By Measuring Background Noise Accurately
Marilyn Monroe was never a size 16. This myth appeared when America switched its sizing system in the 1980s. Roughly a 16 to 18 size in the 1950s is the same as a size eight today. Marilyn’s dressmaker kept her measurements; her 22-inch waist would now be below a modern size zero!
But why change the numbering system?
Sadly, down to our ever-increasing girth, America altered the sizing system so we didn’t feel as fat. Manipulating the numbers also plagues the PAVA system industry, but the background noise level is one area where accuracy is essential. Everything else hangs from this noise level and will hang you if you get it wrong.
- The background noise sets the broadcast level.
- The broadcast level defines the loudspeaker type, number, orientation and power.
- The loudspeaker power demand is the basis for the amplifier power ratings.
- The amplifier power sets the equipment cabinet size, AC power requirements and battery capacity.
Therefore, miscalculating the background noise level means you either overprice the PAVA system, losing you the work, or under-price and see those precious profits (and more) consumed in fixing that early mistake.
The Noise Measuring Process
Measuring PAVA system emergency messages is similar, we’ll cover this process separately, but in brief, there are four main differences between emergency message and background noise dB level measurements:
- The impact of noise on the PAVA system speech level reading.
- The need to separately measure the alert signal and the speech level.
- Dealing with the silences in the speech.
- The small silences between words, the measurement is an average over time, so silences lower that average, which risks tipping you from a pass to fail.
3 dB is a factor of two change: Plus (+) 3 dB is double, and minus (-) 3 dB is half.
So if your noise readings are wrong by 3 dB in either direction, you will either be very expensive (quoting twice the amplifier power needed) or under-powered by half; just the extra space you’ll need to find for the amplifiers and batteries to rectify this lack of power is frightening.
Therefore, the following three-point process will save you from making this costly error. What we’ll be covering is:
- Understanding the space;
- Setting up the SPL (dB) meter;
- Making the measurements.
1. Understanding the space
German writer Johann Wolfgang von Goethe once said, “Architecture is frozen music.” I take this to mean that each space uniquely colours sound. Think back to a mobile phone call when you ask the person, “It sounds like you are in the toilet.” You may not know precisely why you think this, but it is not because the person is speaking differently; it’s how their voice interacts with the room.
This space or room’s uniqueness comes from its shape and size, construction and furnishings and noise composition.
Reverberation is a significant factor in sound system design, and this article is not the place to get into this. In brief, if the reverberation time is long, get some advice. What is a long reverberation time? Clap your hands. If the noise is still around after a second; that’s long. Note this assumes the room is quiet.
Another factor is the frequency range of the background noise
A SPL meter, also known as a dB meter or an SLM (sound level meter), listens to a wide frequency range and then averages the level to give you a single number. Therefore, be careful about how you interpret those values.
If the background noise is primarily speech, then it has all the frequencies of the PAVA system voice announcement, so the SPL meter reading provides the best indication of the background noise level you need to overcome.
If the background noise has a lot of higher frequency content, such as that from machinery whine and high-speed cutters, then the SPL meter might indicate a lower dB reading than you need. Those lower dB readings might suggest the announcements can be quieter. However, be careful as those machines’ whines can be at the critical frequencies (2 to 4 kHz) essential to understanding the speech: You might want to add 3 dB or more to the announcement level to compensate.
Low-frequency noise (machine hums and growls) creates the reverse effect, with the SPL meter reading a higher dB number, but as lower frequencies have less negative impact on voice announcements, you might be able to reduce the broadcasts by 3 dB. The following section outlines our OptiNoise technique, which reveals whether the noise has a lot of low-frequency content.
Understanding the effect frequency has on broadcast levels means you are ahead of many sound system designers, but you’ll have no meaningful noise readings if you don’t set up your SPL meter correctly.
