Talk:Decibel/Archive 1

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Contents

120 dB breaks eardrums?

This sentence doesn't make sense: "In air, sound pressure levels above 85 dB are considered harmful, while 95 dB is considered unsafe for prolonged periods and 120 dB causes an immediate perforation of the ear drum (tympanic membrane)." But the table that follows lists that as the threshold of pain. -- jun

It also makes no sense that 120dB perforates your eardrum, but an M1 Garand makes a 150dB noise at 1m. If that were true, every US rifleman in WWII would have come home deaf.

"thats a very short burst of sound... and alot of them DID come home deaf or with significant hearing loss. even 85db for longer than 8 hours can cause hearing damage" - FYI

Actually the question of whether 120dB(SPL) ruptures eardrums very much depends on the frequency of the sound as anyone who has attended a sound off competition can attest to since low frequency sound (bass) damages hearing at a much higher dB(SPL) than higher frequencies. -Fun fact: that is why earbud headphones generally cause more hearing damage than proper circumaural ones.-

Not the only problem with this section. Perceptions of potential harmfulness vary from country to country. It is not the SPL alone, but also the duration that matters. Thus, in Europe a time-weighted average (TWA) of 80 dB(A) over an 8-hour working day is the level at which some action is required of employers. If that reaches 85 dB(A) then additional action, including a reduction in noise, is required. Exposure above 87 dB(A) (again 8-hour TWA) is prohibited, although this can be achieved by providing personal protective equipment if there is no reasonably practicable means of reducing the noise level. There are additional limits, referring to peak exposures, at 135, 137 and 140 dB(C). 140 dB(C) is generally taken as the threshold for immediate noise-induced hearing loss - but not perforation. The reference for the UK is the Control of Noise at Work Regulations 2005, which are derived from European Union (EU) Physical Agents (Noise) Directive (2003/10/EC). The initial comment is wholly valid, it would be very unusual to perforate an eardrum simply by exposure to a peak of 120 dB(C). --Andygc 14:58, 12 March 2007 (UTC)

I've made some significant edits to this section. I think it provides a more accurate and encyclopedic approach to the section, but it does still need work. --Thesleepwalker 17:02, 14 March 2007 (UTC)

Introductory reasons

Okay, my phrase 'This is a direct result that many of our perceptory organs (like hearing and sight) are logarithmic in nature, therefore the dB is a more natural unit.' has been deleted due to '(rv badly phrased interjection)'. Care to elaborate, Light current?

Bel vs. Decibel, which came first?

The part until the next line has been moved over from Talk:Decibel: Earlier I thought Alexander Graham Bell coined the term "bel" as a measurement of sound and that it was later determined to be so coarse that 1/10th of it proved more useful (the decibel). On seeing someone on this page claim that Bell coined the phrase "decibel" I looked up its history in the Oxford English Dictionary. The OED has their earliest recorded uses in 1928 and 1929; Bell died in 1922.

  • the earliest few quotations the OED has on file for "decibel": "1928 Electrical Communication VII. I. 33/2 If common logarithms are used, the reproduction is obtained in Decibels. 1929 W. H. MARTIN in Bell System Techn. Jrnl. VIII. 2 The Bell System has adopted the name ?decibel? for the ?transmission unit?, based on a power ratio of 10·1... For convenience, the symbol ?db? will be employed to indicate the name ?decibel?. 1930 Discovery Dec. 398/2 The band-pass filter, which follows the low frequency modulator, allows the lower side-band to pass with an attenuation of six decibels."
  • the earliest few quotations the OED has on file for "bel": "1929 W. H. MARTIN in Bell System Techn. Jrnl. VIII. 2 It was further suggested that the naperian unit be called the ?neper? and that the fundamental decimal unit be called the ?bel?, these names being derived from..Napier..and Alexander Graham Bell. 1930 Gloss. Terms Electr. Engin. (B.S.I.) 13 The bel is a unit used in the comparison of the magnitudes of power, voltages or currents at two different points in a network of lines or apparatus."

Clearly the term "bel" was in use before the term for 1/10th of it came about. Yet it seems odd to me that the terms did not see print for almost a decade. Anyone? Specifically, I'd like to know who made the suggestion for the terms "napier" and "bel." I suspect it wasn't Alexander Graham Bell. --Koyaanis Qatsi

I've generalised the article a bit, since decibels are not just used for acoustics (e.g. they're used to measure the gain of amplifiers and loss of transmission lines.) -- DrBob

Think we ought to change the title to Bels? Scienceman123 20:40, 20 April 2006 (UTC)

No. We moved it here from bel because effectively no one uses bels. — Omegatron 20:52, 20 April 2006 (UTC)

Acoustic decibel reference

How likely is it, if I find a claim that a sound is at, say, 120dB, that the reference level is indeed 20 micropascals? Similarly, if author A claims that one sound is so many dB, and author B claims that one sound is so many dB, how safe is it to compare the measures given, if neither indicates the reference level? --Ryguasu 04:38 Feb 26, 2003 (UTC)

Assuming they are both talking about sound in air and that they are not trying to obscure, it's fairly safe. The standard reference for sound in air has been 20 micropascals since the early 1920s. Technically, of course, decibels without a reference have no meaning.

Watts vs watts per square meter

This 0 (zero), decibel level also corresponds to one billionth of a watt, 0.000 000 000 001 watt, roughly a mosquito flying 10 feet away.

Watt would be total power, not related to distance; also http://ccms.ntu.edu.tw/~karchung/decibels/decibels1.ppt says 40 dB. - Patrick 09:52 Apr 16, 2003 (UTC)

List of acoustic decibel levels

It'd be nice to have a list of example decibel levels on this article - Khendon

Decibel vs Bel, separate article for acoustics or not

  1. Do acoustic decibels and general decibels really need to be separate articles?
  2. Since decibel is much more commonly used than bel, shouldn't it be the title of the article? kind of like kilogram being the standard SI unit, even though it is a basic unit with a prefix.

- Omegatron 00:47, Apr 17, 2004 (UTC)

Agreed Omegatron, I think this article should move to decibel - anyone disagree? -- Rissa 00:45, 6 Jun 2004 (UTC)
I am moving it back to decibel. It is much more common. it sounds like the richter scale uses bels without calling it as such, but EVERYTHING else uses decibels. - Omegatron 14:12, Jun 27, 2004 (UTC)

20 micropascals or 2 pascals???

This seems to say that dB SPL is referenced to both 20 micropascals and 2 pascals. I'm sure the standard (used for dBA, etc.) is only one of those. Which is it? - Omegatron 17:43, Nov 23, 2004 (UTC)

20 micro. someone put in 2 N/m^2 without any exponent for some reason. fixed now. - Omegatron 17:51, Nov 23, 2004 (UTC)

which dB?

Please make sure the reference for all values listed as only dB are clear from context. (dB SPL, dBu, etc.) - Omegatron 17:43, Nov 23, 2004 (UTC)

The treatment of sound pressure level appears to be inconsistent with standard reference works across Wikipedia. Both Kinsler and Frey's "Fundamentals of Acoustics" (2nd edition) and Robert Urick's "Principles of Underwater Sound" (3rd edition) indicate that a measured intensity is a level (Urick p.15) or sound pressure level (K&F) relative to a reference effective pressure (K&F pp.125-126). Both of these sources recommend reporting decibels with an explicit listing of the reference effective pressure, like so: "74 dB re 20 micropascals", where the number and units following re is the reference effective pressure. I was working on the decibel article's misuse of SPL as if it specified a reference effective pressure when I realized that several other articles were similarly affected. I don't have time at the moment to correct all of them, but I will drop this comment into the discussion part of each article that I have come across so far with this problem. Level or sound pressure level in both these standard texts simply refer to a measurement in the sound field and are not indications of a specific reference pressure upon which the decibel is based. In other words, "dBSPL" is an incorrect means of attempting to refer to the in-air reference effective pressure. In no article thus far have I seen the "dBSPL" usage tied to an authoritative source. By contrast, the "dB re" formalism is common to both standard reference works that I have cited. I have worked over the "Acoustics In Air" section up to the table of very high SPL values; there is more work to be done to fix the whole article, though. Wesley R. Elsberry 11:23, 9 April 2006 (UTC)
Other sites using the "dB re" formalism: Oceans of Noise (explicit in defining SPL and SIL in terms of "dB re"), SURTASS LFA, NIST listing SPL in terms of "dB re", and Acoustic Impacts on Marine Mammals. But the best thing I've found has to be ASACOS Rules for Preparation of American National Standards in ACOUSTICS, MECHANICAL VIBRATION AND SHOCK, BIOACOUSTICS, and NOISE, which states:

3.16 Unit symbols 3.16.1 When to use unit symbols In the text of the standard, the unit symbol for a quantity shall be used only when the unit is preceded by a numeral. When the unit is not preceded by a numeral, spell out the name of the unit. In text, even when a numerical value is given, it is desirable to spell out the name of the unit. Moreover, the name shall be spelled out when it first appears in the text, and more often if the text is lengthy. Thus, in text write "...a sound pressure level of 73 dB; or "...a sound pressure level of 73 decibels." Do not write "sound pressure level in dB"; the correct form is "sound pressure level in decibels." Do not write "dB levels", "dB readings", or "dB SPL." Levels or readings are not of decibels; they are of sound pressure levels or some other acoustical quantity. Write out the word "decibel" for such applications, and be sure that the word 'decibel' follows, not precedes the description of the relevant acoustical quantity.

The guidelines given for the National Standards clearly excludes the use of "dB SPL". Wesley R. Elsberry 17:03, 9 April 2006 (UTC)

The supposed "better reference" for use of "dB SPL" added to the decibel article ends up being a document that merely includes "dB SPL" in a list of terms. The glossary within the same document does not even list this supposed term, even though weighted decibel terms are defined. The glossary in the file does have an entry for "sound pressure level", which is

Sound pressure level: (1) Ten times the logarithm to the base ten of the ratio of the time-mean-square pressure of a sound, in a stated frequency band, to the square of the reference sound pressure in gases of 20 micropascals (µPa). Unit, dB; symbol, Lp. (2) For sound in media other than gases, unless otherwise specified, reference sound pressure in 1 µPa (ANSI S1.1-1994: sound pressure level).

Notice that the unit is "dB", NOT "dB SPL". The inclusion of "dB SPL" in their list of terms does NOT establish that their usage is correct, and even their own reference of the ANSI standard indicates that their usage is incorrect. SPL refers to a measurement, and is NOT an indication of the reference effective pressure. The ANSI standard referenced makes this clear, as SPL is defined as being used for other reference effective pressures, too. Wesley R. Elsberry 16:22, 10 April 2006 (UTC)

I have changed the article following your comments. Han-Kwang 17:22, 10 April 2006 (UTC)

The relevant SI guide SP811, Taylor 1995 is very clear on this. On p16 (Sec 7.4), the guide states the rule "When one gives the value of a quantity, it is incorrect to attach letters or other symbols to the unit in order to provide information about the quantity or its condition of measurement. Instead, the letters or symbols should be attached to the quantity". Then on p30 (Sec 8.7) it gives an example of the applcation of the rule, precluding the use of "dBm" to mean "dB re 1 mW". The same rule also precludes dBA, dBu, dB SPL and many other similar uses. Thunderbird2 09:43, 9 September 2007 (UTC)

Theatre?

One of the values in the table is "Theatre". That's too vague to be at all useful (to me, at least). An empty theatre? A filled theatre of whispering people? Sitting 5m away from Hamlet giving his soliloquy? A theatre showing a Jerry Bruckheimer movie? (From context, probably not the last one, but that's still a decent range.)

It's not a very explanatory entry, and would be better replaced with something more obvious. 30 dB is fairly quiet, the sort of environment where most people could fall asleep. The table entry probably means a theatre full of people who aren't intentionally making any noise at all (just breathing and rustling about in those uncomfy theatre seats as they wait for the show to start - of course all polite people cease their conversation at that point ;). Unfortunately I can't think of an equivalent value that would be more descriptive at the moment. - toh 19:04, 2005 August 22 (UTC)

Kerbside vs Curbside

The original spelling in this article was by Heron on 18 Nov as Kerbside. This is fine. Unless you wish to remove the entire entry, please leave the spelling in its original sense. Ian Cairns 00:29, 8 Dec 2004 (UTC)

Not that it's terribly important, but the policy seems to be one of consistency for the whole page:
- Omegatron 15:19, Dec 8, 2004 (UTC)

dBA incorrect?

according to whom? - Omegatron 02:57, Dec 12, 2004 (UTC)

Exactly my question. The link is to dB(A), and that's what I've seen the manufacturers use ([1]).--Jerryseinfeld 18:38, 23 Jan 2005 (UTC)
Well yeah, manufacturers use it all the time (not just "historical sources"). Someone is claiming that it's not official, though, because it implies a reference to an "A" unit, like dBV implies a reference to a volt. Either this is a small minority opinion, or it's a new opinion that just hasn't gained a foothold yet, like binary prefixes like "mebibyte". My personal opinion is that it's fine to use it the way it is. It's just a shorthand way of saying it's A-weighted. Like saying miles (survey). In addition, the standard acoustic version is just "dB" with no qualifier (it should be dB SPL or dBSPL or dB(SPL)) which leads laymen to believe that a dB is a unit by itself. I would much rather see people using dB SPL than see them stop using dB (A). - Omegatron 20:08, Jan 23, 2005 (UTC)

Unit symbol dB rather than dBA, dB(A) etc

When making acoustical measurements and determining, for example, an A-weighted sound pressure level, the measured sound pressure is still compared with the reference sound pressure (20 μPa), and the unit symbol should still be dB. This is standard usage in definitions given in modern ISO and IEC standards, and is now mandated for ANSI standards developed by the Acoustical Society of America in their ASACOS rules. The preferred form of expression is "the A-weighted sound level was xx dB" or LA = xx dB.

There are many historical sources which use "dBA", or "dB(A)" or talk about "dBA levels", and these should be understood as indicating A-weighted sound levels, although the unit symbol is incorrect.

Unit symbols such as dBu or dBm indicate the reference value used to determine the level, and thus are correct. But the A denotes a frequency weighting, not a reference quantity.

Potential inclusion in SI

Any citation to support the claim that "(BIPM) has recommended its inclusion in the SI system."?

  • This doesn't even seem plausible to me—it's outside BIPM's bailiwick. More likely the Comité International des Poids et Mesures (CIPM), or the Consultative Committee on Units which is, I think, under the CIPM and advises the CGPM

Also, does that recommendation include the neper as well as the decibel? Gene Nygaard 01:51, 19 Jan 2005 (UTC)

RMS

"dB(0.775 V)—(usually RMS) voltage amplitude"

Is it ever not RMS? Every site I see says it is referenced to RMS. - Omegatron 02:50, Jan 19, 2005 (UTC)

Is this website violating the GNU License??

http://encyclopedia.laborlawtalk.com/Decibel

if so, i find it funny that a "law" site is guilty of copyright violation.

No, it does not appear so. 203.26.206.129 07:09, 13 Apr 2005 (UTC)
Read the bottom of the page: "This article is licensed under the GNU Free Documentation License. It uses material from the Wikipedia article "Decibel"." RoceKiller 10:16, 14 Apr 2005 (UTC)

dB Doubling Versus "Perceived" Doubling

So I always thought that 10dB represented a doubling of the sound intensity but now I find out that this may be technically incorrect. That 3dB represents a true doubling of sound pressure and that 10dB is where the human ear perceives the pressure as doubling. Can anyone confirm this?

It would be nice for the article to clarify this.

I'm not sure of the actual answer, but these are where you should look. I'm going to try to figure it out, too: 3.01 dB corresponds to a doubling of sound power, not sound pressure. How we perceive it depends on amplitude and frequency, and would be derived from the equal-loudness contour. Also see the units of perceived loudness, phon and sone. - Omegatron 16:36, Apr 29, 2005 (UTC)
Wow. You helped me realize there are a lot of confusing, overlapping sound measurement articles. I've created a template to keep them tied together: Template:Sound measurements - Omegatron 16:49, Apr 29, 2005 (UTC)

Here is the story on this: 10 dB in acoustic terms, at 1 kHz corresponds to "twice as loud" Actually it should be 9 dB, but people get sloppy and use 10 dB instead. This goes back to Harvey Fletcher, and the article by Fletcher and Munson (1933) in Jol of the Acoust. soc of Am.

