What is a frequency converter?

A frequency inverter (VFD) is an electronic device that controls the speed of AC induction motors. Before we look at how this works and how it can be used, we should examine the history of motor controls and also how the induction motor itself works. It has always been useful to control the speed of electric motors used in industry. Almost every process that uses a motor benefits from speed control. Not only is the process generally improved, but in many cases (especially with pumps and fans), this results in significant energy savings. Before electronic controls were available, motors were controlled in various ways, such as by controlling the field current on a DC motor with a series of resistors or by using other motors. However, when thyristors, the first power semiconductors, became available in the 1950s, it became possible to control the armature voltage and thus the speed of a DC motor using phase control. These DC drives are still manufactured today and are widely used. However, the challenge has always been to control the speed of the AC induction motor, also known as an asynchronous motor. While a DC machine typically has two wound parts (the field and armature windings) as well as brushes and a commutator, the AC machine has a simple, fixed winding (the stator) and a rotor. The rotor usually consists of conductors formed by casting aluminum or copper in the iron core. There are no brushes or commutators. The machine is therefore cheaper, simpler, and more reliable. It is not surprising that these machines make up the majority of motors used in the industries of the world. So how does it work and why does it need a frequency inverter? Let's start by looking at a three-phase transformer, as shown in Figure 1. If the transformer winding is connected to a three-phase AC supply, a sinusoidal current flows in the primary windings. The current causes a magnetic flux to be induced in the iron core of the transformer, which rises and falls as the applied voltage (and thus the current) alternates, usually at 50 or 60 Hz, depending on the power grid. The changing magnetic flux then induces a voltage in the secondary windings, and when a load is connected (or even if the windings are short-circuited), a current flows. The ratio of the turns of the primary and secondary windings determines the ratio of the primary voltage and the secondary voltage, which is why transformers are so useful. Now imagine we wind the windings, create a small air gap between them, and let the secondary winding, now called the rotor, move freely. This is the basis of the induction motor, as shown in Figure 2. If we connect a three-phase supply to the primary winding - now called the stator - we have transformer action as before, and current flows in the rotor windings (secondary windings). As mentioned above, the rotor usually consists of cast conductors within an iron core with a short-circuit ring at each end. Since this arrangement looks a bit like a circular cage (ignoring the iron, of course!), the motor is sometimes called a squirrel-cage motor. In Figure 2, the conductors are perpendicular to the diagram and the short-circuit rings are not shown. Figure 3 shows a typical motor construction. If we now have a magnetic field and an electric current, we get a force, and according to Fleming's Left Hand Rule, this will rotate the rotor, so we have a motor. However, as the motor accelerates, it begins to "catch up" with the magnetic field, which effectively - alternately - rotates around the stator at the frequency of the three-phase supply. Now we will only get transformer action as long as the magnetic field keeps changing; transformers only work with alternating current. So when the rotor catches up with the supply, there is no more changing magnetic field, so no transformer action, no rotor current, and no torque. Therefore, a standard induction motor always runs slightly slower than the applied frequency. This speed reduction is called slip. When an induction motor is loaded, the slip increases slightly, more current is drawn, and the motor takes on the load. So the speed of the motor is basically dependent on the applied frequency. In a simple induction motor, this speed is typically a few percent lower than the synchronous speed (where no torque is available). By doubling (or tripling, etc.) the number of windings or pole pairs, this speed can be reduced. A motor with one pole pair (a two-pole motor) operated with a 50 (60) Hz supply will therefore operate at 48 (58) revolutions per second or 2880 (3480) rpm. A four-pole machine, the most common, therefore runs at 1440 (1740) rpm. Six- and eight-pole motors are readily available, with special applications requiring more poles. Figure 4 shows the classic torque/speed relationship of an AC induction motor. So if we want to control the speed of the motor, we need to vary the applied frequency. However, if we manage to vary the frequency, we also need to consider the voltage, as the magnetizing current in the stator depends on the integral of the voltage over time. That is, the area under the sine wave curve. If we decrease the frequency, the period or length of the sine wave increases, so does the area underneath it, leading to excessive magnetizing current in the motor. So if we reduce the frequency, we also need to proportionally reduce the voltage applied to the motor. How we do this electronically with a frequency inverter will be discussed in the next article. In the last article, we saw that an AC induction motor runs at a speed dependent on the applied frequency, with a slight reduction in speed known as slip. To control the motor speed, we need to vary this frequency and also control the applied voltage to maintain the optimal flux or magnetic field. Almost all frequency inverters operate on the basis that they take the existing AC supply, convert it to DC with a rectifier, and then convert it back to a variable frequency supply with an inverter. The inverter is the key part of this, so a frequency inverter is sometimes simply called an inverter. Inverters and to a lesser extent rectifiers rely on modern power semiconduct

How do they actually work?

