This blog by Colin Hargis, Control Techniques chief engineer, explains some of the underlying ideas and principles of electromagnetic compatibility (EMC) in AC drives.
It is worth first getting clear the concept of frequency spectra, which is used widely by EMC engineers, but which can cause confusion. One way to help understand EMC effects is to work in terms of the frequency content, or spectrum, of the disturbance. This is mainly because the propagation of radio waves is best understood in terms of the propagation of sinusoidal waves of known frequency. However the most common source of unintentional electromagnetic emission is fast-changing electrical impulses coming from switching operations, such as the power switching in an inverter or switched-mode power supply. It is intuitively obvious that the faster-changing the impulses are, the higher is their frequency content. For regular repetitive pulses the two are linked by the Fourier series, and for non-repetitive pulses by the Fourier transform. You do not have to be a high-powered mathematician to see that if a pulse-train has pulse edges with rise-time of then the corresponding frequency spectrum will contain high levels up to a frequency of at least , and in practice this may be true up to or even more.
The equipment which is most sensitive to disturbance by unintended electromagnetic emission is radio communications equipment. This is because it is intentionally designed to respond to very small signals in the part of the radio spectrum where it works. The international standards which govern limits to electromagnetic emissions are created by an international body, CISPR, whose task is to protect the radio frequency spectrum.
Some other equipment also uses sensitive radio frequency detectors in its operation – for example, capacitive proximity sensors – and they are therefore also inherently sensitive to electromagnetic disturbance.
Most analogue and digital equipment is not intentionally sensitive, but it might malfunction if it is exposed to excessive levels. For example, a simple analogue control input might have a designed bandwidth of only 1 kHz or thereabouts, which is below the radio frequency range. However a very high level of pickup of radio frequency energy might be unintentionally rectified by the analogue circuit and result in a low-frequency error. In a digital circuit, a short voltage pulse induced by an electromagnetic impulse might cause a bit error in an address or data, resulting in a microprocessor crash.
The idea of a ground is important for EMC, but also confusing. Good EMC is best achieved with the help of a physically widespread metallic reference structure, which can be relied on not to have different electrical potentials around it. In a standard industrial electrical panel, this would be the panel backplate. The large area means that high-frequency current can flow in the structure without generating much magnetic field or differential voltage. Usually it is not important for EMC whether this structure is connected to the physical ground (earth), because the length of wire required to connect to ground is so great that its inductance makes it irrelevant for high frequencies. However such structures are usually connected to ground for electrical safety reasons, so engineers tend to refer to them as the “ground” because there is no other single suitable English word. It needs to be understood that it is their local equipotential effect which is beneficial for EMC, not the fact that they are connected to ground.
In many ways the EMC requirements for VSD are no different from other professional electronic equipment. The VSD does however contain a powerful inverter with power switching semiconductors which are intended to generate high and very fast-changing levels of voltage and current. It is this fast-changing voltage and current which tends to generate electromagnetic emission as a by-product, so the VSD is potentially a source of high levels of unintended electromagnetic emission. If wrongly installed a VSD can be a real source of disturbance to other equipment, or “noise”, especially if the other equipment is sensitive (intentionally or unintentionally) to electromagnetic waves.
For example, consider a VSD whose output is a pulse-train in each phase with a repetition frequency of 3 kHz, and pulse rise-times around 100 ns. The frequency spectrum of this pulse-train contains all of the harmonics of 3 kHz. They diminish steadily with increasing harmonic order, but there is still considerable energy content as high as 10 MHz, and unintended emissions have to be considered up to about 60 MHz. (These harmonics should not be confused with power line harmonics, which will be covered in a separate blog.) If you are familiar with radio propagation then you will know that radio waves in the short wave bands between about 5 MHz and 20 MHz can propagate around the entire world – so it is important that we do not release too much unintended electromagnetic energy in these regions.
