arnaud

Well, I have found the answer to the 009 brigthness issue when listening loud or to edgy recordings: I just acquired a refurbished 007mk1 with new headband and earpads! I was shocked by the difference with my mk2 version (SR007A in Japan), it feels similar to the change from 007A to 009! In particular, while the 007A sounds quite a bit more alive for lack of better term than the mk1 version (out of stax srm727A amp), it also has that boomy low end and somewhat of emphasis in the treble. The mk1 is all about smoothness, balance. The bass sounds almost too lean but I appreciate how clean it sounds compared to the mk2 (ported version). In retrospect, I realize why it is the 007A/727A combo that made me jump to the stax world a few years back: upon short term audition in store, it clearly sounds more impressive than the 007mk1/727A. But now that I have the 009 when I need more bass impact, "resolution", liveliness, I feel the 007mk1 is the perfect complement with its polite / non-agressive / non-emphasizing rendering and especially its ability to play loud without any ear strain. The 007A sounds a bit like a failed 009 wannabe now . Anyhow, I should have listened to the mafia long ago and am glad I am now a happy 007mk1 owner . Arnaud

These pictures of the T2 are soo freaking impressive. These chassis are packed but it all smells perfection being so tight and neat. I really wish I had the skills / talent to do that kind of stuff, I can only hope to hear one someday (

Humm, "silicon" gold

I liked the Fostex quite a bit except for one little snag: piercing / harsh sounding trumpet on Miles Davis "Someday my prince shall come". It's not in the recording if I am to trust the many different kinds of headphones / rigs I tried with this sample recording. Might have been my ears conditions on that day but no other gear sounded like that (including a sanity check with my stax rig when I got back home on that day).

Yes, possibly we could model some forms of distortion, but non-linear does not mean magic and you still very much need a mathematical model for whatever you're simulating. Typical non-linear simulations are discrete time domain and with varying physical properties / geometry at each time step. I do all my work using linear vibro-acoustic models so I couldn't help you with any non-linear types of distortion. But things like residual vibration (of the stators or enclosure for instance) are linear behaviors that I could simulate...

Birgir, there are actually other software tools in my company to simulate all sorts of non linear physics (like vehicle crash worthiness, airbag deployment, parts stamping, ...). So getting the effect of stretching / tensioning is feasible, I just can't pull the resources since I do that for free on the side. By far the easiest way to get the effect is inverse method: find out the first membrane resonance from SPL measurement (that until someone can come up with a way to measure electrical impedance), and backout what is the effective stiffness of the diaphragm using a coupled vibro-acoustic simulation under the same air loading conditions (you simply tune the properties until the resonance frequency matches). I think I did it wrong the first time around with my 009 simulation since I used the in-vaccuo vibration simulation and tune the membrane stiffness, then couple it to the acoustics. i plan on revisiting this...

Hehe, if you're actually clueless, you're hiding it rather brilliantly . Thanks for the 101 course review, it does feel like you telling me the same thing as before and I still come back some months later thinking differently, please bare with me . Anyhow, this gets me motivated to give another shot to stax headphone acoustic simulation. Between experience of people like you and wachara and measurement capabilites of someone like purrin, there's got to be a way to come up with realistic model which could then be used to improve upon existing design (or at least help us cut the chase to the key design parameters). I have to say though that there's been so much experimentation already that my simulation work has a steep uphill curve to climb before bringing value .

What I meant was that the diaphragms don't share motion, the "woofer" and "tweeter" common rim is a circular spacer which pins both membranes (one on the outer edge, one on the inner edge. This probably makes no sense from a manufacturing point of view, that wouldn't be a first coming from an acoustics guy . Also, from your post above, it would seem like that stator-diaphragm gap is essentially fixed by the polarization voltage. I somehow though it was adjusted based on the diaphragm tensioning and thickness (it really has to elastically deform to get shorting and my guess was the max excursion occured at the tuned LF resonance frequency, e.g. 50-75Hz for stax estats). Anyhow, at the very least I am curious to see the effect of having two independent membranes. I "just" need to find time to update the model I made.

