But what does mean a reference axis for near field holographic measurement ? All axis are measured.
It means that the software is going to choose a given curve among others to display. I'm not sure that this curve is chosen by Amir. He would have to tell the x, y, z coordinates of the tweeter to the software. Is it the case ? If not, the Klippel is probably detecting the axis where the treble response is the strongest.
It is rather a good thing that all speakers are mesured equally. In doubt, we have to read the listening window curve instead of the on-axis curve.
The fact that the speaker was designed to have a best response in a different direction must be taken into account when we interpret the measurements.
I can't imagine the software would simply choose the position where the treble is strongest; that's often not the best listening position. I do think amir has to give it a reference point, and some of the klippel documents mention the ability to choose an expansion point. But you're right that since it's doing spherical calculations, it seems like adjusting the reference point shouldn't be a particularly big deal. It might just affect the accuracy bit if, say, we're measuring at 10 degree intervals, but the optimal angle is somewhere in between. My concern wasn't just with the bass, but the rising treble as well.
Thats said, I forgot the CTA2034.zip file contained the individual 10 degree angles as well. We see at 10 degrees below the tweeter axis, which would probably be the closest available curve to a measurement starting from the reference axis, the response is closer to what's expected. So that makes me feel a bit better
Thanks for the comparison.
Neumann's measurement should be discarded, as it is self-referent : the speaker was designed to sound flat according Neumann's measurements, therefore Neumann's curve is obviously flat..
Surely this doesn't make sense? Otherwise we should discard every anechoic manufacturer curve, including Harman's, because they all target something roughly flat. It should simply be a matter of whether we find the manufacturer's measurement trustworthy. In theory, flat is flat everywhere, which is why we have microphone accuracy standards and the like.
But there is Indeed a difference between near-field direct measurements, as yours, and Klippel's holographic measurement. One explanation, given above, is that holographic measurement gives an accurate result, while direct near-field captation is an approximation. To confirm this, we need to understand the limitations of direct measurement, if they exist.
We might, if we get the chance, compare with other measurement methods. Half space might not settle the argument, as it is a bit inaccurate in this frequency range, where there is a transition between direct radiation and half-space radiation.
Ground plane (microphone against a concrete plane) might be interesting.
True free-field would be ideal, but it requires a lot of Equipment (a crane to lift the speaker far above the ground).
The bass difference is still a puzzler to me, since as previously mentioned, the Neumann measurements in anechoic chambers should be anechoic down to 100Hz before calibration kicks in. The treble is also more uneven, which makes me wonder a bit, since as far as I recall the Klippel
is using gated measurements as opposed to holographic expansion for the highs. But that probably does not have a meaningful perceptual effect.
I just want to get to the bottom of this mystery if possible, because it makes me think that if the Klippel is truly the most accurate way of capturing bass - which is what I thought myself up to this point and have defended in other threads - then there are a
lot of innaccurate anechoic measurements out there, and a lot of speakers that are supposed to be flat that are not, in fact, flat. This may well be possible, but I want to be sure rather than dismissing years of measurements right away.
On a related note I have access to two rooftops in my building directly across from one another with a big gap in between. I wonder how close to free field conditions I could get this way....