to the system star ground. The order to which = each=20 attachment is made to the buss is important, flowing along with the = signal from=20 input to output. Notice that none of the Grounding Rules are violated. = In an=20 implementation of this example, there may not be a separate physical = buss that=20 can be identified. Rather, the buss is formed from ordinary printed = circuit=20 traces and regular hookup wire between PC boards. It is the structure = that=20 creates the buss.

Figure 4.2-1 also shows separate power supplies = for the=20 DAC and the amplifier. The DAC is mostly digital electronics that tends = to=20 reflect digital switching noise back into its power supply, particularly = on the=20 power common. Therefore it is a good idea to keep this power common = separate=20 from analog signal references or the power common of analog supplies.=20

Now let=92s change one thing and see what happens. I am not an = advocate=20 for an SPDIF interface without a pulse transformer, but there are plenty = of them=20 out there so for this example I will remove the transformer and bring = the SPDIF=20 interface directly into the DAC.

Figure=20 4.2-2, A poorly-chosen grounding scheme.

I=92m sure that = you see=20 that we now have a Pin 1 Problem. Rule 3 is broken by connecting the = input=20 shield to the buss rather than to the system star ground. Rule 4 is = broken by=20 connecting the DAC circuit signal reference to the buss rather than = directly to=20 the system star ground. Okay, let=92s fix the grounding so that those = rules are=20 not violated.

Figure=20 4.2-3, A well-chosen grounding scheme.

In fixing those two = things=20 we have transformed the grounding structure into a star-of-stars. This=20 illustrates an important concept =96 any device that has both = single-ended inputs=20 and outputs (that are not galvanically isolated) cannot use a ground = buss=20 structure. This is because the signal references for both the input = circuit and=20 output circuit must be directly connected to the system star ground and = adding a=20 ground buss between them would form a ground loop. Another example of = such a=20 component is the ubiquitous single-ended preamplifier. Grounding = problems with a=20 preamplifier may well be the source of the myth that a star-of-stars is = the only=20 acceptable grounding scheme.

Let=92s take a look at another = example, this=20 time a power amplifier. Using a ground buss structure in a power = amplifier has=20 been around for a long time and is popular with Japanese amplifier = constructors.=20

Figure=20 4.2-4, a power amplifier using a ground buss.

The first = thing to=20 notice is that the shell of the input RCA jack is not isolated from the = chassis.=20 Rather, it is effectively the system star ground. The order of = attachment to the=20 buss is important with the buss starting at the input jack and ending at = the=20 negative speaker jack. The quietest circuits are connected closest to = the input=20 jack, with the progressively noisier circuits towards the speaker jack. = An=20 implementation of this example will have a readily identifiable buss, = usually a=20 very thick bare solid copper wire. An example of this technique can be = seen here.=20

All of this is not to argue that a ground buss is superior to a = star=20 ground or vice versa, rather both are tools that can be effectively used = in an=20 appropriate situation. Stars are easier to implement because you don=92t = have to=20 worry about the order of connections like you must on a buss. Further=20 information on implementing star grounds and ground busses is in Randall = Aiken=92s=20 article here.

4.3 - Power Supply Chassis

When should you put the power = supply and=20 amplifier in separate chassis? First of course is if the unit is = physically too=20 large for a single chassis. The second reason is for noise immunity - = for=20 example, a phono preamplifier with a sensitive input stage. Or perhaps = you have=20 a huge power transformer that is radiating a large magnetic field and = you need=20 to physically separate it from the amplifier. There is no grounding = reason to=20 have separate chassis.

The next question is =96 how should they = be=20 separated? The answer is simple; and obvious if you think about it. Even = though=20 the two devices are physically in two chassis, they should be thought of = as two=20 parts of a single chassis interconnected with a cable. Design your = grounding and=20 make a map of the grounding and power on a piece of paper. Then draw a = line=20 around what you want in the power supply chassis (or vice verse, what = you want=20 in the amplifier chassis). Anything that crosses the line is in the = cable. This=20 is shown in figure 4.3-1. The cable shield extends the chassis shield = between=20 the two chassis and should be connected to each chassis at or as close = as=20 possible to the connector. The safety ground wire in the cable is = connected to=20 each chassis at its system star ground, thereby extending the safety = ground=20 function to the second chassis. Be careful of the routing of the safety = ground=20 wire that connects the two chassis =96 it should take as direct a route = as=20 possible and lie close to the chassis. Use a connector with a pin that = makes=20 first and breaks last for the safety ground. It would be a good idea to = use a=20 twisted pair for the wires containing power and power common to cut down = the=20 susceptibility to radiated magnetic EMI.

Figure=20 4.3-1 A separate power supply chassis.

The power supply = chassis may=20 contain more than one power supply. Each power supply should have = individual=20 power and power common lines with each power common connected to its = destination=20 in the amplifier chassis. The destination may be a star ground, a ground = buss or=20 an individual signal reference. The alternative of connecting all of the = power=20 common lines to the star ground in the power supply chassis and then = running a=20 single shared power common wire to the amplifier suffers from Common = Impedance=20 Coupling. Also, you may want the different power common wires to go to = separate=20 places in the amplifier.