Spritzer, what I imagine is that both diaphragms are in the same plane and there are really only two stators. The tweeter is in the center and the woofer around it. I guess you're implying that the woofer has to cover the whole surface because you'll get way too much acoustic cancellation between the front and backwave? At least, I could simulate both responses assuming a constant velocity over each respective surface and see what it gives...

I forgot all about this stuff and was confising isobaric with push-pull. In any case, it would seem like these concept only apply for dynamic transducers in piston regime, so again, hard to see the meaning for estat. On the the other hand, having two diaphragms with different tensioning and mass sounds interesting. Like a concentric dual diaphragm between the same stators (or at least each pair of stators getting the same signal). The high frequency diaphragm is much lighter/thinner, and smaller in size (center). The two diaphragms have the same polarization bias but the spacers are different so that the sensitivities are adjusted to get satisfying blend of the two concentric drivers. No xover, the center diaphragm is inherently highpass and the heavier diaphragm on the outer rolls of at HF. You could also adjust the perforations on both stators to get the right amount of damping for each transducer. Nonsense?

Wasn't there a discussion of dual diaphragm for 009 when little detail was known (which turned out to be the multi layer stators)? I am unsure as to what could possibly turn out good. Only thing I saw that had potential is two back to back iem drivers loading the same cavity symmetrically.

Actually, I think the idea is to have double the gain on the HD800, else the comb filter effect would be even more miserable as Tyll showed in the link. It gets more complicated here with the mess of a dual frame / partial obtrusion of the HD800 acoustic screen, but at least it's likely less offending to run the hd650 at a lower gain.

I wasn't expecting much when I asked for measurements, so when I saw Tyll's post, it was like one of the most exciting reads for some time now! Can't thank you enough for taking the time Tyll (or maybe by getting my act together and completing these csds for you - although you may have lost interest by now lol). This response of yours is almost overkill and I fear for ipodpj's garden hose business!

Even these ants keep coming back for the 009, what an atypical endorsement .

Thanks for this, I want to try it. Haven't found a way yet to integrate (the vst filters) with iTunes or another free player for OS X, any idea?

Yeah, it quickly gets silly when you think about all the hard work to eliminate cross-talk in the amp ^ ^

I will let the owners add to this, but the answer to your question is pretty straightforward: - A crossfeed is typically just that: bleeding some of the left channel into the right ear and vice-versa to simulate how each speaker is heard by both ears vs. your headphone experience. You can go to some more complexity by including some delay line (to simulate inter-aural delay for a specific speaker heading) as well as some frequency alteration (to account for the shadowing effect of sound diffracting around the head). But essentially, any implementation using analog active filters is a crude approximation of the real mechanism, not-withstanding the fact that the peculiar geometry of everyone's head and ear (outer & inner) makes it difficult to come up with the right equalization for all. - The realiser will simply record your head-specific relationship between a given loudspeaker heading (say +/-30 degrees in front of you for a stereo setup) and both left / right ears response at the entrance at the ear canal. It is simply recording that "Head Related Transfer Function" by feeding some kind of white noise or sine sweep and recording the response. You end up with an impulse response that will later on be convolved with the music in real time to convert the speakers L/R signal into a "binaural" signal that corresponds to you hearing it for the given loudspeakers position. - You might recall that when playing this "binaural" signal back, you're using headphones, hence the sound is sort of being filtered twice by your outer ear shape. So, you typically need to filter out that effect and equalize out the response of a specific headphone to the exact same microphones position at the entrance of the ear canal. This pawel product discussed above is doing it for some stax headphones and it's using binaural recordings from a dummy head which is supposed to mimic how the average person hears (the issue is with the deviation across the population not the average though...). - Another specific of the realizer is that it's geared toward multi-channel music and movie mixing engineers so, actually, it's not quite recording HRTFs (which are characterizing how a given head hears independently of the loudspeaker characteristics, amps, rooms, microphones) but lumping the dynamics of everything in the chain: e.g. realiser D/A > preamp > amp > loudspeaker > room > your head > microphones > realiser A/D. This is the reason why they refer to PRIR (Personalize Room Impulse Response) rather than HRTFs. By now, you might be just going tldr but I'll have tried as best as I could...

