Frequency response graphs at headphone.com.
Nov 14, 2002 at 3:35 AM Post #3 of 30
Well... I'm no technical guru but I'll give this a shot...

The theory is that human hearing exists within a frequency range-- 20HZ at the low-end to 20 KHz range at the high end. However, the new audio formats (SACD and DVD-A), have an even greater frequency response than the 20 Hz to 20 KHz 16-bit Redbook CD, based on the theory that frequencies well above 20 KHz should be reproduced and can actually affect the way we hear other high-end frequencies below this number. proponents of the new formats insist that, although these high-range frequencies may be beyond our ability to hear directly, they actually affect the way we hear the audible frequencies. Therefore, reproducing these extemely high frequencies is essential to obtaining a realistic sound within the frequency spectrum that it's genrally agreed we can actually hear.
Also, most HT rigs have sub-woofers that reproduce sounds below 20 Hz that can't really be heard but can be "felt" as waves of energy. Yes, we may not be able to directly "hear" an 8Hz tone, we can certainly feel it and be aware of it's existence. Modern DVDs have sounds that are meant to be reproduced by subwoofers on a separate track at frequencies well below the audible 20Hz.

All that said, one of the last things I'd worry myself about is a particular headphone's rated frequency response. First of all, how can you trust a manufacturer's listed frequency response? Yes, it may say it reaches down to 5Hz at the bottom end, but is that with a 10db fall-off? So what?

What good is that last 5Hz of listed bass (with a huge volume fall-off) or that last 10Khz of treble (with a large volume fall-off) if the in-between just doesn't sound right? The vast majority of music happens in the middle, not at the extremes of the frequency response curve.

Tone and timbre are things that can't be measured by anything except our brains and ears and I'd argue they're at least 80% of the ball game.

Did this help?

Mark
 
Nov 14, 2002 at 3:58 AM Post #4 of 30
they put a mic inside a dummy's head and then strapped on headphones.

haven't you notice that all headphones are 20Hz to 20KHz? yeah, even 99 cent ones.

but a c sharp (C#) on a piano doesn't sound the same as a c sharp on a trumpet. hmmm. it's the medium...

a 3dB difference is twice the sound volume. so anything less that +- 0.5dB is really unacceptable. remember the Electro Voice dual speakers where the tweeter was mounted inside the woofer? that's what they have to make for headphones. of course it can be argued that one transducer is doing it all anyway....

hey, i'm still wondering why anyone would want coper cladded aluminum wire (CCAW) in a voice coil bobbin, or an aluminum voice coil core....

if you really look at the graphs closely and compare them you'll see some very interesting things, especially from the impedence resonance 'hump' down. look at those beyer's again. the graph goes up
eek.gif
now lokk at the grado impedence curves. interesting.... and the akg k501's.... interesting....

i'd like to see a headphone based on horn "air squeeze" principles and not dynamic (like speakers) or membrane (like electrostatic).

graphs are nice.

but the people here's opinions are better. review by consensus.
you're in trouble if you're the odd man out, though. but just as no two people see the same colours, no two people hear the same sounds. life is good.

------------------------------------------------------------------------------------
listening to TSOP-MFSB. the soooul train..... yeah, baby.
 
Nov 14, 2002 at 8:59 AM Post #5 of 30
EDIT: The way Headphone.com had made the idealized headphone graph got me thinking and I went back to the literature. I admit outright that I may have judged their approach hastily and wrongly. To reflect this, I've edited this part of my post.

To the original poster.

Frequency graphs measure the amplitude (sound intensity) of the sound making device at each frequency.

An ideal frequency response for a transduce in free field (in an anechoic chamber) would be vertically straight frequency graph at all input volumes.

Did you hear me say at all volumes? Yes, most transducers (loudspeaker cones, headphone elements) FR changes at the volume varies.

Also, I mentioned 'free field'. But nobody listens to anything in an anechoic chamber, right? Indeed they do not.

Reflected sound from surroundings makes measurement more difficult.

When you have a headphone that forms a reverbant chamber between the headphone cup and the head, the measurements are even harder to do right.

To be honest, there isn't even one proper way to measure headphones right so that everybody agrees on it.

Partly due to that, there isn't a single proper way to design a set of headphones that sound right to most people (at least, not a way to do that by measurements).

Some headphones are diffuse field equalized, meaning that their frequency response graphs (when measured with an artificial head near the tympanic membrane or at the concha) approximate those of a pair of loudspeakers (example being several Sennheiser models) in a reverbant room (that has equal sound power in almost all positions).

Then there are headphones that do NO diffuse field equalisation and sound more vibrant, lively and energetic (one example being Grado).

Which one is 'right'? That's a philosophical question, for which there is no one right answer.

Now, getting back to your original question.

How can you use and interpret the data at headphone.com?

Well, for the frequency response graphs the answer is quite simple: you can't - at least not without hearing the phones yourself.

They have made an 'idealized' graph by combining diffuse field and non-diffuse field headphone FR graphs together. I think this is interesting as it makes an ideal graph for a set of headphones that does not exist (?), but which MIGHT even better reflect how a loudspeaker would sound in a real room (as opposed to a diffuse field reverbant room).

Also, they do not tell at which point in the ear they have done the measurements and if they've blocked the ear canal or not (and how).

Furthermore, even if they told us all this information, we still couldn't draw any objective conclusions about the absolute qualities of each headphone (at least not without hearing them).

I have personally seen excellent measurements for various headphones done with the best available methodology (as of 2002) and they do not always correspond with listening experiences of seasoned listening experts.