2. Setting up your SPL meter
We’ll do a deep dive into SPL meters in a separate piece. In short, SPL meters will often offer three weightings: ‘A’, ‘B’ and ‘C’. These relate (with modifications) to the Fletcher and Munson equal loudness contours (again, don’t worry about that here).
- ‘A’ weighting is best for quieter noise measurement.
- ‘B’ weighting is the same as ‘C’ for frequencies above 1 kHz and halfway between ‘A’ and ‘C’ below 1 kHz. It is now an ‘ex’ measurement dropped by all the standards along with ‘D’ weighting.
- ‘C’ weighting is best for louder noise measurement
How to lower your PAVA system quote price with the OptiNoise Technique
Each of these weightings assesses lower frequencies differently (there’s a difference for the higher frequencies, too, but it is tiny). So you can use this difference to your advantage.
Set the meter to ‘slow’ and ‘A’ weighting, then do it again with ‘slow’ and ‘C’ weighting. If the ‘C’ weighting dB number is much higher than that of the ‘A’ weighting number, then the background noise has a lot of low-frequency content.
As mentioned earlier, low-frequency noise interferes less with speech, so you might decide to knock-off up to 3 dB from the announcement broadcast level. A 3 dB reduction means up to a 50% saving on your kit quote.
Now you need to make the noise measurements, but which weighting should you use?
75 dBA is the magic number
If you used the OptiNoise Technique, you’ll have an ‘A’ weighted dB reading. If that reading rarely went above 75 dBA, then use the ‘A’ weighting. The’ C’ weighting is best if the reading is often above 75 dBA.
Note that the letter after the dB tells us which weighting was used for the measurement. Therefore, if you get over 75 dBA, select the ‘C’ setting on the meter and measure the background noise. Then, after the number, you would write dBC.
Importantly, even if the dBC reading is below 75 dBC because the original dBA reading was mostly above 75 dBA, you still record the dBC result.
A small aside: don’t write dBa for dBA; they are unrelated.
I should mention that BS 5839 Part 8 2023 clause 20.1.4 advises you to get help from a “competent electro-acoustic designer” when noise is above 75 dBA. It’s good to know BS 5839 acknowledges the existence of “incompetent” electro-acoustic designers and to avoid those miscreants.
However, noise levels are never steady, so how can we get meaningful measurements?
Time for some more averaging
I’ve already mentioned how the SPL meter averages the noise level across the frequency range; now, you need to average the noise level over time.
BS 4142:2014+A1:2019 (Methods for rating and assessing industrial and commercial sound) and BS 5839 Part 8 2023 recommend LAeq,T.
LAeq,T means set your meter to ‘A’ weighting and measure the average (eq) sound pressure level over a specific (T) time. Both standards assume the noise is mostly below 75 dBA, hence the ‘A’ weighting.
BS 5839 Part 8 2023 oddly has no interest in BS 4142. BS 4142 advises on the measurement duration based on different environments; BS 5839 leaves the measuring time to you, Clause 20.1.2: “The duration (T) of the measurement should be appropriate to the venue/premises and its usual activities.”
My view is to measure the noise for at least 30 seconds and at a time when the place is in normal operation, i.e. if it’s an office with most people working during the day, don’t measure the noise on the weekend or at night.
If some loud noise happens during the measuring time, you need to consider whether that is a regular or unusual feature of the sound-scape (or should that be noise-scape). If it’s unusual, then redo the measurement.
Remember loud noises will bump up the average noise reading, adding to your PAVA system quote price
If your SPL meter cannot measure the average noise level over a fixed period, i.e. it does not have an LAeq option, you need to buy or hire one, as an error here could be an expensive mistake.
You’ve now learned about the space and the meter setup. The only thing missing is your measurement data.
3. It’s time to make the measurements
Your objective is to get as many people as possible to hear the PAVA system broadcasts.
This means that you need to measure the noise close to the people. The microphone on the SPL meter needs to be correctly orientated.
The best way to do this is to mount the meter on a tripod stand.
The SPL meter needs to mimic our ears
No need to add earlobes to the microphone; that would just be silly.
