Any hope for my poor room?

[warrensomebody]

Hi. I'm new to this forum, but I've been interested in room correction for some time, and have recently been trying out REW. What a great tool! A few problems running on my mac (incl. random crashes), but I've gotten far enough with it to ask this question...

Is there any hope for my poor room? I have a fairly high-end system, and I'm a compulsive optimizer, but the room itself has lots of hard surfaces (concrete floors and wall - irregular shape, and large/open to adjoining rooms). I recently upgraded my prepro to a Lexicon MC-12 v5 EQ which was a remarkable improvement in sound quality, but I'm not sure the EQ features are helping me all that much. Since discovering REW, I can now see that I have all sorts of problems...



The red trace is my room response measured without the Lexicon EQ, and the green is with it. All measurements are 1/3 octave smoothed, done with my RS 33-2055 meter, and on my MacBook Pro (I know that a laptop sound card isn't recommended, but it's all I've got right now - actually the calibrated response didn't look too bad to my eyes). The RS meter I suspect has bigger problems, particularly in the 1-2kHz range -- all measurement I take with it (even mm away from a single speaker) show a significant dip there.

You can see that the Lexicon is correctly applying EQ to bring up the 20-40Hz deficiency, but it introduces new problems at 100Hz (this is the maximum EQ setting, which really doesn't sound too good -- the low or medium settings sound like a better compromise).

All-in-all, my system actually does sound good, but looking at these curves, I'm inclined to buy a FBQ2496. However, I'm afraid that even 24 filters/channel might not be enough, or that I'll end up making things sound like phasey mess after getting the response flat. I'd appreciate any advice from the gang here (and "get a new room" is not an option right now ). Thanks!

Warren

[brucek]

Well there are a few issues.

You simply cannot use the Radio Shack meter to do full range measurements. It's for sub only. Ignore any readings past a few KHz....

You would only equalize the low frequencies. The upper frequencies respond to room treatment only. Sometimes a broadband equalizer can be used (such a tone controls) for upper frequencies. You wouldn't pass your mains through an inexpensive parametric eq like a FBQ.

You may as well use a horizontal axis that starts at about 15Hz. Anything below that in this case would be meaningless.

Hard to say what the room response is like until you get a better mic (i.e. ECM8000).

You didn't mention your speakers and sub type.

brucek

[warrensomebody]

brucek,

Thanks for your quick reply. I was planning on picking up an ECM8000 and XENYX 802 baring any "just give up" comments from this forum. :-)

One question - why wouldn't I want to use the FBQ2496 or DSP1124P to correct that big hump around 4khz (assuming a "real" mic tells me it's still there)? I was playing around with the filter EQ controls and was able to place a nice wide filter around it and flatten it out considerably (the large hump, although I could see how trying to flatten all the smaller ones might lead to negative results).

But having said that, there is one bigger reason not to use EQ on my upper frequencies. My system is bi-amped. My speakers are Pipedreams Ref 18's and the subs are dual passive 18" sono-tube like cabinets. I'm currently driving them with 400wpc, and they're place directly behind the towers, approx 15' from the listening pos. I've experimented with moving them around a few feet in either direction, but it didn't make much difference. The subs are near the front wall, but at least a good 15' from the sides. The listening pos is about 2' from the back wall.

One of the things I was playing around with yesterday was whether it was better to go with the active xover that comes with the speaker system, or to let my Lexicon do the xover work. The measurements from above are with the latter setup. However, with the active xover, I have control of the phase with each channel, and I think with that in the loop, I can eliminate some of the 100hz problems. However, with every setup I've tried, the 21, 35 and 53hz dips are there to stay, so I suspect they're room modes. Perhaps it's possible to play the Lexicon's sub "distance" setting to control the phase?

One question about fixing the low-frequency problems with a parametric EQ - would I want to start with a very low target level -- one that's below the bottom of the dips, say 60db -- and then pull all the peaks down to that level, and then boost the bass output level to compensate? Or is there some reason to stick with 75db as the target level, and only pull the peaks down less drastically to that level and leave some of the dips in?