 Jont Allen jun 11, 2006
So an increase of 9 dB means twice the perceived loudness? Why does the article now say 3 dB and what does "twice as loud" mean anyway?. Thunderbird2 20:13, 1 August 2007 (UTC)

modern usage of bel

"According to the World's standardization organizations should sound power level regarding computers and other kinds of IT equipment be expressed in bels (B) instead of in decibels (dB). Sound power level is here expressed in bels for to avoid confusion between decibels for sound power level and decibels for sound pressure level [1,2,3,4,5]. The computer industry is the only product group that uses sound power in bels, even if other product declaration standards tell that one can use bel for stating sound power level for to avoid confusion with sound pressure level measures." - bels for sound power level

This is the only modern usage I can find of bels being used instead of decibels. - Omegatron 14:37, July 25, 2005 (UTC)


Krakatoa dB level

According to article linked site [2]:

310 (Normalized) KRAKATOA VOLCANO ERUPTION-1883 A.D.

The table in this article:

1,000 Krakatoa (1883)

The linked article at least has references listed; I wonder where the reference for the 1000 dB Krakatoa is. It sounds (no pun intended...) like an awful lot anyway, when similar extreme volcano eruptions are also listed at 300 dB there. -- Jugalator 21:55, 8 November 2005 (UTC)

I'm going to change it to 310. 1000 is a ridiculous amount, given the exponential nature of the decibel system. Crovax 04:23, 10 November 2005 (UTC)

Measurements like this need a distance associated with them. Meaningless otherwise. — Omegatron 16:25, 11 November 2005 (UTC)

Here's some trustable measurements: [3], though they are of infrasound. — Omegatron 16:57, 11 November 2005 (UTC)

Talk:Krakatoa#Loudness

Definition

Somehow I didn't notice that the definition was changed from a power ratio to an acoustics measurement. [4] I think it should be a general power ratio. — Omegatron 22:36, 8 November 2005 (UTC)

Intensity and pressure

I deleted the statement "Neither ear drums nor microphones can convert sound intensity. We hear the pressure variations.", on the grounds that it is meaningless. There is only one physical phenomenon--molecules of air move and displace the eardrum, leading to detection of "sound". One can choose to characterize the motion of the molecules by their pressure or by the intensity of the wave. Which you choose is irrelevant. One can't even say that the response of the ear is proportional to pressure rather than intensity, since the response is logarithmic and log(pressure) is proportional to log(intensity).--Srleffler 06:42, 28 November 2005 (UTC)

You can and should say that the ear responds to sound pressure rather than intensity. Although sound pressure is proportional to intensity in a plane progressive wave, it is not always true. Sound pressure is a scalar quantity and intensity a vector. So, in a perfectly diffuse sound field (equal amounts of sound travelling is all directions, approximated by a highy reverberent space) you may have a high sound pressure, but very low sound intensity in any direction. RNG 2006-06-02—The preceding unsigned comment was added by Richardng (talk • contribs) .
Interesting. You may be right. If this were going to appear in the article, the scalar-vs-vector distinction would need to be explained.--Srleffler 15:32, 2 June 2006 (UTC)

dBZ?

Where would dBZ fit in or is it an alias for something else here? [5] defines it as (under "Base reflectivity"):

decibels of Z, where Z represents the energy reflected back to the radar

Cburnett 23:08, 1 December 2005 (UTC)

"dBZ" an alias for "dBr", perhaps? Cburnett 23:09, 1 December 2005 (UTC)

Correct explanation of dBm and dBu

dBm (or dBmW) and dBW are independent of impedance. 1mW is 1mW regardless of the impedance it driven into. RF engineers typically use dBm to measure the power into a 50Ω load.

dBu (or dBv) is dependant on 600Ω when you are trying to relate it to dBm, because dBm = dBu when the load used is 600Ω. 600Ω is not important to audio, rather it originated from the telephone companies because it is the characteristic impedance of telephone wire when stretched miles (or kilometers) over land. Somewhere along the way (I have no clue why) audio started using the same measurements even though the characteristic impedance of a microphone cable is rarely 600Ω let alone consistent or even long enough for this to matter. I suspect this came from the fact that some of the first audio engineers came from the telecom industry or were educated in universities teaching the practices used in the telecom world. The use of dBu in audio is diminishing greatly because it is being substituted for the more appropriate dBV as output and input stages are being properly designed for voltage transfer and not power transfer.

Image:Circuit.jpg

Image:p-v2r.jpg

Image:v-pr.jpg

Image:0755VRMS.jpg

This is also the first time I've heard that the "u" in dBu means "unloaded." I've been told that the "u" was used to remove the "v" to avoid confusion with dB(1mV). Can anyone confirm this? --Babar77 11:17, 11 December 2005 (UTC)

