Tuesday, November 17, 2020


Variable speed drives have always raised challenges for the condition monitoring personnel due to its varying speed and load conditions and the necessity of fixing the operating parameters before data collection or use the techniques such as order tracking etc. to diagnose the faults. Here we are discussing an interesting case, where an abnormal sound, similar to a damaged bearing or rub sound, arose from the motor during the commissioning run, after being a standby for more than 6 months. Different tests were performed before concluding the results, which are explained below. The motor was in continuous service until stopped a few months back due to the decline in demand.

The subject motor drives a single stage radial flow blower.


Abnormal sound noticed from the motor when operated in the minimum speed range during low load. Vibration spectrum collected from motor showed raised noise floor, around and above 270 Hz up to 3500 Hz, the majority of the peaks observed above 2000 Hz range.

Pic.1 Vibration spectrum

Run-up test was done to identify the changes in peaks. The speed was increased from 3150 rpm to 3490 rpm. It can be seen that some of the noise below 1000 Hz being reduced with increase in speed.

Pic.2 Run-up test cascade plot

The coast-down test was performed to rule out the electrical noise from the vibrations.

Pic.3 Coast-down cascade plot

During the coast-down or ramp down, it was observed that the peaks above 2000 Hz immediately vanished. A detailed study was done to identify the cause of these peaks. For understanding that the working of VFD is briefly described below.

Working of a VFD

VFD module consists of input diode rectifiers, DC bus and IGBT inverter. The AC current supplied to the VFD module is converted into DC current by diodes, which is smoothened out by the capacitor of DC bus. The smoothened DC current is converted into the AC current of required frequency by pulse width modulation, achieved by using the IGBT.

Here, I am not going to discuss in detail about the working of input and DC bus sections. Our concern is regarding the output current and what component of this current could be leading to the vibrations in the motor. To understand this, let us discuss how IGBT works.

Pic.4 VFD module

As mentioned, IGBT or Insulated Gate Bipolar Transistors varies the frequency by pulse width modulation. The pulses are developed by simply “opening” and “closing” the circuit, just like by using switches. Similar to normal switch operation, when the switch is closed, the current flows, when the switch is open, the current flow stops. Now, to control the width of the pulse, the timing of each pulse have to be varied. When the pulse width is more, the voltage output is more and vice versa. Hence, to obtain a sine wave output, the pulse width at the centre of each half-cycle will be more than that at the ends. The timing of each pulse is varied by controlling the opening and closing of transistors.

If you remember the basics of the transistor, which consists of collector, emitter and gate terminals, where the main current passes through the collector and emitter (which is built as normally open or normally closed) and the gate signal triggers the connection between the two (collector and emitter), it will be much easier to understand (If you do not have the basic idea, I strongly recommend to refer some articles on transistor working). The chip, which provides the control timings based on the program, activates or deactivates the current flow through the transistor. Thus, we get output pulses.

Pic.5 Pulse width modulation and generation of the voltage wave

We have an idea now on how the voltage is varied (i.e. by varying the pulse width). However, the frequency is varied by increasing or decreasing the time interval between each pulse. As the time between the pulses is increased, the frequency decreases as time decreases, frequency increases. (Marking this as important as it explains how our abnormal sound is appearing)

Now, getting back to the output current, as discussed before, the output sine wave is not a pure AC sine wave but contains many pulses, which varies in width. The harmonic distortion in the wave depends on the number of pulses per cycle and the gap between each pulse, as described in the pic. 6 below. Note that the same 20 pulses per cycle are used here and only the timing is varied in comparison to pic. 5. (Excuse me for the rough hand sketches)

As the number of pulses is increased, the harmonic distortion decreases and vice versa.

Pic.6 Varying frequency by controlling the timing between each pulse

The number of pulses per second of the IGBT output (the carrier wave) can be any of the following, 2 kHz, 4 kHz, 8 kHz or up to 16 kHz. The more number of pulses, the better the output. However, as the number of pulses are increased, the heat developed in VFD will be high and requires sophisticated cooling and thus the requirement of a larger size VFD module.

Coming back to the motor sound, in our case, the switching frequency was identified to be 2 kHz, which is observed in the spectrum. The raised noise floor was an indication of excited natural frequencies, probably the stator lamination or winding resonance at the carrier frequency. Hence, recommendations were provided to check and increase the switching frequency to 4 kHz (Can be done only in limited models) or to change the speed control from scalar mode to Direct Torque Control (DTC) mode (which was available in our ABB ACS880 model). DTC mode has several advantages over the conventional V-HZ (scalar) control, due to the absence of separate voltage and frequency controlled PWM. Approximate sinusoidal flux and current can be achieved through DTC. The recommendations were implemented and the audible sound vanished. Note that we are just changing the exciting frequency and shifting it away from the natural frequency of laminations. In our case, we reduced the distortions and thus the resonance amplitude. In case we increase the frequency, the generated tone is shifting above our audible range and may still be observed in the frequency spectrum. Even before, during the test, when the speed was increased (increasing the frequency of current), the audible sound slightly decreased, thus confirming the doubt.

So, when you hear an abnormal sound from VFD motor next time, do not rush into conclusions. Take your time, do the tests and confirm the actual problem. The same switching frequency was supplied to similar motors in the plant, but only this particular motor produced the abnormal sound. It was noticed that only this motor utilized scalar control mode where other motors operated in DTC mode. Hence, our second recommendation was implemented, which reduced the sound.


  1. Found your post interesting to read. I can't wait to see your post soon. Good Luck for the upcoming update. This article is really very interesting and effective. eaton vfd

  2. Thank you for sharing

  3. This is my first time i visit here and I found so many interesting stuff in your blog especially it's discussion, thank you. study health sciences in university of New Brunswick Canada

  4. Thanks, Please share article which details cascade plot and coast down plot