Enhancing the dynamic EQ (dEQ)

 I use the Dynamic EQ (dEQ) feature of my the Behringer DEQ 2496 (DEQ) I use for the subwoofers to allow me to have flat bass frequency response with somewhat limited 13" ported subs and a small room.

The dEQ lets me keep the bass flat until the bass level gets too high for that, then gradually roll off the lowest frequency response in the least perceptible fashion.  This lets me play "anything" including bass torture tests like Bass Connection.

For a given size woofer, the required excursion quadruples (or something like that) for each having of frequency.  So the maximum output level falls quickly as the frequency gets lower.  Bass torture tests have lots of bass below 30 Hz where it gets difficult to reproduce and you could damage your subwoofers (as the album warns you).

One limitation of the dEQ is that it apparently triggers on the full spectrum output level, not merely the output level in the band which is being rolled off.  I don't want high output in the midrange resulting in the low frequencies being cut.

I get around this (sometimes, now) by having the dEQ programmed into the DEQ used for the subs only.  So in this DEQ unit, "full spectrum" is limited to below 125 Hz by the crossover (a miniDSP OpenDRC) immediately prior to it.  This still means that a full output at 125 Hz would cause very low frequency reduction when it isn't needed.  But it's better than triggering on the 20-20kHz maximum level.

When I say the dEQ apparently works this way, I'm not sure, because actually the dEQ seems to work in very mysterious ways.  One way this is apparent is in the "level" and "attenuation" meters within the dEQ function itself.

The "level" meter seems straightforward enough, it's the full spectrum signal level (I think).  But the "attenuation" seems to start BEFORE the level exen exceeds the set threshold.  You start seeing the attenuation meter go down (down means more attenuation) before the level has even reached the threshold.  If you are only just reaching the threshold, you may see as much as 5 or 10dB of attenuation in the attenuation meter.

That has long bugged me.  I don't want to be attenuating long before I reach the set threshold.

I explored that issue a bit today.  I played the first track of Bass Connection ("Pure & Perfect Bass") under different conditions to see how much, if any, attenuation is being applied.

It does seem like attenuation starts WAY before the threshold is reached.  OTOH, it appears to be a relatively small amount of attenuation, not the 5 or 10dB the tiny indicator suggests, more like a fraction of a dB in many cases, and probably not that important in practice though still annoying in principle.

For all the tests below I had the output level set to -25dB (-15dB in the chairside EQ, and -10dB in the Tact preamp which precedes it).  This is about -15dB lower than I would actually listen to the album at, but I can simply set the Threshold level lower to compensate for that (at least I think so).  A low level means I can play the track over and over without worrying about damaging anything.

I read the resulting output levels with the level meters set to read the Digital Output (after the dEQ has done its magic).  Sadly these meters do not record anything like an "average" level, only the Peak RMS levels and the Peak Peak levels.  This doesn't entirely show the work the dEQ is doing, because the dEQ doesn't even cut in until after the signal has exceeded the Threshold for the Attack time.

Suppose you had a bass note that was constant in level (as many of those on Bass Connection seem to be) and exceeding the Threshold level I have set and lasts for 1/2 second.   Since I have the Attack Time set to 41 msec, the first 41msec of that signal will go through unaltered, and only then will the dynamic filter be switched in.  If this note is the only thing that is happening, I may not have changed the peak level at all.  The only case where the peak output level will have changed is if there are other notes at the same time, or the bass level isn't entirely constant but gets louder after the first 41 msec.  (That is possible because 41msec represents a full waveform at 25 Hz (this recording goes WAY lower than that) or a half waveform at 12.5 Hz.

So measuring just peak levels is only marginally telling us the attenuation story.  But it's the easiest thing to do, so I did it.

Here are the levels at different threshold settings.

Threshold = -7dB  (There is no output even remotely close to this level, as you can see below, the peak levels are around -37dB in the subwoofer EQ.)

Left Peak/Peak, Peak RMS, Right Peak RMS, Peak/Peak in dB

-30.9, -36.3, -35.9, -31.4

Threshold = -17dB

-31.3, -36.6, -36.6, -31.8

Threshold = -27dB  (my current setting, but note I am playing -15dB below normal)

-32.0, -37.2, -37.3, -32.4

Threshold = -37dB

-33.1, -38.6, -38.6, -33.4

So, when the Threshold is 30dB above the signal, to the point where the Threshold is right at the signal level, there may be about 2dB of loss, and presumably it gets steeper there right at the Threshold.  Not as big a deal as it looks like on the indicator (where it looks like there could be as much as 10dB peak attenuation on below-Threshold signals) though it's weird that it operates this way I think.  You would think that, being digital, they could make it do NOTHING until the threshold is exceeded.