In order of descending importance, the key connections to a VSD for electromagnetic emission are these:
All of the above is concerned with minimising the unintended emission from the inverter, because that is the feature which sets apart the VSD from other common low-power industrial electronic devices such as instruments, transducers, PLCs, motion controllers and SCADA systems. The VSD also has signal ports carrying analogue and digital information, which must be connected correctly in order to avoid errors caused by unintended sensitivity to incoming electromagnetic energy. The requirements for connecting these ports are no different from other electronic devices, but because of the adjacent powerful inverter it is more likely that failing to follow the instructions will result in disturbance.
The rules are summarised in this diagram:
The connection from the inverter to the motor must be made with screened (shielded) power cable. The screen must be directly attached (clamped or otherwise held) in contact with the grounded metal parts at the inverter, and the metal body of the motor. “Pigtail” connections of the screen should be avoided because the small inductance of the pigtail will be enough to cause a significant noise voltage to appear on the cable screen.
There must be a filter in place. Depending on the application, the internal filter might be sufficient. The filter provided in CT drives reduces the power line emission to a level low enough to avoid interference to most general-purpose electronic equipment. If there are strict limits in place, for example where a strict emission standard must be met, or there is known to be sensitive communications equipment nearby, then an additional filter must be fitted at the power input. The filter must be of a design intended for use with VSD, and the installation instructions must be followed precisely. This means that the filter must be fitted close to the inverter and on the same metallic structure as the inverter and the motor cable screen, in order to ensure a “ground” connection with minimal self-inductance.
The incoming power connections to the filter must be kept separate from the “noisy” inverter and its power output connections.
For general-purpose I/O circuits, primarily 24 V logic and 0-10 V analogue systems, there are generally no special requirements. These circuits are not particularly critical because they are intentionally designed with relatively low bandwidth, so they are unlikely to respond to high-frequency induced disturbing signals. It is usually simplest for the drive 0V control terminal to be connected to ground at the drive, but it must also be connected directly to the corresponding 0V or “ground” terminals of the other items of equipment which need to exchange control data.
Occasionally the system design requires the 0V connection to be isolated from ground. The “floating” 0V gives rejection of any interference caused by differing ground voltages between items of equipment. This is used for example in traditional process control analogue circuits with 4-20 mA signalling. However this approach has to be followed meticulously, and problems often arise where one particular item of equipment cannot tolerate the inevitable noise voltage which will be induced into the 0V connection.
In precision motion control applications it is quite common to provide balanced analogue inputs, i.e. there is a “+” and a “-“ terminal separate from 0V. If correctly used with a twin core screened cable in accordance with the installation instructions these can give excellent noise immunity – the differential input circuit rejects low-frequency disturbances, and the screened cable rejects the high frequencies.
Digital data sources such as shaft encoders are very critical because they have a wide bandwidth, making them susceptible to high-frequency interference. Correctly installed screened twisted-pair cable gives good immunity.
Finally, very high bandwidth data circuits such as Ethernet using unscreened twisted pair (UTP) actually offer very good immunity, largely because of the isolated transformer coupling at each node which avoids circulating noise current and also gives excellent common-mode rejection.
You can see from this section that there is a great variety of control circuit types, and the rules differ between them. If you wish to learn more, I suggest a good textbook at the end of the blog.
A screened cable would appear to be a fairly banal component. Most electronic engineers would regard it as commonplace and not worthy of detailed study. However screened cables are more complex than they might appear. Consider for example the following guidelines which you might read in EMC recommendations:
Guidelines 1. and 4. directly contradict 2. and 3. There are other contradictory guidelines, for example the use of a “pigtail” to connect a cable screen is commonplace in some industries, but forbidden in others.
All of these guidelines have valid rationales in some circumstances, but for a screened cable to do its job in the presence of electromagnetic energy, guidelines 2. and 3. are wrong. The cable screen will only reject both electric and magnetic field induction if both ends are correctly connected. A future blog will explain this further.