Let's start with the rectifier. Figure 5 shows a three-phase rectifier consisting of six diodes connected as a simple load with a capacitor and a resistor.

Fig. 5: Three-phase rectifier with capacitor and resistive load

The diodes conduct only in one direction (in the direction they point); the capacitor stores energy somewhat like a battery, the resistor acts as a load. When we connect a three-phase supply to the inputs on the left side, we start pumping current into the capacitor, and the voltage builds up, leading to a current flow in the resistor. You can trace the conduction path from any phase to another through an upper diode, the capacitor/resistor, and then through a lower diode.

In the steady state, the voltage across the capacitor sits fairly close to the peak of the input sinusoidal voltage. Now, the diodes only conduct when the input voltage is higher than the voltage across the capacitor. Consequently, there is a short current pulse through each diode in succession, resulting in the characteristic "Twin Peaks" waveform of the current in each of the three phases, as shown in the figure. If we use a single-phase supply, we only need four diodes, and we get a single peak per half cycle, so we need a larger capacitor to fill in the gaps in the voltage peaks.

In single or three-phase systems, these current pulses can have consequences for the supply, as we will see later. However, we now have a relatively smooth DC voltage across the capacitor. If we remove the resistor and instead connect an inverter, it starts to look like a frequency converter (Figure 6).

Fig. 6: Frequency converter power range

Now onto the business part. We have six Insulated Gate Bipolar Transistors (IGBTs). These act as very fast power switches. They have diodes parallel to them for reasons that will become clear.

We can now turn on an upper and lower IGBT and provide a current path across any two motor terminals. To turn on an IGBT, we simply apply a few volts to the gate (shown unconnected here). Then the IGBT conducts in the arrow direction. Depending on which IGBTs we turn on, we can generate a positive or negative current path through the motor. Therefore, we can convert AC to DC.

What we don't do is turn on an upper and lower IGBT that are directly on top of each other, as this would cause a short circuit to the intermediate circuit. Instead, by turning on and off the IGBTs in a carefully controlled sequence, we can build up a three-phase current in the motor windings. By varying the time for which we turn the IGBTs on and off, we can control this current. This is because the motor current does not change very quickly, so by increasing and decreasing the on times, we can build a sinusoidal current in the motor at practically any desired frequency. This is, of course, what we want to do to control the motor speed. Using the same technique, we can control the effective voltage, which in turn controls the magnetic field.

This on-time control is called Pulse Width Modulation (PWM) and is simplified in Figure 7. Turning on and off the six IGBTs provides current paths to the motor and allows a three-phase sinusoidal current to flow, rotating the motor at the desired speed.

Fig. 7: Pulse Width Modulation

By switching our IGBTs several thousand times per second (usually between 4 and 16 kHz), we can build up a rather nice current waveform, as shown in Figure 8.

Fig. 8: Voltage and current at the motor

Note that the output voltage consists of many pulses and not a nice sinusoidal wave. The motor smooths the current to a slightly jagged sinusoidal wave, but the voltage still consists of the PWM waveform of the IGBTs. This can lead to problems, which we will address later. However, the motor is happy with the jagged current and rotates at the required speed. The motor current is phase-shifted with the "average" voltage due to the motor power factor.

So what do the diodes do in the inverter? Well, the current in the motor does not change very quickly, so if we turn off an IGBT, the current must continue to flow, or there will be issues. The diodes automatically provide this current path by turning on or commutating the current. Hence the name commutation or freewheeling diodes.

Inverters are very difficult to control – they have been described as a short circuit waiting to happen, but modern power semiconductors are quite robust, and fast, powerful digital signal processors enable reliable and precise control of turning on and off.

By the way, the power part of the frequency converter is fully connected to the AC power supply and operates with DC voltages from 300 V (with a 230 V AC input) up to 600 - 900 V with industrial three-phase power supplies. Therefore, the internal isolation between the control circuits, customer interfaces, and the power part is safety-critical.