At this point I put another DEQ unit on the bench with my Tek oscillator feeding it and watching the output on my Rigol scope.

I saw essentially this same pattern...slight attenuation starts right away (but only for frequencies at or below the dEQ cutoff frequency) but stays pretty small until the threshold is exceeded, then it increases as programmed.

I also proved that stacking two identical dEQ's has essentially double the effect, allowing one to get past the -15dB limit of one dEQ.

It almost seemed like signals above the cutoff frequency are simply ignored (and therefore not counted towards the Threshold) but I can't prove this without feeding simultaneously or sequentially above and below cutoff signals.  I can feed 200Hz to my dEQ at any level and it doesn't get attenuated at all, but I can't tell if that high level 200Hz would cause a concurrent or subsequent 20 Hz to get attenuated.  THIS is where it would be useful to have a Arbitrary Waveform Generator, and IIRC the Sigilint has a built in function to add harmonics at any level to any sine, so it would be easy to program a 20+200 Hz, or sequences.  I suppose I could also do such things with software, somewhat less conveniently.

 ****

I do now plan to use at least two identical dEQ's to get a potential 30dB of total attenuation.  15dB doesn't seem to be enough for bass test recordings anyway.

It occurs to me now that putting each dEQ in this scenario needs to have a compression ratio 1/2 of what I would use if there were only one dEQ.  So as I was using 6:1 ratio with 1 dEQ, I should now use 3:1 for each dEQ with two dEQ's.  Since I can start the compression at a lower threshold without running out of compression later, I could also try lower ratios.

I want to test this to see if it works as expected. 

The test DEQ is apparently set to high input levels (not -20dB).  So even at full output from my Tek oscillator, I was unable to get it to -0dB even peak, but only -2dB and -5dB in RMS.  I decided to leave it there and call it 0dB on both Flukes I had connected to input and output.

Even with the threshold lowered all the way to -60dB I was unable to get more than 18dB of attenuation with two dEQ's operating at 20 Hz.  But with only one dEQ operating, I could only get about 9dB.

Raising both corner frequencies above about 700 Hz reached the maximum attenuation of 29.9dB and no further changes made any difference.  So the attenuation is limited by (1) corner frequency (relative to signal for which I'm using "worst case" 20 Hz), and Max Gain (which I've set to the largest negative number, -15dB), regardless of how you set other parameters.

I could raise the corner frequency..I may have excess bass energy above that anyway, but it doesn't have excursion limits like the deep bass, so I've felt disinclined to mess more with it (now measuring fairly flat).  Instead, I could have...3 dEQ's, possibly getting me to attenuation of -27dB, close enough to -30.

Changing the parameters at that point made no difference.  Even with 1:2 compression ratio it was maxing out long before -60dB.

With 3 filters, and the corner frequency set to over 1000 Hz, I can get to -45dB attenuation (!!!)  I wasn't sure that was possible on this unit in which mosts things are (damnedably) limited to +/- 15dB.

Now setting the cutoffs to my less-harm-in-theory 79.5 Hz corner frequency (since J5 if not long before) for all 3 dEQ's I get a fantastic -28dB.  Bear in mind this is a worst case setup (exceeding threshold by almost 60dB) that will never actually occur.  It's probably necessary, therefore, to have this much "on tap."  I'm not sure if my system can pass -28dB sines at 20Hz, let alone -18dB, and I have to consider limits down to 17 Hz or so as well.

Now, with 3 dEQs, probably each one should be no more than 1:2 ratio, giving a total of 1:6 when combined.

So, I tried this setup, with 3 dEQ's, signal at -2/-5 as before, and different threshold levels.  At the 0dB threshold level the signal is just 2-5dB BELOW threshold, but there is still attenuation (undesireable in my view, but perhaps needed for some technical reason).

dEQ's, Threshold, Output Level

3,      0,   -7.8

3,   -10,   -10

3,   -20,    -14

3,   -30,    -19

3,   -40,    -24

I'm still trying to understand this, but clearly under these conditions the Threshold needs be set at -30 to get a reasonable amount of attenuation before 0dB, then you get -2dB for the first 10dB above threshold (referenced to the attenuation at 0dB) and then 4-5dB of attenuation for each additional decade in the threshold setting.  And a threshold of -40dB would be better.

