Originally Posted by cyberbri
Any idea why the frequency response looks like this in the highs?
I'm glad you asked that
. Here is a similar graph of my left main speaker (a Phase Technology Velocity series tower), measured at the listener position & height with RS digital meter pointing almost vertical.http://www.avsforum.com/avs-vb/attac...id=73756&stc=1
What's happening? I believe the "hairy" appearance of the graph at midrange to high frequencies (all frequencies greater than about 300 Hz) is due to comb filtering effects. If you zoom in on any part of the frequency scale, you will see the "hair" is actually narrow, deep minima (or "nulls") at many distinct frequencies. The nulls occur when, at a particular frequency and room location, the amplitudes of the direct sound wave (from the speaker) and all the reflected sound waves (from the walls or even from furniture in the room) sum to zero, or close to zero.
I don't think the "comb filtering" at high frequencies is as annoying to our ears as the graph might suggest. That's partly because each null occurs within a very narrow frequency range, and partly because the nulls are due to reflected sound that is time-delayed (at the listener position) relative to the direct wave. The time delay makes the reflected waves less audible than the earlier direct wave. (That's why we're told to concentrate on treating "early reflections" in room acoustic treatments; the early reflections affect our perception of the sound more than the later reflections.)
Notice that your graph appears flatter than mine below 200 Hz, and significantly flatter below 100 Hz. That may well be the result of the room treatments (absorber panels and cushion): I haven't gotten started with room treatments yet.
The next graph presents some evidence that the comb filtering is due to "reflections" that arrive later than the direct sound from the speakers (if my interpretation of the data is correct). This is a graph of exactly the same measurement as before (same speaker, microphone position, and REW data set); the only change is that I reduced the pre-ref and post-ref time domain widths (under Impulse Response controls) from the standard -125 ms to +500 ms, to a range of only -3 ms to +3 ms, which (I think) eliminates even "early" reflections.http://www.avsforum.com/avs-vb/attac...id=73757&stc=1
The broad peak at 6kHz, followed by a steady rolloff toward higher frequency (in all the graphs being considered here), may be due to the non-flat frequency response of the RS meter. The frequency response corrections that we're using apply only to the low bass region, so we don't know what the RS meter frequency response looks like in the top two or three octaves.
Suppose you want to measure the "inherent" or "room effect free" frequency response of a speaker over the full 20 Hz - 20 kHz range? I believe there are some inexpensive microphone models with relatively flat response and known correction curve from 20 Hz to 20 kHz. But how can the "room effects" on the measurement be minimized? In addition to the time domain narrowing trick (that I already described), the ratio of the direct to reflected sound can be increased by placing the microphone close to the speaker, rather than at the listener position. In fact, the microphone can be placed very close to each driver or "port" on the speaker, and a "near-field" measurement can be done for each driver or port. I've seen these types of "near-field" measurements in audio magazines.
For example, here are some measurements from a recent Stereophile review of a speaker with several drivers and ports, the Monitor Audio RS6 (see the color curves in Fig. 3):http://stereophile.com/floorloudspea...or/index3.html
And here is a picture of the speaker being reviewed:http://stereophile.com/floorloudspea...tor/index.html
Anyone else have thoughts or information on doing "full-range" (20 Hz - 20 kHz) speaker measurements at home? This topic may have been discussed at some point on the Speaker (rather than Subwoofer) forum, but I haven't tried a search yet.