Why would dBV be more appropriate than dBu?
I heard the same about the u being a replacement for lowercase v, though maybe "unloaded" is true, too. — Omegatron 00:50, 5 January 2006 (UTC)
I'm not an electrical engineer, but dBV seems like a more straightforward, intuitive unit, being based on the ratio of the voltage to 1 V. dBu, on the other hand, is based on the ratio of voltage to the odd value of 0.775 V. If you want something that agrees with dBm, just use dBm.--Srleffler 03:17, 5 January 2006 (UTC)
I am an electrical engineer who designs professional audio equipment, and I can verify what Srleffler said is true. dBV is more appropriate because you can do the Volts to dBV conversion in your head, easily. It reduces stupid arithmetic errors when designing under the gun and makes more sense. It's also getting harder to find test equipment that displays Voltage in dBu, and half the time it's wrong. The old HP8903B audio analyzer marks it's dBu measurement as dBm, which is completely wrong because it does not guarantee the load is 600Ω. This has caused A LOT of confusion in the office over the years and forced me to teach this on countless occasions.
Srleffler, I'm not sure if you are saying to make a voltage measurement using a power unit, but in case you are or any other reader is thinking the same thing I must say this is incorrect. If you are talking power use dBm, and if you are talking voltage, use dBu. This is because they have different scales and it is technically accurate. If it were up to me, I would remove dBu all together or make a small note of it for historical reasons only. It may have been very useful back in the day, but it needs to die - trust me. --Babar77 17:37, 5 January 2006 (UTC)
Hmm.. I am a (relatively inexperienced?) electrical engineer who designs professional audio equipment, and see dBu much more often than dBV. Just curious why someone would say that one is inherently better than the other.
Your explanation makes sense. Head calculations and conversions mean nothing to me, though; that's what computers are for.  :-) — Omegatron 18:58, 5 January 2006 (UTC)
Sorry for the confusion, Babar77. When I wrote "just use dBm", I meant that if one wants something that agrees with dBm, one should give a power ratio, not a voltage ratio (allowing that power ratio can be determined from voltage measurements). In general, I despise all use of dB to represent anything other than a power or intensity. If it were up to me, I would abolish not only dBu, but also dBV. Alas, convention and habit sometimes leave us stuck with inappropriate or confusing units. --Srleffler 02:13, 6 January 2006 (UTC)
I didn't say that dBu is gone, it's just going away and fairly quickly where I work. Largely because the only test equipment with dBu is audio analyzers, but not all. Try finding a fluke meter or any DMM these days that has dBu. Plus, even if you are using a calculator, it's still a lot easier to use dBV - less key strokes. Most of the newer mixing consoles I've seen don't even label what dB they are using.
Srleffler, if it were up to me, I would expand the use of dB, or some sort of logarithmic scale. I find using dB to be as generic as using percent, and easier. In fact, because most - if not all - of the human senses are logarithmic in nature, it makes sense to use it whenever you are dealing with human perception - like audio or light. Personally, I would go nuts if I had to design all my audio amps (voltage or power) using linear scales. I think it is safe to say the rest of the audio industry would agree, because any audio guy I've met speaks dB quite fluently. In fact if the consumer world used dB instead of Watts when spec’ing audio amps, people probably would not be getting ripped off when they spend the extra 50 bucks on a 250 Watt (24.0dBW) amp instead of buying the 225 Watt (23.5dBW) amp. Labeling the gain on a microphone preamp (which is voltage) would be a nightmare. Imagine having to label the knob from 3.16x to 1778x instead of 10dB to 65dB. Or trying to say your noise floor is 316,227 times lower than your max signal as opposed to saying your dynamic range is 110dB. I wish LED manufacturers would spec the intensity of an LED in some sort of dB instead of the linear "lumens" or "candela," because your eye is logarithmic as well. Anyway, enough of my rant - dB the planet! :-) --Babar77 06:54, 6 January 2006 (UTC)
Logarithmic units are great. What I object to is the unfortunate custom that has arisen in engineering, of calling any logarithmic ratio "dB". It leads to unnecessary confusion and mistakes, like mixing up when to use 10·log and when to use 20·log. Decibels were originally quite specific, a measure of relative power. The 20·log form only really makes sense in this context. Once you get away from dB as relative power, it's not always clear whether to use 10x or 20x. Physicists use logarithmic expressions of values all the time, we just don't call them all "dB". It's more common to simply plot log(x), or to plot x on a logarithmic scale. Anyway, this is just a minor rant, and I'm not seriously proposing to change the units everyone uses now, after the fact. I at least applaud efforts to be clear by specifying dBV, for example, when the quantity expressed is a voltage ratio that doesn't correspond to an actual power ratio (due to differing impedences). --Srleffler 07:24, 6 January 2006 (UTC)
By the way, does that microphone preamp go from 10dB to 65dB, or from 10dBV to 65dBV?--Srleffler 07:27, 6 January 2006 (UTC)
dB is being used as the generic logarithmic scale in the absence of anything else. I think the mistake was made when people agreed on the 20x for Voltage dB. They should have just left it at 10x for everything, I think it actually would have made more sense. BTW- I knew you were a physicist, you puritan! :-) That mic gain is dB not dBV because it is unitless and relative to the input signal, not a constant voltage. It's like saying, take whatever is coming in and add 20dB, or multiply it by 10. Not take what ever signal is coming in and make it 20dBV, or 10Vrms. If that was the case you would never hear a change in the volume of anything. In other words, a whisper and a shout would sound the same volume. BTW - That's called compression, which would take a whole other paper to explain completely.--Babar77 07:41, 6 January 2006 (UTC)
Just had a thought. Srleffler, you want to get together and create a generic logarithmic unit that can be used for anything? We could say that it would always be relative to 1 of the whatever the linear unit is. We can call it SrBabs or Wiki's. So the calculation would be 10*log(whatever) = 0 SrBab or 0 Wiki. So if we biked 100km, we could say that we biked 20 SrBabs(km) or 20 Wiki(km). BTW- I'm not poking fun at you, I'm just having fun in general. --Babar77 08:00, 6 January 2006 (UTC)
About dBu, dBV. one case wich always confuses people is the Vpp vs. Vrms. In most of audio systems, if not all, AC voltage is referenced as RMS -> Vrms = (Vpp/2) * sqr(2) and this is only true for sine. I have sen many crazy references for 0 dB? levels (0.25V, 0.7V, 0.775V, 1V and even 1.4V). the actual problem came when input impedences where doferent form 600 ohms. thats started in time of Tubes, as tubes are hi impedance sources/drains. so if you load typical comon katode circuit with 600 ohms signal will be really low (output impedance for 12ax7 tube in tipical circuit is about 100K). The only mesurment witch case ir .775V on 600 ohms and twice the voltage 1.55V and not 1.4V as in many in US made equpment, if ballanced (between +/- signal rails). IMHO —The preceding unsigned comment was added by 84.237.187.179 (talk) 08:40, 4 March 2007 (UTC).
I wish I could access this:
Logarithmic units: a need in acoustics
A H Davis 1934
Abstract. In view of the confusion which prevails in the use of logarithmic units, the paper suggests the use of a new unit, to be named the "brig", for the ratio of two quantities, together with certain subsidiary changes, particularly in the nomenclature of acoustics. — Omegatron 17:08, 6 January 2006 (UTC)

Back to the u in dBu

  • [This reference originally was labeled dBv (lower-case) but was too often confused with dBV (upper-case), so it was changed to dBu (for unterminated).] [6] copied? copied? copied?
  • dBu
    The reference level is 0.775 volt rms across any impedance. See dBm above for the origin of the value. The "u" stands for unterminated.     [7]
  • This reference originally was the dBv and was often confused with the dBV, so now it is called the dBu (for unterminated), [8]
  • The origin of the index of dBu is from "u = unloaded" and of dBV is from "V = 1 volt". Some say:
    The "u" in dBu implies that the load impedance is unspecified, unterminated, and is likely to be high.     [9]
  • dBu represents the level compared to 0.775 Volts RMS with an unloaded, open circuit, source (u = unloaded). [10]
  • The corresponding unit for use in circuits where the exact impedance is unknown or irrelevent is the dBu. The "u" stands for "unloaded". [11]
  • the new term dBu (u meaning unloaded) [12]

Summary: I have no idea. — Omegatron 15:28, 25 January 2006 (UTC)

Origin of bel

  • "The bel as a unit of level was originally devised by engineers working in telephony as a way measure the loss of signal amplitude in telephone wires as a function of the length of wires. Developed in 1928, the bel (named after Alexander Graham Bell) was based upon a psychophysical law first stated by Gustav Fechner in 1860 and thought to be true at the time the bel was devised. The purpose of Fechner’s law was to summarize the relation between the growth of stimulation (a physical event) and the growth of sensation (a psychological event)." [13]
  • "The decibel is derived from the less frequently used unit the bel, named in honor of Alexander Graham Bell (1847–1922). Two flux densities differ by 1 bel (10 dB) when the larger is 10 times greater than the smaller. It is to be noted that the logarithmic nature of the response of sensory organs, described in the Weber–Fechner law, underlies the definition of the bel." [14]
  • "It was noted many years ago (Fechner, 1860) that the sensitivity of the ear to changes in intensity was not related linearly to either intensity or pressure. It was believed then that the ear's sensitivity to sound intensity or sound pressure was an approximately logarithmic relationship. Initially, it was proposed that a new measure of intensity be utilised which was derived from the log (base 10) of the ratio of two intensities." [15]

Interesting. I didn't know Fechner's law was first. Should be in the history section. — Omegatron 03:37, 6 January 2006 (UTC)

For a review with historical discussion, look at: Allen, J. B., (1996) "Harvey Fletcher's role in the creation of communication acoustics" J. Acoust. Soc. Am., 99(4), pp. 1825--1839

For a modern review of Fechner and Weber, and some "new" insights on this topic, look at: Allen, J. B. and Neely, S. T. (1997), "Modeling the relation between the intensity JND and loudness for pure tones and wide--band noise," 102(6), pp. 3628--3646

 June 11, 2006

Expressed in negative?

Out of curiosity why do some stereo amplifiers express units of decibels in negatives. Example, -90bB is louder the -10dB, topping out at 0dB. --Trode 19:58, 10 January 2006 (UTC)

-10dB should be louder than -90dB, not the other way around. It just depends what the reference is, i.e. what power 0dB corresponds to. If the 0dB reference is chosen to be the maximum power the amplifier can put out, then every volume setting will be a negative number in dB.--Srleffler 22:33, 10 January 2006 (UTC)
I was wondering that too, and I found the answer here: http://tomclegg.net/decibel 143.252.80.110 15:29, 18 January 2006 (UTC)
The first consumer stereo instance I remember seeing the 0 dB reference level was on a THX pre-amplifier. The idea is to calibrate your stereo with a pink noise reference source and an SPL meter at the 0 dB level. That way when I say I watched a certain movie at the -11 dB level you will have an idea how loud it was. Supposedly the audio playback volume at THX dubbing and THX movie theaters is set to the 0 dB level. Well in theory that's the idea, in practice though I've never been to a public THX theater that is anywhere close to 0 dB. About -7 dB is more like it which is still really loud. (Spectrogram 00:10, 27 January 2006 (UTC))
Obviously when you turn a volume control clockwise it should get louder! As Srleffler points out, if 0 dB is at the maximum point, then the negative values are a measure of attenuation before the amplifier. This is the case described by tom clegg and Spectrogram. But some volume controls do have decreasing values as you move clockwise, so the -90 dB position would result in more amplification than the -10 dB position. In this case the values should be dB referenced to some standard (eg. dBu or dBV). For example, the gain controls of most Soundcraft mixing desks are marked with values of dBu. As far as I understand, the idea is that if you have an input signal of -90 dBu, you turn the knob to -90 dBu and it will provide 94 dB of amplification bringing the signal up to +4 dBu, or standard operating level. Trode, I'm not sure which case you are talking about? -90 dB is louder, but topping out at 0 dB? Pediacycle 05:15, 9 August 2007 (UTC)

No way beyond 196!