WITH a threshold of -40dB, here would be the signal in, attenuation, out levels (computed from above data):

Signal (pk pk), attenuation at 20 Hz

-42dB, -8dB,

-32dB, -10dB

-22dB, -14dB

-12dB, -19dB

-2dB, -24dB

0dB, -25dB (extrapolated)

So in effect a -40dB threshold is going to keep the output level below -25dB regardless of how high the input is.  BUT if I crank up the input to make up for the 8dB loss at threshold, that would make it -20dB, which is about where the danger zone at 20 Hz starts.

If instead I set a -30dB threshold, I get this

-32dB, -8dB

-22dB, -10dB

-12dB, -14dB

-2dB, -19dB

0dB, -20dB

So a -30dB threshold is going to keep the output level below -20dB, but correcting for the 8dB initial loss would make it only -12dB, which is clearly too much output.

Now I'm not sure how much I need to correct for the 8dB "initial" loss as the signal is at threshold.  At even lower levels, there will be less than 8dB loss.  So the question is how far below threshold do I make it flat.

I suspect the answer is I'm going to make up for about half of that loss because the bass is usually lower than Threshold, so figure the max output levels to be -21dB and -16dB respectively.

As long as I'm not doing really slow sweeps at full power, -30dB threshold might be OK.

Now one thing bugs me here.  Why am I only seeing about 4-5dB loss for each 10dB above threshold?  That actually looks like a 2:1 ratio, not the 6:1 ratio I though I was getting from 3 dEQ's each having 2:1.

But I now think the dEQ's are not in parallel, triggering off of the input level, but in series, so each dEQ triggers off the output from the previous dEQ.  What's happening is that at first only 1 dEQ is operating.  Then, when it runs out of max minimum gain, then there are 2 dEQ's operating, and so on.  So I'm not getting something that's stalling out with just 8dB or so of total attenuation, but can keep on compressing over a wider range of input levels, BUT not exceeding the original ratio.

Dialing in a steeper ratio, I might be able to keep the threshold as high as -30dB, and still keep output below -20dB, along the same lines as I was getting to -16dB previously (a correction of +4dB).  I think it's good not to set the threshold any lower than -30dB, my ad hoc setting was -27dB.

AND, maybe I would only need 2 dEQ's rather than 3, reducing the initial loss from 8dB to 5dB.  As I am doing things here, I don't need anything like the -45dB max attenuation I know is possible from 3dEQ's.

AND, there's a whole other set of possible ideas possible by using 2 or 3 dEQ's but with different thresholds and cutoffs.   In some range, say -40 to -30dB, we might only attenuate the very deepest bass, say below 30 Hz.  Then, from -30dB to -20, we attenuate seriously up to 40 Hz with an 80 Hz crossover.  THEN, -20dB and above the cutoff frequency is set really high so it acts more like a regular compressor.  (The DEQ's also have a DYN module that does simple compression without regards to frequency, and I used to use that too.)  In the case the the dEQ's in series we have to be mindful of how that works also.

A sort of frequency-broadening compression is what I've long thought well designed subwoofers use in their driver protection circuitry, if they have any.  The higher the output level, the greater the range of frequencies which is rolled off.

I also looked at having just two dEQ's again, and at 1:2 ratio it's very unsatisfactory.

So here's 1:5 ratio at -30dB Threshold

dEQs,  attenuation,

3,   -24.1dB

2, -18.9dB

It looks like I am actually getting the 5:1 ratio (signal is at -2dB, remember) all the way to the top, whereas with 2dEQ's I'm only getting 5:1 ratio for the first 20dB, then no more attenuation, though it might actually be spread out over that range.

So this looks like the winning package for now:

-30dB threshold, 1:5 ratio, 79 Hz corner, and 3 dEQ's

Most of these are simply from ad hoc goodness, like 1:5 is normally said to be as high as you want to go (and this is the bass, so more indifferent), and 79 Hz corner means the attenuation really kicks in below about 40 Hz, where it is needed.  The need for 3dEQ's being the part determined experimentally.

