Saturday, September 10, 2016

Adjusting EQ: Lather, Shave, Rinse, Repeat until Satisfied

I might be inclined to say I have no method when it comes to adjusting crossovers and room equalization.  When I finally have the opportunity to do it in a big way, I just jump right in.  I lose track of time, and perhaps 48 or 72 hours later, perhaps after I've reached a plateau of some kind, or more likely just run out of time before I have to go to bed to be ready for work the next day, I call it quits.  Mainly, and that's part of it, because often I keep trying to call it quits, but there's always one more problem that critically needs adjustment also or the whole enterprise was in vain.

And so audio adjustment often keeps me up later at night than intended, though that's sometimes just because of hours of reveling in good sound after a new plateau has been reached.

I sometimes write down my adjustments so I have a trail marker to get back to later.  I've often wished I had these trail markers when I didn't.  So I'd write them down more often except that it always seems that the moment I have everything fully written down I discover another area that seriously needs adjustment.  And even more often, I find in the process of writing things down that there must have been some things I had forgotten to set back to their best position, or hadn't finished setting.  So the writing everything down often takes the longest of all, and can't fully be done until everything else is finished, which is why I don't get around to it as often as I think I should.

I also hesitate to "save" settings on my Behringer DEQ 2496 equalizers.  It seems as often as otherwise when I "save" settings I end up restoring some previous settings instead, or causing some other irreversible change.  So generally I'd prefer to have settings written down before saving them.  And since I can hardly ever get to the point of writing them down without immediately finding something else that needs changing, the saving part is rarest of all.

One of the tricky things is that when you have more than one DEQ and/or DCX unit stacked together, you end up turning the adjustment knob on the wrong unit before you realize it, or pressing the wrong button, etc.  Heck you don't even need a pile of units to do that, just working one unit can lead to unintended and hard to reverse changes.

You might think after all this time (I've been playing with spectrum analyzers and equalizers since around 1979) I'd have a more systematic and conclusive method.   Perhaps precisely because I've been at it so long that I don't.  Things like Room EQ Wizard didn't exist 40 years ago.  What did exist was oscillators and 1/3 octave analog equalizers.  I didn't have a 1/3 octave EQ until much later, but sometime around 1978 a friend of mine lent me his oscillator to sweep my system.  He said it had come from James Bongiorno and was very low distortion or something.  Whatever, it was very unstable and you could only sweep slowly or it would do weird things.  But it clearly revealed what I remember to this day: 30dB peaks!  Sharp peaks and valleys even within a single octave, such as 30-60Hz.  Certainly no octave equalizer was going to be able to fix those.  And yet people had these automatic octave equalizers.  Bah!  Even something like the automatic 1/6 octave EQ function of the Behringer DEQ 2496 is not something I'd bother trying.

I did try using the Tact automatic system a few times, but I didn't like the lack of control over the process.  I would have been happy to limit the correction below 200 Hz, but it was not until many years after my unit was made that they started offering that feature.  With the electrostatic speakers there was a great impact on the highs with even the smallest movement of the measuring microphone.  When you are actually listening, you have two actual ears at different positions, and it works differently and you adjust your head position automatically.  Anyway I think it gets much less complicated at the lower frequencies where you have the longer wavelengths making small movements unimportant.  Now I have the elctrostats more less towed in toward the listener so there is less position dependence too.  Perhaps now, finally, Tact might actually work reasonably well.

Also REW.  I have done exactly one REW measurement, and I was not impressed with the results, even turning my EQ off I hardly saw any of the features in the room response I correct for.  I haven't done more REW runs but I intend to.  But I think one issue is that most automated response correcting systems deal with shallow time intervals, not the long response I see using my oscilloscope.

So, while I look forward to comparing methods, such as comparing REW with "my method" if you can call it that, what exactly is "my method" today?  Well it has even been evolving, rationalizing even as I started thinking about writing this.

Basically, it's what every method does: Lather, Save, Rinse, Repeat until Satisfied.

