The Basics of Electrical Sound File Analysis, Pt. 1

January 30th, 2015

When it comes to monitoring the performance of electrical machines, ultrasound is one of the best strategies for detecting faults. But in order to do so, analysts need to be able to understand .wav files and the basics of using Spectralyzer.

One of the first steps to using Spectralyzer is bringing down your frequency range, as the default is too large. Ideally, the range should be locked in between 900 to 1000 Hz, while the relative scale will fluctuate with the quality of the recording. When it comes to selecting the number of samples, it is a matter of preference preference. However, it is important to understand that a lower number of samples will make it difficult to identify faults, while higher ones can generate false positives. The best solution is a happy medium at 81/92.

Identifying a tracking fault
The best way to identify faults is to look at the “spectral footprint.” Each type of fault will manifest itself in a unique way that is identifiable in the ultrasound .wav file. Also, it is important to listen to tonal quality, because these fluctuations, such as peaks at 60 Hz, can identify a problem as one that is electrical or otherwise. Finally, be sure to watch for frequency – if the file has intermittent peaks or a continuous wave, the kind of fault will be much different.

One of the most common faults that emerges is tracking, which is essentially a discharge of energy along an electrical insulator. There are any number of causes for this kind of fault, including contamination and moisture, but one of the biggest byproducts is carbon, which can be identified by the Ultraprobe.

In the .wav file, this will translate into high frequency noise in between harmonic markers and drastic frequency peaks. By looking at these markings and listening to tonal sound, you can first identify that the problem is electrical in nature. Typically, these energy discharges are the “snap, crackle, pop” sounds in the file. Generally speaking, the footprint for a tracking fault is consistent in amplitude, but not uniform in intensity or time. By slicing down the time domain for a sound file, an analyst can better identify the frequency of discharges.

It is important to understand that when looking at the ultrasound recording, dBs indicate how close to the source the probe is from the fault. This means that where one positions the Ultraprobe plays an integral role in identifying the nature of a fault. The angle of the energy discharge can change over time, so make sure to get readings from around the entire machine.

Reporting faults
Once the fault has been identified, one should put a visual image to the sound file. By using the report feature on Spectralyzer, analysts can show the what the fault looks like from a time standpoint.

Tracking faults can ultimately lead to worse faults. For instance, tracking can inevitably lead to destructive corona or arcing. However, these faults are far more destructive than tracking faults, so by being able to identify and report these faults sooner, analysts identify issues that, when acted upon, can prevent further harm to the machine.

A great strategy for reporting faults is offering a comparison. By taking two recording that are a few months apart, an analyst can export images of both files. While the frequency and severity of each file may differ visually, the same footprint should be identified in each recording, which can provide further confirmation about the fault. By reporting the fault as such, an analyst can further confirm the electrical shortcoming, and hopefully, prevent machine failure before it gets any worse.

Suggested Ultraprobe Instruments for Electrical Inspection and analysis:  Ultraprobe 10,000; Ultraprobe 15,000

©2018 UE Systems