I then tried replacing the felt with a thin, adhesive-backed rubber damping material, like you might use to pad something or eliminate a rattle. In this case, I didn’t bother running a test with a solid sheet of material, because that would have blocked the driver entirely. This proved to be a useful way to fine-tune bass response. The more of the damping material blocking the driver (for example, with the damping material perforated), the more bass roll-off was measured. However, the effect at frequencies above 300 Hz was modest.
For me, speaker design has always been easy to understand: Put a couple of drivers in a box sized (and maybe ported) to suit the woofer, then use a crossover circuit to tune the drivers’ response. Headphone design is more difficult to grasp. Most headphones employ just one driver, with no added circuitry. How, then, do headphone designers achieve such a wide variety of sounds?
In my work reviewing and measuring headphones, I’ve seen many of the techniques headphone designers use to tune their products. Some designers have shared their secrets with me. Still, I wanted to learn more, so I decided to experiment on my own, and to publish my results here.
This article is intended as a guide for those curious about the inner workings of headphones, and for DIYers who want to attempt to tweak their headphones for better sound. This won’t do anything to help you buy the right headphones – for that, there are plenty of good sources, including my reviews on SoundStage Xperience and the reviews on The Wirecutter, where I serve on the headphone testing panel.
All text and images in this article © 2016 Brent Butterworth. This work is free for anyone to read and use, but it may not be copied or distributed without my permission.
To conduct my experiments, I built a test jig -- basically an ersatz headphone -- that would allow me to change the acoustic parameters of a headphone enclosure, then measure how those changes affect the headphone’s frequency response and tonal balance. The test jig consisted of a conventional 40mm, mylar-diaphragm dynamic headphone driver taken from a busted set of SOL Republic Tracks HD headphones, mounted into an enclosure made from ABS pipe with a wood front baffle and a removable wooden back. (I later replaced this with a 40mm biocell driver a headphone manufacturer gave me.) I then simulated ports by drilling four holes each in the front and rear baffles, and I plugged them with pieces of dowel rod that I could later remove. I used an ear pad from some old AKG K240 headphones to interface the jig with my G.R.A.S. Model 43AG ear/cheek simulator, which was connected to an Audiomatica Clio FW 10 audio analyzer. I used the 43AG’s clamping mechanism with painter’s tape around the ear pad to maintain consistent positioning of the jig on the ear/cheek simulator. The test jig is pictured above.
Changing the headphone driver in the test jig had a huge effect on the measured response. (The biocell driver was about 8 dB more sensitive than the mylar driver, so I reduced the drive level to make them roughly equal and give me the ability to compare the responses easily.) The big, overwhelming difference is that the biocell driver has a large peak centered at 3 kHz. That’s a good thing in general, because that 3 kHz peak is common in headphones, to compensate for the natural resonance of the ear canal. It’s generally considered to be necessary to mimic the sound of real speakers in a real room. There’s also a little extra kick at about 420 Hz.
The miniature dynamic drivers used for over-ear and on-ear headphones typically aren’t open in the way a conventional woofer is. Instead, the basket (the frame of the driver) has vents in the rear. These vents are often damped with paper that’s roughly as thin as tissue paper but stiffer. I’ve been told by a couple of headphone designers that the number of damped versus undamped vents has a large effect on a headphone’s response. The 40mm driver I used was supplied with nine of its 10 vents covered with paper, so I used a razor knife to open up the vents three at a time, then ran measurements with four, seven and 10 vents open. This did have a relatively large effect on the response. The change going from one to four vents open had the most dramatic effect, introducing the beginnings of a broad resonant peak in the 400 to 700 Hz range and a dip between 2 and 4 kHz. Opening up more vents intensified the effects. This appears to be an easy way to increase a headphone’s sensitivity.
As you can see if you examine some headphones (especially on-ear models), drivers are often covered completely or partially with thin foam. In my tests, using a piece of foam about 3/8-inch (10mm) thick, covering the driver completely (the red curve on the chart) had a large effect: greatly reducing bass response, introducing a narrow dip at 3.5 kHz and a broad peak at 6 kHz, and nearly eliminating the high-frequency resonances seen in most of the other measurements. Putting a ½-inch (13mm) hole in the middle of the foam, centered with the driver, nearly eliminated the effects on the bass response but retained almost all of the effects at higher frequencies. Opening the hole to 1 inch (25mm) nearly eliminated all the effects of the foam; compared with the result using no foam, the foam with the 1-inch hole produced a mild bass boost of 1 to 1.5 dB between 60 and 90 Hz and roughly -2 dB dips centered at 150 and 320 Hz.
