#### ThomasC

Senior HTF Member
I'm a little confused about how frequency response is measured. For example, here's the frequency response for the SVS SBS-01 speaker:

68Hz - 20kHz ± 3dB

I understand the first part (68Hz - 20kHz), but I've read a lot of explanations about the decibel fluctuation part and I'm still not clear on what it means. The glossary that is stickied at the top of the basics index has a definition of frequency response, and this is a snippet:

"A measurement of 20 to 20,000 Hz ±3 dB means that frequencies between 20 Hz and 20,000 Hz can be reproduced no more than 3 dB too loud (twice as loud) or 3 dB too soft (half as loud)."

This is taken from a Wikipedia entry on audio system measurements:

"The signal should be passed at least over the audible range (usually quoted as 20 Hz to 20 kHz) with no significant peaks or troughs. The human ear can discern differences in level of about 3 dB, so peaks and troughs must be less than this. Modern equipment is capable of less than ±1 dB variation over the quoted frequency range. Rapid variations over a small frequency range (ripple), or very steep rolloffs are considered undesirable as they can correspond to resonances associated with energy storage which produce delayed echoes and hence colouration on the sound."

So does that mean that when I'm listening to something, the audio signal may be fluctuating within a 6 dB range and I won't be able to tell that the volume is changing very slightly?

I was reading an article on Bose which sparked my curiosity about fluctuation - from what I've read, the typical fluctuation for a Bose speaker is ±10 dB. The Wikipedia article says that rapid variations or steep rolloffs can cause colouration - is colouration what people hear when the Bose speakers achieve a quasi-surround effect?

Thanks for any help.

#### JeremyErwin

Senior HTF Member

Bose designs weird, unconventional loudspeakers. It's reputation, for good or ill, is based on the 901, which had lots of tiny 'full range' drivers-- eight on the rear, and one on the front. Most of the sound the listener heard was reflected off walls-- analogous to a concert hall environment, perhaps. This created a wide, though perhaps inaccurate sound-stage.

Stereophile's review, which sadly, does not include information of frequency response, essentially opines that this is wishfu thinking, and the Bose speakers essentially simulate a comb filter, a technique well known to recording engineers, who used it to mix pseudostereo sound from a mono source.

Most acoustic measurements are ideally taken in a anechoic chamber. But

The problem is that Bose's design is contingent on having a room response, and so anechoic measurements would seem to reflect, quite poorly, real world response.

Of course those are the 901's (which interestingly enough came with custom equalizers), and not the am-15, which presumably operates on still another strange acoustical theory. It surely can't be as simple as building a really cheap, tiny speaker? After all, there are patents involved. Maybe they're "design patents".

#### ThomasC

Senior HTF Member
Thanks, but I don't think that helped much in my understanding of fluctuation.

#### Jeff Gatie

Senior HTF Member

It's simply a matter of further defining the frequency range of a speaker. If a speaker is "flat" along it's frequency range, it is a better performer than one that is not flat. Given a car, what is a better determination of performance -

Car 'A' goes 0-155. Period
Car 'B' Goes 0-60 in 4.5 seconds and 0-120 in 12.2 seconds, with a top speed of 155.

Car 'B' obviously has a stricter method of performance evaluation applied and thusly you can tell more about it's performance. Now let's do it with speakers -

Sub 'A' has a frequency response of 16Hz-150Hz. Period.
Sub 'B' has a frequency response of 16Hz-150Hz +/- 3dB at 105 dB from 1 meter in an enclosed area.

Which sub has the stricter performance evaluation applied? Which sub is more likely to be performing sufficiently at 16Hz vs. one which could be outputting a barely audible -40dB signal at 16Hz? Which one can we definitively say has a "flat" frequency response from 16-150Hz? Which of these numbers are usable for choosing a product and which would you be cautious of using as a criteria for purchase?

That's all the +/- 3dB rating is for. It is simply a method used to standardize and quantify the measuring of frequency response.

