Industrial Noise & Vibration Centre

+44 1753 698 800

Low Frequency Noise Complaints – do more with less…

dust extract system low frequency noise attenuation

Posted on: Dec. 1, 2021

Updated. A brief guide on how to evaluate low-frequency noise problems quickly and without spending any money - or simply using your smartphone.

Evaluating low-frequency noise complaints is one of the most common (“bane of my life”) issues that EHOs and the Environment Agencies contact us about. Using the right approach, we can sort most of them within a few days. The following is a brief, practical guide detailing solutions to the most oft encountered initial evaluation problems using a tool that won’t cost you a penny.

Case Study: low frequency fume extract system noise evaluation

This weld fume fan extract system was the cause of numerous complaints about a very low-frequency hum / woofle / throb. Frequency analysis of audio recordings showed the problem to be a 35Hz tone plus first harmonic. As this did not tie in with fan speeds etc, we immediately knew that it must be a resonance. One site visit later and the problem was resolved by altering the duct acoustics and inserting an aerodynamic flow control plate to improve the system efficiency. The tones were eliminated and the broadband noise reduction was very successful at a cost <£1k. More detailed information on fume and scrubber extractor fan system noise attenuation.

An entirely zero-cost way to analyse this type of problem is detailed below.

Low Frequency Noise Complaint Evaluation

quickrecord

dB(A)

Forget dB(A), it is irrelevant as it progressively filters out low-frequency sound (-39dB at 31Hz). If I had a £ for every extensive and entirely inappropriate set of logged dB(A) plots carried out as part of a low-frequency noise evaluation…

Audio Recordings

The best option. Record samples with plenty of gain for subsequent listening and analysis. Ideally, we like to get 2 or 3 recordings when the noise is bad and a further couple when it is not there or less prominent. That’s it: no prolonged sets of logged data.

record-speed-change

I can’t hear anything on play-back

The problem with very low-frequency sound is that you need speakers the size of a wardrobe or an unusually good set of headphones. The best results in practice can often be obtained with a good set of in-ear buds that are well screwed into your ears. However, you can use our free QuickRecord software utility to play back at a higher speed.

We would usually use a x3 or x5 speed increase to shift a hum at perhaps 40Hz up to 120Hz or above where you can clearly hear hums etc. whilst using kit that ordinary mortals can afford…

low frequency-spectrum-plot

Install the utility, open a wav file and hit play. Pause, click the button next to play and select your preferred speed, un-pause to listen further. Email or call us if you’d like a copy of QuickRecord – consult@invc.co.uk. Note: QuickRecord usually runs fine in Windows 7 or 10, but it is supplied “as-is” with no support.

Alternatively, you can download the open-source Audacity software which allows you to achieve something similar.

Here, after loading the sound file into Audacity, I’ve selected “Effects” and “Change Speed”, and a 200% increase which will double the frequency on playback (the file duration displayed is halved to match). Don’t save the file!

Is there a “real” hum and is it audible?

You can use the spectrum capability in Audacity to take a quick look for any tones. Revert back to 100% speed and then select a section of the wav file time trace (left click and drag). Then select the “Analyze” tab and “Plot Spectrum” from the drop-down list of options. This displays a simple spectrum.

Spectrum noise analyser smartphone app

Note that the log X-axis allows you to view the low-frequency end of the signature more easily, but you lose the ability to spot harmonics by eye (they are evenly spaced on a linear X-axis). In this case, the dominant 35Hz tone can clearly be seen.

Alternatively, there are many free android and iphone smartphone spectrum analysis apps available that you can use to detect tones live on-site or even play back previous recordings into the phone for analysis. In the latter case, select "line-in" as the source in the analysis app (if it is available). Our recommended free Android app guide >>

Be aware that, although the measured frequencies of any tones are completely accurate, the amplitudes are not. Not only are they uncalibrated (level), phone microphones have a roll-off at low frequencies that varies with the type of phone. However, you will be able to see if there is a tone (and at what frequency) and by how much it protrudes above the broadband background noise on either side of the tone. More detailed information on frequency analysis and using the precise frequencies to diagnose noise sources.

complainant fan tonal noise masking 264Hz

Whilst you could spend a fortune on a high-end sound level meter with analysis software that calculates the BS4142 compliant tonal content, in practice it is easy to do this by eye. Simply draw an imaginary line on the broadband amplitude either side of the tone spectrum as shown here ("A" weighted and linear).

The simple rule of thumb is that if the tone is c 5dB above the line, it will be audible. At about 10dB above the line then complaints are likely.

…and back to the reality of the case study
low-frequency-noise-control.jpg

This plot is an overlay of the noise before and after treatment, showing the effect of the low-frequency noise reduction (tones) and much of the broadband noise using our own, more sophisticated software.

Contact us if you need help with analysis of a specific problem or the interpretation of results. Or you can simply email sample recordings or data to us (remote diagnosis and control - a no-cost service) The most common causes of low frequency noise problems are fans, chillers, burners, compressors etc. We have developed low cost, elegant engineering noise control measures for all of these sources – take a look at a range of environmental noise diagnosis and control case studies.