Measurement of the output voltage with variable speed motor drives

Variable speed motor drives, also known as inverter drives, are increasingly used in industrial installations. At the time of installation and maintenance, the measurement of the output voltages often yields unexpected results. Why is this, and how can you prevent it?


Variable speed motor drives

In conventional electrical machines, which are directly connected to a one- or three-phase power supply system, the speed can only be regulated to a very limited extent, if at all. A possible solution is to use an external gearbox. This, however, is generally bulky, very noisy, costly and susceptible to wear. The use of new semiconductor components, which are designed for high currents and voltages, has cleared the path for the development of so-called motor drive controllers with adjustable speed, or “inverter drives”. These devices allow a speed control within a large range and also offer low electrical losses and a constant torque that is independent of the actual speed of the machine As a result, variable speed motor drives are establishing themselves more and more in industrial installations, where among other things they provide the following benefits:

  • No wear, because asynchronous machines are used;
  • Effective control;
  • High efficiency.

At the time of installation and maintenance, the measurement of the output voltages often yields unexpected results. The following explains how this can happen and how you can use a measuring instrument to achieve correct measurement results.


Generation of the variable output frequency

There are several ways to generate the variable output frequency. The first models were called self-regulated or machine-synchronised thyristor converters. Today, they are still used in inverter drives for high performance. However, better alternatives have been developed when less power is required.

Thyristors can only be turned off at the zero point of the mains supply. This is why the output voltage of these inverters is not a continuous sine wave, but is always with phase cutting (such as shown in Fig. 1.) A change of phase angle of this phase cut controls the output power, which can change the speed of the machine and at the same time, can reduce the available mechanical performance. Unfortunately, these inverters do not allow any desired modulation of the output signal waveform.

Even using additional circuits, this problem can only be partially solved, and at the same time, this results in high costs. The emergence of gated power semiconductors has made a completely new approach to speed control possible. These semiconductors can be switched on and off, making them well suited for “interruption” in a DC power system. Fig. 2 shows the basic structure of such a drive.

Figure 1: Output voltage of the thyristor converter. The unfiltered output voltage shows substantial phase cuts.


The one- or three-phase power input is connected to a series of rectifiers, which feed into an internal

DC intermediate circuit. The DC voltage is buffered in a large sized storage capacitor charged to a voltage Ub: Ub = Ö2 * U mains »  1.41 * U mains


Figure 3: Output voltage for each output line


The DC voltage is then connected to a series of double-sided switches that alternately connect each of the three machine connections with the positive or negative bus line. In addition, each branch of the switch can be set to an inactive (i.e. non-conductive) state, thereby floating the corresponding machine connection.

The switches are all controlled from a central control unit, which generates the drive pulses with which each of the six switches is actuated at the right time. The switching speed is variable and this determines the output frequency. The order in which the three outputs are actuated is determined by the direction of rotation of the machine. The control unit is set so that the output frequency can be varied over a wide range. Since the speed of the machine directly depends on the supply frequency, it can be efficiently controlled. Figure 3 shows the resulting output voltages for the individual output lines. On each machine, there is a positive pulse, which is the period in which the connector is de-energised, then a negative pulse and once more a period in which there is no drive voltage.

In this simple example, the no-load output voltage at each of the outputs is either +1/2 Ub or zero (potential-free) or -1/2 Ub, where Ub is the bus voltage. Since all three outputs are connected in the same way, the mean value for each output is half of the DC bus voltage.

If the above waveform were applied to a low-pass filter, the output would be similar to a sine wave that has the same fundamental frequency as the square-wave voltage of the control circuit (see Figure 4). Low-pass filters, which are suitable for the power levels encountered in motor drives, would however be large and expensive: alternatives have therefore been developed.


Figure 2: Fundamental design of a variable speed motor drive


Alternatives to low-pass filters

An alternative to low-pass filters is the result of a further improvement in power electronics. In actual used systems, the positive and negative pulses are not normally the result of the generation of a single pulse with the desired polarity. Instead, all pulses are generated by switching on and off the same semiconductor switch repeatedly at a much higher pulse rate and a varying pulse duty factor (see Figure 5).


Figure 4: Output voltage — direct, and via a low-pass filter

The trick now is to vary the pulse duty factor so that the current (but not the voltage) has a sine wave through the machine winding. The induction of the machine windings then acts as a low-pass filter through which a sinusoidal current flows, due to the pulse-width modulated voltage.


