Professional audio systems are usually built with symmetrical transmission technology, while HiFi-Systems are wired asymmetrically. What is it, what is the difference, why are there two different systems and what happens when you combine them?

First some basic electronics:

Current always flows in a circle, which is why it is called an electric circuit. So if you have an electric Signal If you want to transfer data from A to B, then you need two lines: a forward and a return line. This even applies to alternating current. Although the direction of current flow changes constantly, the current that flows in one of the two lines must flow back in the other at all times.

The information can be contained in the stream in different ways. In the simplest case, it can be contained in the current or in the voltage. There are also more complicated cases, e.g. B. the information on a carrier frequency be modulated. This is e.g. This is the case, for example, with the TV cable. That shouldn't interest us here. The most common case with audio signals is that the information is in the voltage.

If you want to transmit a voltage, you make the source low-impedance and the destination high-impedance. This way the signal arrives with the least loss. (By the way, it is exactly the opposite in the case of electricity transmission.) For example, the output resistance of a hi-fi device is usually well below 100 ohms, while the input resistance is 10 kOhm or more.

The input resistance lies between the forward conductor and the return conductor. Any cable and plug resistances that are always present are connected in series, and as long as they are very small compared to the 10kOhm, they play practically no role. At least that's the theory.

With symmetrical transmission, each outgoing conductor has its own return conductor, so the two wires always appear in pairs and are often even twisted together. This ensures that any interference from radio signals or magnetic fields has practically the same effect on both wires. The useful signal is obtained in the receiver by determining the voltage difference between the two lines, and with this subtraction an interference signal, which is present on both lines in the same way, is eliminated again. The symmetrical transmission is therefore quite immune to interference.

So for symmetrical transmission we need two wires per signal and a subtractor in the receiver to determine the voltage difference between them.

In the case of asymmetrical transmission, on the other hand, one tries to save. If you want to transmit several signals at the same time, you can use a common return line for all signals and thus save wires. If you connect this return line to the internal reference potential (the so-called signal ground) in every device, then you can save on the subtractor, because the useful signal is already on the right one level is related.

However, there is a problem with this. This system now requires that the internal reference potentials are the same in each device. Any difference would appear in the receiver as if it were superimposed on the useful signal and indistinguishable from it. With an asymmetrically connected system, you have to ensure that there are no significant differences in the level of the signal grounds of all devices involved. The ground connection must be as low-impedance as possible, while the individual feed lines are not as sensitive in this regard. Achieving this can be harder than you think.

One could perhaps think that the equality of the potentials can be established simply by connecting the signal grounds with one another with a cable. But every cable has one Resistance, since there are no room-temperature superconductors (yet). Plugs have contact resistance, especially the cheap ones, and this increases over time due to corrosion. And even if none of that were the case, there is still the line inductance that causes an impedance. In a word, the longer the line, the less chance one has of achieving ground level equality.

Unbalanced transmission is therefore primarily an option when the cable lengths are short and when you are looking for the most cost-effective solution. These are exactly the requirements that one finds in hi-fi technology. Every last tenth of a cent has been saved there and the devices involved are usually in the same place.

Different rules apply in professional audio engineering. There you are dealing with larger and more extensive systems, where there is no hope from the outset that you can ensure balanced ground levels. So symmetrical transmission is used where it is not required. The extreme example of this is analogue telephone technology, which has been working symmetrically for years. It's about distances of many kilometers. You wouldn't stand a chance with asymmetrical technology. The interference caused by different ground levels would be many times higher than the useful signal.

50 years ago, a hi-fi system consisted of a turntable and a Radio with amplifier passed, often built into the same chest and plugged into the same outlet. In such a case, there is little to fear with asymmetrical cabling. Nowadays, however, hi-fi systems are often more extensive. There will be TV sets, DVD-Players, computers and set-top boxes crammed together and wiring may run throughout the home. Because the antenna line is also unbalanced, the ground wiring actually extends beyond the antenna system. In addition, z. B. in computers, the signal ground is connected to the protective conductor. This means that the protective earth also has an effect on the reference level. In such a widely ramified mass system there are almost always some vagrant currents flowing around, e.g. B. those that are generated by the induction effect. In an unbalanced system, it is very difficult to keep these interference currents out of the useful signal.

In the ideal balanced system, the ground wiring is separate from the signal wiring. This way currents flowing in the ground wiring are irrelevant. The balanced connection between two devices consists of three wires: supply, return and ground. The ground connection is carried out as shielding off to protect the internal signal lines from radiated radio signals. Basically, the shielding of the cable can be viewed as a continuation of the metallic housing of the device. For this reason, the ground conductor (shield) in the cable is also connected to the housing ground of the connected devices, using the shortest possible route, so that HF ​​interference cannot even get inside.

If it wasn't for the shielding of RF signals, then the shield and the ground connection would be completely unnecessary. It plays no role in the signal transmission itself. For this reason it is also wrong to connect the ground wire in the cable to the signal ground of the device with balanced connections. The signal ground only plays a role as a reference point within a device, it is not required externally. Every connection to the outside offers only one gateway for interference signals. Inside the device, however, the signal ground is connected to the chassis ground at a single point. This has its reason in the (un-)sensitivity to interspersed radio signals.