BTW, my claimed 20 Hz above turns out to be only 15 Hz because my old Tek oscillator is not calibrated.  I noticed that the amount of attenuation increases down to the lowest frequency I can measure, about 12 Hz.  So the numbers above need to be corrected by another 2dB or so to account for the fact the test was actually done with 15 Hz instead of 20Hz, if we wanted to get so-much attenuation at 20 Hz.

I could give up a little max attenuation with only 2 dEQ's.  That might be better for anything but bass test records.  1 dEQ looks useless.

Deploying this much dynamic EQ means I will probably have to modify my regular EQ to compensate for how much loss it causes at normal level, whatever that is.  In future I may have to look at Pink noise at multiple levels to see how much the dynamic EQ is affecting it.

Without something like dynamic EQ, one would have to have different bass settings for each record--like Dick Burwen does--lowering the bass substantially for "bass test" records, but possibly raising it for others.  In principle the dynamic EQ applies the change needed as needed from moment to moment.  I wouldn't be surprised if Dick Burwen doesn't use that also because he's been making dynamic devices of many kinds for a long time.  But his own 20,000 watt speaker system hardly needs it, the entire listening room is an extension of the bass horns.

Why is this needed for most people?  Because basically we can't easily build woofers to have the same dynamic range as other speakers, especially at 30 Hz and below.  There is a high order increase in the amount of air that needs to be moved at lower frequencies.  "Typical" speakers cheat by not having much response below 30-40 Hz or so, they just don't bother.  If you are serious about getting subsonic bass which you can feel, etc, at 20 Hz and below you need to have some way of limiting the signal when you are approaching the reasonable limits of your speakers/room/everything setup.  Dynamic EQ is thereby essential for subwoofering, but the dynamic EQ that may already be built-in to the subwoofer is only for protection, not to limit the sound to where it "sounds nice," which be be somewhat room dependent.  Therefore, dynamic EQ is basically a requirement for serious subwoofering, or even just good full range woofering, unless your bass horns are as large as a room.

Update July 12

The living room subwoofer DEQ currently has a single dEQ, with Threshold at -27.  With three dEQ's, there's not really any reason I couldn't set the threshold to -27 or even -26 and still hold the maximum level below -20dB, according to data above.  With two dEQ's, I'm not sure there's any way to do that, except possibly with higher ratio than I've been using.  With one dEQ, I can barely keep the maximum level below -10dB, and -20dB is simply impossible to hold as only 15dB of total attenuation is possible even with very high corner frequency.

So I think I use 3dEQ's with threshold at -26, as out-of-the-way as possible.

That will be saved as the new DDD (3 dEQ's) and DD (2 dEQ's).  J5 already had one DEQ set the same except threshold at -27 and I am leaving it alone.

Yesterday J5 was doing wonderfully.  I heard the most stirring realization of Amused to Death by Roger Waters (in 24/96 download)...I hadn't previously realized how complex the album is, how spectacularly deep and rich the bass is, and the explosions...  Waters was the bass player in Pink Floyd before going solo and making this album, which I now feel even improves on Dark Side of the Moon, my previous favorite.  

DDD seems bulletproof with Bass Connection.  I can crank it up to 0dB on the chairside EQ (Tact set to 0dB als) without strain.  Everything but the bass gets louder higher than about 6dB.  This is unreal.  I'm used to everything going to hell before -9dB on this album (with no dEQ or dYN).

Maybe a bit TOO bulletproof.

Watching the attenuation indicators in the 3 dEQ's makes me think it might even be compressing TOO MUCH at first, then trailing off.  The first dEQ applies 1:5 compression on the signal, and then the second dEQ applies 1:5 compression on that, etc.  Eventually the first dEQ runs out of compression and just the second two are operative, and eventually they all run out of compression.  Perhaps the thresholds need to be staggered to keep the compression at a relatively constant level.  That needs bench testing.

I'm trying out DD on Bass Connection.  It's more mind blowing at -9dB playback level, I think, and doing fine otherwise, though there's a bit more wall shaking.  Outside the listening position, and/or when are aren't in it for the mind blowing, DDD might be better.

The staggering issue applies to DD as well as DDD.