A power supply chassis can service more = than=20 one amplifier chassis. In this case a cable with the complete set of = power,=20 power common, safety ground and shield, should run between the power = supply=20 chassis and each amplifier chassis. If a power supply in the power = supply=20 chassis serves two (or more) amplifier chassis, its power common should = be=20 connected to the destination in each amplifier chassis.

4.4 =96 Input Switching

There was a time when it was the = vogue to=20 switch both wires, signal and ground, of single-ended inputs. I don=92t = know where=20 this came from =96 perhaps it was just a copy-cat of switching balanced = inputs. Or=20 perhaps it was an attempt to keep a myriad of dirty grounds from = corrupting the=20 signal reference. In any event, now that we have cleaned up the grounds, = it may=20 not necessary to switch the grounds. However there still are a couple of = cases=20 where switching both the ground and signal may help:

Switching = the=20 grounds helps with cross-talk and bleed-through between the inputs. = Cleaning up=20 the grounds should help this. If there is still a problem you can = connect (with=20 a relay) a 50 Ohm resistor between the signal and signal reference of = all inputs=20 except that which is selected.

AC leakage current flowing = through the=20 interconnect shield may pose a problem, particularly in Class 2 and SLB = devices.=20 Remember, when the interconnect shields are not switched, all of the = grounds in=20 the system are connected together in a mesh.

4.5 =96 Input Jacks

The RCA jack of a single-ended = connection is=20 usually isolated from the chassis and the shield is connected to the = star=20 ground. It may be beneficial to connect a 10nF ceramic capacitor from = the shell=20 of the RCA connector (shield) to the chassis. This provides a direct = path for RF=20 noise from the shield to the chassis right where the shield enters the=20 chassis.

4.6 =96 Volume Controls

Volume controls are potentially a = source of=20 hum and noise from two causes: first, by radiated coupling into the = resistive=20 element, and second, by inappropriate choice of circuit connections. The = radiated coupling is easily solved by ensuring that the body of the = control is=20 grounded to the chassis through its metal mounting bushing. =

Volume=20 controls on non-galvanically isolated interfaces are unique in that they = are in=20 the output circuit of one component and also in the input circuit of = another=20 component. They work by common impedance coupling between the two = circuits.=20 Therefore keep in mind that there are two separate signal loops, an = input signal=20 loop and an output signal loop. It is important to keep the two loops = separate=20 in the grounding system. This is best accomplished by connecting the = reference=20 end of the volume control potentiometer to the system star ground. =

Figure=20 4.6-1, Volume controls on a non-isolated input.

When the = input is=20 galvanically isolated it is no longer necessary to connect the volume = control to=20 the system star ground. It should be considered as part of the amplifier = circuit=20 and referenced to the signal reference of that amplifier.

Figure=20 4.6-2, Volume controls on isolated inputs.

4.7 - Grounding Transformers

Different kinds of transformers = have=20 different grounding requirements. Sometimes a transformer will have an = internal=20 shield between the windings and this will give you a hint how the = transformer=20 should be grounded. The shield is to inhibit capacitively coupled AC = leakage=20 current between windings and between windings and the frame. Sometimes = the=20 shield is connected to the frame and other times it has a separate wire. = What is=20 on the winding being shielded will tell you where to ground the frame = and the=20 shield. The idea is to get the AC leakage current noise back to where it = came=20 from in as short a path as possible. So, for example, the AC leakage = current=20 from the primary of a power transformer came from the mains power line = and=20 should be returned to earth, so the transformer frame should be = connected to the=20 safety ground.

The frame of the power transformer in a class 1 = component=20 (with chassis connected to safety ground) must be connected to safety = ground -=20 just make sure that there is a good contact between the transformer = frame and=20 the chassis. It doesn=92t make sense to ground the power transformers in = a class 2=20 component because since there is no connection to safety ground, there = is=20 nowhere to bleed the AC leakage current. Also, perhaps the transformer = is not=20 intended to be grounded as part of the isolation design. Therefore, = grounded or=20 ungrounded, the power transformer in a class 2 device should be left as = it is.=20

Some audio output transformer circuits provide better = performance with=20 the transformer frame grounded while others work better with the = transformer=20 frame ungrounded so you will need to experiment to determine which is = better for=20 a particular case. The leakage current in a power amplifier output = transformer=20 came from the output stage power supply and thus, if the transformer is = to be=20 grounded, the frame should be isolated from the chassis and a wire = attached to=20 the frame should be returned to the power common of that output=20 stage.

Speaking of output transformers, the negative terminal of = the=20 output should be connected to the power common of the output stage. =

The=20 output windings of line output transformers and the input windings of = line input=20 transformers don=92t need to be grounded. Looking back on the chapter on = transformer coupled interfaces; you will remember that grounding one = side of the=20 interface creates a single-ended interface and the grounding schemes are = covered=20 there.