It's just what you choose to do the different steps.  For lather I currently use 3 things:

1) Oscillator (I like Genrad 1390, though mine needs a little adjustment, and the gap between bands around 200Hz is troublesome.  But log sweep is Essential.)  Basically you sweep up and down and hear where the high spots are so you can shave them down.  Now with bass you don't have to be exactly at the listening position, and this is convenient, since I do most of my adjusting lying next to the controls a couple feet from the listening position, though I do most of my actual sweeping at the listening position now, I have power and signal cables exactly the right length to do this (part of the new setup, I can just pull the oscillator from the rack and set it next to listening position, taking up all the slack in the cables).  It wouldn't be a bad idea to do sweeping elsewhere, but I only did listening to booms elsewhere.  More on that later in this story, as on other points I'm only touching on now.  The Genrad oscillator is far more stable and nice to sweep than James Bongiorno's old oscillator (did I throw it out??? or give it back, I hope I gave it back to my friend, because now I know it was more intended for distortion measurements than speaker sweeping.  Perhaps un-handled Sumo Charlies never got properly alighned because I tossed Bongiorno's unstable oscillator.  My role in audio history: a fluke.).

2) Pink noise testing.  This can't show you where the actual peaks are good enough for adjustment, but of course it shows the overall level which should be as close to flat as possible.  With a system like mine, if you can eliminate the room peaks--mostly peaks--you are almost flat.  Then there are the valleys...  I'll get to those later.  But the flatness I get from merely nulling peaks always amazes me.  I'm nowadays throwing generally fairly high Q notches at the response curve, and presto all the weird looking irregularities in response go away rather than getting worse.  Also, nulling peaks seems to virtually erase room decay peaks also.  Your octave EQ won't do that.

3) Music torture tests.  With some of these, if you can even play them at an intelligible level you are doing good.  In this most recent round, I've been using what came up as I was flipping through my album list first: Bass Ecstacy by Bass Erotic (actually, one of my rarely disclosed favorites).  It can get awful after listening to over and over, if not just once, but it's the best thing ever for adjusting the bass, because there's a constant bass line which is actually quite tuneful it it's not just all boom, boom, boom because of modes.

4) Real music, ultimately, of course, though I consider Bass Ecstacy as "real" as anything.

Then, once I have found an irregularity, usually an alarming peak--of which there are an alarming number in the bass response in my living room--I usually fix it by dialing in a parametric equalization function.  Both the DCX I was using and the DEQ I am allow you to dial in around 10 or 10 parametric equalization functions (PEQ) per channel, defined by their central frequency, bandwidth or Q factor, depth in dB, and direction (increase or decrease, but increase is rarely useful, and even while I am often tempted to dial in increases, even as little as I do I almost always end up cancelling them).

As I implement the crossover functions for each range with parametric EQ's, requiring two EQ's for a 24dB Linkwitz Riley (the DEQ conveniently has "HC" and "LC" responses which are 12dB/octave Butterworth responses, which can be added together to make 24dB/octave Linkwitz riley response functions) I need two PEQ's for that, leaving me 8 more PEQ's to fool around with.  8 is just barely enough and I could imagine using more.

When I first set up the new DEQ for the bass, as I recounted in the previous post, I essentially copied over the PEQ's I had previous set in the DCX.  However there was one small catch, and that was that the DCX specifies Q factor where the DEQ specifies bandwidth.  At first I did not know the conversion and guessed very wrongly, planning to reset later.  When I did find the conversion it became clear I could not translate many of my higher Q settings exactly.  And, almost unbelieveably, they cannot be set to exactly the same frequencies at 0.1Hz resolution.

I ended up resetting the bandwidths just hours later because my guesses were so wrong, I made every notch too narrow and the bass was horribly sucked out.  BTW the rule is fairly easy to remember, a 1-octave bandpass or cut filter has Q of 1.414, and successively narrower filters have proportionately higher Q's plus a bit more.  So half octave is 2.871--just a bit more than twice 1.414, third octave is 4.318, sixth octave is 8.651--just a bit more than twice third octave.  Tenth octave, which I had guessed to have a Q of about 10 actually has a Q of 14.4, which is a bit more than 10 times 1.414.  Since I've spent most of my hobby dealing with Q's in the neighborhood of 0.707, I couldn't imagine such high numbers were needed to represent the narrow bandwidths.

So neither the Q's nor the frequencies could be copied over exactly.  So that's one excuse, but mainly every time I go through these settings I can't help editing them a bit, even as to what seems more logical or reasonable or consistent with recent listening and/or measurements.  I allow myself to do a tiny bit of editing without immediate oscillator sweeping to re-test.  About 1.5 dB, or that much at a time.