Is there any way to see what's going on with the phase with REW before and after these filters are in effect? Thanks,

Warren

[brucek]

why wouldn't I want to use the FBQ2496 or DSP1124P to correct that big hump around 4khz

Well, generally we consider the mains signal to be pristine and should be 'messed with' the least of all the signals. To put that signal through an A to D convertor and then pass it through a DSP and then back through a D to A convertor and through line amps on input and output is a crime. You won't go to jail - but you should . To add insult to injury, it's a $180 device attached to your mains signal. Did I mention that you would land in jail if you did it.

Great rigs for a sub though.

Oh my, oh my, oh my......... I just looked up your speakers. Pipedreams Ref 18's are $50,000 a pair. OK, jail isn't good enough. Holy cow.

Perhaps it's possible to play the Lexicon's sub "distance" setting to control the phase?

Of course, it amounts to almost the same thing. A phase control is a very reasonable approximation to a time delay.

would I want to start with a very low target level -- one that's below the bottom of the dips, say 60db -- and then pull all the peaks down to that level, and then boost the bass output level to compensate? Or is there some reason to stick with 75db as the target level, and only pull the peaks down less drastically to that level and leave some of the dips in?

Yeah, you have the right idea. You have to realize that you can go too far with cutting and you end up with a signal that's too low feeding the sub, and then trying to compensate with the sub amp and run out of headroom.

When you set up REW, calibrate the target and levels to 75dB and then once the measurement is taken, you can move the target up and down to include as much or as little of the measured signal for REW to consider as you like. Set a new target level while watching your measurement plot, and then Find Peaks and if you don't like what you see, clear the peaks and lower the target and Find Peaks again, etc, etc. Then let REW suggest its filters and then you can add your own and tweak until you like the corrected waveform. Enter the filters and the actual measurement will be very close.

Do all this with the mains off (in stereo). Then turn the mains on and see what the crossover looks like. Adjust the phase (or time delay of the sub) to get the best response. Simply adjust and measure over and over. Takes seconds.

Hopefully that waveform you showed above didn't include your 2 18" subs.......................

brucek

[clubfoot]

warrensomebody you will need an ECM8000 to measure/correct the upper frequencies. Move further away from the back wall and remeasure with the sub only, it will surprise you. Once you have a good sub response that follows the target curve, you can tackle the mains. I know it sounds insane, but set your mains to small and reposition them to obtain the flatest response, then recheck with the sub and use the variable phase control to smoothen the xover region.

[warrensomebody]

brucek & clubfoot - Thanks for both your replies.

Bruce - I completely agree that I probably don't want a $180 device in my mains signal path, but I'm probably in voilation of the hi-fi laws and code of conduct by having a digital preamp, and for usually preferring my mp3 collection to the pile of vinyl and plastic. But as long as I can avoid jail time, I'll keep optimizing!

(And don't get too excited about the cost of my speakers -- I got a very good deal on them used.) (Getting them shipped out was a whole other story, though...)

I'm actually away now, visiting family, so it will be a few weeks before I can take any more measurements... but I did take some of just the sub before I left. I know what you mean when you say you hope the waveform didn't include the subs -- but it did, and I was a little depressed by that. It could be the size of the room, the placement of the subs, or the phase -- I'll have to experiment some more when I get back.

I've included a graph of just the subs with and w/o the EQ (green, and red respectively - 1/3oct, 80hz cutoff, response averaged over 4 sweeps +/-1ft from the main listening position). I'd appreciate any insights you might have into what's going on with them.

And clubfoot - I don't quite follow your comment "set your mains to small..." do you mean set my prepro to think they're smaller speakers which will make it easier to evaluate their response? Would that be different/better than just turning the subs off to get them positioned?

Warren

[warrensomebody]

Sorry, one more graph (I couldn't resist)...

I've removed the 1/3 octave smoothing and also superimposed a single sub's nearfield response (RS meter about 1" above the sub's top driver). I'm surprised at how quickly the response rolls back off below 63hz -- could this be my prepro? I also had a nearfield graph with the EQ in, but it was identical, so I omitted it here.

Also, I happened to notice that the 100hz response of the sub alone doesn't look nearly as bad as in the full-spectrum graph in the original post, so maybe this is just a phase problem where the ranges overlap. I think I understand what clubfoot was suggesting now. But in the unsmoothed sub-only graph, below, it appears that I have a serious nulls at 82, 95 and 100hz which are adding to the problem.

That's about all I can show you until I get back.