There is mention of 220 and 300 decibels in the article. This is nonsense, since the absolute maximum is 196 in earth athmosphere, sea level. This is where air becomes vacuum in the dilutated phase of the soundwave and thus cannot be increased further. 195.70.32.136 13:34, 26 January 2006 (UTC)

I agree that the numbers are dubious, however, I don't think there's an inherent limit to the sound pressure level. Of course vacuum is a negative pressure limit, but the only limit in the positive pressure direction would be the point where air becomes liquid? So you could have asymmetrical waves with much higher pressures, right?
  • Yet another amusing calculation reveals that with a sinusoid of 196dB SPL, the rarefying part of the fluctuation reaches vacuum. This is the theoretical limit on sinusoidal pressure fluctuations in normal atmospheric pressure, then. (Compressive impulses can, of course, reach much higher SPLs; cf. the hydrogen bomb.) [16]
  • Sound is technically at its upper limit at 194.09 dB. Above this level it should be called a shock wave. Sound — Omegatron 15:20, 26 January 2006 (UTC)
I have always thought the upper limit in one atmosphere was 196 dB SPL and anything higher would be considered an overpressure. A sealed chamber can have greater than 1 ATM so in that special case a higher max SPL would be possible. Anybody know what the max SPL in sea water is? Base SPL levels are very different in water than in air. Liquid dB SPL's probably would make a nice wiki section topic if it isn't someplace else already. (Spectrogram 23:40, 26 January 2006 (UTC))


The math:

  1. vacuum = 0 Pa = 0 atm, so pressure difference is 1 atm = 101,325 pascal [17]
  2. L_\mathrm{p}=20\, \log_{10}\left(\frac{p_1}{p_0}\right)=10\, \log_{10}\left(\frac{{p_1}^2}{{p_0}^2}\right) \mbox{ dB SPL}
    where p0 is the reference sound pressure and p1 is the sound pressure being measured. Sound pressure level
  3. Reference for SPL is 2×10−5 Pa
  4. L_\mathrm{p}=20\, \log_{10}\left(\frac{101,325}{2\times10^{-5}}\right)= 194.09 \mbox{ dB SPL}


I changed the article regarding values above 194 dB SPL. The list at makeitlouder.com from which the values like 200--240 dB for explosions, tornadoes, earthquakes apparently are taken uses various nontrivial approaches to get these numbers as a means to compare violent processes, for example the total released energy per second (including heat), the wind speed compared to the velocity of the air inside an acoustic wave, and so on. It makes no sense to put these numbers into a dB SPL comparison table as an absolute truth. Han-Kwang 12:24, 1 April 2006 (UTC)

I put in "Although pressures higher than 2 atm are possible in air", but is that correct? If a sound wave is at 194 dB SPL, the rarefactions will be at vacuum, but where will the compressions be? I assumed it would be 2 atm in a linear manner, but I could be wrong. — Omegatron 16:00, 1 April 2006 (UTC)
In order to get wave propagation in a medium, you need a linear compressibility, i.e., dp/dV = K (K is a constant). In a gas, the actual relation is IIRC dp/dV = K/V^(gamma+1), where gamma is typically 1.6 for a diatomic gas. If you imagine a two equal slabs of gas stacked on top of each other in a confined space, and expand one of the slab with a factor 1.9, then its pressure will be about 36% of the original. The other slab will be compressed with a factor 10 and its pressure will increase with a factor 40. So a true vacuum inside the rarefactions is impossible. Put in a different way: if you assume a sinusoidal pressure change over position, then you need to change the number of gas molecules in the system, since the decrease in density at the lower parts of the sine wave is larger than the gain in the higher parts. Since this is impossible, the wave must have a shape that differs considerably from a pure sine. Because of this asymmetry, this type of disturbance (shock wave) changes shape during propagation, unlike normal sound waves. Anyway, this is more about the definition of sound. As you said yourself above, it's not called sound anymore. I feel that sound is something with a well-defined frequency spectrum that is reasonably conserved during propagation, and that is not the case with a shockwave. Hence I object against the use of the term sound pressure level for a shockwave. The question is where you put the boundary between normal sound and shock waves, which is a matter of taste, but I'd guess that it will end up +/-6 dB from the 194 dB that you get from the naive calculation discussed before. Han-Kwang 19:42, 2 April 2006 (UTC)

-3 db

-3dB = 20 log (1/sqrt(2)) not 20 log (.5)

This is a very bad approximation in the table since -3dB is such an important value

This is an egregious error

Scott Curran

I'm not sure where you're looking. 20 log(0.5) would indeed be wrong. -3 dB is 10 log(0.5). Perhaps the tables of ratios at the very end of the article confused you? The ratios are power ratios not voltage ratios, since dB (as opposed to dBV) is a measure of relative power not voltage. --Srleffler 22:24, 4 February 2006 (UTC)

hrrm... what would -3 dB be close to? a pin drop? soft music? (Please respond on my talk page.) tinlv7 21:51, 25 September 2006 (UTC)

Clarifying factor of 10 vs 20

I tried to revise a bit of the definition to make the difference between intensity and pressure equations a bit more understandable, but when I saved the equation did not render and the entire page seemed to become bold font, so I reverted to the previous edit. If someone wants to give this another try, please do.

In electrical circuits, the dissipated power is typically proportional to the square of the voltage V, and for sound waves, the transmitted power is proportional to the square of the pressure amplitude p. Effective pressure is related to intensity by the following equation:

I = p_e^2 / \rho_0 c

Substituting a measured voltage or pressure and a reference voltage or pressure and rearranging terms leads to the following equations and accounts for the difference between the multiplier of 10 for intensity or power and 20 for voltage or pressure:

V_\mathrm{dB} = 20 \log_{10} \left (\frac{V_1}{V_0} \right ) \quad \mathrm{or} \quad p_\mathrm{dB} = 20 \log_{10} \left (\frac{p_1}{p_0} \right )\ ,

etc. The added equation is on p.125 of Kinsler and Frye's "Fundamentals of Acoustics" 2nd edition, 1962. Wesley R. Elsberry 18:46, 12 April 2006 (UTC)


I think it's better to put all the mathematical derivations somewhere in a separate section. The equation pe^2/rho0 c isn't very helpful in the current form, i.e. without explaining what all the symbols mean (and you appear to change notation from p to p_e). OTOH, it would distract from the main point if you do explain all the symbols at the current location. Han-Kwang 21:26, 12 April 2006 (UTC)
This issue came up because someone altered the article page to ask about why 10 was used in one place, and 20 in another. While putting a question in the article is the wrong way to go about things, that person did have a point. If we defer the derivation, then we have people wondering why there is suddenly a different multiplier in the equation to obtain a decibel value. On the other hand, I can see that it would be nice to be able to offload the gory details to its own section. I'm not sure how we handle this to make it right for everyone. I'll note that I really liked how Omegatron handled putting in the info as it is now, with the explanation of the terms in the new equation added. Wesley R. Elsberry 00:14, 13 April 2006 (UTC)

Thank you for using correct dimensions!