Some line interface transformers have a center tap which = may be=20 grounded for a balanced interface. I would recommend grounding the = center tap of=20 only the sending end =96 grounding both ends may create a ground loop. = Even though=20 the primary winding is from the output driver, I would not recommend = grounding=20 the center tap to the output drive power common because this would allow = the=20 possibility of a Pin 1 Problem if the other end were also grounded. = Rather, I=20 would recommend using the system star ground in this case.

4.8 =96 Ground Isolation

Some people believe that it is = necessary to=20 isolate the system star ground from the chassis and safety ground in = order to=20 have a hum-free audio system. However, if all of the components in the = system=20 have their grounding implemented properly, there is absolutely no need = for=20 ground isolation, Although isolating the grounds may eliminate a ground = loop, it=20 does come with two penalties: First, since the signal reference is not = directly=20 connected to the chassis, the chassis is not an effective shield for the = electronics. Second, since the power common is isolated from the safety = ground=20 and connected to the signal reference, any AC leakage current from the = power=20 supply may flow through the signal reference to get to the safety ground = in=20 another component.

If you must isolate the grounds; never, ever, = for any=20 reason, disconnect a safety ground or fail to provide a safety ground in = any=20 equipment that you build. First, it is unsafe and second, there are = equally=20 effective methods of isolating grounds that do not come with the safety = hazard.=20 Figure 4.8-1 shows two such methods.

Figure=20 4.8-1, Ground isolation

First is to provide a =93ground = lift=94 switch=20 between the two grounds to be isolated. This does such a good job of = isolating=20 the grounds that it precludes current from an electronic short circuit = from=20 blowing a fuse. This is illustrated in figure 1-3.

A better = solution is=20 to provide a Safety Loop Breaker Circuit (SLB). This circuit will = allow the=20 current from a fault to flow to the chassis and also provide ground = isolation=20 under normal, non-fault conditions. In his article, Rod Elliott states = that a=20 Safety Loop Breaker Circuit may not be legal in some places so you = should check=20 this out before you use this circuit.

The best solution for = ground=20 isolation is to employ galvanic isolation on problem interconnections. =

A=20 component with a properly designed grounding system will not have any = internal=20 ground loops. Therefore, there is never any reason to isolate grounds = within a=20 component.

Let=92s take a closer look at how the Safety Loop = Breaker=20 circuit works to see why I say it is a third choice, behind fixing the = ground=20 loop problem or galvanically isolating the ground loop with a = transformer. I=92ll=20 start with dissecting the problem so as to better understand what is = going on.=20

Ground Loop Suppression

Figure=20 4.8-2

Figure 4.8-2, The ideal, as shown in the left hand = circuit,=20 is to have the signal transferred unchanged from the sending circuit to = the=20 receiving circuit. However, if there is noise in the system, as shown in = the=20 right hand circuit, the signal reference rides on the noise and the sum = of the=20 signal plus noise is presented to the receiving circuit.

Figure=20 4.8-3, Showing how the noise is added to the signal by common impedance=20 coupling.

Let=92s say, for example, that the interconnect = shield=20 connecting the two signal references has an impedance of 1 Ohm and the = level of=20 the noise is 1 Volt. The 1 Volt across 1 Ohm will cause a noise current = of 1 Amp=20 to flow, and the 1 Volt of noise will be presented to the receiving = circuit.=20 Note that I refer to the impedance in the loop rather than the = resistance in the=20 loop. This is because the noise spectrum is spread across a wide = frequency range=20 and the impedance at higher frequencies is often more significant than = the=20 resistance. For simplicity, I assume for all of these examples that the = noise=20 source has an internal impedance of zero Ohms. Of course it will have a = (very=20 low) finite impedance of some internal ground wiring. Refer to Chapter = 3.1 for=20 an example of a noise generator.

Figure=20 4.8-4, Adding impedance to the noise loop.

In this example = I have=20 added 9 Ohms of impedance to the noise loop for a total of 10 Ohms of = impedance.=20 The impedance may be added in the component with the noise or in the = attaching=20 component. The noise voltage is still 1 Volt; however the 1 Volt across = 10 Ohms=20 of impedance causes only 0.1 Amp of noise current in the loop. The 0.1 = Amp of=20 current through the 1 Ohm impedance of the interconnect shield develops = only 0.1=20 Volt of noise to be presented to the receiving circuit. We have thus = reduced the=20 noise that the receiving circuit sees by a factor of ten. Note that none = of the=20 noise has disappeared; the remaining 0.9 Volts of noise is dropped = across the=20 added 9 Ohms of resistance. There is an important difference between the = two=20 circuits in figure 4.8-4. In the left hand circuit, the added impedance = is under=20 the noise source so that signal references are close to safety ground; = therefore=20 the noise voltage on the ground system is low. One signal reference = (being a=20 class 1 device) is at ground while the other is 0.1 Volts above ground. = In the=20 right hand circuit, the added impedance is in the attaching component so = that=20 the signal references are raised above ground. One signal reference is 1 = Volt=20 above ground while the other is 0.9 Volts above ground. In both cases = there is=20 0.1 Volts between the signal references but the left hand circuit has a = much=20 lower ground noise. The noise on the ground system is not important to = this one=20 signal because it is common mode, but noise on the ground system may = become a=20 problem if other connections in the system allow this noise to flow to = other=20 areas. We=92ll see some examples of this later. Although it is a trade = off, it is=20 best if noise on the individual signal as well as noise on the ground = system are=20 both kept as low as possible. Therefore, when there is a choice, it is = better to=20 install the added impedance in the circuit containing the noise source. = We have=20 been talking about isolating grounds; however we really are isolating = the noise=20 generator =96 the ground isolation is an artifact of that.