The basic process of re-adjustment (not assuming a blank slate, but a previous set of fairly correct adjustments, which is virtually always what I am doing) goes like this:

1) Tune the oscillator into the very center of the most bothersome resonance.

2) If there is no PEQ in that frequency range, this is easy.  We create a new PEQ, very deep and narrow, and sweep it up and down to most complete cancel out the resonance.  Then adjust the PEQ depth back to normal volume level for this frequency range.  Then adjust the Q as to what seems right, and test by sweeping up and down to see that the center of the new PEQ isn't surrounded by new peaks or valleys.

3) If there are PEQ's in this range, a decision may be needed as to which PEQ to coopt, or to create a new one.  If an existing very close one is available, the settings are memorized and then it is made more narrow and deep to sweep as before, then adjusted like a new PEQ.  Then a re-sweeping is done to be sure this fixes the overall problem without making matters worse.  Often a compromise PEQ frequency is chosen.  These things are very strange.  PEQ's interact among themselves and resonances in whack-a-mole fashion, so you squeeze down at one point and the resonance pops up again on this side or the other where there wasn't much resonance before.  But often you can do a pretty good job just by moving the center frequency around to exactly the right compromise point and making the PEQ bandwidth wider.  Other times it may obviously better to create two seperate PEQ's, though nowadays I can't do much more of that, since I'm reaching the 10 PEQ limit of my DEQ.

When I started playing with DEQ units and PEQ's in 2005, I studiously avoided high Q PEQ's.  There's some sort of audiophile truth or myth regarding high Q filters having time smearing properties or something.  But the situation may well be in many cases that if you are cancelling out a high Q resonance, you need a high Q cut to properly deal with it, and result can be close to perfection in the time as well as frequency domains.  About 16 years ago REG of TAS (their most respectable commentator IMO) countered the audiophile myth by pointing something like this out, that (well, according to him) room resonances are typically of the minimum phase variety, and minimum phase filters can be perfectly cancelled by the proper counter-filter, and most importantly that doing so fixes the time domain as well as the frequency domain simultaneously.  And I have observed things like this--except maybe not so close to perfection, but it does generally seem to work this way.  So you are best trying to deal with these things in an engineering-like fashion, computing the needed change and applying it, rather than fiddling around the edges with audiopile tweaks.  And so, like REG, and Linkwitz, and many others, I am a believer in the systematic use of EQ to correct ultimate speaker and room response, and not a believer in the "simpler is better" type of mythology that eschews solutions like digital EQ.  (WRT the minimum phase of room modes...I didn't believe it at all when I read it, but it has turned out to be sufficiently true to make it a good starting point, even in my listening room which inconveniently-for-theory adjoins other parts of the house on two side, further complicating any major modes.)

In fact my recent opinion may be that the higher Q the filter the better, all the less for it to mess up the general frequency response curve, if it doesn't need to be messed with generally.  And quite often it seems, even using the highest resolution scanning possible, with a hand-swept oscilloscope, that cancelling out a deep resonance cancels out the hump surrounding it also, and that's preferable than trying to cancel out the hump with a wide filter, which usually doesn't work very well, only cancelling about half of the hump when you think you are cancelling it exactly.

And the reason why these high Q filters work best is simple--room modes are particularly high Q resonances!  Even when they seem to have broader impact, a room mode is fundamentally a notch resonance.

Now as many times as I've EQ'd the living room, I had really not fully come to terms with 32 Hz or what it means.  The response at 32 Hz is quite a bit recessed at the mid-front listening position.  I've sometimes tried to boost it in one channel or the other, the left channel seeming to handle the boost better.  Just a little boost simply because 32.70 Hz is C1, a very important musical fundamental no good system should miss.  But then within a week I usually end up cancelling the boost after bloat or house rattling has become a problem again, and removing recently added boosts is the first line of defense against house rattling.

I should have though more fully about what the room modes are.  The length and width of the room are about 15.5 feet and 13.5 feet.  That 15.5 foot width creates a major mode at 2Wf or about 34 Hz.  Since the woofer is at the side of the room, the mode causes cancellation in the center and maximization at the front and back walls, leaving to maximum room rattling and discomfort in background listening.