After skimming through the article, I'm delighted to see that the expressions are dimensionally sound. That is, the decibel values aren't confused with the powers themselves, and the arguments of the logarithms are all dimensionless and nice. I was fearing a dimensional mess, since that is how it usually is, in my experience. (The index "dB" is even written upright. Great!) Bromskloss 16:58, 26 April 2006 (UTC)

We do our best.  :-) It certainly wasn't always this good...
I'm sure there are still a few problems somewhere. If you can find them, we would be happy to know. — Omegatron 18:27, 26 April 2006 (UTC)

real world examples

Since I'm not really familiar with this technical topic, I'll suggest this here first. I think a list of real world examples would be useful for lay people, similar to the one at the bottom of this page: [18]. I'll probably add something in a few days if no one else does. Thanks. --W.marsh 20:55, 17 June 2006 (UTC)

Yes this would be good! Could you add it please? But please dont copy it verbatim!--Light current 00:03, 18 June 2006 (UTC)

Nitpicking

In the Stevens 1957 quote, is it "100 db" or "100 dB"? Was the original capitalization different? — Omegatron 15:55, 16 July 2006 (UTC)

Definition

This section was a lot better a while ago, when it explained the 10/20 difference and power/intensity ratios and voltage/pressure ratios all in the same place. Now it's just confusing. Can someone try to get it back to the shape it was in? — Omegatron 12:33, 20 July 2006 (UTC)


194 dB

I am writting this in reference to the 200 dB that was labled as the limit a sound wave can't exceed. When the information above it clearly states that the threshold is 194.09. I understand the table is in multiples of 10... but it is not correct to state it as 200 when it is 194. Decibels are logarithmic so it does make a substantial difference. I changed it to 194 so that it matches the data given previously.--Tobyw87 02:55, 12 August 2006 (UTC)

Decibels on MythBusters

A MythBusters episode featured a car subwoofer that generated 161 dB at 16 Hz. It blew out the sunroof, but the car survived.

Mergefrom Sound power level

...for obvious reasons. Pan Dan 22:07, 9 September 2006 (UTC)

Please dont merge these two articles. While sound power level is a physical quantity, decibel is a unit of measurement - a totally different thing. Instead, improve the articles to reflect that fact. --84.179.128.129 13:26, 25 September 2006 (UTC)
Actually, isn't sound power a physical quantity, and sound power level a unit of measurement? — Omegatron 16:18, 25 September 2006 (UTC)
No, not at all. Sound power level is a logarithmic physical quantity, derived from sound power, which is a physical quantity, too. The possible units of measurement for all kind of levels are Neper or Bel resp. Decibel. --84.179.145.11 12:31, 2 October 2006 (UTC)

After more carefully reading Decibel, Sound power, and Sound power level, it is clear to me that this was an ill-advised merge proposal. My apologies. I'm going to remove the tags. Pan Dan 15:10, 7 October 2006 (UTC)

A practical example

A fictional 2 way speaker (A box with separate driver for high ("Treble") and low("Bass") ) has the following specs:

High driver: 92dBSPL @ 1W @ 1m. A Low driver: 86dBSPL @ 1W @ 1m. B

Now if we want to match the output of the two speakers so the sound is "equally loud" we need to do the following:

Get the difference between the two by subtracting the sensitivity:

                           Difference in sensitivity = A-B 
                           = 92 dBSPL - 86 dBSPL
                           = 6 dBSPL
    • As we concluded earlier this 6dB difference requires that we double the power delivered to the low driver.**

This is incorrect. A 6 dB SPL difference requires not double but four times the power.

    • Since a doubling in [power] relates to 3dB, we need to adjust the cross-over unit in this system so that the [gain] of the Low signal is 3dB more than the Highs. If there is no crossover you can always adjust the Amplifier's output to be 3dB more.

This is incorrect. We need to adjust the crossover or amplifier ouptut by 6 dB = difference in sensitivity.

It matters not whether it is SPL or power. Mathematically, it works out that a change in SPL in terms of dB equals the same dB change in the power output of an amplifier. Thus a 6 dB increase in SPL requires a 6 dB increase in amplifier power.

Another way to view this is that a 6 dB change in SPL requires twice the voltage input to the loudspeaker (voltage being equivalent to pressure). Doubling the voltage from a power amplifier equates to 4 times the power output. A 3 dB increase in amplifier power results in 1.414 times change in its voltage output, and thus only a 3 dB increase in SPL.

Chuck McGregor Senior Technologist Eastern Acoustic Works (EAW) 12.38.109.253 15:46, 11 September 2006 (UTC)

Need help with Fin Whale article

I'm revising a section of the Fin Whale article that deals with the animal's vocalizations. One of the sources cited states that the vocalizations have been up to 186 decibels relative to 1 μPa at 1 meter (and can be heard for hundreds of miles). I'd like to add a comment for the reader to put that into perspective and am having diffculty wrapping my mind around the scientific information in this article, especially the fact that the sound is being made in water and the μPa is different . If I was a theoretical diver in the water next to one of these animals when it made these sounds, what accurate statement or comparison could I add that would illustrate the magnitude of the volume of these vocalizations (i.e., would be louder than standing next to a jet engine at full thrust, etc.). Let's assume that these vocalizations are at frequencies within the normal range of human hearing. Any help would be appreciated. Neil916 (Talk) 16:52, 25 October 2006 (UTC)

Never mind, I found my own answer. Google knows everything! Neil916 (Talk) 17:28, 26 October 2006 (UTC)

Example vs Acoustics

Note, the example given in the Example section does not agree with the Acoustics section.

Example in 6 dB per Bit section inconsistent

The example quotes a dynamic range of 90 dB for 16-bit linear PCM, which is inconsistent with equivalent calculations giving 96 dB on pages Signal-to-noise ratio and Quantization noise. I believe the problem is that the quantization noise range is of amplitude 0.5 LSB, where this example assumes 1.0 LSB. --ElectroTorx 12:00, 6 December 2006 (UTC)

"6 dB per bit" is an approximation. Also, the amount of noise depends on the signal, so estimates vary as signal models vary. — Omegatron 15:30, 26 February 2007 (UTC)

The chapter and the previous explanations are just not right. Where to begin? When determing a signal-to-noise ratio or dynamic range of a system, a full-scale sine signal is compared to the background noise or the power of the lowest sine signal that can be represented within the system. So there is no ambiguity in the comparison signal.

Also, the article's claim that the LSB has some special properties, i.e. "the first bit (least significant bit, or LSB) produces residual quantization noise (bearing little resemblance to the source signal)" is just not true. To have a meaningful definition of noise you need the LSB to act pseudo-randomly. In systems with very few bits (and white noise levels much lower than the LSB) this is not true, and thus e.g. speech encoded into 3 bits doesn't sound only noisy, it sounds harsh, and completely different from an analog signal with masses of noise, and any theoretical S/N figures just aren't useful because the noise is not white in nature. All kinds of funny things also happen in the frequency domain.

However, as bits are increased the LSBs starts more and more to act like white noise and meaningful S/N figures BOTH can be calculated AND they are useful. Any signal communications book show, and with good reason, that the signal-to-noise ratio of an N-bit PCM system is roughly (1.77 + 6.02N) dB. Thus the S/N ratio of a 16-bit system is roughly 98 dB (yes, 98, not 96, and definitely no 90). Simple demonstration Matlab / Octave code follows:

> q=sin(0:0.1:1000000);q2=round(q*32767)/32767;
> qdif=q-q2;
> 20*log10(sqrt(sum(qdif.*qdif)))-20*log10(sqrt(sum(q.*q)))
ans = -98.096

Dynamic range is more open to debate, but this is the first time ever that I've seen DR defined as something else than the S/N ratio in a linear PCM system. There are no references either. I am going to check my sources, and when I've done that, I am going to correct this part of the otherwise informative page. RealLeo (talk) 10:12, 11 December 2007 (UTC)

Understanding Decibels

Since decibel is so widely used, and each usage varies from the next, you must assume it will be very confusing to someone who is unfamiliar with the topic and all the different usages.

Stating that decible is a ratio is entirely incorrect. Decibels are defined by a pair of related formulas involving a pair of related quantities. The fact that division happens to be used in both formulas does not make decibel a ratio any more than the fact that logarithms are used makes decibel a logarithm.

Since decibel has such varied usage, and hence "definition" per se, it makes more sense to focus on the core of what it is, how it is correctly used (which formula), and then wander down the path of various usage examples and their syntax.

Since this is my first time editing a wikipedia article, I did not invest the time to do wide sweeping changes to the intire article, instead I just concentrated on correcting the first paragraph. If you think the change of direction is warranted, please let me know and I'll continue.

