Quote:
Originally Posted by jcx 
well we can compare Tyll's measurements - presumably plotted with 1 kHz normalization - which is where Stax specs the sensitivity - with only 1 dB difference we hardly need to keep track of the offset of the graphs
http://www.innerfidelity.com/images/StaxSR007SZ31576.pdf
http://www.innerfidelity.com/images/StaxSR009.pdf
the 007 does look like it could sound a few more dB lower than the 009 - while nearly matching at 1 kHz (from Stax published specs) there are some few octave regions of the 007 frequency response that are 3-5 dB lower than the 009
while 3 dB is "clearly noticeable" it is often claimed -10 dB difference is required for a subjective "halving" of the loudness
also good to read both parts of the Repeatability article for Tyll's measurements http://www.innerfidelity.com/content/expert-tests-innerfidelitys-headphone-measurement-repeatability-and-reproducibility
Suppose headphone placement and equipment precision is already fully addressed.
You want to measure sensitivity. You have several ways to do that. For instance: a) 1khz sine wave load at 100Vrms; 2) white noise load at 100Vrms; 3) first you measure 100Hz sine wave load at 100Vrms, then 200Hz sine wave at 100Vrms and so on until 20.000Khz and then you normalize your 200 dots using the human hearing equal loudness curve (http://hyperphysics.phy-astr.gsu.edu/hbase/sound/eqloud.html#c1).
Similarly, you have several ways to measure frequency response. For instance: a) White noise load at an arbitrary power output (say 10Vrms); b) a sweep sine tone from 20Hz to 20.000Khz with 2 seconds length at an arbitrary volume level (I believe that when the amplifier plays the 20hz peak it may not output the same power when the sweep sine achieves the 20khz peak seconds later). Now imagine that you want those measures done with 100Vrms and then with 300Vrms (at this output power level we would probably start to clip the microphone event with phantom power).
What happens with both SR-007 and SR-009 with that several methods?
When you have a white noise load, you won’t have much free power. Then the amplifier is a significant variable. Since the SR-009 has a large driving area, a different diaphragm material and probably thin spacers, it won’t demand too much power from the amplifier at the extremes. But if you measure it at higher output levels, you could achieve a region where the SR-009 starts to misbehave (large diaphragm excursions reaching closer to the stators, perhaps phase problems also?). On the other hand, the SR-007 would demand more power from your amplifier (smaller driving area, although it has a not perforated outer section; different diaphragm material; perhaps thicker spacers). With SR-007 you can achieve a region where the amplifier starts to misbehave. If the amplifier have a 10% influence with SR-009, then the SR-007 would have say 20%. And white noise is not the same thing as a real load (music; that sometimes have more HF and other times have more LF content…)
When you have a sine sweep the amplifier is a negligible variable. You have all the power available to play a pure sine wave. But then need to be sure that your normalization is correct, which is not an easy task, at least to me (I am not good with mathematics). And that normalization might be the same as a real load (music; that sometimes have more HF and other times have more LF content…).
I am not an engineer, I have not done such measures and that is my limited mental model. I am sure I am failing seeing this sensitivity thing. But when you throw your ideas to scrutiny you want people to say “you are wrong here and you are wrong there”. Of course only people that made those measurements would have the real experience to reach the right conclusions…
So to me, for now, such sensitivity specifications seem not as extensive or precise as we would need.
Now imagine that the vast majority of your customers not only are concerned with volume safety levels, but also like classical. If you build an amplifier with huge power outputs you need to pay a lot for freight. Then what would be wiser for Stax to sell? (i) an SRM-T2 with an SR-007 or (ii) a headphone with more sensitivity and amplifiers with lower power output? I think the second option, since not only the total purchase price is lower but also more money spent by costumers stay at your link in the supply chain (less freight and you sell more companion amplifiers).
Can your costumers afford both SR-009 and SR-007 and a high power output amplifier? A minority can. Then imagine no SR-009 and a harder to drive multi-layer metal alloy stator with thicker spacers, the traditional diaphragm and the same transparency. Sure it would be better for rock and roll and bass corporeal impact, but then Stax would sell fewer headphones and fewer amplifiers.
Is this reasonable?