#### Cees Alons

Moderator
Senior HTF Member
Because supposedly the human ear won't hear much difference (if any at all) if the sound level happens to be 3dB too high or too low, they tell you the range of a "flat" frequency curve within those limits (+ or - 3 dB).

Generally a frequency curve is flatter than that, and the 3 dB is mainly used to define the end-points (e.g. below 68Hz, in your example, the level will be lower than 3 dB below the rest, while above 20KHz it will start getting that low too).

Of course they can still misrepresent the truth, e.g. by plainly lying, or when the curve isn't that neat (e.g. slightly sloping) and they give cleverly chosen begin- and end-points.

Your fear of "fluctuations" is unfounded, fortunately: the differences may occur at different frequencies (a specific frequency might be amplified 1 dB less than another specific frequency), but the amplifier is never supposed to have different amplification at any of the frequencies in the domain over time (well perhaps a very long time, when it gets older).
Thus if you would input a steady test-tone, it will come out steady and is not supposed to fluctuate between -3dB and +3 dB.

Cees

#### ThomasC

Senior HTF Member
Thanks, everyone! That clears it up for me.

#### Jeff Gatie

Senior HTF Member

Good point Cees. I automatically look at the endpoints as where the +/- 3dB applies 99% of the time and didn't see the confusion that the newbie may have thinking there is 6dB "fluctuation". Thanks for clearing it up that for most quality speakers, the +/- 3dB is almost always in the minus direction and almost always located at the ends of the freq. resp.

#### JeremyErwin

Senior HTF Member

Imaging is pretty important in a speaker as well.

#### chuckg

##### Supporting Actor
I don't think any of the responses above truly give an answer to the question. No offense, but nobody defined any of the qualities under measurement.

Simply:

Frequency is the "note." A piano usually hass 88 keys, with the low notes around 40 or 80 Hertz (vibrations per second) and the highest notes up around 8,000 Hertz. You can easily imagine that there are other notes on other instruments, and infer that there are sounds whose frequency is between notes on a conventional instrument. There are sounds (like saying "ess") that make sounds much higher in frequency than the basic note played. The human ear is generally able to hear frequencies from 20 Hertz to 20,000 Hertz.

Loudness is measured in deciBels. The Bel, named after Alexander Graham Bell, is the base unit, and a deciBel is one-tenth of a Bel. Going up from 1 Bel to 2 Bel doubles the perceived loudness - an increase of 10 deciBels (not 3). Goiing up another Bel again doubles the perceived loudness. To increase loudness by 3 dB requires twice the electrical power input. To make your sound system play twice as loud requires nearly ten times the amplifier power.

Now, putting those two things together and dividing by the answer to the ultimate question of life, the universe, and everything, we can get some idea what the frequency response curve of a loudspeaker really means.

You want each note to be played at the appropriate loudness, but that is difficult to do. Some notes wind up louder than others, some quieter. Speaker engineers have decided that they will measure the loudness of each frequency, and give you a short report on how well the speaker performs.

Low notes are much more difficult to reproduce than high notes, and so often the frequency response report will begin at some frequency higher than 20 Hertz. Most folks consider that a difference in loudness between two notes of 3 dB will largely go unnoticed, so they tell you the range of frequencies that the speaker can produce given that limitation. All speakers will reproduce low notes, but it takes bigger speakers and much more power to make the low notes as loud as the high notes.

Nearly any speaker can easily reproduce the higher frequencies, so the report goes up to 20,000 by convention. Some speaker makers lately have been adding "super tweeters" that reproduce even higher frequencies, but in my opinion (and many audiophiles would disagree) this is just hype.

Some speaker makers will decide that having a lower dB variation is more important than reporting the lowest reasonable frequency that the speaker can reproduce, so you'll sometimes see something like "80-20,000 Hertz, +- 1.2 dB" That doesn't really mean that the speaker is really better than one that says "45-20,000 +-3dB", since some of the lesser loudness may be in the 45-80 Hertz range.

To truly see how "flat" a speaker is, one needs a graph of frequency versus loudness.