So I can't boost this frequency at all, it generates discomfort when the room is rattling, and the resonance also induces a kind of haze which smears everything, including the midrange, even when there is no rattling.

In fact, as of today, since I finally figured this out yesterday, I've gone to cutting 32.7 slightly, about 5.5dB with 1/7 octave bandwidth.  This precisely seems to remove the wall rattling that occurs in sine wave sweeping even at what seem pretty reasonable levels elsewhere.  I could use as much as 15dB cut, but 5.5dB is a working compromise which works up to my "test" sweep level, which is louder than actual bass in most music.  Usually I try to limit compromise cuts that compromise the listening position for elsewhere to 3dB, but this case is extremely important.

Making this cut did something very good in the room resonance character.  Where previously there was a consistent resonance around 32 Hz which decayed less than half the speed of decay elsewhere, like a sore thumb, now the decay is pretty much in line with everything else.  So it is actually helping the music everywhere, even if it is causing suckout of 32 Hz.  BTW I am only applying this cut in one channel, the right channel, which seems to work because that channel lines up with the important right wall which leads to the kitchen doorway and the water heater tank and HVAC coils.  I so hate it when the heater closet is rattling, I don't think it's good for the equipment in there.  I also wonder if recent wall rattling was responsible for some loose paint around the hall ceiling.

The added midrange clarity is huge.  Of course I've always known you can simulate midrange clarity with overall bass reduction.  But then you miss the bass.  Currently the only significant cut below flat response at the listening position is the narrow 32 Hz cut in the right channel.  Generally the bass is just fine, strong and especially clear now also.

In fact as compared with the electrostat-bass style alignments (flat response in the subwoofer region, no bloated "room curve") I've been using recently, the current adjustments have way more bass power impact than most of them (except the horrid boosty ones that had to be turned back within days).

Not to mention bass impact, which is the best ever.  In fact overall clarity, impact and transparency, and I can play up to real listening level now, reaching 80dB peaks in splA and 90dB peaks in splC.

Some are going to say this is way too loud, but some cars are louder on the road.  On the street outside my home you'll never measure above 40dB splA from my system, which means I meet noise codes 24/7 at max volume, 0dB, inside.

But it certainly does sound loud, and I can play much louder than before without a kind of haze likely caused by resonances kicking in.

BTW another key frequency is about 38 Hz.  That's caused by the side wall distance.  I've been nulling that for years without realizing why.  It peaks at the listening position because the subs drive it from both sides, making the center the peak, and I sit right in the room center.

I still don't understand why I have to null at 45Hz, I've been doing that since I started setting EQ's manually in 2010.  45Hz or nearby has always needed a deep and sometimes wide null, though now less wide as I also null at 38 and 52.

Modes aren't just front to back as we think of them, they're side to side, opposite corners and so on.  There's really no simple explanation, it's just the totality of reflections inside a very complex shape, with multiple rooms and walls of varying flexibility and so on.

Now at least I understand why 32Hz is weak, and trying to make it better quickly goes awry.  This is perfect justification for getting the Bag End E-Trap, the most space effective solution to sell defined modes, and I don't have any space for 4 feet of damping in all wall corners, or more than just a few, and that would still hardly absorb 32 Hz.  An active bass trapping system is essential IMO and it's a wonder they aren't more standard.  Most are still trying to tweak their way to heaven rather than engineer it.

The above shows the current uncorrelated pink noise frequency response.  I would have no technical difficulty in getting flat or even highly elevated response below about 40Hz, but I simply have to cut it off or let it be cut off because of how the 32 Hz depression results from the fundamental room mode and if I boost the response there the entire room shakes.  But still I have useful response if not flat response to 18Hz or lower, as determined with oscillator.  (Sweeping with the oscillator it's clear the mode cancellation cutoff begins just below the highly suppressed peak at 38 Hz.)  When I can fix the room mode with an active cancellation device and/or passive damping or other means, then I may have flatter response below 40Hz.

The above also shows the result of many successful equalization victories, such as attacking the 100Hz bulge.  100Hz is where the crossover frequency so I wondered about the origin of the bulge.  I was trying to suppress the bulge by  dialing back the subs, but even dialing back to -15dB at 100Hz didn't completely help.  Then I got the bright idea (and actually this should always be done) of measuring the panels and the subs separately.  Most of the bulge was actually coming from the panels, and I now think it's primarily the floor to ceiling resonance.  Since I'm sitting within 3 feet from the floor, I get much of the elevated boundary response at the listening position.  This was a rare case when I chose to apply EQ to the panels.  I do dial back 100Hz at the subs also, so the acoustic response for both seems about correct, about 6dB down from plateau at the crossover point, according to my ears.