Set the meter at ear height or an average ear height if some people are sitting and others standing/walking. Our ears are generally in the horizontal plane (parallel to the floor), so the meter should be the same (the meter points at the wall rather than the ceiling).
Once you are in position, you want to avoid interfering with the measurements.
Stay back from the meter
If your meter can have the measurement time set, you can move as far away from the meter as you like. If you need to use a stopwatch, stay closer, ready to stop the noise logging once the time has elapsed. Try to stand at least a metre away. We recommend you do three tests at each measurement position and take the average (I’ve not realised how much averaging goes on with this process). That leaves recording your results correctly.
Your results are not just the final value
When you were at school, you needed to show all your workings. In this context, you need to record all the following information:
- Location: This needs to be quite precise so the measurement can be repeated. In addition to the distance from things, mention physical elements (preferably immovable elements, so avoid references such as“next to the Christmas Tree” or “near Isabel Edwards’ desk” ). Nowadays, taking a couple of photos from different angles is best; you just need to know which photos go with which measurement.
- Date and time.
- Meter orientation.
- Measurement duration.
- General activity whilst recording the background noise. Remember that loud noise issue I mentioned earlier. If a door is in regular use and bangs closed, then your measurement will always include it, so state that in the notes, i.e., the door shut noisily twice during the measurement. Why? Later, you might decide this should not be a sound level factor because the client has fixed the door or it’s not a fire exit route, so it’s not used during an evacuation.
- Meter make and product code.
- Date the meter was last calibrated.
- Day note. Perhaps it was someone’s birthday (adds noise), a company meeting (at the office so busier than usual, or off-site so quieter than expected), or a machine was being serviced, so the background noise was lower.
Planning is a necessary evil
Just as the whole PAVA system design process, getting a meaningful number for background noise needs quite a lot of planning and note-taking.
Nowadays, you can download an SPL meter app for your phone with impressive capabilities. You might think I am against them, not at all. I think they are of great benefit, but use them wisely, understand what the numbers mean and, like Marilyn Monroe’s dress size, don’t blindly trust them.
If you’ve got 5 minutes spare, read my “No B.S. Guide to STI”, not the infection, just the Speech Transmission Index here.
Must you measure the background noise?
BS 5839 Part 8 2023 says, “To achieve a satisfactory level of intelligibility the VAS [voice alarm system] should be designed to broadcast a speech signal level that exceeds the background noise level anticipated under emergency conditions.” Also, it underlines the importance in Annex A with “… it is important that the most realistic ambient noise level is taken as the design datum.”
Can you use 65 dBA for the PAVA system speech level?
It is commonly thought that Part 1 of BS 5839 just requires sounders to be 65 dBA. Whereas Part 8 is more demanding for PAVA systems, clause 20.1.1 says,
“NOTE 1 A signal-to-noise ratio of at least 10 dB is usually needed to obtain an acceptable level of speech intelligibility.”
Also, unlike a fire sounder, speech level varies widely. Therefore, the measuring process is more challenging than simply waving a dB meter in the air.
Even so, Part 1 of BS 5839 2017 isn’t quite a pushover on this.
Clause 16.2.1 2) recommends,
“where the sound pressure level of background noise is greater than 60 dB(A), the sound pressure level of the fire alarm signal should be 5 dB above the sound pressure level of the background noise.”
Part 1 Annex B Page 133 has a helpful table of typical sound level ranges. For an open plan office, the noise can be between 50 and 70 dBA, so any office with a background noise of 56 dBA or above means PAVA announcements of 65 dBA would be inadequate.
However, Part 8 Clause 13.2b) says: [Where the acoustically distinguishable space is a simple acoustic space] The sound pressure level of voice messages should be greater than 70 dB LAeq, measured over a period of not less than 10 s or the emergency message length, whichever is longer.
How did we do? Spotted a mistake? We want to provide as much value to the reader as possible here so please contact us with feedback or queries.