[Wayne A. Pflughaupt]

Hi Warren

Welcome to the Forum! I’ll chime in on some of this.

Is there any hope for my poor room? I have a fairly high-end system, and I'm a compulsive optimizer, but the room itself has lots of hard surfaces (concrete floors and wall - irregular shape, and large/open to adjoining rooms).

Actually, the only real problem is the hard surfaces. Irregularly shaped rooms, rooms that are open to others, etc. – those situations often, if not usually, minimize the effects of room modes. The only down side to the open room situation is that it may require more potent subs to fill the extra space. Honestly, from what I’ve seen over the years, the people with the worst bass problems seem to be the ones with shoe-box dimensions.

But having said that, there is one bigger reason not to use EQ on my upper frequencies. My system is bi-amped. My speakers are Pipedreams Ref 18's...

Not sure I follow. Anyone with a subwoofer technically has a bi-amped system. Or did you mean your Pipedreams are bi-amped?

...and the subs are dual passive 18" sono-tube like cabinets.

Something is really fishy here. I can’t imagine a pair of 18” subs that can barely get below 30 Hz. If they aren’t in drastically undersized boxes, it looks like some placement experimentation is in order. If they are not co-located, that could be dragging down extension, too. Wait a minute, that doesn't sound right! I mean it's costing you extension.

One question - why wouldn't I want to use the FBQ2496 or DSP1124P

brucek mentioned the situation using a cheapie EQ with a system like yours, but since your pre-amp is digital, the DEQ2496 might work for you, since it had S/PDIF ins and outs. If your pre-amp has the right connections to accommodate the DEQ, you might be able to use it without the extra DA/AD conversions.

Since discovering REW, I can now see that I have all sorts of problems... All measurements are 1/3 octave smoothed

Personally I would use 1-octave smoothing for evaluating the mains. Otherwise response just looks too scary and needlessly freaks people out. (For equalizing purposes, I actually prefer my old-fashioned 1/3-octave real time analyzer!)

Does the Lexicon have any provisions for manually tweaking its equalization? If so, that’s all you need, not any outboard equalizer. With 1-octave smoothing, you can easily identify any general irregularities you might have. You don’t want to be chasing every little jagged irregularity – you can’t hear those anyway, but broad ones, you can.

In your top-post graph, for instance, a couple of 1/6-octave filters might make some audible improvements for the peaks at 300 and 450 Hz. (If there is no audible improvement [often best noted with pink noise], then don’t use the filters there!)

And especially the broad section between 2-15 kHz. Yes, I know, the upper frequencies are sagging because of the Radio Shack meter. But let’s pretend that was your actual, true response, for the sake of this discussion. Flattening that broad area by a few dB with a wide filter would bring a noticeable improvement.

Or, let’s say that your true response doesn’t sag at the top, but is flat from 7 kHz out to 20 kHz. In that case, your upper end would be a bit hot, so a basic shelving filter to reduce that area would make an improvement.

The thing to keep in mind with EQing the mains is that it’s totally different from EQing the subs. With the subs you want to eliminate as many peaks and valleys as you can, and we often apply numerous finely-tuned filters below 100 Hz to accomplish it. By contrast, with the mains it’s more of a “general” thing than surgical precision. You don’t want to chase every little irregularity – it’ll sound like when you’re finished.

One question about fixing the low-frequency problems with a parametric EQ - would I want to start with a very low target level -- one that's below the bottom of the dips, say 60db -- and then pull all the peaks down to that level, and then boost the bass output level to compensate? Or is there some reason to stick with 75db as the target level, and only pull the peaks down less drastically to that level and leave some of the dips in?

Probably not the best tact. Look at your first graph, for instance. Disallowing 20 Hz, your lowest points are hitting about 60 dB. Your highest peaks are hitting about 78 dB. That would be an 18 dB gain reduction in your signal. You might end up with the signal so low that it won’t drive your sub amp to its maximum output. For instance, if your sub amp has output meters, and they only hit half-way up with the amp’s gain all the way up, your amp isn’t delivering its maximum power output.

And clubfoot - I don't quite follow your comment "set your mains to small..." do you mean set my prepro to think they're smaller speakers which will make it easier to evaluate their response? Would that be different/better than just turning the subs off to get them positioned?