The 220Hz depression is almost certainly caused by the primary panel reflection to the wall behind it and there's not much I can do about it.  I applied EQ to lessen a huge peak above 500Hz and that makes the depression seem much smaller now in relative terms, and the whole spectrum is beginning to look reasonably flat.  I'm not sure what causes the 500Hz peaking, it could be another effect of the primary panel reflection with the wall behind it.

In the highest frequencies, 20kHz looks right in place with no peak before it at 18kHz, but no huge peak at 20kHz either.  THAT took a lot of work too, and the super tweeters now sound much better with a kind of quasi 6th order response curve I cooked up.  It sounds far better also, with no metallic sound ever.  They often have little sound at all actually, but that is by design.  I still have the benefit of less beaming in the sound as you move around.  Super tweeters are part of my magic portfolio.

Overall there is very little boost, if any, at low frequencies as compared with the midband.  This is the "electrostatic bass" sound which sounds real to me and not bloated.  And meanwhile I couldn't make it have much less output through the bass above 40Hz without it sounding seriously thin.  The point of best sound is often surprisingly narrow.

Now the above is something people usually don't measure...the response in the adjacent room.  This is the frequency response from the kitchen table.  In some ways, you might consider it as good as or even better than the living room response.  It's very nice actually, and here with a bit of a boost at 32 Hz rather than a 6dB drop.  Of course if I attempted the cure the 6dB drop in the living room the peak in this response would be much greater, and it gets worst near the kitchen passageway and the adjacent heater closet.  Anyway, this adjacent room response is surprisingly good I think, and it matters because I most often listen to my living room system in the background...from the kitchen table.  I could play the kitchen system, which is high end too.  But when I'm on the computer at the kitchen table, I prefer my music a bit more distant, with the living room being just right.  I know, for all my talk about preserving sonic information listening to the stereo in another room is the opposite: a form of information filtering.  But indeed that's what I want mostly, I just don't want a system that filters unconditionally.

Since I can never seem to get the numbers fully written down without starting to change them, and it's a lot of work to write them down, I'm just going to take pictures of the DEQ screens, starting with the two subwoofer PEQ's, left and right, followed by the panel crossover and then the super tweeters.


Looking at all the notches I've added to the sub response, you might wonder, as I have at times, if I wouldn't be better simply setting the subwoofer level lower, because it seems like I'm notching out most of it down to about 16Hz.  I've tried setting the overall level lower, but you get sweeps that are just as uneven as with the overall level higher, since most of the response below 100Hz is from the subwoofer, largely due to the 100Hz LR24 crossover.  There simply is no substitute for nulling out the resonances, and you should note that the nulling isn't even at all, it's concentrated where the resonances are, unfortunately, concentrated.

My previous setup had a grevious error that had existed for many months, the left sub was out of polarity.  The polarity difference had started when I replaced the subwoofer amplifier.  Originally I compensated at the panel of the DCX by inverting polarity there.  I remember doing that.  But when I replaced the DCX crossover about 10 months ago, I may have forgotten to carry over that setting.  Or also I had been fooling around with polarity inversions, and possibly I got confused as to which way is the right way.

Sadly the DEQ's have no polarity adjustment, but I am using the digital phase control on the subwoofer at 180.  I've always distrusted these controls because I wonder what they do at intermediate positions.  In general people should be adjusting delays to get time alignments between speakers, not applying frequency-dependent "phase" adjustments.  I think when set to 180, however, it simply inverts polarity, which is what I need.

Needless to say, fixing the subwoofer polarity had a profound impact on focus, it is so much better now.  People think of subs operating below the crossover frequency where hearing has little directionality, but they also operate above the crossover frequency to some degree.

Why me, I wonder.  Why am I saddled with a living room with 33Hz suckout when I am a bass loving organ music lover and 33Hz is a key note?

It does look like the "large room" recommendation has some merit, in fact my living room is smaller than some home theater standards recommend.  It looks like walls should have a minimum length of 18 feet or so, to put the primary resonances well below 33 Hz.

