Differential .v. single-ended transmission
This page discusses two concepts in analogue signal transmission: “single-ended” and “differential”. Analogue signals are used to convey the output of an analogue instrument to a digitiser. While digital signals are relatively tolerant of interference, analogue signals can be disrupted and altered by electromagnetic waves in the environment. This document explains the problem and describes a solution. It then provides a brief introduction to twisted-pair cabling before discussing interfacing between Güralp differential equipment and non-Güralp single-ended equipment.
James Clerk Maxwell's equations show how magnetic and electric fields are related. Michael Faraday used them to show that, if the flux lines of a changing magnetic field pass through a coil of wire, a current is forced to flow around the coil. This process is called induction and we say that the resulting current has been induced by the changing magnetic field. The current is proportional to the number of turns of the coil, the strength of the field and the rate of change of the field.
It is very easy, when connecting electrical equipment, to inadvertently create something that functions as a coil of wire, even though it typically has only one turn. Consider the diagram below, which shows a piece of electrical equipment, A, sending a signal to another piece of equipment, B, along a shielded cable. Both devices share a common ground point and the outer shield of the signal wire between them is also grounded to the chassis at each end, as indicated below by the small green wires in the diagram:
Note that A is transmitting a voltage, relative to ground, that represents the wanted signal. B is measuring this voltage, relative to ground, in order to recover the signal. It is clearly important that both devices share a common ground. One might naively suppose that the shield of the cable and the two connections to a common earth point would ensure this.
However: the shield of the signal wire, the chassis of each piece of equipment and the two earth lines form a coil of wire with a single turn. The area of this coil is shown hatched below:
If any lines of magnetic flux were to pass through this area, as shown below, a current would be induced to flow around this loop, in accordance with the Maxwell-Faraday equation. Because the shield of the signal wire has non-zero resistance, Ohm's law says that there must necessarily be a voltage across it. The two pieces of equipment no longer share a common ground potential and the voltage appearing across the shield will directly affect the signal received at B.
This is the basis of most electromagnetic interference. When the "coil" involves different cables and different pieces of equipment, all of which should be at ground potential, it is known as a ground loop. System designers must pay particular attention to avoiding ground loops. As we shall see later, however, the "coil" does not need to be particularly large for this effect to occur: the gap between two adjacent cores of a single cable can be enough when there are strong electromagnetic fields, such as during a thunderstorm.
In a single-ended transmission system, an analogue voltage, relative to the transmitter's ground, is used to represent the signal to be transmitted. This voltage, , is fed to a line-driver, , which is a type of amplifier. The amplified signal, , is applied to the wire. At the other end of the wire, the voltage (relative to the receiver's ground) is sensed using a receiver, , which is another type of amplifier. The receiver's output, , feeds the subsequent circuitry, which we will consider to be a digitiser for the purpose of this discussion (although the technique is used in many fields).
In an ideal world, this arrangement would be sufficient but, in an environment where unwanted electromagnetic noise is present, this noise can interfere with the transmitted signal and appear at the receiver. Consider the example below, where a spike (an unwanted voltage transient) has been induced onto the wire. It is clear that nothing stops this spike from being passed straight to the receiver and on to the digitiser . The digitiser has no way of telling the difference between the instrument's output and the induced noise.
One solution to this problem is to transmit two copies of the signal using two separate wires. In the diagram below, the line-driver has two outputs, a non-inverting output and an inverting output . The non-inverting output transmits the signal just as in the single-ended scheme and is normally identified with a '+' sign (although it still carries the entire signal waveform: both the positive and the negative portions). The inverting output transmits a negated or inverted copy of the signal and is normally identified with a '-' sign (although, again, it carries both negative and positive portions of the signal waveform) and/or a small circle on the schematic symbol.
These two outputs are each fed to separate wire cores in the transmission cable. The receiver has two inputs: again, one is non-inverting and the other is inverting. The non-inverting input is normally identified with a '+' sign while the inverting input is normally identified with a '-' sign and/or a small circle on the schematic symbol. The receiver effectively subtracts the two signals from each other.
We can show how this helps reduce interference by using some simple mathematics. If the signal, as a function of time, is represented by S(t), the line-driver sends S(t) on the non-inverting core and and −S(t) on the inverting core. The receiver subtracts −S(t) from S(t):
so the original signal, , is recovered.
