Hi everybody,
I've been approached by several people to chime in. Great discussion! I’m also very happy the guys at Soundstage did these independent measurements in the anechoic chamber of the NRC. I think speakermeasurements.com is the best source of independent measurements on the web. Great stuff! The results of their measurements on the 8c mostly correspond with our own, although there are also a couple surprises.
I’m pretty sure this is the result of the anechoic chamber not being anechoic at low frequencies. You need very thick layers of fiberglass to damp long wavelengths effectively. At low frequencies even in anechoic chambers you have reflections and room-modes. The speakers don’t have the dip.
I’m not sure how to explain the excess in energy between about 100 and 400 hz. It’s not the result of a voicing decision. We believe in accuracy. Flat response and at the very least smooth directivity are the most important factors in that. We aim for a flat on-axis response and minimal variation within any likely listening window. In the NRC measurements the response in the low midrange also is a bit less smooth than our own. I don’t know why.
Perhaps the platform the 8c stands on can in part explain the rise at low frequencies? It’s probably not the whole story though.
View attachment 25789
The measurements graphs in our spec sheet are based on measurements we did in the anechoic chamber of the Technical University of Delft for mid and high frequencies, spliced to a combination of near-field and ground-plane measurements for the bass. The anechoic chamber is a cube with internal dimensions of 10 meters. What I love about the anechoic chamber is that you can do a single measurement and get a clear picture of what happens across the entire audio bandwidth. It’s very convenient. However, even in this large anechoic chamber below about 200 hz you start seeing the effects of reflections.
View attachment 25790
I spent a lot of time in the anechoic chamber, but since about a year we have access to a large night club, with a concrete floor and a high ceiling. I can get a very clean window of approximately 70 ms, for a frequency resolution of about 14 hz. I actually get cleaner results than in the anechoic chamber, by combining ground-plane measurements for the lower frequencies and semi-anechoic measurements for higher frequencies, with the speaker on a high tower. Although extremely convenient, an anechoic chamber is never completely anechoic. Reflections are much reduced in level and number, but they are still there. In the bass you also see the effects of standing waves. Combining ground-plane measurements and semi-anechoic measurements is more work, but the results are ultimately better.
The two woofers are still almost point sources up to their crossover point at 100 hz. They are simply much smaller than the 3.4 meter wavelength of a 100 hz tone and they have negligible horizontal directivity. In free-field conditions the woofers have a more or less spherical radiation pattern, when you put them in front of a wall that becomes a hemi-speherical radiation pattern, with a directivity index that is close to that of the cardioid midrange.
With a ground-plane measurement, you do indeed have to be careful in your interpretations. If you measure like this...
View attachment 25792
… it seems as if the baffle is twice as large as it really is. Because the ground works as an acoustic mirror, the microphone ‘sees’ two speakers on top of each other. As a result of the seemingly larger baffle, you’ll measure more output in the lower midrange than the speaker has when it’s on a stand. However, if you place the speaker like below and you tilt it so that it points straight at the microphone…
View attachment 25793
Source:
https://www.stereophile.com/features/105kh/index.html
… you’ll get proper results. Ground-plane measurements then actually give you the exact same results as anechoic free-field measurements up to about 2 khz. Above that, the level gradually starts drooping because the microphone is not exactly flush with the ground-plane and you do actually get a bit of a reflection that gets more and more out of phase with the direct sound as frequency goes up.
Now on to distortion. The 8c is a relatively compact full-range loudspeaker that offers low distortion at normal and somewhat elevated listening levels. At high levels distortion can become audible on some program material. This is the price we choose to pay for a cardioid radiation pattern in a speaker this size. However, 96 dB at one meter at a single frequency from a single speaker is very loud. In practice, in rooms with some reflections a pair of 8c’s can be played quite loudly and they do actually sound clean at quite high levels. But if you’re into very loud music, they certainly have their limits.
Distortion in the bass is a bit higher than our own measurements. We use very low distortion Wavecor subwoofers in the 8c, so I'm not really sure why the distortion rises so quickly at around 50 hz. I also don’t know what causes the narrow peaks in the NRC’s distortion measurements at higher frequencies.
Actually, during the design of the 8c I also expected to have to tilt the midrange a bit because a flattish power response might sound too bright. However, the 8c is the first speaker I ever designed that didn’t require any voicing at all. It sounded very neutral straight from the lab. Its tuning is purely based on measurements.
The additional measurement to capture the output of the woofers is done at 180 degrees, with the mid and tweeter on. What you see is the bass (LR4 lowpass), then the very good rearward rejection in the midrange up to about 550 hz and then the somewhat higher level (although still quite low relative to on-axis) between 550 hz and say 1 khz. In the latter range directivity is the result of baffle shading. These results correspond quite well with our own measurements:
https://dutchdutch.com/wp-content/uploads/2018/11/8c-Spec-Sheet-2019.pdf
Please note that at precisely 180 degrees rearward rejection is a bit less good than just off the rear axis. That is because right behind the speaker the diffracted sound around the left and right baffle edge are in phase and thus louder than slightly off-axis. This is particularly easy to see in the horizontal directivity plot in our spec sheet.
That’d be really great!
I agree. Fact is that no anechoic chamber is completely anechoic. You still have to take into account its limitations.
Doug Schneider of Soundstage on 8 April posted the following:
On the topic of measurements, we had a couple editors of publications chime in that they think that including measurements might confuse or put off readers. Is that true? Would readers, as well as manufacturers and distributors, etc., like to see measurements accompany reviews or not? Yes (for measurements) or No (for none) is sufficient, but if you'd like to explain your stance, go ahead.
An interesting discussion followed. Check it out! Personally I’m much in favor of measurements, plus interpretation by someone who knows what they’re talking about.
Please let me know if you guys have any more questions.
Cheers!