Of course, it's also an opportunity for me to solve universal problems.  Most people in the world cannot have mansion style living rooms not to mention dedicated THX specification listening rooms.

Now some audio-simplicity scolds would say I need to rethink this differently.  I need a whole different arrangement (as if I hadn't thought of that).  I need no sub (no thanks!!!) or one sub placed in different locations (where?) or maybe lots of small subs with no EQ.

Yes I can see that things might be different if I put the subs in the plane of the listening position, but that would destroy the functionality of the living room as an entertaining room.  I have to work within the constraints of multi-purpose rooms, as most people do.  Again, another "opportunity" to be a leading problem solver.  I've thought long and hard about different arrangements and multiple subs and nothing other than what I'm doing works very well for my room as both a listening and an entertaining room--except for the use of a resonance canceling device or devices in the back of the room, and that's the tack I will be taking.  I've enlisted EQ to do all that EQ can do, but one thing it can't do is cancel a suckout caused by the primary room mode.

I am also perhaps unfortunately constrained by my choice of subs.  The PB13 Ultra are Big, and I have two of them in a small room.  Having such large subwoofers creates additional constraints on where they can be located without destroying the multi-functionality of the room.  Things would be different if I had much smaller subs, and perhaps something like the $2k "swarm" of 4 small woofers I that REG of TAS touted recently.

To defend this choice, I've always wanted to do real subsonics down to at least 16 Hz.  You'd need a lot of small subs for that.  I've always wanted the sub to have far more capacity than needed, so it wouldn't sound strained or be limited in dynamic range.  It's amazing how much air moving subwoofers have to do for real subsonics.  If you really want to do subsonics well, no cone driver of any size will ultimately be enough.  The best subwoofer ever made was based on an electronically controlled fan.

So this puts me in the sorry spot of reaching for 16 Hz and finding that makes it difficult for me to do the more prosaic 32 Hz also.  In fact this points out that one can (and I quickly did) reach the point where it's the room response that is not only the limiting fidelity factor, it's the limiting factor in dynamics AND fidelity.

So now I can recount the sad tale of when I first got the PB13, I couldn't make it sound good at all.  It was a one-note charlie, and just buzzed and rattled with every frequency I put through it.  But then I discovered that all these buzzes and rattles actually were coming from my room, and not the subwoofer.  And that set me down the path I'm on today, of using EQ extensively.

Actually this also brings to mind my first subwoofer purchase in 2005.  I wanted a more audiophile-recommended subwoofer, so I got a PSB subwoofer from a dealer in Austin, which required me to rent a van.  This was for my bedroom.  I set it up and got nothing but rattles and buzzes.  I rented another van and brought the two subwoofers (one never opened) back to the dealer.

I seemingly had less trouble with the PB10, a very inexpensive sub from SVS, though by that time I had figured out that most buzzing didn't come from the sub itself.  Finally I got SVS's best cylinder sub for the bedroom, and that has worked incredibly well.  By contrast the PB13's have been very problematical in the living room.  I sometimes wonder if cylinder subs would work better, but I think the main problem is the living room dimensions.

NOW there is one more trick I can try with my 2496 DEQ.  I can apply dynamic EQ, that only notches out 32Hz when it goes above some level.  I think I'm going to try that.  But I've also ordered a Bag End E-Trap now, and I think that's the main way I'm going to fix C1 on my system.

Update: Well sucking out the major mode is immediately obvious, the stereo clearly speaks with less authority in the room.  The stereo has to stand up to the plate and make the room swing just a bit.  I had to compromise again, dialing back by 3dB so there are now only 3dB and 2dB cuts.  Dynamic eq might work if set up properly, or perhaps a Levinson-esque approach of presetting the 32Hz cut for the music at hand.  For regular music just a little cut for taming the room decay, for torture bass discs a larger cut to keep the room from exploding.

Also there was a certain lack of sweetness which made me remove the two notches around 500 Hz that seemed obviously necessary from sweeping.  Will have to determine what these resonances are caused by.  Now the panels only have the 100 Hz cut (and more could arguably help) and the 4kHz intentional roll off which might actually just barely get below flat from 4-6k.  The rest on the panels is au naturale for now and the 500 Hz may be a natural thing and explanation for the damping panels in back.