The first thing I wanted to see was the difference in using the same driver in open back and sealed back environments. This produced less difference than I expected, and the differences I did see were not the ones I expected. To my surprise, the open back measurement showed more of a pronounced resonant peak (centered at 220 Hz) than the sealed back measurement did. I was also surprised to see that the open back measurement reduced bass response below 60 Hz by only about -1.5 dB. It also reduced midrange output by about 2 dB between 500 Hz and 1.8 kHz, and seemed to have a strong damping effect on the driver’s high-frequency resonances at 5.7, 6.8, 10.5 kHz.:
Changing the internal volume -- and thus, the resonant frequency -- of a speaker’s enclosure is one of the best-known ways to tune a speaker. I assumed it would have huge effects on a headphone, but the difference, while important, was less than I expected. To change the enclosure’s volume, I started by putting balls of Blu-tak poster adhesive (also often used to stick speakers to their stands) in the enclosure. After putting in about 1 cubic inch of Blu-tak, I ran out, so I added a couple of AA batteries to more dramatically reduce the enclosure volume.
Contrary to popular belief, reducing enclosure size can actually increase bass response, depending on the driver and the internal volume of the enclosure. This is partially what happened here, and it helps explain how some very small on-ear headphones, like the Beats by Dre JustBeats, can produce much more bass than a larger, audiophile-oriented headphone like the PSB M4U 1. The most dramatic reduction in enclosure volume bumped the bass up between 70 and 180 Hz by as much as +4 dB, while reducing response by about 1.5dB between 30 and 40 Hz, and creating a dip of about 3 dB in the midrange response between 250 and 420 Hz.
Replacing the soft felt with hard, stiff felt didn’t in most cases change the effects observed above. A full covering of hard felt introduced a stronger bass roll-off compared with the soft felt, but when any part of the hard felt was open (with half covering, a ½-inch hole or perforations), the effects were almost the same as with the soft felt. As with the soft felt, the hard felt was useful in reducing high-frequency response.
The response curves produced by this “headphone” didn’t look as clean as what I’ve measured from commercially available headphones, but it didn’t matter, because I only wanted to see the changes in response that different tuning techniques would produce. This setup also wouldn’t allow me to test the effects of some other important tuning parameters, such as changing the driver and enclosure materials, but it would allow me to experiment with some of the acoustical tuning methods I’d seen headphone designers employ. These include:
Sealed back vs. open back
Driver rear damper adjustments
Mylar vs. biocell driver
Front (baffle) ports
Foam baffle damping
Soft felt front damper
Hard felt front damper
Thin rubber front damping material
Open-cell foam earpad
In each of the following charts, I’ll show you a graphic representation of the physical tuning adjustment I made to the jig, with the physical part I changed shown in red. A frequency response chart shows the effect of each adjustment.
Tinkering with the stuffing (or damping material) inside a headphone enclosure is, from what I’ve seen, the tuning technique most commonly used by enthusiasts when they tweak their headphones. I tried it using recycled denim stuffing (a material with excellent acoustical absorption, sometimes used in speaker measurement to eliminate spurious reflections) and standard open-cell foam of the type found in craft stores and often used in manufacturing of acoustic treatment devices. The chart shows the effects of filling the enclosure about one-third full (red curve for denim, purple curve for foam) and completely stuffing the enclosure (green curve for denim, orange curve for foam). These changes didn’t produce major acoustical effects, but stuffing the enclosure with foam did flatten the response and also damped some of the high-frequency resonances from the driver.
Although they decrease isolation (i.e., they allow the entry of outside sounds into the headphones), rear ports are a relatively common tuning technique. For example, Beyerdynamic’s Custom One Pro offers adjustable tuning by operating in closed-back mode or allowing the user to open one to three ports. (Many closed-back headphones also have tiny ports included only to equalize the pressure on the driver when you put the headphones on.) The chart here shows the results with four ports measuring 1/8 inch (3mm) in diameter. As you can see, the effects are significant but not dramatic. With this driver and enclosure, opening up any of the ports boosts the bass by +2 to +4 dB. With the ports open, a boost around 200 Hz is created, with a corresponding dip at 350 Hz; these changes become more pronounced as more ports are opened. The original ports in were 5/16 inches (8mm) in diameter, but this produced a big difference when I opened up just one, and almost no difference when opening up more ports.
