Why do frequency response graphs bunch up on the X axis?

Discussion in 'Amps and Cabs [BG]' started by Rockin Mike, Jun 24, 2014.

1. Rockin Mike

May 27, 2011
Most graphs I've ever seen have the lines evenly spaced on the X axis, like this:

Why do frequency response graphs have them bunched up like this?

And how do they select the frequencies where they bunch up (in this case 100Hz, 1kHz, and 10kHz). Is that arbitrary or is it defined by some physical fact?

I see this type of graph for speakers, microphones, and eq controls.
Some of you smart folks help me understand them better please?

2. CL400Peavey

Nov 7, 2011
Grand Rapids Michigan
The graphs are logarithmic. When you go up an octave you double the frequency. So an octave up from 50 hz is 100 hz. A 50 hz increase. An octane up from that is 200hz, not a 50 hz increase. If they used a linear scale, the visual representation would have all the lows scrunched at one end, and the highs would take up the whole graph.

HolmeBass and pglaser01 like this.
3. Stealth

Feb 5, 2008
Zagreb, Croatia
Your best bet is to read up on logarithmic scales. In acoustics, some measures such as loudness and frequency are best explained through logarithmic scales since the relations between the two measured items aren't linear, like CL400Peavey said, frequencies behave like that.

If you look at the standard A 440 Hz tone, its octave is at 880 Hz, the second octave is double at 1760 Hz and so on - so the higher you go the bigger the spacing - thus the "bunched x-axis." You're using a linear measure (Hertz) to display something that behaves logarithmically, so you need a log x-axis.

On the other hand, the y-axis looks linear, but it's using a logarithmic unit of measure, a decibel. A gain of 10 dB is ten times louder than a signal without gain at 0 dB - yet a gain of 100 dB makes it 100 times louder, 30 dB would be 1000 times louder and so on. Here you've got a logarithmic measure to display something that behaves logarithmically to begin with - so in the end the notches (10, 20, 30) are spaced linearly apart. This same graph would be bunched up if you converted those decibels to watts (as a power unit).

4. Downunderwonder

Aug 26, 2009
New Zealand
Saves on wasting graph paper, lots and lots and lots of graph paper.

The bunching up of the intervals reminds you not to plot or interpret in linear mode.

5. will33

May 22, 2006
austin,tx

It's generally accepted that 3db is a "small but noticable" change and 10db is twice as loud in how humans percieve sound. In a loudspeaker, if it's -10db at some frequency, that isn't contributing all that much to the output anymore as it's only playing that roughly half as loud as the rest of the bandwidth. -20db would be almost useless as far as contributing to the overall output. The vast majority of output increase in most any speaker happens in the first 100 watts, with the need to double power for every 3db increase from there.

The rest of it though about the octave doubling is right on.

6. Stealth

Feb 5, 2008
Zagreb, Croatia
I agree, there's always the perceptive factor about it, combined with the fact the human ear is not equally sensitive to boosts and cuts.

Power is one thing, loudness is quite another. I stand corrected.

7. will33

May 22, 2006
austin,tx
+1

Small differences in the mids where the ears are sensitive amount to large percieved difference in tone and feel.

8. dincz

Sep 25, 2010
Czech Republic
You could replace the labelling of the X axis (frequency) with musical notes e.g. E0, E1, E2 etc and there would be no "bunching", but it would then be less meaningful to non-musical types. Electronic circuits respond to signals in a "per octave" way not in a "per Hertz" way, so the logarithmic X axis labelling keeps both engineers and musicians happy - usually.

Stealth likes this.
9. hazmatt

Jun 3, 2012
san diego
This. Human cortical representation of frequencies is most detailed for the range which makes up human voice for spoken language. This means we are most discriminating for sounds in the 300-3000 Hz range. It happens that a logarithmic scale like the Y axis here both compresses audible range onto a reasonable sized graph, and still preserves the area that makes the largest difference in our perceptual range.

Stealth likes this.