That's how this goes, yes I admit even for measurement head like me listening and guestimating is the final word, until the next measurement, and so on.

Later Update:

Wow!  Dynamic EQ is having your cake and eating it too!  No need to attenuate at the primary 32Hz mode, except dynamically after 20 ms (or less, perhaps) and and after a particular threshold.  It doesn't sound unnatural, it sounds far more natural than just notching out 32Hz.  It brings back the "authoritative" sound of the stereo to at least let it excite the room mode up to some level, even if limited above that.

I was really missing 32 Hz, which indeed appears to be a very important fundamental to be missing.

Still, it would be better not to need this form of dynamic compression, so I'm looking forward to playing with the E-Trap.  What the E-Trap can't solve may be relegated to compression.

Here is a incredible Room Mode calculator, amroc.  To get numbers similar to what I actually measure, I have to enter length of 17, width of 13, and height of 9.  The actual numbers are 16.5, 14.5 and 8.5 (average).  The actual numbers are probably affected by the hall on one side and the kitchen passage on the other boosting the effective length, and the width being reduced by bookcases and furnishings. Still it clearly shows the complexity and a very similar resonance spectrum to what I am dealing with in any case. It's not all in my imagination.

Nyal Mellor, a major regular at What's Best Forum, has written a good short intro to room acoustics and room correction here.  This is straight ahead, no BS.

Art Noxon, seller of acoustical damping treatments, makes his point about the limitations of EQ, and as other such sellers sometimes do, goes a little OT.  EQ doesn't change the fundamental resonances and their impact on the sound...well actually it does change their relative impact on the playback sound, by making the resonances far less objectionable to musical playback, and to a degree working "with" the resonances rather than ignoring them.  And also, there is such a thing as a sound cancelling speaker, and I'm beginning to think something like this is key to playback in small rooms.

It's possible the most excellent cancelling would be done from a system that operated from the playback source.  Swarm type subs do this.

Come to think of it, cancelling the primary mode is simple in these terms: put a 0 degree phase woofer in the antiphase of the resonance!

I see issues however.  First I'd want this to be a specially restricted woofer, that only speaks to the resonances, and doesn't cross over at the front sub crossover frequency of 100Hz because that would introduce too much imagine blurring spurious out-of-plase high frequency sound.  It could just cross over much lower, or be tuned.

Being tuned in this way, it doesn't have to "sense" sound through a microphone.  (And, it also has the advantage of more time to "think.")  But other than the increased time, the same FFT type convolutions need to be done, either on a "sensed" signal or an electronic one.  If one is just reading the signal as it comes, it doesn't matter fundamentally whether the resonance cancelling satellite sub is reading a microphone or reading the stereo audio signal.  If the microphone is reasonable accurate.  And what is sensed is what actually is, rather than what one predicts one needs to act upon, which is actually too impossible to calculate.

So I point this out so as to make it clear that a speaker with the sound signal isn't necessarily totally better off than one that sense sound in a modal corner, for example.  Though, in the long run, subwoofer satellite resonance cancelling speakers could take advantage of both.

I do regret the rather high price...but I'm being a pathmaker here.  Most people are way behind in discovering the need for things like this.  If there was a mass market for modal cancellers there would be less expensive options.  Though I'm thinking the Bag End company has good experience for making devices like this.  They famously made an "Infrasub" with response to 8 Hz.  Such a speaker would be impossible with ordinary cones (electronic fans could do it) except with extremely good limiting.  So the speaker has 8Hz response most of the time, giving you the ambiance, but not the full 8Hz special effects of a real life explosion, for example.  That kind of thing has it's place, and in fact that's another thing that it looks like all good systems need: dynamic EQ to keep the system and room from exploding.  So you get the full everything up only until you can't.

And then the nagging voice about the money, yes the money.  I have a clear need for 32Hz response, as a bass lover, organ music lover, rock music lover, symphonic music lover.  Circumstances have cheated me of that response, but there's a sensible workable fix, the Bag End E-Trap, it just costs $1650 (incl shipping).  Others might spend a fortune on tweaks to fix the problem and get little or no real results.  I'm headed right at the heart of the problem and taking an unusual but apparently workable solution, and there are no other workable solutions that wouldn't mess up my life.  This is an apparently necessary companion to my pair of $1750 subs.

No comments:

Post a Comment