Originally Posted by Nevod
Measuring headphones is way easier than I have expected! REW is a very straightforward tool, no need to wander around in menus.
Then I wanted to turn the mic off so it doesn't interfere with music and went into Windows sound settings. Appeared that I didn't check the "+20dB mic boost" option, enabled it, and.. Not only loudness boost, but now it picked high frequencies normally. Ran tests, and the fall-off was mostly gone. There is still some, but I'm sure it's now more due to the headphones than the mic. Then I read on some forum that the "mic boost" is actually phantom power, and these mics indeed work without it, albeit poorly. But soundcard's power supplies aren't good as well, they cut off HF. Realtek sound cuts at 12kHz, Creative Audigy at 14-15 kHz, so it is actually better to always use external phantom power and an line in instead of mic in.
Measuring headphones with REW is surprisingly painless.
One clarification I wanted to give - there is a difference between providing phantom power, and providing microphone preamplification. Preamplification is needed to bring a microphone output up to line level, whilst phantom power is needed for specific types of microphones, like condensers. These work in a similar way to electrostatic headphones, but backwards - they need a constant voltage bias in order to operate at all. I also want to give a warning at this point that there are two levels of phantom power that are usually used - 9V and 48V. When professional audio interfaces say 'phantom power', they nearly always mean 48V. This is probably bad news for an electret mic like the Panasonic WM61a. However, electrets mics will still require preamplification in addition to their phantom power, or their output level will be very low, and the signal to noise level will also be atrocious. I believe the microphone boost in windows handles both the 9V phantom power and the preamplification for most 'consumer' soundcards.
[Apologies in advance for the rather long post - I wanted to get as many of my thoughts down in one post as possible]
In other news, I've been working with BMF over the last few months while he developed his latest set of mods (I've mostly been helping with the measurement and analysis side of things, so I can't claim much credit for modding innovation). However, there are a couple of fundamental differences between the mods I am running now and the mods that BMF has put forwards, not out of choice, but out of necessity. I thought I'd share.
Firstly, I had removed material from the ear-side of my stock T50RPs quite a while ago. The black felt over the driver was removed, leaving the driver square bare apart from the fine mesh over it. I had also removed the foam 'dust cover' on the Shure 840 pads. This means that there is almost no material between the driver and the ear apart from the magnets and that very fine mesh. I did this because I wanted to see if it would help to bring out the treble in the stock headphones, which are quite dark sounding.
It definitely brought out the treble, but not in an ideal way, and once the rest of the BMF mods were complete, it left an ugly spike at around 9kHz on the frequency response. I started experimenting with different materials to put earside in order to replace those two components that I removed, and I eventually settled on a square of microfibre cleaning cloth directly over the driver (cut exactly to fit the driver square, and taped down being careful not to block the baffle vents). Over this, I placed an oval of 'floppy' craft felt, 2mm thick, cut to the size of the baffle, and held in place just by the pads themselves.
Here's a graph showing the difference it makes on the frequency response:
The difference in the high frequency response is easily visible - which one you will prefer is up to you, but I personally find it much nicer to listen to with the replacements in place. There may also be a small difference through the 500Hz to 3kHz range, with the earside replacements bringing that down slightly to a flatter response. The difference is small enough though that you might chalk it up to just variation in placement of the headphones on my head. And although I didn't do many repeat measurements, it is my feeling that the difference in bass response is mostly due to headphone placement, and not to the change in ear-side mods, whereas the high frequency change has been consistent across multiple tests.
I briefly tried a sandwich of felt -> a thin slither of cotton wool -> felt, and it measured marginally better than my microfibre -> felt setup, but it was very thick and pressed against my ears somewhat uncomfortably. You may wish to try this one yourselves.
The second fundamental difference between my headphones and BMFs is in the white paper on the back of the driver. I have what we cautiously believe is the less porous paper. Because of this, in my mods, I do not apply material directly to the back of the driver paper. Applying a thin strip of porous plastic tape, one grid square wide, and 3 long, can bring the upper midrange up in level if you wish, and if you have the same driver paper as me.
