A later blog will discuss the screened power cable recommended for connecting the AC VSD to its motor. In both cases the purpose of the screen is to prevent unwanted electromagnetic coupling between the circuit inside the screen and other circuits outside. A key difference is that the motor cable screen is there to protect external circuits, whereas the signal cable screen is to protect the circuit inside it from interference by electrical noise outside it.
Screened cables are commonplace in electronic systems and are generally taken for granted. However they are not as simple as they seem, and are frequently misused and misunderstood. Fortunately modern electronic circuits have generally good immunity to electrical noise, so systems usually work despite bad practice in cable management. However when using variable speed drives it becomes rather more important to use correct practice because the inverter generates quite high levels of electromagnetic noise, which can cause disturbance to the associated control circuits if they are not properly arranged.
Part 1 looks at general principles of screened signal cables, and part 2 looks at some more specific practical details.
Some common questions
There are various rules promoted for the management of screened cables which have arisen for good reason, but they can be conflicting and confusing. Here are some common questions which I hope to answer:
- Must I connect both ends of the screen?
- Must I not connect both ends of the screen??
- Must the screen be connected to ground (earth)?
- Must I worry about ground (earth) loops?
- How long can the ground pigtail be?
- How do I connect through a terminal block?
- How do I connect a balanced (differential) analogue circuit?
- What about Ethernet? Can unscreened cable work?
A few critical terms
In the following explanation:
The ground is the safety ground or earth (PE) in a mains-connected system, which is ultimately connected to the protective bonding network of the building and to physical ground (earth) below. When signal circuits are connected to the ground and the connection is not made for safety reasons, this may be referred to as the functional ground, as distinct from the safety ground.
The signal return or common or reference connection in a system is referred to here as the “reference pole”. In Control Techniques equipment this is referred to as the “0V” connection. This is often connected to ground, but it need not be. Some balanced data circuits may not have a reference pole.
In an electrical panel, the main mass of the metal construction is referred to as the “chassis”. This is usually connected to ground for safety reasons, but for electrical noise considerations it is more important that it comprises a widespread conducting surface which is unlikely to have different electrical potentials around it.
In a balanced or push-pull signal circuit the signal lines are referred to as A+ and A-. Depending on the design, there may or may not be an associated 0V or chassis connection.
“High frequency” means broadly a frequency in the radio communication range, well above the cable cutoff frequency, e.g. above about 50 kHz or thereabouts. In variable speed drives such high frequencies occur as a side effect of the very fast switching of the power semiconductors.
What is electrical noise (interference?)
Electrical noise here refers to the effect of unwanted interaction of electrical circuits. All electrical activity results in electromagnetic fields which can induce unwanted electrical signals into nearby circuits. Generally the effects tend to be worst for frequencies in the radio range, because the rapid change of voltage and current enhances the unwanted coupling. Signal circuits may be sensitive to high frequency interference, either because they use high frequencies themselves (e.g. serial digital data links, encoder data) or because they have unintended sensitivity to high frequencies well beyond their intended bandwidth (e.g. analogue inputs). A well-designed signal circuit will have its bandwidth tailored to the requirements of the application, so that it is not unnecessarily sensitive to fast-changing disturbances. However high levels of disturbance outside the intended band can still cause errors because of non-linearity. This is why for example it is quite common to hear interference in a sound system caused by a mobile (cell) phone.
An important feature of this kind of noise is that it can cover an extremely wide range of frequencies. Interference can occur from mains frequency sources of 50/60 Hz right up to mobile phone and other radio frequency regions of around 2 – 5 GHz. This is a range of 8 orders of magnitude, and rules which work well for some frequencies may be ineffective or even counterproductive at others. This is why rules for EMC and for screened cable management can sometimes appear to be conflicting – they may have been intended for threats in specific frequency ranges.
Note that another kind of electrical noise is the thermally generated random noise which exists inherently in all circuits at temperatures above 0 K. This is only of interest for highly sensitive radio receiving equipment and is not covered here.
How screened cables work
The screened cable has one or more signal cores surrounded by a continuous screening conductor. A coaxial cable has a single inner core surrounded by a screen and is mostly used for radio frequency applications. The screen is most commonly made from a braid of fine wires, which may be supplemented by a conductive foil. Less commonly, the screen may be solid metal, and may include magnetic material such as ferrite.
The purpose of the screen is to prevent external electromagnetic energy from inducing an unwanted signal into the signal circuit. An electromagnetic field comprises related magnetic and electric fields together. To help in understanding its operation, we can first consider separately the effect on electric fields and magnetic fields. For a circuit to be immune to electromagnetic interference it must be immune to both electric and magnetic fields.
Electric field screening
This is the simplest mechanism to understand. Figure 1 shows an electric field E from an outside noise source impinging on a screened cable in a simple single-line (unbalanced) signal circuit connecting a signal source to a signal load. The field terminates on the screen conductor and does not penetrate to the inner conductor, so no interference occurs.
In the absence of the screen, the electric field would induce current into the signal circuit whenever it changed. This would cause a transient error, i.e. noise, in the received voltage, by an amount depending on the impedance of the circuit – the higher the impedance, the higher the error. Normally the source is designed to have a low impedance in order to minimise the error voltage caused by electric field ingress.
Figure 1: Electric field screening mechanism
The ground connection is shown as optional in Figure 1, because in principle it is not needed in order for the screen to work. What is essential is that the reference poles of the source and the load must be connected to the screen so that the signal voltage exists on the internal conductor relative to the screen.