In Figure 5, the upper curve shows the output voltage for only one output line, thereby clearly indicating the effect of the variation of the pulse duty factor. The lower curve shows the effective output voltage per internal clock cycle T on a relative scale

This indicates that the effective output voltage is sinusoidal. However, the actual output voltage of the motor drive is much more similar to the upper curve! In contrast to the above-mentioned thyristor circuits, the drive circuits can now be used consistently in switching operation. Energy losses in the semiconductor switches are therefore minimal, resulting in high efficiency and low heat generation in the drive module.


Voltage measurements

Although the improvements in efficiency and the possibilities of speed control in these motor drives are obvious, a problem remains for the installation and service personnel. The output voltage of the motor drive should generate a sinusoidal current via an inductive load, but the applied voltage has a completely different waveform. Direct measurement of the output voltage can therefore lead to unexpected results, because the measuring device, used as a multimeter, is designed for mains frequencies, i.e. 50 or 60 Hz, in contrast to a sinusoidal curve. The output voltage of the variable speed drive is, however, a square wave with high frequency and constantly changing duty cycle. The peak amplitude of the square wave is immutable for this. Furthermore, two polarities must also be taken into account! Most multimeters respond to the applied peak or peak-to-peak voltage or the mean value of the voltage; they are therefore calibrated to show the RMS value of the amplitude of the sine wave. Additionally, most multimeters use a two-phase rectifier for AC voltage measurements at their input to ensure that voltages with both polarities contribute to the reading to the same extent.

If we look at the output voltage in Figure 5, the average voltage per cycle T (after rectification) is directly proportional to the duty cycle of the waveform and the bus DC voltage, and therefore changes continuously due to the varying duty cycle. Within a half cycle of the output current, the average voltage is then: Uaverage = d * Upeak = d * (1/2 Ub) where: d = duty cycle, which switches from 0 to 100% and back again.

The resulting measured value in volts can differ considerably from the expected value at the machine terminals (e.g. it is shown in the display of the motor drive itself, which shows the effective output voltage calculated from the internal control electronics). To illustrate the possibilities of incorrect measurement values, we have tested a range of different brands and models of multimeters, all under exactly the same conditions, with the same motor drive and the same settings. The measured values (see Table 1) vary between 143 V and 1000 V.


Achieve correct measurement results!

To calculate the output voltage in this special situation, the specific application of the motor drive must be taken into account.


Figure 5: Output voltages for pulse-width-modulated motor drives.


The driving power for electrical machines is generated by the current flowing through the machine windings, while in principle, the applied voltage is only required to allow this current to flow. Variable speed motor drives exploit this fact, in which a high-frequency, non-linear voltage is applied. This produces a sinusoidal current in the machine windings, whose frequency is determined by the control unit and which has the polarity of the switching voltage.

Therefore, to calculate the effective output voltage, only the fundamental components of the applied voltage need to be taken into account. In order to achieve this, a large number of samples of the applied voltage is taken and a detailed image of the voltage signal waveform is stored in the digital memory of the measuring device; the fundamental component can then be calculated and displayed based on this.

The screenshot in Figure 6 shows both the peak-to-peak amplitude as well as the effective output voltage of a motor drive in the boxes at the top of the screen.

The effective output frequency of the displayed signal waveform can easily be determined here: a single cycle takes about 6.3 divisions, the time axis division corresponds to 5 milliseconds, so a single cycle takes about 31.5 milliseconds. The output frequency is then 1/31.5 milliseconds = 32 Hz

Alternatively, the cursors can be used to highlight a cycle of this output signal waveform and the frequency can then be read off directly.

This is exactly what the last three measuring devices listed in Table 1 do. This includes the Fluke ScopeMeter from the 190B and 190C series, in which all incoming voltages are digitised at a high sample rate and a digital image of the waveform is stored for further analysis.


Digital multimeter models

Measured value


1 10.01V
2 154.2V
3 157.6V
4 170.1V
5 187.1V
6 193.6V
7 204.3V
8 215.3V
9 237.93V
10 254V
Fluke 41B 143V
Fluke 43B 143.3V
Fluke Series 190 144V

Table 1: Alternating current measurements using various digital multimeters


The ScopeMeters from the 190 Series are fitted with a specific pulse-width-modulation-voltage measurement function for such applications. This feature allows the ScopeMeter 190 Series to analyse the digitised signal and to calculate the fundamental component frequency. This has the same signal waveform as the output current of the motor drive. The RMS value is calculated based on this signal waveform and displayed as a pulse-width-modulation-voltage value.


Figure 6: Output of a motor drive



Variable speed motor drives provide a number of advantages to designers and users of machines. However, measurement of the output voltages and frequencies may complicate the work of the maintenance technician and the installer of the machine drive. Only measuring devices specifically designed for the measurement of these output voltages provide reliable readings that agree with the (calculated) display on the motor itself.