It is particularly interesting now if you want to connect symmetrical with asymmetrical devices, or if you want to switch to symmetrical in the meantime when connecting asymmetrical devices to avoid ground problems (e.g. ground loops). This is where the detail devil often strikes, because the different mass types are not clearly differentiated. Because of an unfortunate ground connection you can lose the complete advantage of the symmetrical technology. So you have to use brains here. The symmetrical technique even has a bad reputation with some, precisely because it is easy to make such mistakes. Incidentally, such mistakes are also made with pleasure by device manufacturers, who should actually know better. So you can find z. B. Many devices where the ground on a connector for balanced signals is not connected to the chassis ground but to the signal ground, contrary to what I wrote above.

So how do you combine symmetrical with asymmetrical if it ever becomes necessary?

The easiest way to do this is with transformers. All four combinations Unsym->Unsym, Unsym->Sym, Sym->Unsym, Sym->Sym can be solved with a transformer (even the same transformer). He does not need his own power supply and he can handle several hundred Volt Voltage difference between the two sides. That would be ideal if there weren't a few downsides too: a transformer has increasing harmonic distortion at low frequencies, and counteracting it is inevitably expensive. In a word: Good transformers cost a lot of money. In addition, they have a considerable weight and Volume, at least when compared to other electronic components (e.g. transistors). Who, however, 250 euros for a Cinch- thinks the cable is a bargain and doesn't need to flinch at the price of a good transformer.

Unfortunately, the quality of a transformer cannot always be recognized from the published data. It is particularly interesting how the distortion factor behaves at low frequencies. An indication of the distortion factor at 1kHz says little. The transformer should also be well shielded, e.g. B. by a mu-metal cap.

Transformers are normally offered as a component for installation in devices. This is of course rather uninteresting for the normal user. Designs that can be looped into a cable connection are more suitable. Examples are the Monacor FGA-40, or the much better and more expensive Lundahl LL6810-phmphm. Both have the cinch plug that is usual in hi-fi, so that they can be used in particular for the connection Unsym->Unsym, i.e. for separating ground loops. The Monacor model is stereo, Lundahl mono, so you need two of the latter for stereo.

If you don't want to use a transformer for one good reason or another, the possibilities unfortunately fan out into a number of cases, so that you have to study the problem a little more closely.

The difficulties have to do with the fact that there are different ways of technically realizing a symmetrical input or output. Depending on which of these variants is present in the specific case, the connection between the symmetrical and asymmetrical devices must be different. It is therefore necessary to know some of the technical details of the devices involved; simply specifying symmetrical/unsymmetrical is not enough. So I have to go a little deeper to describe it correctly.

First to the connectors used

The connector used in the professional field for symmetrical signals is the XLR-Plug. The standard pin assignment is 1:housing ground/shield 2:hot(plus) 3:cold(minus). The 6,35 mm stereo jack plug is also used, although it is only used here for a mono signal. Here is the pinout Tip:Hot Ring:Cold Sleeve:Ground.

Unfortunately, especially with XLR, there are a number of devices that use different assignments. That is the first cause of problems. Often will open Pin 1 is not the housing ground, but the signal ground, which I have already criticized above. This problem also applies to the jack plug. In addition, XLR sometimes swaps hot and cold.

The terms hot and cold are to be understood as follows: Hot is the "normal" signal, i.e. the lead, so to speak. Cold is therefore the return line. The designations + (plus) and - (minus) are also used, but that is a bit confusing, because we are dealing with alternating voltages that can be positive compared to ground or each other and negative again a little later. I will therefore stick to the terms hot and cold.

First of all, we assume that the connections of the devices are correctly wired. The problems that arise when the manufacturers have made mistakes will be discussed later.

A Sym–>Sym connection is very simple. You simply connect hot to hot, cold to cold and the cable shield at both ends to pin 1 (on XLR). The wires for Hot and Cold should be twisted together in the cable. A cable configured in this way provides the best results. The housing grounds of both devices are connected to each other via the cable shield so that interference currents flow through the housing and do not penetrate into the interior of the device. This connection applies to all variants of symmetrical input and output circuits in the devices involved.

If one of the devices (or even both) makes the mistake of connecting the signal ground to pin 1 instead of the housing ground, interference currents can penetrate the device and become noticeable in the useful signal. In this case it may be necessary to disconnect the shield at one end of the cable, or a possibly existing ground lift switch (with this the connection between the protective conductor connection of the device and the signal ground in the device can be separated. This makes it possible to eliminate the hum loop to be removed. However, one accepts that the sensitivity to radio interference increases). However, this can make the connection more susceptible to RF interference. Another trick is to only connect the cable shield to the metal connector housing, but not to pin 1. With a bit of luck, the metal housing of the device socket is connected to the device housing, which means that you have the connection to the housing ground again. Maybe you can even correct the wrong wiring in the device, but of course you have to observe the warranty conditions.

