Wow. We didn't expect or intend for the response to our measurements and comments to be as controversial as they turned out to be. In posting those measurements, as well as my opinions and assumptions about them, we simply wanted to add to the discussion. We certainly didn't post them to stir things up, or to say "I'm right, you're wrong."
- When we first saw Tyll's measurements (and his accompanying impressions), we were--like many of you--surprised. Based on my own subjective experience with the MDR-Z1R, and our observations of the general subjective consensus here on Head-Fi, Tyll's article seemed to us to offer a contrasting assessment.
- With our curiosity aroused, we then measured our prototype, as well as another Head-Fi'ers production unit. While they weren't exactly alike, we think it fair to say they were similar, and neither exhibited the magnitude of the treble peak shown in InnerFidelity's measurements. To be clear, we did not say our measurements are right or Tyll's are wrong. We simply stated that neither of the MDR-Z1R units we had on hand exhibited the treble peak to the magnitude present in InnerFidelity's findings with the one he has on hand.
- Our measurements also exhibited significantly lower harmonic distortion, particularly in the bass range, compared to InnerFidelity's. Again, are we saying his measurements are inaccurate or wrong? No. Nor are we saying that our measurements are more accurate or right, relative to InnerFidelity's. We are saying our measurements of the two units we have here showed less total harmonic distortion than Tyll's measurements showed for the unit he had there. Tyll felt the bass distortion he measured looked problematic and assumed/suspected that the distortion he was seeing might be due to excursion-limiting from a small magnet, and I can understand (given what he measured) why he would venture a guess at possible reasons for the distortion he was seeing, because...
- ...when we review and/or measure, we all have to make a modicum of reasonable assumptions. For example, if we're sent a sample unit (and most of us who get review units usually only get one of a particular model), we assume it's representative of the ones others are purchasing. We, therefore, assume our impressions of the headphone are impressions of a unit representative of the ones others are purchasing. When we measure the headphone, we assume the same about our measurements. Tyll measured a large treble peak and also heard it. We don't doubt him.
- Discussions of compensation curves came up as one possible cause (as there can certainly be differences between compensation curves--we use the KEMAR DF compensation that was provided to us), so I was asked to post uncompensated measurements. I posted uncompensated frequency response measurements of prototype X009 (since that's the unit I'll be sending to Tyll).
- Today, we wanted to see (using the production unit, since it seems a bit more consistent in terms of measured performance) if we could coax this one into producing a treble peak similar to the one InnerFidelity was seeing and hearing from their unit. To do this, we put the headphone on the head, and we tried to maximize its output around 10 kHz. To get it to peak required gentle nudging while watching the monitor in real time--even very small movements can substantially change levels ~10 kHz. Focusing only on maximizing output ~10 kHz--independent of anything else that we'd usually consider as we position the headphone--we were not able to get the treble peak level that InnerFidelity measured in their uncompensated plot, but we were certainly able to get the ~10 kHz level significantly up versus our normal measurements. Here's InnerFidelity's uncompensated plot (red arrow pointing to it below--it's the bottom plot). (Note that I truncated their X axis down to 20 kHz for an easier comparison (our X axis only goes to 20 kHz.)):
- Here's our attempt to maximize ~10 kHz on both channels. Please note that there may be a couple more left-channel (blue) plots than right-channel (red) ones, as we had an easier time coaxing it out of the left channel from positions that looked more normal, so on a couple of the measures we focused solely on the left. Here's what we got:
- Again, focusing only on maximizing the output ~10 kHz, we were able to get a higher reading than with our typical measurements, though still not as high as InnerFidelity's measurement shows there. In doing these, the 30 Hz square wave comparison (a technique we first saw Tyll do in a video years ago, and that we and others have since regularly employed) was not the same for the two channels in any single measurement, and more poorly formed than when focusing on getting the square waves as right as the headphone allows. Again, this is because, for this exercise, exacting overall measurement placement wasn't the main focus, maximizing output at ~10 kHz was.
- We then decided to try something else, which is to do a hybrid between our normal headphone placements for measuring and the above. Specifically, we mounted the headphone for measurement and worked placement as we normally do (including Tyll's square wave tests and other checks). However, once we had it set, we'd then move the earcups in small increments to increase output ~10 kHz (monitoring the measured level in real time) before running the sweeps. The difference from the test above was that we limited the placement and movement to keep the 30 Hz square waves intact and equal between channels (or relatively close anyway). Doing this, we were able to squeeze a few more decibels out ~10 kHz (though not to the extent that we were able to above), with a mild increase in right channel variance down low in a couple of measurements (versus the left), in terms of both frequency response and THD. Here's what we got when doing this:
Our conclusion? Again, it is simply too soon for anybody to be drawing firm conclusions as of yet, assuming any firm conclusions will ever be drawn. Again, our frequency response and THD measurements of the two MDR-Z1R's that we have on hand continue to differ from InnerFidelity's measurements of the one they have. This could very well be due to different measurement rigs, or different MDR-Z1R units, or both. We look forward to measuring (and listening to) the unit that Tyll had, and we are sure he looks forward to doing the same with ours. We expect there to be more illuminating discussion, too, but hopefully with fewer snits in britches.