It appears doubled in amplitude, so the apparent sensitivity of the instrument is doubled. This is why the published sensitivities of traditional Güralp instruments are given in a format like 2×1000 V/ms⁻¹. The sensitivity of an instrument described like that would be 1000 V/ms⁻¹ when used with digitisers with single-ended inputs but 2000 V/ms⁻¹ when used with Güralp digitisers or any other digitisers with differential inputs.
Now consider what happens if electromagnetic noise is induced in the cable, represented below by the spike at and . If we consider the noise as a function of time, we can represent it as N(t). The noise is added to the signal on both cores of the cable so the non-inverting core now carries S(t)+N(t) and the inverting core carries −S(t)+N(t). When these are subtracted in the receiver:
and it can be seen that the noise is eliminated from the output at :
Twisted pair wiring
For differential transmission to work perfectly, it is important that both cores of the cable pick up identical noise. This is physically impossible because some magnetic field lines will inevitably pass between the wires, inducing electric currents which add together to produce a noisy signal at the receiver:
This situation can be improved significantly by twisting the two cores together within the outer cable sheath. Any field lines passing between the wires now cause opposing currents in alternate twists:
Because the induced currents are of equal magnitude but opposite direction, they cancel each other out, causing significantly reduced noise at the receiver:
Cables made in this fashion are called “twisted-pair” cables. If several pairs co-exist in the same cable, they are typically twisted with different pitches. (The pitch of a twisted pair cable is one half multiplied by the number of times the two wires cross each other within a given distance - normally one metre.) Using different pitches reduces the amount of signal which is inductively coupled from each pair to its adjacent pairs.
The twisted pairs can be enclosed by a wrapping of foil or a woven wire mesh, which provides additional protection from interference. This is called the screen or shield. Such cables are classed as “STP” or “Shielded Twisted Pair” cables. A single screen can surround all of the pairs or each pair can have its own screen. Twisted pair cables without a screen are classed as “UTP” or “Unshielded Twisted Pair” cables.
UTP: Unshielded twisted-pair cable with four pairs
STP: Over-all foil-screened twisted-pair cable with four pairs and a separate ground wire
When using shielded twisted pair cabling between two pieces of equipment which have independent ground connections, the screen should be connected to ground at one end only; by convention, this is the source end. This prevents the creation of a ground loop.
Voltages, values and specifications
Most Güralp instruments generate signals which vary between positive 10 V and negative 10 V with respect to signal ground. When transmitted differentially, this is normally described as “±10 V differential”.
When the signal on the non-inverting wire is +10 V, the signal on the inverting wire is −10 V and the voltage difference between the two wires is thus +20 V. Conversely, when the signal on the non-inverting wire is −10 V, the signal on the inverting wire is +10 V and the voltage difference between the two wires is then −20 V. The difference between +20 V and −20 V is 40 V so some sources refer to the exact same transmission arrangement as “±10 V differential”, “±20 V peak-to-peak differential” or “40 V peak-to-peak differential”.
Using single-ended equipment
Although Güralp instruments have differential outputs, they can still be used with other manufacturers' digitisers which have single-ended inputs. Similarly, other manufacturers' instruments having single-ended outputs can be used with Güralp digitisers which have differential inputs, although the situation is a little more complicated.
To use a Güralp instrument with a digitiser with single-ended inputs, simply ignore the inverting output. Connect the instrument's non-inverting output to the digitiser's signal input and the instrument's signal ground to the digitiser's signal ground . Do not connect the inverting output to anything.
To use a non-Güralp instrument with single-ended outputs with a Güralp digitiser, connect the instrument's output to the non-inverting input of the digitiser and connect the instrument's signal ground to the digitiser's signal ground . Do not connect anything to the digitiser's inverting input .
The diagram below shows the recommended connections between a Güralp 6T, which has differential velocity outputs and single-ended mass-position outputs, to a generic six-channel digitiser with single-ended inputs.
The following diagram shows the connection of three generic uniaxial geophones to a Güralp digitiser. (In practice, the use of pre-amplifiers, local to each instrument, with programmable input impedance and differential outputs would provide significantly better performance.)