In practice, depending on the design of the source and load, they might not be able to tolerate electrical potentials on their reference poles, so it is common practice to connect the screen to ground. Notice that there is only a single connection to ground in Figure 1, and for simple electric field screening it does not matter where the connection is made. However when the field E is varying in time, a current flows to ground because of the changing electric charge. Once a current flows we must also consider the magnetic field effects. As the frequency increases the current associated with an electric field also increases, so the arrangement of Figure 1 is really only successful in excluding low-frequency electric field interference such as from 50/60 Hz mains.
Magnetic field screening
The magnetic field screening effect of a screened cable is a little more difficult to understand, but equally important. Wherever electric currents flow there are associated magnetic fields which can induce electrical potentials into circuits when they change. Figure 2 shows a magnetic flux B, originating from an external current-carrying circuit, linking the same circuit as in Figure 1.
Figure 2: Magnetic field screening mechanism
When the magnetic field changes it induces a potential into the conductor which is proportional to the rate of change of magnetic flux linked by the conductor, shown here as EB1 for the screen and EB2 for the inner conductor.
The induced potential would represent a transient error in the received signal, i.e. noise, except for the fact which is illustrated in Figure 2:
The exact same voltage is induced into both the inner and outer (screen) conductors. So EB1 = EB2.
The reason for this is that the magnetic flux which links the screen conductor has inherently also to link the inner conductor.
The voltages EB1 and EB2 shown in red are equal but opposite in the signal circuit, so they cancel in the load.
Provided nothing happens to unbalance the two induced voltages, the cancellation is very exact and the screened cable gives excellent protection from changing magnetic fields.
Notice that in Figure 2 neither the source nor the load are connected to any other circuit, i.e. they are galvanically isolated. In this case no current can flow in the screen and there is nothing which could cause an error between EB1 and EB2.
In practice even with galvanic isolation there is stray capacitance so that some current can flow at the higher frequencies. However any current which flows in the screen causes a change in the magnetic flux which also links the signal conductor. The cancellation mechanism still operates.
Magnetic field screening
In Figure 2, it is shown that the voltage induced by the external magnetic field is identical in both the inner and outer conductors. Another source of voltage which is not induced equally is simple resistive voltage drop. Figure 3 illustrates a situation where the sending and receiving ends both have connections to their local chassis or ground, and a ground difference voltage ED causes a current ID to flow in the screen. The difference voltage might be caused by a variety of effects in the complete system, essentially it is the sum of the various noise voltages which are gathered by the cable screen acting like a receiving antenna for electromagnetic waves of all kinds, as well as voltage drops caused by circulating stray currents such as at mains frequency.
There is also a particular source of ground difference voltage in drive systems using a motor shaft encoder. Despite the use of screened motor cable, the motor body may have a significant noise voltage relative to ground because of the fast PWM pulses in the motor winding and motor cable. If the shaft encoder has a metal body fixed directly to the motor body then it is difficult to avoid a ground differential voltage in the encoder cable screen.
Figure 3: Effect of screen current
The current ID causes a voltage drop in the screen, with two components:
- An inductive component IjwL where L is the cable screen inductance
- A resistive component IR where R is the cable screen resistance
The inductive component is caused by the magnetic field induced by the current. The magnetic field also links the inner conductor, so it contributes equally to EB1 and EB2 and it does not disturb the received signal.[1]
The resistive component does not appear in EB2, so it appears in series with the signal and causes an error.
Notice that whereas an electric field would cause an induced current, so that the effect would be proportional to the impedance of the circuit, here there is an induced voltage. Making the impedance of the signal source lower does not reduce the error. When magnetic field induction is the main source of interference the best technique is to use a current signal, and this is the reason for the widespread use of the 4 – 20 mA current source method in process control systems with very long signal cable runs.
At high frequencies where the inductance of the cable dominates its impedance, IR is relatively small. Also, because of skin effect, the effective resistance is less at high frequency since the current flows mainly in the outside of the screen, not the inside. The upshot of this is that at lower frequencies the cable screen becomes less effective. This can be measured as a screen cutoff frequency, below which it is ineffective. It tends to be in the range of 1 kHz to 10 kHz for commonly used cables [see page 62 of the reference for example].
Figure 3 also highlights the effect of “pigtails”, i.e. lengths of wire used to make screen return connections. You can see that the current ID flows in the pigtails, and any voltage drop there in the pigtail inductance appears in series with the signal. The point here is that this is an inductive voltage drop which does not appear in both conductors, so it is not cancelled by the screened cable. The pigtail is harmful to the screening capability of the cable at higher frequencies.
Types of cable screen
The traditional cable screen is a braid of fine copper wires, with a coverage approaching 100% (i.e. minimal “windows” in the braiding). Some data cables use a metal foil or a metallised plastic foil, either alone or with a braid.
To be effective over a wide range of frequencies the screen needs to have maximum coverage, low resistance, and good lengthways continuity between braids so that current can flow along the outside with minimum voltage drop and minimal mixing with the current on the inside. Foil alone tends to have a rather high resistance and is not effective, but when combined with braid it can help in separating the inner and outer conducting surfaces.
Reference
Henry W Ott: Electromagnetic compatibility engineering: Wiley: ISBN 978-0-470-18930-6
Another recommended book
Tim Williams and Keith Armstrong: EMC for Systems & Installations: Newnes: ISBN 9780750641678
[1] It takes some thought to understand this properly. All of the magnetic field caused by the screen current has to link the inner conductor. Not all of the magnetic field caused by inner conductor current has to link the screen.
Browse all Technical blog posts