Thus = the=20 conclusion is that the added impedance helps reduce the noise by = inhibiting the=20 noise current in a ground loop; however it does not eliminate the = problem as=20 would galvanic isolation or fixing the ground loop problem. None of the = noise is=20 eliminated =96 it is just moved to a part of the circuit where it will = have less=20 of an impact on signal.

Well, if 9 Ohms is good, why not add 90 = Ohms or=20 900 Ohms of impedance? This would further reduce the noise level; = however there=20 is another, more important, factor to consider =96 that of safety. If an = electrical fault occurs there can be a very high voltage present across = the=20 impedance. We need to keep the ground impedance low so that most of the = fault=20 current will flow through the ground circuit rather than through an = external=20 route, for example, a person.

Safety Loop Breaker Circuit

The Safety Loop Breaker circuit=20 presented in Rod Elliott=92s article addresses this issue of ground noise = differently.=20 Instead of adding impedance to the ground loop, Rod=92s circuit = subtracts voltage=20 from the point in the circuit where it is inserted. This is accomplished = with a=20 pair of diodes (inside a bridge rectifier) in parallel opposition; that = is, the=20 anode of each is connected to the cathode of the other. These diodes = prevent the=20 voltage across the Safety Loop Breaker circuit from being greater than 1 = diode=20 voltage drop (0.6 Volts). As with the simple resister example shown = above, none=20 of the noise has disappeared: the diodes will drop 0.6 Volts and any = remaining=20 noise (over the 0.6 Volts) is distributed across the impedance in the = circuit.=20 If the voltage across the circuit is less than 0.6 Volts, then the = diodes will=20 not conduct and the circuit works by adding impedance to the loop with a = 10 Ohm=20 resistor across the circuit. This resistor also limits the impedance = between=20 safety ground and the internal ground buss. Also, In the event of an = electrical=20 fault that places a high voltage on the internal ground buss, the diodes = will=20 keep the voltage on the buss from exceeding 0.6 Volts. There is a 100nF=20 capacitor across the circuit to reduce RFI problems. Let=92s take a look = at how=20 all this works.

Figure=20 4.8-5

Figure 4.8-5, We need to think in terms of voltages = here. The=20 Safety Loop Breaker circuit drops 0.6 Volts of the 1 Volt noise voltage, = leaving=20 0.4 Volts to be dropped across the impedance in the circuit =96 the = interconnect=20 shield. In the left hand circuit, the Safety Loop Breaker circuit is = under the=20 noise source so the Safety Loop Breaker circuit drops 0.6 Volts, leaving = 0.4=20 Volts on the signal reference. The other signal reference is at 0 Volts=20 (ground). In the right hand circuit, the Safety Loop Breaker circuit is = in the=20 attaching component so that the signal reference has the 0.6 Volts of = noise that=20 is dropped by the circuit. The other signal reference has 1 Volt of = noise on it.=20 In both cases there is 0.4 Volts between the signal references but the = left hand=20 circuit has a much lower ground noise.

What would happen if we = installed=20 a Safety Loop Breaker circuit in both the component with the noise as = well as=20 the attaching component?

Figure=20 4.8-6

Figure 4.8-6, Here is an example of there being not = enough=20 voltage around the loop for the Safety Loop Breaker circuits to drop 0.6 = Volts=20 each. The voltages and current are now established by the impedances in = the=20 loop. The noise on the interconnect shield between the two components = has been=20 reduced by almost a factor of ten and the noise on the ground system, = 0.48 Volts=20 on the right and 0.52 Volts on the left, is about half-way between the = two cases=20 shown above that each have a single Safety Loop Breaker circuit. =

These=20 examples with a single noise source and only two components are pretty = simple.=20 More complex cases with multiple components, multiple noise sources and = isolated=20 Safety Loop Breaker circuits are shown below.

Multiple Ground Loops

Now, strange things happen when we add = additional components to the system. This is where we see the effect of = ground=20 noise in the system.

