[Wayne A. Pflughaupt]

Hi Warren,

I hadn't really looked at or thought about that nearfield response very carefully, but now I see that my bass response begins falling off at frequencies below ~63hz with almost the same slope as above 63hz.

I’m by no means an expert in this, but I’m not sure how much stock you can take in near-field response, especially one that’s not taken in a highly controlled environment with precision measuring instruments and protocol. No matter what you do or how you do it at home, you can’t fully remove the room’s influence on the reading, because low bass is omnidirectional. IMO, listening-position readings are what matter most. Of course, if you sit a foot away from your subs when you’re listening to music...

I guess that makes sense. I get a little confused between the effects of diminished headroom (SNR?) and the ability to compensate with available output power of the amps driving the subs.

Practically speaking (i.e. in everyday use), “headroom” is the difference between your normal system demands and your maximum available output. Reserve output, if you will. So “diminished headroom” would be anything that would consistently demand more output, thus narrowing the gap to maximum available output.

SNR (signal to noise ratio) is typically a noise-floor rating – i.e., how quiet your equipment is at idle.

I mean if I have a 400wpc amp driving the subs, do I have twice as much headroom to play with as a 200wpc amp?

Nope – doubling available power typically gets you about 3 dB of headroom. To put it another way, raising SPL by 3 dB requires a doubling of amplifier power.

I assume that "small" and "large" relates to the high-pass frequency of the mains crossover. With the Lexicon, you get to set this as a frequency, so I have mine set to 80hz since the sub's low-pass freq is 80hz. Is that "small" enough, or are you suggesting picking an even higher frequency?

Yes, “small” settings refer to high-passing the mains, although such a setting typically simultaneously invokes a low pass in the sub at the same frequency (depending on the equipment in question). Generally, settings between 80-100 Hz are the most commonly used.

I guess what I meant was that the Pipedreams system uses an active crossover, so in order to do room correction on both the bottom end and top end, I was thinking I would need 2 parametric EQ units. But now I realize that (a) I could probably just put a single EQ upstream from the crossover, or (b) I could do what I already said I had been experimenting with -- removing the active crossover in favor of just having the Lexicon perform the crossover duties.

Okay, so the mains are bi-amped. Not sure I would replace the stock crossover with the Lexicon. In addition to the slopes possibly not being the same, it’s possible the Pipedreams crossover employs some corrective equalization and/or level-matching. Many passive crossovers do this, as do active speakers, so it’s not unreasonable to expect that a dedicated electronic crossover might, too.

[Re minimizing EQing the mains to a general correction:] I see. Is that because people have better ability to distinguish resolution at the lower frequencies, or some other reason?

Hmm – how much time do you have? People have written dissertations on this; I’ll try to give the Reader’s Digest version. Let’s take another look at your top chart:

Notice that response is much more ragged for the upper frequencies, and much less so at the low frequencies. If you switched your REW graph to a resolution finer than the 1/3-octave you were using here, it would “look” even worse, especially at the upper frequencies.

Chasing virtually every deviation in bass response with precision equalization is in the realm of practical possibility, because they are much fewer. In other words, we do it because we can. With the upper frequencies, that’s just not practical. For instance, take a look at these highlighted (i.e., unscreened) sections of your graph:

EQ filters 1.jpg

Focusing on the red baseline, notice that in these two places you have ragged response that’s gradually ascending and descending (again, the above-7 kHz response is descending because of the RS meter, but that’s irrelevant for the sake of this discussion). What would it take to flat-line these sections? Well, to start it would take a filter for each peak or dimple. That would flatten the line, but you still would have the slope.

The problem with EQ filters is that they cut a perfectly symmetrical path, like this:

As you can imagine, it’s problematic for a symmetrical filter to deal with a slope. It would take at least a few filters just to accomplish it, most of them essentially equalizing the equalization.

So, in dealing with these two highlighted problems in response, what it would require is multiple filters for each location – probably no less than 6-7 per. That’s a lot filters trying to address response sections that are not even an octave wide. Obviously this isn’t terribly practical, even if it were doable. Fourteen or so filters burned for a small section of response: How many dozens of filters would it take to deal with everything above 100 Hz?

Let’s take a look at another problematic example:

EQ filters 2.JPG

Here we have a plateau in response that falls off very sharply on the left, and gradually on the right. Problems like this are virtually unequalizable with symmetrical filters. You can see with the green line what an EQ attempt looks like: it didn’t eliminate the plateau – it just notched a hole in the middle of it. Which would sound better there, EQ or no EQ? Who knows? You’d have to listen and see.

Look again at the green (after EQ) line, and imagine what would have happened if the filter had been spread wider, to cut down more of the plateau: In trying to achieve our objective, it would have also deepened the low points on either side of it.

That’s a prime example of what can happen with willy-nilly equalizing of the mains: Causing as many new problems as you solve. That’s why you keep things general, addressing mainly broad trends, and perhaps a few egregious dips or peaks. Most importantly, you have to be able to look at a graph and recognize what you can deal with, and what is better left alone. It’s no wonder mains equalization has such a bad reputation: I’d venture that most of the people who have attempted it and badmouth it afterwards are rank amateurs who have no idea what they’re doing.

On top of everything else, much of the raggedness you see with full-range readings is the result of room reflections, and the resulting phase shift (i.e., identical signals reaching the evaluating mic at slightly different times). This is why your speakers always sound much better than they look on “paper.”

There are other issues relating to minimizing EQing the mains, compared to the drastic manipulations we give to subs, that I’ll only briefly touch on. Such as, the drivers in full-range speakers are light duty compared to sub drivers and can’t take aggressive equalizing nearly as well. So are the amplifiers in receivers – it’s not going to take much to have them running out of headroom and clipping (although that might not apply to your situation). And unlike with subs, clipping in the mains is eminently audible.

BTW, what do people think about this Behringer gear's digital performance, e.g. DAC linearity, jitter, etc. I know I probably should expect much given the price range, but on the other hand, all this digital equipment has gotten amazingly good over the past few years. Just curious -- I'm pretty convinced from what I've seen here that its benefits far outweight these concerns.

You say that digital gear has gotten pretty amazing in recent years. brucek says you’ve answered your own question. Both are certainly supportable viewpoints. We have at least one staffer here, Ilkka, who I believe uses a DEQ in his system, and he’s been happy with it. If you Google “deq2496 review” you’ll get some other feedback.

The thing is, though, I don’t know if the people who like the DEQ are using it in systems as high end as yours. That’s something that certainly needs to be considered.

Regards,
Wayne