I’ve seen some manufacturers (Torque Audio and Phaz, for example) offer headphones with foam earpads covered with perforated material. The upside of the perforated covering is that you get a more open sound, as with open-back headphones, but it almost completely sacrifices acoustic isolation from surrounding noise. They do, however, have a huge effect on the sound because they reduce or eliminate the seal between your ears and the headphones. In this case, I tried some open-cell foam with a 2-inch (50mm) hole cut in the middle., then tried using duct tape to seal half and the all of the interior walls of the hole. Then I added more duct tape all over the bottom of the foam. All of these experiments dramatically reduced bass, while introducing a large peak (+6 to +10 dB) around 1.5 kHz and greatly increasing output between 3 and 6 kHz. The more tape I added, the more of the bass I got back, but the more of the high-frequency boost between 3 and 6 kHz I lost.
Front baffle porting, with ports arranged around or adjacent to the driver, is a commonly used tuning technique that can work better than rear porting because it doesn’t allow outside sounds to come in the way rear ports do. The Torque Audio t402v allows four user-adjustable tunings by operating in sealed mode or with three user-selectable port sizes. Adding and adjusting front baffle porting, using four ¼-inch (6mm) holes, produced large changes in frequency response. Bass was reduced by an average of 8 to 16 dB; the more ports open, the less bass. The front ports also introduced strong resonant peaks between 500 and 1000 Hz, the frequency of the resonance depending on how many ports were open.
A few speaker designers have confessed to me that it’s now possible for an untrained but technically savvy person to build a competently engineered loudspeaker, using software available free or at low cost, with parts from Parts Express and other sources. The tuning of the enclosure is the result of well-understood mathematics, and the crossover circuit can be adjusted to eliminate most of a speaker’s flaws.
It should be obvious from my experiments above that the same can’t be said of headphones. Headphone tuning balances a number of complex acoustical factors, including the ones I explored plus ones I didn’t explore, such as driver and ear pad design and the effects of the front grille covering the driver. And while we know speakers sound better if they measure flat on-axis and produce a gradual, gentle treble reduction off-axis, we’re still not entirely sure what constitutes the ideal frequency response for headphones.
My experiments don't even come close to being a complete guide to headphone tuning; there's plenty of room here for some enterprising doctoral candidate to do in-depth research on the topic and maybe create a definitive, scientific guide to headphone voicing. And of course, I haven't even touched on the topic of in-ear headphones. Still, my experiments did produce some general hypotheses that might be useful for DIY headphone tweakers, or even those just starting out in professional headphone tuning. Some of these conclusions might not hold true, or as true, with substantially different drivers and enclosure sizes.
1) Changes made in front of the driver, between the driver and your ear, have a much larger effect than changes made behind the driver. (This also makes me wonder about the efficacy of using exotic materials for driver enclosures.)
2) Changes to enclosure volume and stuffing have minor effects, while changes to rear porting have greater but still relatively subtle effects.
3) Opening some of the dampers on the back of the driver can provide a useful midrange boost.
4) Adding ports on the front baffle is a precise way to make major changes in the bass response.
5) Adding soft or semi-hard damping material in front of the driver reduced both low- and high-frequency response, without affecting midrange or lower treble.
6) Adding solid material to block part of the driver is a precise way to make minor adjustments in the bass response without affecting treble.
Soft felt is commonly placed in front of headphone driver baffles to tune the headphone’s response. In this case, I used thin, flexible felt fabric of the type typically used for arts and crafts projects. Covering the driver completely or halfway with felt greatly reduced the output below 800 Hz. Using felt with a ½-inch (13mm) hole in front of the driver introduced a more gradual roll-off below 400 Hz. Using felt perforated with 12 ¼-inch (6mm) holes had almost no effect on the measurement. The felt also had a significant result on high-frequency response; using the perfed felt, for example, reduced output between 6 and 12 kHz by -3 to -5 dB.