The Anaxilus template should be fine. The thickness of the velvet I'm using is 1-1.5 mm (hard to measure), but a little thicker (if that exists) wouldn't hurt. For the ring I would recommend to use just one layer, the rear frame bar deserves 2 or 3 layers, one attached on top of the other by means of a strong double-sided adhesive tape or foil, e.g. carpet tape (also to keep the fabric pieces in shape). It's a good idea to add some less sticky adhesive tape to the underside with contact to the plastic frame to facilitate the removal.
I've decided for black since it provides considerably higher light aborption. Thus it's not unlikely that it also provides better sound absorption than white (let's say by 5.75%), since light absorption happens by diffraction within the fibre structure, and the same could be valid for sound waves.
.
I hope my EQ curve for the velvet-modified HD 800 is of use for you (at least as a nonbinding template considering the chance that your individual HRTF may differ from mine):
This in view that my two pairs require significantly different compensation curves. Above curve is for the older pair. The newer pair, although brighter, needs a stronger increase of the upper treble.
The equalizer used is xnor's Graphic Equalizer for foobar2000. (While I guess your «PEQ» stands for «parametric equalizer».)
Yes, sure! It wasn't meant just to cure the 6-kHz excess, but for fighting some of the brightness, coldness and diffuseness of the original HD 800 which I attributed to reflections within the earpieces. Of course there was the hope that it would reduce the 6-kHz hump, and it was indeed successful in this respect; but the reflection damping affects the whole high-frequency range, since with the available space and the possible thickness of the damping material you can't hope to dampen much more than high-frequency reflections. Well, after all that's the most important range in an open headphone, since larger wavelengths will escape the critical zone quickly. As you can see, there's still the need for some EQ compensation at 6 kHz. Most likely because the reflections here aren't the real source of the resonance. It's probably the magnet system behind the membrane that's to blame, as it represents a serious acoustic obstacle – and a logical source of reflections. An ideal headphone has nothing in front and at the back of the membrane (not even a human ear, which will reflect the sound back to the driver/membrane, which will reflect it back to the ear, etc.), so that it can freely breath and won't suffer from compression effects.
I'm not into measuring, so can't answer your question. Measuring headphones is problematic anyway, that's why I was never interested in it, the more so as I have a hard time believing someone starting with it can get reliable results, let alone better ones than the established testers. Different measuring arrays produce different curves, and then there's the problem with the (Harman) compensation curve which isn't standardized, add to this the own individual head-related transfer function (HRTF) which would require an overlayed compensation curve – but since the latter isn't known, any imaginable resulting curve will be hard to deal with. You can try it and try to correct it in the areas you find inappropriate – but you can just as well solely rely on your listening impressions. Of course having a passably trustworthy EQ curve at hand is better than none, as it gives a better starting point than building one from scratch. In this case using the Sonarworks curve (for an umodified HD 800) as another reference might be a good idea – if you don't fully trust my own implementation.
I'm sure it does matter, and cotton is far softer (= less self-reflecting) than polyester. I think the velvet I'm using is made of cotton.
That's a valid approach. However, equalizing won't help with the inevitable reflections on the bare surfaces within the earpieces. So don't underestimate the merits of this modification – if you want a signal as accurate as possible also in the time domain.
And, just for fun, your first "like."
That's a legitimate question. The answer is «yes». Let's call it convenience. Since there's no published EQ curve I can rely on, I have to experiment on the basis of one that's passably close. For that I consider a ⅓-octave equalizer with fine 0.1-dB steps a better tool than a parametric equalizer. It's easier (for me) to find what sort of compensation is necessary, step by step. Finally reproducing the curve with a parametric equalizer would be the ideal scenario. But here's where convenience calls for its right. And in fact the resulting sound is already excellent enough. If there were some aggressive narrow-band peaks slipping through the cracks I would seriously have to reconsider my approach, though.
![DT1990-ChSd-1024x555[1].png](https://cdn.head-fi.org/a/12230656.png)
