Rms Spl And Peak Spl Spectrum - Sound Level Spectrum

[ROOTS]

Hello,

I have a clarification in Sound Level Spectrum.

We are manufacturing Automotive DC Electric Horns and when we measured the Spectrum we achieved the following results;

1. Peak SPL : 108.8 dBA
2. RMS SPL : 112.1 dBA
3. Mean : 85 dBA

My doubt is whether the RMS SPL should be lesser or greater than the Peak SPL. My graph shows with RMS SPL as greater.

Pl. clarify me whether the graph is correct and also to arrive the formula for calculating RMS SPL, difference between RMS SPL, Peak SPL ansd overall SPL.

I have also attach the graph for your easy reference.

Pl. help.


Regards
M.Prabhakar

[Miner]

I have a small amount of experience with sound analysis of equipment. In that experience, the RMS has always been less than the Peak (How can it be the peak if it is not the max?). My guess is that you need to open your scale up to see the full spectrum. You are using 0 to 5000 Hz, but the frequency of the peak may be 20,000. Also, sound spectra of equipment will often show multiple peaks at different frequencies. The more complex the equipment, the more peaks.

Also, learn about harmonics and sidebanding. Your file appears to show a lot of sideband frequencies. Make certain that you have high quality microphones for your analysis, and that you have calibrated your microphones to your software. If you have not calibrated the microphone to the software, the amplitude measurement is meaningless.

Definitions:
- SPL = Sound Pressure Level
- RMS = Root Mean Square (This one method of determining an "average" SPL) This is analogous to Ra in surface roughness. You should check how your software determines this. Is it the RMS for the entire spectrum, or the RMS for the range of frequencies investigated?
- Peak SPL = the maximum SPL observed. You should check how your software determines this. Is it the max for the entire spectrum, or the max for the range of frequencies investigated?
- Overall = Check how your software defines this.

[Tim Folkerts]

Originally Posted by ROOTS

Hello,

I have a clarification in Sound Level Spectrum.

We are manufacturing Automotive DC Electric Horns and when we measured the Spectrum we achieved the following results;

1. Peak SPL : 108.8 dBA
2. RMS SPL : 112.1 dBA
3. Mean : 85 dBA

My doubt is whether the RMS SPL should be lesser or greater than the Peak SPL. My graph shows with RMS SPL as greater.

Pl. clarify me whether the graph is correct and also to arrive the formula for calculating RMS SPL, difference between RMS SPL, Peak SPL ansd overall SPL.

I have also attach the graph for your easy reference.

Pl. help.


Regards
M.Prabhakar

Like Miner, I also know a bit a bout sound - just enough to be dangeraous!

The horn appears to have a "fundamental frequency" around 360 Hz, at then several "harmonics" at multiples of this frequency. As I interpret the graph, it basically shows the loudness of each of these peaks.

What you are most likely interested in is the "total loudness", since that is what people hear. In this case, you would want to add all the values. I expect that the "RMS SPL" listed above is the RMS sum, not the RMS average.

Two things to note when doing such a sum. First, decibels are a logrithmic measurement. Adding 109 dB and 105 dB (approximately your two biggest peaks) is only about 110.5 dB. Adding another two peaks at 103 and 102 raises it another 1.2 dB to about 111.7. Given the roughness of my esitmates, and adding in the rest of the smaller peaks, 112.1 dB sounds about right for the RMS total.

Secondly (and perhaps less importantly) "dBA" is slightly different than "dB". dBA tries to take into account that the ear hears some frequencies better than other frequencies. This would adjust the sum slightly, but probably not much in this case.

Remember, this is based on theory, not experience. I have taught sound in physics classes, but never done industrial measurements or had to calibrate equipment.


Tim F

[BradM]

Boy, for two ole boys with just a little bit of experience, nice posts!

I would have hated to see the length of it if they were your hobbies!

Roots, not sure if you may want to Google that one, and maybe post in some other Internet discussion threads specifically for sound/audio. Although, the level of expertise on the Cove has so far been nothing short of Amazing.

Also, maybe you could try some of the schools with Acoustical Engineering programs.

[ROOTS]

Dear Mr.Miner / Mr.Tim Folkerts,

Thankyou very much for your kind feedback.

Mr.Miner for your question, the RMS shown here is for the entire Spectrum. (ie. from 0 to 5000 Hz). But however, now after seeing both of your feedback I am clear and will act accordingly.

And one more question Mr.Tim, You have mentioned that the RMS shown in the graph is RMS Sum and not RMS average. When we have to consider RMS average? in other words at what type of measurement RMS average to be considered? Can you Pl. reply

Once again thankyou for both of you for your kind help.

Regards
M.Prabhakar

[Tim Folkerts]

Originally Posted by ROOTS

And one more question Mr.Tim, You have mentioned that the RMS shown in the graph is RMS Sum and not RMS average. When we have to consider RMS average? in other words at what type of measurement RMS average to be considered? Can you Pl. reply

M. Prabhakar,

One common place where an RMS average is used is in AC electricity. In the US, the standard voltage for outlets is around 110 V AC. However, since the voltage goes positive and negative in AC, the actual "average" is zero! Instead, what is normally quoted is the RMS Voltage. You square (S) the volage at each instant (say every 0.001 s), then take the mean (M), then take the square root (R). So you have the Root of the Mean of the Squares. For a typical sine wave used in AC electricity, the peak value ends up being 2^0.5 times the RMS value.

Besides getting rid of the "average being zero" problem, the RMS AC voltage provides the same power as the equivalent DC voltage.

So 110 V(RMS) would go from about +155V to -155V, and light up a light bulb just as well as 110 V DC.


Tim F

P.S. Given that RMS means "root mean square" it is perhaps a bit of a misnomer to talk about RMS without taking an average.

Now that I think about it, you could also take the RMS average of the original pressure measurements. To get the FFT (fast fourier transform) graph you presented, you must have collected pressure data at fairly high speed (certainly higher than the 5000 Hz at the top of the graph). Instead of processing the data to get the graph, you could just take an RMS average of the raw pressure data. This RMS average could be used to calculate the dB level directly, and it would be higher than the dB level of any of the individual peaks.

[ROOTS]

Dear Mr.Tim,

Thankyou for your immediate response. You have very clearly and nicely explained me the RMS. Now I got some idea to work. Thankyou very much.

Regards
M.Prabhakar

[ROOTS]

Dear Mr.Tim F,

Goodmorning to you.

One more clarification in the RMS SPL.

As you said SPL in complex frequencies gives RMS Sum is understood to me. But what we call the individual SPL at each frequencies. (ie. my question is we have checked the Spectrum from 1800 to 3550 Hz and found the SPL as below;

1. At 2000 Hz = 81.8 dBA
2. At 2500 Hz = 94.2 dBA
3. At 3200 Hz = 111.3 dBA

RMS Sum = 111.38 dBA

What we call the SPL at 2000 Hz, 2500 Hz and 3200 Hz. Whether it is RMS SPL or Peak SPL or maximum SPL.

Please clarify.

Regards
M.Prabhakar