For home theater, we generally recommend running the main speakers set for “small.” If they’re set to “large,” and you have speakers with a lot of bass output, that’s not good. For one thing, if you intend to EQ the subs, you have a problem: The mains will have the same bass peaks and valleys the sub does. But the bass the mains are putting out will not be equalized, so it will negate any equalizing you’ve done to the sub.

I know folks who have highly-capable main speakers feel they’re “not using their speaker’s full potential” when they roll the lows out of them, the price for utilizing that capability is very often inferior bass performance.

I've included a graph of just the subs with and w/o the EQ (green, and red respectively - 1/3oct, 80hz cutoff, response averaged over 4 sweeps +/-1ft from the main listening position). I'd appreciate any insights you might have into what's going on with them.

Well, my insight would be that the Lexicon can’t equalize subs as well as the BFD 1124 can. Better keep it to the mains, if that’s possible.

Regards,
Wayne

[warrensomebody]

Wayne A. Pflughaupt wrote:

Hi Warren

Welcome to the Forum! I’ll chime in on some of this.

Hi Wayne,

I really appreciate all the prompt replies from you guys. This forum is great.

Not sure I follow. Anyone with a subwoofer technically has a bi-amped system. Or did you mean your Pipedreams are bi-amped?

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.

Something is really fishy here. I can’t imagine a pair of 18” subs that can barely get below 30 Hz. If they aren’t in drastically undersized boxes, it looks like some placement experimentation is in order. If they are not co-located, that could be dragging down extension, too. Wait a minute, that doesn't sound right! I mean it's costing you extension.

Yes, I see that something really is fishy now. That last graph I put together was a real eye opener (the one with the yellow line). 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 can't imagine what might be doing that, but I'll play around with different equipment (or see if there's some filter setting in the preamp that's accidently in effect) once I get back.

brucek mentioned the situation using a cheapie EQ with a system like yours, but since your pre-amp is digital, the DEQ2496 might work for you, since it had S/PDIF ins and outs. If your pre-amp has the right connections to accommodate the DEQ, you might be able to use it without the extra DA/AD conversions.

Cool idea. I believe it has S/PDIF in and out, so I'll investigate this solution. It would be great to avoid the extra DA/AD conversions.

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.

Personally I would use 1-octave smoothing for evaluating the mains. Otherwise response just looks too scary and needlessly freaks people out. (For equalizing purposes, I actually prefer my old-fashioned 1/3-octave real time analyzer!)

Does the Lexicon have any provisions for manually tweaking its equalization? If so, that’s all you need, not any outboard equalizer. With 1-octave smoothing, you can easily identify any general irregularities you might have. You don’t want to be chasing every little jagged irregularity – you can’t hear those anyway, but broad ones, you can.

The Lexicon doesn't provide much for manual tweaking beyond tone control and a tilt compensation across the frequency spectrum.

In your top-post graph, for instance, a couple of 1/6-octave filters might make some audible improvements for the peaks at 300 and 450 Hz. (If there is no audible improvement [often best noted with pink noise], then don’t use the filters there!)

I'll need to get my hands on one of the Behringer units and see what things sound like with tweaks in that region. I'm not sure what flat pink noise vs pink noise with frequency humps will sound like, but hopefully it will be obvious which one sounds right.

And especially the broad section between 2-15 kHz. Yes, I know, the upper frequencies are sagging because of the Radio Shack meter. But let’s pretend that was your actual, true response, for the sake of this discussion. Flattening that broad area by a few dB with a wide filter would bring a noticeable improvement.

Or, let’s say that your true response doesn’t sag at the top, but is flat from 7 kHz out to 20 kHz. In that case, your upper end would be a bit hot, so a basic shelving filter to reduce that area would make an improvement.

Makes sense.

The thing to keep in mind with EQing the mains is that it’s totally different from EQing the subs. With the subs you want to eliminate as many peaks and valleys as you can, and we often apply numerous finely-tuned filters below 100 Hz to accomplish it. By contrast, with the mains it’s more of a “general” thing than surgical precision. You don’t want to chase every little irregularity – it’ll sound like when you’re finished.

I see. Is that because people have better ability to distinguish resolution at the lower frequencies, or some other reason?

