Thanks Stefano-M,
Yes -- It is important when comparing measurement graphs like these to understand exactly what is being displayed. These types of graphs can be deceiving, make a loudspeaker look *worse* than another simply because the techniques being used to take the measurements work better for some and not others. That being said, the NRC measurements are still very valuable, once you are armed with the knowledge on how to interpret the data.
Further evaluation of the phenomenon discussed at Ecoustics can be examined here:
Comparing the on-axis graphs of the NHT classic 3 with the off-axis graphs. This is a sealed enclosure, no port.
Classic 3 On-axis:
Classic 3 Off-axis
Notice the bass extension at 50Hz between all the various measurements remains the same.
Now lets compare the bass extension of Paradigm Signature S2, a front ported loudspeaker.
Signature S2 On-axis
Signature S2 Off-axis
Now let us compare to a rear ported speaker, the Sierra-1.
Sierra-1 On-axis
Sierra-1 Off-axis
Examining the graphs more closely, you will notice that with the Paradigm, the bass extension begins to slightly decrease with the off-axis measurements. This is because the port tube, which is on the front baffle of the speaker, is now moving further away from the measurement microphone, subjecting the output of the port to more absorption, due to the effects of the chamber
Looking at the graph of the Sierra-1, on the off-axis measurements, you will notice that the bass extension actually increases the greater the off-axis angle. On the on-axis graph, 50Hz is 9.5 dB down. On the off-axis graph, at 75 degrees, 50Hz is now 6.5dB down, a difference of 3dB (considerable). This is because at 75 degrees off-axis, the mouth of the port is much closer to the microphone, resulting in more information from the port reaching the mic rather than being absorbed. Bass extension will continue to increase as the mouth of the port approaches on-axis (0 degrees)
The rather large "bump" at around 25Hz and then the dip at 35Hz is also an effect of a rear ported speaker being measured in an anechoic chamber. You will see something similar to this on the majority of rear-ported stand mounted speakers tested in the NRC chamber. According to this NRC graph, the Sierra-1 is -10dB at 25Hz, now THAT would really be something
Technically, if the speaker is -3dB at 44Hz (which is accurate), the speaker would be -27dB at 22Hz (anechoic). A ported loudspeaker rolls off at 24dB octave (there are minor exceptions to this but it holds true in general) With that in mind, the on-axis full-range graph posted on our website is far more accurate with regards to expected bass performance below 100Hz than is the measurement taken at the NRC. (our graph indicates -25dB at 22hz, very close to the -27dB I mentioned above)
Measurements are indeed an important aspect of performance, but they do require a complete understanding of exactly how the measurement was obtained and precisely what you are looking at. Much of what I do involves interpreting measurements; validating the results, evaluating potential problems and much more.
For example, just from the simple on-axis measurement of the Sierra-1, I know what the approximate width of the front baffle is. Look at the on-axis graph. Notice the dip at approximately 3.5kHz. Examine the off-axis graphs, notice that the same dip is not there. This tells me that the dip is not caused by the tweeter itself, it is a reflection and because the dip disappears at as little as 15 degrees off-axis, I know that it is not audible, even though it looks "ugly". Experience tells me that this is an edge reflection, an effect caused by diffraction. 3.5 kHz equals a wavelength of ~3.875 inches. So the output of the tweeter is reaching an edge at 3.875 inches, or in other words, the tweeter is spaced approximately 3.875 inches from an edge. Since the tweeter is in the middle of the baffle, 3.875 * 2 = 7.75 inches (the width of the front baffle). While not 100% precise, if we then subtract half of the 1/4" radiused edges on either both edges, which makes the front baffle appear larger than it is on measurements, we end up with precisely 7.5" Which is the precise width of the speaker
Again, there is much information that is revealed from measurements but you have to know how to evaluate it, how it was obtained and most importantly, a quick glance is not nearly enough time to draw any conclusions....
Hope some of you can find this at least somewhat interesting