Figure=20 4.8-7

Figure 4.8-7, A third component with a Safety Loop = Breaker=20 circuit added to the system. This example is an extension of that shown = above in=20 figure 4.8-6. Component 1 and component 2 are exactly the same with 1 = Volt of=20 noise being generated in component 1. The difference is that a third = component,=20 exactly the same as component 2 is added to the system. The Safety Loop = Breaker=20 circuits in components 2 and 3 are in parallel, effectively lowering the = impedance in the loop that is presented to the noise generator in = component 1.=20 The result is poorer noise reduction. Not only has the noise on the = shield of=20 the interconnect cable between components 1 and 2 increased, but the = noise on=20 the signal reference of component 2 is passed along on the shield of the = interconnect cable to component 3. Note that the total noise impacting = the=20 signal across the system is the sum of the noise between components 1 = and 2 plus=20 the noise between components 2 and 3. So the total noise is 0.09 Volts = (0.06 +=20 0.03) =96 almost twice the amount present without component 3. Because = this noise=20 is on the ground system, it is always present everywhere in the system = even if a=20 different signal is selected to be active (for example in a preamp). = Okay, so=20 what would happen if the added device were a class 1 device with the = reference=20 directly connected to the safety ground?

Figure=20 4.8-8

Figure 4.8-8, A third component, which is a class 1 = device,=20 attached to the system. This example is the same as that in figure 4.8-7 = except=20 that component 3 is a class 1 device. The noise performance in every = area is=20 worse than if component 3 had a Safety Loop Breaker circuit. This is = because the=20 class 1 device effectively nullifies the Safety Loop Breaker circuit in=20 component 2. Now, an interesting thing happens if we swap component 2 = with=20 component 3. The components are the same, just their position has = changed.=20

Figure=20 4.8-9

Figure 4.8-9, The noise on the shield of the = interconnect=20 cable between components 1 and 2 is about twice that as in the case = shown in=20 figure 4.8-8 but the total noise as seen by component 3 is about the = same (0.4=20 Volt + 0 Volt versus 0.21 Volt + 0.19 Volt.) The important difference is = that=20 now the ground system is clean. Note that the Safety Loop breaker = circuit in=20 component 3 is redundant and it doesn=92t make any difference if it is = there or=20 not. We are back to where we were in figure 4.8-5.

The conclusion = from=20 the above is that adding a second Safety Loop Breaker circuit in the = attaching=20 component can help the noise on an individual interface, but at the = expense of=20 noise on the ground system. Adding additional Safety Loop Breakers = circuits to=20 the system may actually be detrimental. Including a class 1 device in = the system=20 provides a clean ground system at the expense of increasing the noise on = the=20 individual interface.

Multiple Noise Sources

Let=92s now take a look at what = happens when=20 there is multiple noise sources in the system.

Figure=20 4.8-10

Figure 4.8-10, Two interconnected noise sources. = Because the=20 two noise sources are similar but not identical, the noise voltages will = sometimes add and sometimes subtract. The power of the two noise signals = will=20 directly add, while the voltages will add as the square root of the sum = of the=20 squares of the two noise voltages. The result is that each noise source=20 contributes 0.7 Amp of noise current and the combined effect produces = 1.414=20 Volts of noise across the 1 Ohm resistor.

Figure=20 4.8-11

Figure 4.8-11, Nothing unusual here; one Safety Loop = Breaker=20 circuit helps a little and two circuits help a lot more. As with a = single noise=20 source, things get interesting as we add additional components to the = system.=20

Figure=20 4.8-12

Figure 4.8-12, A single Safety Loop Breaker circuit = with two=20 noise sources. This example, as well as the following example shown in = figure=20 4.8-13, is the same as two circuits shown in figure 4.8-5 back-to-back. = In this=20 example both components 1 and 3 have 1 Volt noise sources and component = 2 has a=20 Safety Loop Breaker circuit inserted into both ground loops. The diodes = in this=20 circuit clamp the voltage across it to 0.6 Volts; therefore each cable = shield=20 has 0.4 Volts of noise. Note that the two noise voltages do not add in = this case=20 because the total voltage is limited to 0.6 Volts.

What if the = Safety=20 Loop Breaker circuits were in the components with the noise rather than = in the=20 attaching component?

Figure=20 4.8-13

Figure 4.8-13, Safety Loop Breaker circuits in the = noisy=20 components. The difference between this example and the previous is that = component 2 is now a class 1 component with its signal references = connected to=20 safety ground. The signal references of components 1 and 3 are 0.4 Volts = away=20 from this ground. Thus, the ground system in this example is a lot = quieter than=20 that of figure 4.8-12. This is most important if component 2 were a = preamplifier=20 with other components attached. The noise on the ground system is seen = by and=20 thus affects all components in the system.

Okay, but what if = component 2=20 also had a Safety Loop Breaker circuit; wouldn=92t that be even = better?

Figure=20 4.8-14

Figure 4.8-14, An example of the interaction of = three Safety=20 Loop Breaker circuits. The analysis of this gets messy because the two = noise=20 sources interact through all three of the Safety Loop Breaker circuits. = The=20 result is that you trade off a little noise on the interfaces for a lot = more=20 noise on the ground system.