J.C. Roberts 11:25, 27 January 2007 (UTC)

Joe 05:29, 24 August 2007 (UTC)Even the ham radio manuals contribute to dB confusion. In my opinion, after preparing for tests in three levels of ham radio (technician, General, and Amateur Extra), the American Radio Relay League's manuals help to keep the public mystified on the concept of decibels. I found explanations in all three manuals severely wanting, and for such a basic concept to radio, pretty inexcusible. There are several websites that provide a much better framework, IMO, which are easily found just by typing in the words "calculating decibels." It seemed to me that one of the problems in the Amateur Extra manual was an illustration that was scaled in a way that did not show the correct relationship, making it difficult for even someone with good mathematics training to follow the examples in the text. Woe to the test-taker that draws an exam question on calculating decibels, if they rely solely on the test manuals, IMO!

120 causes ear drum rupture?

How is that possible? The M16 being shot is 160 decibels and you hold it right up to your ear. I don't seem to recall a rash of broken ear drums being mentioned as a side effect. I find this number difficult to believe. Maury 13:12, 12 February 2007 (UTC)

Is it bureau or committee?

The introduction states "the International Committee for Weights and Measures (BIPM) has...". However BIPM refers to the International Bureau for Weights and Measures, not the "Committee". Could someone please determine which organization the author was referring to. Hanjabba 20:55, 23 March 2007 (UTC)

BIPM stands for Bureau international des poids et mesures. CIPM stands for Comité international des poids et mesures.
See section 3 of http://www.bipm.org/utils/common/pdf/CCU15.pdf — Omegatron 22:08, 23 March 2007 (UTC)
According to the .pdf above, the article should be referring to the "committee" (i.e. CIPM). The in-line link for "International Committee for Weights and Measures", however, directs you to the bureau. I will change this link so that it redirects to the committee article instead. Feel free to revert if I am doing this in error.

Hanjabba 02:48, 24 March 2007 (UTC)

Power and voltage measurements

I was going to tidy the following paragraph, then I realised I didn't understand it:

"It should be clarified to the novice that although 3dBW in power is a 10-fold increase(decrease) - when working in voltage the formula becomes 20 x Log of (Vout/Vin) and hence 10 fold VOLTAGE increases yield 6DB(mv et al) as opposed to only 3dB with power formula). Ref; that formula for power is E-squared (over R)."

Doesn't this contradict the earlier statement that, for power measurements, there is "approximately a 3 dB increase (decrease) for every factor 2 increase (decrease)"? And, in the same vein, don't the earlier definitions imply an approx. 6dB voltage increase for a doubling of voltage? Where does the 10-fold increase come from? I'm not clear if this paragraph is attempting to re-explain what's described in the "Definitions" section, or trying to say something new. Matt 20:29, 8 June 2007 (UTC)


I agree. This section is factually incorrect, and should be revised or removed. It clearly contradicts the definitions stated previously, and only adds confusion to this article, especially for the "novice".

To correct this, the references to 3dB and 6dB could be changed to 10dB and 20dB respectively. In addition, the 3dBW reference in the first sentence is an absolute power measure, not a relative one as is suggested by "10-fold increase(decrease)". To correct this, either the W should be eliminated from dBW, or the increase should be stated as being relative to the Watt, and not an increase in general.

To further clarify this, I would suggest the use of +/- signs to indicate a purely relative increase or decrease. For example, the first sentence could be rewritten as, "... although +10dB and -10dB in power refer to an increase and decrease in power by a factor of 10, respectively ..."

SethHoyt 19:12, 15 June 2007 (UTC)

  • Since we both think it's wrong, I have for now removed it and replaced with a note that the definitions are different for power and voltage, referring readers back to the Definitions section where that is explained. IMO any further (corrected) examples should go in the Definitions section, otherwise it gives the impression that the meaning is somehow being redefined specially for dBu, dBV etc., over and above what was explained earlier. Matt 01:38, 21 June 2007 (UTC).


Unit vs. dimension

Regarding the formula XdB = 10log10(X / X0), the article says "X0 is a specified or implied reference level (in the same units as X)". This isn't quite correct as it is not meaningful to say that X or X0 "is" in any particular units. For example, if I handed you a piece of string and asked you in what units its length was, what would you answer? What the article probably should say instead is that X0 should be of the same dimension as X, so that X / X0 is dimensionless. I did make this change but was struck back by an anonymous editor and rather than engage in an edit battle over that, I figured I'd take it up with everyone here. So, what do you say? —Bromskloss 09:04, 12 July 2007 (UTC)

Precisely so, but alas not relevant to the purpose of an article intended in part for people who don't know or care what a dimension is. The ignorant will think they know what a "unit" is, and as far as it affects the meaning of dB, they'll be close enough, while if they mistake the meaning of "dimension" they may be way off. If the concept needs further precision, then it also needs a longer explanation. If string were power, cutting away half a string would attenuate it by 3 dB regardless of whether it were measured in millimeters or furlongs, and that's the point this passage is trying to communicate. Jim.henderson 21:03, 12 July 2007 (UTC)
In my view the current text "which must have the same dimensions as X" is likely to confuse more than it illuminates. I believe that many people (especially people who do not already know what a decibel is) will not know what "dimension" means in this context, and will associate it with "two-dimensional", "three-dimensional" etc. On the other hand, it is far more obvious what "measured in the same units" means. I cannot conceive of any practical application of the formula where the units of measurement are not known. Indeed, as it is written, the units must be known (even though they "cancel out") in order to supply the actual numerical values of X and X_0 to the formula. The possibility of knowing the ratio but not fixing the units is a theoretical one that I don't believe we need to trouble readers with at this point in the explanation. Matt 02:26, 17 July 2007 (UTC).
The distinction between "dimension" and "unit" is only one of abstraction, but here it is a crucial one. If people understand the concept of "unit" (eg, 1 m) it is a simple step to generalise this to "dimension" (any distance, whether measured in metres, inches or lightyears). I think we should stick with "dimension" and explain more carefully what it means (here conceptually, and in detail in the dimension article). The main problem I see with Matt's argument is that the units are irrelevant, precisely *because* they cancel. Thunderbird2 11:08, 17 July 2007 (UTC)
Yes, but in order to actually perform the arithmetic indicated by the formula you need actual numbers, which entails identifying an actual unit of measurement, and moreover the units of X and X_0 must be the same. Let's take the example given of length. Suppose that X is 3 metres and X_0 is 10 feet. X and X_0 have the same dimension (i.e. length), but the calculation 3/10 results in garbage because the units of measurement are not the same. Matt 22:07, 17 July 2007 (UTC).
Hi Matt. The way I see is like this. The correct calculation is not 3/10, but (3 metres)/(10 feet)=~0.984. The ratio is dimensionless and therefore doesn't depend on the units used. Thunderbird2 18:58, 18 July 2007 (UTC)
Well, that's true of course. However, I was referring to "actually performing the arithmetic". To numerically evaluate (3 metres)/(10 feet) you need to either convert metres to feet, or feet to metres (or you could convert both to, say, cubits, but that would be perverse!). That is, you must make the units the same. Maybe we'll have to agree to disagree here. Bottom line is that I believe saying the units must be the same adds a ton of clarity — and the loss of some theoretical generality that is not particularly important in the context is a price well worth paying. Maybe the dimensionality issue could be made into a separate point somewhere, so that it doesn't intrude into the explanation for people who simply want to understand how to use the formula, but is there for people who want it? Matt 02:30, 19 July 2007 (UTC).
I agree except for one detail. Instead of saying "to achieve this the units must be the same" I would be prefer "this can be achieved by making the units the same". Thunderbird2 20:12, 19 July 2007 (UTC)
How to perform a division (or addition, multiplication etc.) is something I don't think we should talk about at all in the article. Help on unit conversion and how to arrange the numbers on a paper for calculation, the reader would have to look for elsewhere. —Bromskloss 14:40, 22 July 2007 (UTC)
I entirely agree that the mechanics of long division and unit conversion are not topics for this article. But, bearing in mind that this is a general-readership encyclopedia, do you not feel that someone unclear about how to calculate decibels, and coming to this article for information, will find a statement that the "units should be the same" far clearer and more helpful than a statement that the "dimensions should be the same"? I certainly do. How would you feel about a compromise statement that retains the mention of "dimension" but somehow also works in a statement along the lines that Thunderbird2 has suggested? Matt 20:31, 22 July 2007 (UTC).
How about saying that X and X0 should both be powers (if that's true)? That would be both simple and correct. —Bromskloss 10:56, 23 July 2007 (UTC)
I would argue that the most general case would be energy rather than power. Or, better, a quantity that is proportional to energy. In other words, if you double it, you also double the energy. Power qualifies, because if you double it (and keep the time duration fixed), you double the energy Thunderbird2 16:24, 23 July 2007 (UTC)
The intro to this part specifically says that this calculation applies to power or intensity. So, X and X_0 needn't just measure power, they could, also, for example, measure power per unit area. I'm not aware that energy ratios are generally measured in decibels. How about the following:
...where X is the actual value of the quantity being measured, X0 is a specified or implied reference level, and then XdB is the quantity expressed in units of decibels, relative to X0. Which reference is used depends on convention and context (see later in this article). X and X0 must have the same dimension (that is, must measure the same type of quantity), and must as necessary be converted to the same units in order to calculate the ratio. The reference level itself is always at 0 dB, as shown by setting X = X0 in the above equation. If X is greater than X0 then XdB is positive; if X is less than X0 then XdB is negative.
Matt 11:05, 26 July 2007 (UTC).
I agree that power is more common. For the wording, I suggest replacing Matt's "... converted to the same units in order to calculate the ratio." with "... converted to the same units before calculating the ratio of their numerical values." Plus I think the word "dimension" is better in the plural (dimensions). The rest looks fine. Thunderbird2 15:28, 26 July 2007 (UTC)
Agreed, and done! Matt 23:54, 26 July 2007 (UTC)
Great. Thanks for taking the trouble Thunderbird2 10:25, 27 July 2007 (UTC)