Probably not the best tact. Look at your first graph, for instance. Disallowing 20 Hz, your lowest points are hitting about 60 dB. Your highest peaks are hitting about 78 dB. That would be an 18 dB gain reduction in your signal. You might end up with the signal so low that it won’t drive your sub amp to its maximum output. For instance, if your sub amp has output meters, and they only hit half-way up with the amp’s gain all the way up, your amp isn’t delivering its maximum power output.

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. 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?

For home theater, we generally recommend running the main speakers set for “small.” If they’re set to “large,” and you have speakers with a lot of bass output, that’s not good. For one thing, if you intend to EQ the subs, you have a problem: The mains will have the same bass peaks and valleys the sub does. But the bass the mains are putting out will not be equalized, so it will negate any equalizing you’ve done to the sub.

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?

I know folks who have highly-capable main speakers feel they’re “not using their speaker’s full potential” when they roll the lows out of them, the price for utilizing that capability is very often inferior bass performance.


Well, my insight would be that the Lexicon can’t equalize subs as well as the BFD 1124 can. Better keep it to the mains, if that’s possible.

Well, I can see that it can't! If only I had discovered this forum a few months ago, before my last upgrade. :-)

Regards,
Wayne

Thanks Wayne. I really appreciate your help.

Warren

[brucek]

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,

Asked and answered by yourself.......

I see that my bass response begins falling off at frequencies below ~63hz with almost the same slope as above 63hz.

Yeah, you need to start with the subs. In fact, start with a single sub and no mains and find the right location and settings that give it some bottom end. Then worry about the equalization of the one sub and then add the second....

If the subs pass through that active crossover, it may have a subsonic filter in it. You may need to use REW to do frquency response of it by itself. We can tell you how to do that...

brucek

[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:

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:

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

[warrensomebody]

Wayne A. Pflughaupt wrote:

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...

Hi Wayne - I guess I was taking cues from a brief foray into speaker design I did a few years ago, where measuring the characteristics of the drivers up close and individually was a way of eliminating the effects of the room, etc. But you do have a point -- there's no way to really eliminate the room from by doing this, especially for bass response, and the response at the listening position is what really matters.

Wayne A. Pflughaupt wrote:

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.

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 get your point, but it sounds like maybe we're saying the same thing here. Talking about headroom in terms of dB is really just talking about power using a particular measurement system (a logarithmic one). As I recall, 3dB is not only a doubling of power, but a doubling of perceived loudness.

Wayne A. Pflughaupt wrote:

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.

Good point. It hadn't crossed my mind that the Pipedreams active xover might do something beyond simply controlling the bass cutoffs and slopes, i.e. have some pre-set EQ built in. My main reason for trying to eliminate it have been (a) quest for system simplification (yes, I know I have a digital preamp, but if I can reduce 2 boxes down to one...), and (b) the thing is completely undocumented (switches on the front labeled A, B, C, D; 2 slope settings but no description of what the db-per-octave numbers actually are, etc, etc). I've tried taking measurements while adjusting all these things, but there are literally so many combinations, and my graphs are so bumpy (as you can see), that it's really hard to tell exactly what all these things do. (And the manufacturer only seems to want to sell me a newer xover!)

Wayne A. Pflughaupt wrote:

...
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.

Aah... but that's the real trick -- to know what to listen for when you're not an experienced sound engineer. It's one thing to think you have a good sense for whether music played over a system sounds good or not, but completely another to judge system anomolies from test tones and pink noise. Of course, the music is ultimately what we're after... but I guess I just can't help myself when it comes to getting things "right". :-)

I read a great suggestion from one of the users on this forum -- set the sweep time to fairly long, so that it goes slowly enough to audibly hear where the peaks and valleys are. I'm going to try that when I get back, and see if there's any obvious adjustments needed in the mains range (e.g. that curious bump from 3-4khz). I think it might be cool if REW had a mode that let you control the sweep frequency manually (kind of like turning the dial on an FM tuner) so that you could really get a subjective impression of what's a problem and what's not. But a slow sweep might be good enough -- certainly a huge improvement over the test tones on that old Stephen Court and Alan Parson's Sound Check CD I used to use!

Wayne A. Pflughaupt wrote:

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.”

Ok - I'm getting a better idea of the problems and diminishing returns from attempting to EQ the mains. I'll be very interested to get back to some measurements with an ECM8000, etc, soon, but given that I think they sound good, I'm going to be very disinclined to put an EQ in there. Thanks again for all your input.

Warren