You can add or remove Safety Loop = Breaker=20 circuits at different points in the system but the problem is that both = noise=20 sources contribute to each other through shared impedances and need to = be=20 isolated from each other so that there are no shared impedances. =

The=20 conclusion from the above is pretty much the same as with a single noise = source:=20 Adding a Safety Loop Breaker circuit in the attaching component in = addition to=20 that in the noisy component can help the noise on an individual = interface, but=20 at the expense of noise on the ground system. Including a class 1 device = in the=20 system provides a clean ground system at the expense of increasing the = noise on=20 the individual interface.

Okay then, would it help to put a = separate=20 Safety Loop Breaker circuit on each interface?

Figure=20 4.8-15

Figure 4.8-15, Separate Safety Loop Breaker = circuits. At=20 first glance this looks pretty good =96 each cable shield has only 0.05 = Volts of=20 noise just like the circuit in figure 4.8-6. However there is a problem = lurking=20 internally in component 2 that will become obvious if we re-draw that = portion of=20 the circuit.

Figure=20 4.8-16

Figure 4.8-16, Internal ground structure of = component 2. We=20 see that although the noise voltage on each signal reference is pretty = good,=20 that we have 0.68 Volts (0.48 Volts x 1.414 noise addition factor) of = noise=20 between the two signal references. This is far worse than with a single = Safety=20 Loop Breaker circuit and is almost the same as with no Safety Loop = Breaker=20 circuit at all! By the way, this example shows what the result can be = when=20 violating Grounding Rule 4 =96 the signal references must both be = connected to the=20 system star ground. Okay, let=92s just connect the two signal references = together=20 =96 that should fix the problem. Nope =96 what we now have is the same = as the=20 circuit in figure 4.8-14 except that the two 10 Ohm impedances are now = in=20 parallel making 5 Ohms. This reduces the effectiveness of the Safety = Loop=20 Breaker circuit such that the combined noise voltage on the references = is now=20 0.51 Volts and the noise voltage across each interconnect shield = impedance is=20 0.28 Volts =96 just about the amount of a single Safety Loop Breaker = circuit.=20

Thus the conclusion is that you can do no better than with a = single=20 Safety Loop Breaker circuit in a component.

The conclusions about = a=20 Safety Loop Breaker circuit are:

The circuit does not eliminate, or even reduce, noise =96 it moves = the=20 noise. The noise may be moved to a place where it does not affect the = signal.=20 However, the noise may be moved to a place where it does affect the = signal;=20 where the noise is moved depends on where the circuit is located.=20
A circuit located in a device causing the noise always helps.=20
In a system containing only two components, the device causing the = noise=20 and the attaching device, adding a circuit to the attaching device = also helps.=20
In a system containing more than two components, adding a circuit = to the=20 attaching device hurts the noise performance. This is because some of = the=20 noise is moved to the ground system.=20
It hurts the noise performance to have more than one circuit in a = device.=20 This is because the noise will be moved to the internal signal = reference in=20 that device.=20
Thus the conclusion is that the Safety Loop Breaker circuit helps = reduce=20 the noise by inhibiting the noise current in a ground loop; however it = does=20 not eliminate the problem as would galvanic isolation or fixing the = ground=20 loop problem. None of the noise is eliminated =96 it is just moved to = a part of=20 the circuit where it will have less of an impact on signal. =

4.9 - Construction Summary

Follow the grounding rules in section 4.1.=20
Make a ground map and carefully design the ground structure using = stars=20 and busses.=20
Power supplies are two-terminal devices. Do not tap into the = internal=20 circuitry.=20
Use Safety Loop Breaker circuits only where they are needed. =

Chapter 5 - Conclusions

Audio Component Grounding

Most grounding problems are caused = by=20 something being connected to the wrong ground or ground current flowing = where it=20 is not needed. =93Wrong ground=94 implies that there is more than one = kind of ground=20 and I differentiate grounds as follows:

Safety ground =96 this is the separate (green or green/yellow) = wire in the=20 power line going back to the circuit breaker panel. It is connected to = earth=20 at the panel.=20
Chassis and cable shields =96 These provide protection from = electrostatic=20 fields.=20
Power common =96 This is the 0 Volt reference from the power = supply.=20
Signal reference =96 This provides a point of reference for the = signal in a=20 circuit.

These grounds need to be connected together in = a very=20 specific manner as part of the design of a component grounding = structure.=20 Grounding should be designed as carefully as any other part of the = component and=20 I encourage you to make a map of the grounding structure to show up any=20 potential problems. Here are some rules to help you plan your grounding=20 structure.

Rule 1: Each of the following must = be=20 connected to the system star ground by one and only one route.