Yes. After all that foofraw, I am pleasantly surprised that the idea could be presented so neatly. Sometimes it takes very careful thinking to say something simple. Jim.henderson 05:32, 31 July 2007 (UTC)

Perceived loudness

The article says:

"...a 3 dB increment (approximately double the sound power) is also approximately double the perceived noise to an average human observer in the normal hearing range of frequency and sound power."

I'm not convinced about this. The "dBA ratings" section says, for example, that a soft whisper at 5m is rated at 30 dB, and a normal conversation at 60 dB, so by that reckoning the normal conversation ought to sound about 1,000 times louder than the whisper. That factor seems too high. At http://www.phys.unsw.edu.au/jw/dB.html it says "Experimentally it was found that a 10 dB increase in sound level corresponds approximately to a perceived doubling of loudness" which seems more reasonable to me. Does anyone else have a view? Matt 20:10, 1 August 2007 (UTC).


As was discused further up the page #dB_Doubling_Versus_.22Perceived.22_Doubling, a 10 dB change in sound pressure will be commonly perceived as an approximate doubling / halving of 'volume.' I don't believe this could ever be measured exactly though, because it depends on individual perception of volume. And it may be dependent on frequency?

However, the introduction now reads:

"...a 10 dB increment (approximately double the sound power) is also approximately double the perceived noise to an average human observer..."

which is incorrect. A 10 dB increase in power is 10 times the original value. Pediacycle 04:07, 9 August 2007 (UTC)

This discussion would make more sense if someone can define "perceived loudness". I have a feeling I deleted the relevant sentence about a week ago, mainly because it referred to "pressure" as a measure of "volume". (Sorry, I didn't notice this discussion at the time). But if someone can define it (perceived loudness I mean), I don't so why a remark like this shouldn't be included. Thunderbird2 11:26, 17 August 2007 (UTC)
Since it's a perception question, it is psychoacoustics. As I understand it, such questions are investigated with a bunch of ignorant test subjects. Play two notes, and they say, "Yes, the second is twice as loud as the first" or "No, the first is more than twice as loud than the second" or whatever, with variations in pitch, timbre, sequence, open air, earmuff headphones, etc to level the playing field or sounding board. When about as many test subjects off the street say it's more than twice, as say less than twice, then that's a statistical, psychoacoustic measure of "twice as loud". And as I understand it, the usual result is, the median power ratio at which they say so, is generally around 9 or 10 dB with a roughly Gaussian distribution on a dB scale (and much more skewed on a linear power or linear pressure curve) and a standard deviation of 2 or 3 dB. However, I've got no reference for it. Jim.henderson 00:07, 18 August 2007 (UTC)
It suggests in the loudness article that loudness should not be measured in decibels, so perhaps the sentence - if reinserted - belongs there rather than here. Thunderbird2 09:47, 18 August 2007 (UTC)

Ah. Thank you. Well, actually

Loudness, a subjective measure, is often confused with objective measures of sound pressure such as decibels.

So, among other things, it's saying dB is an objective measure of sound pressure. Anyway it shows a slightly poorly labeled chart relating subjective loudness to frequency and objective amplitude. Surely, somewhere, simpler tests can be discovered by which the term "twice as loud" was explored by similar subjective, statistical testing. Jim.henderson 00:47, 19 August 2007 (UTC)

That sentence (not yours - the one from the loudness article) is poorly worded. The decibel is not a measure of pressure or anything else; it is a unit in which measures of pressure (like sound pressure level) can be expressed. (Although I'm sure you can find similarly poorly worded sentences here too.) The main point is that a statement about loudness belongs in the loudness article. Do you agree? Thunderbird2 10:01, 19 August 2007 (UTC)
I think that a short statement about the relationship between decibels and perceived loudness is quite appropriate for this article. "Loudness" is, after all, probably the most familiar application (or misapplication) of the decibel for the average lay reader. A link could be provided to the loudness article for further detail (though I don't really understand that article; the relationship between phons and decibels looks linear - is that right?) Matt 03:26, 23 August 2007 (UTC). —The preceding unsigned comment was added by 86.134.10.96 (talk)

this page could be better 67.131.76.100 19:44, 10 October 2007 (UTC)


Link removed

I got a notification that the link I recently put was removed by a moderator, and that I shouldn't resubmit it, but rather discuss it here. The link I suggested blind tests a 1dB difference in audio levels.

One decibel blind test

Most links in the Decibel section tend to cover the same theoretical issues. In contrast with these, I thought it could have been nice to _intuitively_ illustrate what a 1db difference. I appologize if submitting this link offended someone. I was beleiving (and I still do ;-) this link was a nice contribution. —Preceding unsigned comment added by 81.240.0.50 (talk) 23:37, 27 December 2007 (UTC)

Ah, I edited too quickly, and hadn't seen that you also had edited too quickly and entered the link address incorrectly as http://http://www.audiocheck.net/blindtests_level.php?lvl=1, and so I reverted it, as it caused a link to a linkfarm.
My apologies; I will put back the correct link in due course (or you can do so if you feel like it). Oli Filth(talk) 23:44, 27 December 2007 (UTC)
I now understand the band link issue. Indeed, I possibly made an error when pasting the url into the code... hence the double http. Thanks Oli for your watchfulness! —Preceding unsigned comment added by 81.240.0.50 (talk) 08:46, 28 December 2007 (UTC)

dBFS and RMS

Is there really a requirement that dBFS be instantaneous? RMS measurements would make sense, too, such as measurements of a noise floor.

  • Here is a document that refers to noise floors in dBFS. An instantaneous measurement of noise would be meaningless, since noise has an amplitude that varies randomly and can theoretically hit any value on rare occasions.
  • "The level of inherent noise in our ADI-1's 20-bit ADC is about -100 dBFS (RMS unweighted)" [19]
  • This document has a lot of "dBFs(RMS)" measurements

— Omegatron 19:14, 8 March 2007 (UTC)

I'm sure that dBFS can be either a peak or rms measurement, depending on the context. I think that when the original author said "this sample" they were refering to the aforementioned 0 dBFS measurement, not dBFS in general. 0 dBFS is of course the maximum instantaneous (peak) level that can be represented in digital audio, the only signal with a 0 dBFS RMS level would be a perfect square wave. Pediacycle 05:43, 17 August 2007 (UTC)

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.