DIY Modification for the Sennheiser HD 800: "The Anaxilus Mod"
Why would one of the world's best headphones benefit from a tweak?
For the same reason we have aftermarket car parts to modify stock cars: manufacturers have reliability requirements that make producing hot rods problematic. The modification I'm going to show you will not last as long as your headphones and you'll probably have to make another one within five years. This wouldn't be acceptable to a manufacturer. Sennheiser could surely make a better part than I, but in principle having some sound deadening padding adhered to the inside structure is a bit messy.
It also appears to me that the mod shown here does something similar to the removable liner in the HD 800, so Sennheiser engineers have already done things to address damping in the headphone. My guess is that the stock HD 800 is a terrific compromise of many, many issues --- most of which none of us will ever know or appreciate --- but in sum I find it the highest expression of headphone engineering to date.
That said, the stock HD 800 remains a tad hard on my ears, and many headphone enthusiasts have expressed similar sentiments. Over the years, a tweak for the HD 800 has been developing in the enthusiast community at Head-Fi.
I have prepared a companion .pdf booklet for this article that includes the Anaxilus Mod template, frequency response comparison graphs, and full data sheets for all measurements done in this article. It also includes four recent measurements of the stock HD 800, which were done on different days. These are labeled B thru E.
Evolution of the Anaxilus HD 800 Mod
|iPodPJ --- It all started when iPodPJ put a a little strip of T-shirt material into his HD 800. Much hilarity ensued and the thread is worth a read for that reason alone. But the fact remains: it's his little tweak that was the first evolutionary step to tame the brightness of the HD 800.|
JaZZ --- Wanting more bass than he was experiencing, JaZZ started modifying the HD 800 by putting some fabric into the headphones to block some of the escaping sound and get some more compression for the low notes.
His attempts to make changes and play around with sonic characteristics of the HD 800 inspired Arnaud (who is an acoustics engineer) to make some attempts. Their dialog helped JaZZ to eventually conceive the 3-layered velvet mod shown in the photo to the right.
I also think he's most correctly identified the problem these mods are addressing:
"...the main effect of the mod is the reduction of inner reflections, the reduced treble intensity is the result of reduced parasitic transient corruption"Close enough. I'll go into more detail later in the article, but I believe the metal ring around the driver is acting acoustically as a reflecting surface and helping to promote a resonant mode that is storing a bit of energy. It is this very slight ringing at 6-8kHz, I believe, that some people find objectionable.
Arnaud --- After being inspired by JaZZ's mod, Arnaud started playing around with some of the killer acoustics analysis software he gets to use for work. First in a post in JaZZ's thread, and then after starting his own thread, Arnaud tells us about getting a Melamine sponge as a small gift at work and then goes on to describe it's very cool acoustic properties.
Shown in the images to the right, he built a computer model of the HD 800 driver and enclosure, and then ran a finite element analysis of the acoustic response inside the chamber at 6500Hz both with and without his little block of Melamine foam in place. You can see that the distribution of sound is more even with the foam, which removes a hot spot near the rear of the driver. Here's Arnaud's explanation:
This is a plot showing the distribution of sound pressure level in the earcup in the baseline configuration, you can see reflection off the frame and observe that sound field is dissymmetrical. I am not sure about this but have a feeling this contributes to the HD800 "fuzziness." The soundstage is very wide and deep but placement is not so precise in stock configuration. I think reflections off the frame can be a cause for this.
Like JaZZ, Arnaud too slowly evolved his solution. Instead of velvet, he used 5mm thickness Melamine foam around the driver. Both reported significant improvements in reducing the biting characteristic of the HD 800.
The "Anaxilus Mod" --- The problem with the above mods to the Sennheiser HD 800 is that they require quite a bit of patience and skill with a blade to cut into proper shape. Head-Fi members Anaxilus and Purrin were chatting one day, and decided to do something about it. I sent a message to Anaxilus to get the full story, here it is in his words:
I/We had read about and tried the Jazz mod and purrin mentioned Arnaud's measurement simulations so we talked about the HD800 more and I said, "fuck it, let's cut some shit up." Purrin scavenged a bunch of stuff he had lying around (i.e., stole craft materials from his kids) and after much cutting, the Creatology Foam worked best to our ears so we stuck with it. Eventually it went from just a basic ring to include a precision trapezoidal area then also to a black version of the foam for aesthetics.
Purrin has also been a great source of information for headphone hobbyists by developing a measurement system that produces cumulative spectral decay (CSD) plots. CSD plots (sometimes called waterfall plots) show you how quickly acoustic energy decays after you turn the sound off. At the top of the plot, the headphone is excited with all frequencies; the audio level is shown both in elevation (z axis) and in color (red for high volume level, to blue and then black for low level).
In the plots to the right, the bottom axis is frequency and labeled left to right: 1kHz, 5kHz, 10kHz and 20kHz. Time is on the Y-axis and is labeled at right in one millisecond intervals. The stock HD 800 performance is very good --- much better than most headphones --- but we can see a bit of stored energy in the unmodified headphone as a turquoise blob centered at about 5.5kHz. After modification, the headphone shows a significant reduction in the amount of energy being released there. However, a couple of blips at about 9kHz and 15kHz appear after the mod.
It was after seeing these CSD plots that I decided to give it a go. Let's have a look at what I found ...