All signal references=20
All power commons=20
Shields of non-galvanically isolated single-ended inputs and = outputs=20
Safety ground and chassis. The safety ground and chassis should be = thought=20 of as a single entity.=20
The connection may be direct, or indirect through a = star-of-stars=20 or buss. This is expanded upon below.=20
The safety ground and chassis may be connected to the = system star=20 ground through a Safety Loop Breaker Circuit.=20
The =93one and only one=94 part of this rule precludes ground = loops. There is=20 no excuse for a ground loop within a single component. =

Rule 2:=20 The shield of a balanced input or output (XLR pin 1) must = be=20 connected to the chassis at or as close as is possible to the=20 connector.

Rule 3: The shield of a single-ended = input or=20 output that is not galvanically isolated must be directly = connected to=20 the system star ground.

The shield is the signal reference = in the=20 cable

Rule 4: Any circuit associated with an input or = output=20 that is not galvanically isolated must have its signal reference=20 directly connected to the system star ground.

Rule = 5:=20 The mains safety ground must be directly connected to the = chassis.=20

From IEC 60950, =93The wire is terminated with a closed = loop=20 connector which is fixed to the earthing stud or screw with a star or = lock=20 washer and a nut. Other parts of the product that need to be earthed are = connected by closed loop connectors to the same stud and locked with an=20 additional nut. It is important that the earth wire from the power = supply cord=20 is located at the bottom of the stud and locked with its own nut. The = earthing=20 stud must not be used for any purpose other than earthing. It cannot be = used,=20 for example, for the mechanical fixing of parts other than the earth = conductors.=20 Its mechanical structure must also be such that it cannot be loosened = from=20 outside the device. For example, it cannot be a post fixed with a screw = from=20 outside the product.=94

Rule 6: Each signal reference = must=20 be directly connected to its power reference.

That is, no = circuit=20 may have its signal reference connected to its power common = through=20 another circuit=92s signal reference or power common. This rule allows = for a=20 star-of-stars with the signal reference and power common directly = connected=20 together in a star and that star connected to the system star (either = directly=20 or through a buss).

Rule 7: Circuits may be = grouped=20 together with their signal references forming a buss.

The = order of=20 the grouping is not arbitrary. Just as the signal is routed along, stage = to=20 stage, the associated signal reference can be routed with the signal = between=20 stages. Keep the signal and its associated signal reference electrically = close=20 together; they should be treated as a pair. This minimizes the risk of = noise=20 being injected into the signal reference.

One end of the buss = should be=20 connected to the system star ground, either directly or by a star of = stars.=20

Audio Component Interconnection

Provide a dedicated branch power line, or at least a benevolently = loaded=20 one, for audio components.=20
Plug all audio components into the same power strip or power = outlet.=20
Provide a separate branch power line for computers, TVs or any = other=20 devices having switching power supplies.=20
Provide isolated interfaces for connections between:

a. The = audio=20 system and other devices, like computers or TVs.

b. Pin 1 = Problem=20 devices and any other audio component.

c. Class 1 component and = a class=20 2 component. Isolation may not be needed for this = case.

d. Class=20 1 component and an SLB isolated component. Isolation may not be = needed=20 for this case.=20
Loops aren=92t bad =96 it depends on what is on the loop.=20
Use shielded heavy gauge twisted pair interconnect cables. =

Understanding, Finding, & Eliminating Ground Loops = in Audio=20 & Video Systems by Bill Whitlock has a lot of great information = on=20 solving interconnection problems in audio systems.

Bibliography and References

Books

[1] Grounding and Shielding Techniques in = Instrumentation,=20 Ralph Morrison.

This book is complete and explicit with lots of=20 examples. It is very technical and not a casual read but is essential to = thoroughly understand the subject. It is now in its 5th edition and is=20 expensive, although earlier editions are available used. Page references = in this=20 article are from the 3rd edition.

[2] Solving Interference = Problems in=20 Electronics, Ralph Morrison.

Page references in this article are = from the=20 1st edition.

[3] Noise Reduction Techniques in Electronic = Systems, Henry=20 W. Ott.

Like Morrison, this book is an excellent reference for = the=20 subject material. Page references in this article are from the 1st=20 edition.

[4] Electromagnetic Compatibility Engineering, Henry W. = Ott=20

A significant update and expansion of [3] Get this book even if = you have=20 his earlier one.

[5] through [10] Reserved for future = entries

Publications

[11] Journal of the Audio Engineering = Society,=20 Volume 43, Number 6, June 1995. This whole issue is on shields and=20 grounds.

[12] Susceptibility in Analog and Digital Signal = Processing=20 Systems, Neil A. Muncy JAES, Volume 43, Number 6, June 1995 Pages = 435-453. Muncy=20 introduced the =93Pin 1 Problem=94 in this paper

[13] Balanced = Lines in Audio=20 Systems: Fact, Fiction, and Transformers, Bill Whitlock. JAES, Volume = 43, Number=20 6, June 1995 Pages 454-464.

[14] AES48-2005 AES standard on=20 interconnections =96 Grounding and EMC practices =96 Shields of = connectors in audio=20 equipment containing active circuitry.