It gets even more complicated with connections between balanced and unbalanced devices. For this I have to explain the individual circuit variants. First the inputs:

1. Transformer balanced input

A transformer is used internally. Transformer and transmitter are actually the same thing; in English e.g. B. There is only one word for it: Transformer. Here Hot and Cold are connected to the primary winding of the transformer.

If you want to connect an unbalanced output to this input, then simply connect the unbalanced ground to the coldConnection and the signal with the hot connector. Pin 1 remains unused.

A more elegant variant (better HF shielding) is possible when using a slightly more expensive triaxial cable: Here you can connect the outer shielding to pin 1 at the input. On the unbalanced output side, it remains unconnected. The inner shield connects unbalanced ground to cold.

Commercially available adapter plugs between cinch and XLR are wired incorrectly for this purpose because they connect the unbalanced ground to the housing ground on the symmetrical device and to cold. This creates a ground connection that can lead to ground loops. It would be better if the manufacturers of such Adapter would install at least one ground lift switch, with which the connection to the chassis ground can be interrupted.

2. Input with differential amplifier

This cheaper and therefore more common variant uses an electronic differential amplifier. There are some circuit variants that do not need to interest us here because the differences have no influence on the wiring. This differential amplifier "calculates" the voltage difference between hot and cold, which eliminates an interference signal that occurs on hot and cold at the same time. The decisive characteristic of the differential amplifier for this is the common-mode rejection. A high common-mode rejection means a high insensitivity to interference signals.

The same applies to connecting an unbalanced output to this input as to the transformer-balanced input. Here the difference between the ground levels of both devices is compensated by the common mode rejection of the differential amplifier. The distortion values ​​of a differential amplifier can also be better than that of a transformer, especially at low frequencies.

With this input, the difference between the ground potentials of both devices may only be a few volts, while with transformer-balanced inputs the difference may be hundreds of volts without causing any problems. In the vast majority of cases, however, you are dealing with differences of less than one volt, so that a differential amplifier is an option.

Now for the output circuits.

1. Transformer balanced output

A transformer is built in analog to the input. In this case, Hot and Cold are connected to the secondary winding of the transformer.

If this is to be used to drive an unbalanced input, then again connect Cold to the unbalanced ground and Hot to the signal input. The shield connection (pin 1) remains unconnected, unless you use a triaxial cable, as described above for transformer-balanced inputs. Alternatively, you can also use a single shielded pair of wires, with hot and cold being connected to the pair of wires. The screen is only connected to pin 1 on the output side.

Here, too, commercial adapters cause problems again.

2. Fully balanced output

That's basically two outputs, one carrying the inverse signal of the other. The normal signal is on Hot and the inverted signal is on Cold. So the difference is twice the normal signal.

For a purely balanced connection, this type of output is actually the best. Unfortunately, this type of output cannot be correctly connected to an unbalanced input. It remains only to connect the chassis ground of the output to the ground of the unbalanced input. The cold output remains unconnected. This is not entirely satisfactory because chassis ground is not a good signal reference for an unbalanced signal.

Ironically, there is an advantage here if the manufacturer mistakenly put the signal ground on pin 1.

Commercial adapters that connect the cold connection to the shield are even dangerous here because they short-circuit the cold output, which can damage it if it is not short-circuit-proof.

3. Cross-coupled balanced output

This circuit tries to emulate the behavior of a transformer a little better. You can connect one output to ground here, then the other output simply produces twice the voltage. This avoids the problem of damage described with the fully balanced output.

The connection to an unbalanced input here goes from cold to ground and from hot to the signal input. Pin 1 is again unconnected, except when using triaxial cable or shielded pair wire.

4. Impedance balanced output

Here is just the cold output with the same impedance terminated like the output resistance of the hot output. So there is actually no signal at the cold connection. When connected to a balanced input, the matched impedance means that noise affects the hot and cold equally, so they correctly cancel out at the receiver.

A connection to unbalanced inputs is again unsatisfactory here. The only option is to connect the chassis ground at pin 1 to the unbalanced ground, with all the disadvantages that this has. Compared to a fully symmetrical output, at least no damage can occur here if cold and shield are connected to each other.

5. Mass compensated output

Here the cold connection is again terminated, as with the impedance-symmetrical output. Here too, there is no signal at the cold output. However, the voltage at the cold pin is used to correct the output voltage at the hot pin. In other words, the cold connection acts like a sensor line, which is used to determine the reference level at the receiver so that it receives the right signal for it.

When connected to a balanced input, the behavior is the same as with the impedance-balanced output. When connecting to an unbalanced input, you connect Cold to the unbalanced ground and Hot to the signal, as with the transformer-balanced output.

So you can see that there is no cabling variant that can be used equally for all balanced output circuits if you want to connect an unbalanced input. This unfortunate state of affairs has probably contributed a great deal to the fact that symmetrical technology has so far not been able to spread appreciably in the hi-fi sector. Some companies that dare to take the step pay for it with a higher effort for customer support.