A quick note of caution before I forget: I live in the UK, and bought some transpore off Amazon. It says 3M Transpore on the box, and on the inside of the roll of tape, but it is in fact not the same thing as BMF's transpore. It is much closer to what he calls micropore - a porous cloth based tape. This does not work at all in headphone mods. I bought a porous plastic tape from a sports first aid website instead - from pictures I've got, it looks identical to what BMF and the US market calls Transpore, though it has no brands printed on the inside of the roll.
If you have removed the ear-side material, or you have the more porous white paper and you wish to try BMF's mods, I suggest you give my variants a go too, which are included in the spoiler below. If you are not in this position, I would just go with BMF's mods straight up - his measure very slightly flatter than mine do, without the 1dB humps every now and again through the midrange.
Warning: Spoiler! (Click to show)
Newplast flush mass loaded in baffles
Stock cup vent felt in place (all vents open on the outside)
Acoustipack lite in cup floor wells (same as BMF)
Paxmate pyramids superglued on top of the Acoustipack (same as BMF)
Cotton on top (this needs to be tuned by hand - cotton varies in density from different companies, and it is very difficult to measure out accurately)
Stock white driver paper, with nothing stuck to it.
Black felt removed over driver
All 4 baffle vents drilled out
Microfibre square over driver (careful not to cover baffle vents with it)
2mm thick floppy craft felt over that
Shure 840 pads with dustcover cut away
Being able to measure the headphones is very useful. Some things like measuring out equal amounts of cotton per cup can be very difficult, and getting the same cotton wool as someone else can be difficult in the first place. Whilst you could tune by ear and make the very strong argument that if you can't hear any difference, it doesn't matter, ultimately it is nice to be able to quantify what each mod is doing. So, I'm going to post some measurements I've taken recently, of my finished mod (in the spoiler above), and a couple of other headphones I own to give a reference point - hopefully ones some of you will be familiar with. I'm sure you all know this already, but I'll say it again just in case - you can't really compare measurements taken by different people, especially in the upper midrange and higher.
I've written my own measurement software to do the analysis you see below. There's a couple of good reasons I'm not making it public: 1) It's not robust, and has not been written to be user friendly. I wrote it so I could test out some different mathematical techniques and plot styles, and to test that: 2) REW gives the same results with the same inputs, and is much easier to use. There's nothing 'funny' going on behind the scenes in REW, so I would stick with that.
My Panasonic WM61a microphones are mounted to the back of a pair of cheap IEMs. I did this so that I could do live monitoring of what the mics pick up, though in practice I do not actually do this. The microphones do not face directly out of my ears, and point rather upwards and outwards than straight out. They claim to be omnidirectional, but I cannot say for certain what effect this has. I use the phantom power battery box that others have been suggesting, and record into a Focusrite Saffire 6 interface (which can only provide 48V phantom power, hence the battery box).
BMF 8.1 (Arleus variant (I still recommend you build BMF's version, if possible))
(note the imbalance in high frequencies - this is present in stock form, and in all my mods. I think it is inherent in my headphone. However, I'm quite happy with the general flatness of this curve.)
(Two types given for this - a 3D waterfall style one, and a spectrogram style one - I believe the latter shows more clearly what is happening, but some Head-Fi members might be more used to the former)
(Note also the general speed of these headphones from the initial impulse is fantastic, with a touch of ringing on the left channel at 1kHz. There's also a bit of residual muckiness in the midrange up to 1kHz compared to the other two I'm posting, but this is about 30+dB down and probably not very audible. It is an area we could try to improve on in future though)
(Also generally very good, though a touch of ringing on the 2kHz peak)
(Note the characteristic Grado ringing at 2kHz, 4kHz, and others at high frequencies, but the response in the midrange is pretty darned clean, which I believe gives Grados their characteristic 'fast' sound)