[15] Electro-magnetic=20 compatibility =96 =93The art of grounding=94 , Brent Hertz
AES = preprint 3041 (G-1)=20 February, 1991. Grounding in the studio environment.

[16] EN = 60065 -=20 Audio, Video and Similar Electronic Apparatus, Safety = Requirements

[17]=20 through [20] - Reserved for future entries

Articles on the Internet

Interconnection of Balanced and Unbalanced = Equipment Bill=20 Whitlock=20
Hum=20 & Buzz in Unbalanced Interconnect Systems, Jensen AN-004, Bill = Whitlock=20
A=20 New Balanced Audio Input Circuit for Maximum Common-Mode Rejection in = Real=20 World Environments Bill Whitlock=20
Understanding, Finding, & Eliminating Ground Loops = in Audio=20 & Video Systems Bill Whitlock=20
Subtleties Count in Wide-Dynamic-Range Analog = Interfaces=20 Bill Whitlock=20
Sound = System=20 Interconnection, Rane Corp. Technical Staff=20
Grounding and=20 Shielding Audio Devices, RaneNote 151, Stephen Macatee=20
Considerations in Grounding and Shielding = Computer-Controlled=20 Audio Devices, Stephen Macatee=20
Pin 1 = Revisited, Jim Brown=20
SCIN: = Shield=20 Current Induced Noise, Jim Brown=20
A Ham=92s=20 Guide to RFI, Ferrites, Baluns, and Audio Interfacing, Jim Brown. = A=20 different perspective on the problem. A particularly good coverage of=20 filtering RFI.=20
Power and Grounding for Audio and Video Systems: A = White Paper=20 for the Real World =96 International Version, Jim Brown. An = excellent=20 introduction to the subject, from the professional audio perspective, = but a=20 lot applicable for the advanced hobbyist.=20
Ground Loops, D. Self. Ground loops =96 how they = work and how=20 to deal with them.=20
Star=20 Grounding, Randall Aiken=20
Common Mode Choke Theory=20
The=20 Two-Point Earth Method. Inside the Sakuma Amplifier.=20
Earthing=20 Your Hi-Fi =96 Tricks and Techniques, Rod Elliott. A good = description of the=20 Safety Loop Breaker Circuit.=20
Blocking Mains DC Offset, Rod Elliott=20
Earthing. Short introduction to the requirements = for a Class=20 1 device.=20
Ground Loop Problems and How To Get Rid of Them, = Tomi=20 Engdahl=20
Ground and (safety) Earth, Eric Juaneda. Good = coverage of=20 parasitic currents.=20
International = Safety=20 Standard for Information Technology Equipment, IEC 60950=20
When good grounds turn bad =96 isolate! Thomas = Kugelstadt.=20 Analog Applications Journal, 3Q 2008, page 11.=20
So You Thought Your Amplifier Was Balanced? Andy = Grove and=20 Peter Qvortrup. Good introduction to the advantages to balanced = circuitry, and=20 particularly transformer interfaces.

Revision = History

May 29, 2009 - 1.0
Base=20
June 1, 2009 - 1.1
Expanded on computer attachment In = Effective=20 Interconnection Schemes.=20
June 13, 2009 - 1.2
Re-structured the Effective = Interconnection=20 Schemes section.=20
June 29, 2009 - 1.3
Minor editorial changes
Added = Conclusions=20 chapter=20
July 3, 2009 - 1.4
Changed Input Switching section.=20
July 9, 2009 - 1.5
Added section on Safety Loop Breaker = circuit=20
July 20, 2009 - 1.6
Restructure chapter 3 for better = readability=20
July 30, 2009 - 1.7
Add reference to t.i. Analog = Applications=20 Journal.=20
September 7, 2009 - 1.8
Correct typo=20
April 23, 2010 - 1.9
Restructure chapter 3 to eliminate = duplicate=20 heading number
May 1, 2010 - 1.10
General cleanup
fix typographical=20 errors
fix several figures
update references

Comments on the article = "Audio=20 Component Grounding and Interconnection"

=20 Posting Rules

You may not post new threads

You may post replies

You may post attachments

You may edit your posts

BB code is = On

Smilies are = On

[IMG] code is=20 On

HTML code is Off

Trackbacks are=20 Off

Pingbacks are=20 Off

Refbacks are=20 Off

Forum=20 Rules

=20 Similar Threads
Article	Article Starter	Category	Comments	Last Post
Anyone=20 tried component video cable for audio use?	brents82	Everything = Else	9	6th January = 2010 03:02 AM
Bryston=20 quad-complementary interconnection	Ragnwald	Solid = State	33	11th July = 2008 04:55 PM
New=20 to car audio and could use some component/sub = suggestions	Stewart8989	Car = Audio	6	14th July = 2006 08:09 = AM

=20 Similar Threads

Article

Article Starter

New To Site?	Need = Help?
Search=20 Privacy Statement=20 Contact Us =	Frequently Asked=20 Questions=20 Mark Forums Read=20 Log Out dmfraser66 =