So, to make a long story short: listen to the headphones you want and then look at the graphs to see if they correlate with what you hear. If they don't, then trust your ears, not the graphs.

regards,
Halcyon

PS I'm not saying that headphone.com has done something worthless - on the contrary, the test is very interesting, but from a stand point of headphone measurement techniques it's not a final say (nothing ever is, is it?). The measurement techniques and philosophies vary quite a lot and due to this, so do the measurement results.
 
Nov 14, 2002 at 9:18 AM Post #6 of 30
The problem with low end response is that physics start to get in the way. You see, whenever you drop a half the frequency you quadruple the travel of teh unit. In other words, the distance a cone will travel at 200Hz is 4 times what it takes to reproduce 400Hz. Or 1/4 of what it takes to reproduce 800Hz. B&W produces certain high-end speakers that have no baffles, correct word I think, around their midrange cones as the y cross it over at the exact frequenct they wanted so that it's travel is limited and improves the sound it produces. This explains why tweeters are so small and need nothing around them!

I owned a speaker box in my old car that had two 15" woofers on top, and when I played back some Bach organ peices you could see the speakers moving over an inch up and an inch down trying to reproduce just the 18Hz note on the CD. Drop that frequency down a little lower and you start seeing that the speaker travel becomes so large it's nearly impossible to reproduce without damaging/destroying the voice coil.
 
Nov 16, 2002 at 1:07 AM Post #8 of 30
Quote:

Originally posted by Matthew-Spaltro
in other words. The flatter the line. The flatter the frequency response.


Yes. Any deviation above or below teh line represents a boost or loss in the given frequency. Now, depending on your musical tastes that might prove to be favorable. For those who love midrange to be more pronounced, getting a set of cans which has that particular characteristc will blend with what you like. For those who like bright presentations, find a can that emphasizes high end notes. Etc, ad nausium.

I knew I guy who had a big monster SUV and installed something like 6 tweeters and only one small 8" sub because he loved the high frequency stuff more than bass. Same goes with selection of cans, find what part of the music spectrum you like and find the cans that match up to that preference and you'll have a winner that works for you.
 
Nov 16, 2002 at 1:35 AM Post #9 of 30
Yes, but the current Headroom graphs,for instance,make grado headphones look like they have a ridiculously elevated treble. Grados are lively but not distorted like those graphs would lead you to believe. The new graphs recently published in Tylls rants regarding smoothed headphone data,which incidentally,feature a comparison between several grado phones, are much more like what I hear. These graphs are referanced to 1khz like the graphs you see in stereophile for speakers frequency response. Atleast in this type of comparison there is a standard all phones can be referanced to.
 
Nov 16, 2002 at 4:03 AM Post #11 of 30
it probably means that the cans weren't broken in, or they were mis-placed on the dummy's head. and yes, a +10db boost beyond 5k would tend to make them "brighter". but remember about the two dips that are supposed to be there to make headphones listenable. kinda makes you not believe the grado advertising where it states that they're within 0.5dB of each other, though.

now compare the sr80 graph. would you have everyone here believe that the sr80 is better than the sr325? not likely. WHERE the dips occur determines the overall sound.

look at the akg k501 graph - it's how many dB down from the impedence hump? now look at some beyers. measure bass response from the impedence hump frequency (usually about 50 Hz). one of them has a graph where it shoots up. hmmm. do you really think it produces that kind of bass? i don't think there's a headphone that can produce a 20Hz note/tone. and just adding +5dB at 20Hz will not make them flat, due to "cone" excursion. if the cone can't extend far enough you'll get distortion.

graphs are not only references (most 0dB lines intersect at 1KHz to 5KHz), but are relative points. nowhere is the power stated on the left side. the graph could very well be 0.5 watt, 1 watt, or 2 watts, etc. (i doubt the grados can take more than 1 watt). so therefore are all the headphones being driven to the same volume level, ie, 98dB or 100dB? and 98dB is loud! if it's a power wattage reference standard, like 0.825watts, the headphones will exhibit different responses due to sensitivity. you probably get better gas mileage at 50mph than 10mph, don't you?

use the comparison function to compare 4 or so headphones. it'll give you a better idea.

now compare an open headphone to a closed one. can you really tell which one will sound better?
 
Nov 16, 2002 at 4:16 AM Post #12 of 30
Quote:

Originally posted by wallijonn
it probably means that the cans weren't broken in, or they were mis-placed on the dummy's head. and yes, a +10db boost beyond 5k would tend to make them "brighter". but remember about the two dips that are supposed to be there to make headphones listenable. kinda makes you not believe the grado advertising where it states that they're within 0.5dB of each other, though.


Wallijohn,

The matching of the two drivers is that they are within .5db of each other, volume wise, not frequency response wise.
 
Nov 16, 2002 at 4:57 AM Post #13 of 30
I've noticed... what do the Sennheiser 580/600 do best? Take a look at the headroom graphs at how unusually smooth they are from about 500 hz to about 5khz. I think that's the hot-spot for human hearing. I think this makes for a tremendous head-start in making a good-sounding headphone.


Quote:

Originally posted by markl
Well... I'm no technical guru but I'll give this a shot...

The vast majority of music happens in the middle, not at the extremes of the frequency response curve.

Did this help?

Mark


 
Nov 16, 2002 at 5:43 AM Post #14 of 30
compare the dt931's to the senn 580, 590,and 600. go to the "product measurement" section.

i don't like looking at "normalised" graphs. i prefer 'frequency response' graphs.

so why isn't the world in love with the dt931's instead of the senn 600's?

btw, headphone.com states that the dt931's lack bass. see the actual frequency response graph and not the normalised one.
 
Nov 16, 2002 at 5:44 AM Post #15 of 30
SE, at what frequency, 1KHz? i'd rather have drivers that were matched at 100Hz, 1KHz, and 7.5KHz. then you'd have something.
 

Users who are viewing this thread

Back
Top