You're missing the point about the realiser. I agree that adding a speaker coloration to your headphone isn't a fantastic idea to start with (and I would have allowed to simply try to obtain actual HRTFs rather than raw data lumping the whole reproduction chain). However, acquiring personalized binaural response goes a long way toward making the out of head experience all the more believable. As long as you're calibrating it with a well setup system (both room treatment and speaker performance), it is a reasonable compromise toward getting binaural experience with all your stereo recordings using headphones. The swiss product mentioned here appears to be a more traditional active crossfeed filter, albeit based on an actual dummy head HRTF + headphone specific compensation for the same head. Ignoring the fact you don't listen like the dummy head (still better than nothing for those recordings that have really unnatural L/R separation), another major challenge is to approximate such measured HRTF with an analog filter (there's a reason why people use DSPs for this and simply use measured data / convolve in real time).

A foam modded HD800 with p1-u is indeed very special for me as well, probably quite complementary to a 009 rig. I sold my HD800 for a Stax rig (O2mk2) after direct with comparison with modded HD800/p1u and haven't gone back since. The two rigs (same source) were rather similar tonally but a tighter bass with the HD800 while the Omega 2 excelled in the mid / highs, low level detail.

I suspect not all downsampling tools sound the same because of the anti aliasing filter used. In my experience, it has been difficult to objectively qualify the benefits of high res. recordings in the sense that it typically sounds better but it's subtle enough I can't really put the finger on what is lost when decimating / downsampling (using iTunes). On the other hand, the jump in quality from CD to HDTracks version in case of a remaster is typically quiet noticeable. Lastly, in some instances though it feels more like a different presentation than improved one and I suspect this may the result of a remix rather than remaster? I use this website to correlate what I hear with the recording dynamic range and it's typically meshing rather well: http://www.dr.loudness-war.info/

One question for me is if I wouldn't get annoyed with the speaker style rendering after a while / for many recordings (I got used to headphone imaging over the years, I no longer have a speaker rig). I could see myself using it for movies but main usage would be stereo speaker emulation. For those who really favor speakers over headphones, it's probably a no brainer. I'd try to get reference PRIRs from AIX or such place but it's also difficult to know if I'd be ok with the added coloration / loss of resolution. The coloration bit might be less of an issue for a reference audio rig. The loss of resolution is more concerning for the obsessive / compulvise audio maniac in me , for the same reason I don't use traditional equalization actually. Also, I have some high res music which sounds very good (may be mostly due to the mastering quality though), but it then feels bizarre to feed it to something that processes at 24/48 resolution, the PRIRs themselves being acquired at 16/44 resolution it would appear (I understand though it's most likely just fine given the tiny mics quality, room background noise and all). Like others, I see this as another toy (in the sense of how I would use it, I consider it an amazing product in regards to the technology involved) and am unsure if I wouldn't do better looking into improving the audio rig instead (source or amp). I'd be interested to hear others experience with 2ch audio.

When I see how prolific you guys are, I am just wondering if the amp makers like Stax are doing anywhere near the same amount of R&D? For one, it seems like there's a large collective brainstorming going on constantly in such threads. Can't someone just bring these solutions to the masses (e.g. non E.E. / soldering beware people like me) for a profit or this just spoils the fun?

There you go, that's what I very vaguely remembered, thanks...

Is there a simple way to understand this current spec you're talking about? Is it like the current available if you were to drive a typical load at full voltage swing? Also, how do these 10-25mA figures for your BHs and over KGSSHVs relate to something like the stock 727 amp? I thought, I may have read somewhere it was more like 5mA for such amp but I don't quite recollect it.

To this day, I still could experience anything else than stax gear, but I can say with certainty that the sr009 does not sound half bad with the stock 727 amp. While it clearly wasn't optimal with the O2mkIi (uncontrolled bass, falling apart on complex musical passages), the amp seems to do pretty well with the 009. In particular, I really don't feel my rig is artificially bright (I can't listen to some average recordings at high SPL though). For the 323S, I only could compare shortly, but one thing I remember is that it didn't seem to render the low end as well as the more expensive 727 amp ( http://www.head-fi.org/t/531743/new-listening-impressions-of-stax-c32-prototype-and-shipping-sr-009/780#post_7502031 ).