If I calibrate at 7.35V as 0dB input, and the corresponding 6.92V output as 0dB output, at 20Hz (actual frequency this time), with triple dEQ's at -26dB, I get

In, Out

0,  -20.6

-5,  -22.8

-10,  -24.9

-15,  -27.0

-20,  -29.6

-25,   -32.7

-30,  -36.0

-35,   -40.2

-45,   -47.8

-55,   -57.1

Even at 55dB, there appears to be 2dB loss to the compressor, but making that the reference, I can recast this as dB loss to compression per 5dB (ad hoc interpolating the last two where it hardly matters), this gives me:

bsCenter, bs-(q-p)/(5/bs)

-50,  0.35

-40,  1.2  (unsafe interpolation because underlying change)

-32.5,   0.8

-27.5,   1.7

-22.5,  1.9

-17.5, 2.4

-12.5, 2.9

-7.5,  2.9

-2.5  2.8

With 1:5 compression, there should be 4dB compression in each 5dB.   Within the working range, there's more like 2-3dB compression, which is around 1:2, about half what I dialed in, but it seems to work pretty well anyway.

Obviously the compression isn't being front loaded at the threshold around 25dB and running out of space later.   In fact it's doing pretty well from the threshold on down, with the compression increasing nicely from -32.5dB on down.  If anything, it starts too slowly.  It looks like staggering the thresholds would only make it worse.

I would prefer not to have any compression prior to threshold, but this is what I've got.

(It took a lot of re-calculation and checking to get this correct because it's so mind bending.)

I'm now wondering what it's like with only 2 dEQ's, which I might sonically prefer.

After having done the measurements, it appears that with 2 dEQ's, the situation is not much different than with 3 dEQ's, just with a bit less attenuation.  At the 0dB I'm using here (actually -2dB peak on the Behringer) I'm getting a total of 15.6dB attenuation for 2dEQ's, as compared with 20.6dB for 3dEQ's.  However, as we back off all the way to -35dB signal level, there is still a 1.7dB difference favoring 2dEQ's.  So the differential improvement in useable compression is only about 3.2dB over the range from 0 to -35dB with 3dEQ's compared to 2dEQ's.  This is spread out evenly, with no particular concentration near the threshold.  In effect, there doesn't seem to be as much improvement was expected with 3dEQ's, and neither one has a bunching-up-near-threshold problem that I had expected based on the Behringer documentation (which now doesn't seem quite accurate in explaining the dEQ) and the visual compression indicators.  Though there is very slightly less compression near 0dB with 2dEQ's, whereas there is an increase in compression with 3dEQ's, this isn't a particularly large effect and doesn't continue to the next band.

Here are the compression levels in each band defined by the center levels between two level measurements

Mean Level,   Band attenuation 2dEQ's,   Band attenuation 3DEQ's

-2.5,    2.0dB,   2.8dB

-7.5,    2.4dB,   2.9dB

-12.5,  2.5dB,   2.9dB

-17.5,   1.9dB,  2.4dB

-22.5,  1.4dB,  1.9dB

-27.5,   1dB,     1.7dB

-32.5,  0.9dB,    0.8dB

Roughly speaking, 2dEQ's reaches 1:2 compression at maximum but is typically less than that, while 3dEQ's typically reaches or slightly exceeds 1:2.compression.

I'm now thinking 2dEQ's is better by not being much worse in compression, and having almost 2dB less undesired compression at threshold.  It's worst weakness is that the compression does fade slightly near 0dB, but just slightly and nothing like the complete loss of additional compression I was expecting from the display and documentation.

If that final -2.5dB level compression could be improved slightly using staggering or ratio change or Fc change or something, 2dEQ's would be better still.

I next tried varying the ratio from 1:2 to 1:100.  It surprisingly makes very little difference.  Going from 1:5 to 1:20 only raises the the total compression with 2dEQ's from 15.1dB (threshold=-26dB, my new standard) to 15.9dB, a 0.9dB increase.

I'm going to have to explore this more to be sure, but it also seems that varying the ratio changes how the compression ratio ramps up from the threshold to 0dB.   I think it starts more slowly with the lower compression ratios, and more quickly with the larger ones, but only to a very small degree.  In either case, the compression that is occurring close to 0dB remains about the same and pretty close to the average compression from the threshold (as in the two full examples above).

I think this means that both the Behringer documentation and the indicators in the DEQ itself are basically wrong in their description of how it works.  The documentation and tiny indicators within the dEQ mode itself suggest that the compression ratio you select influences how steeply the compression occurs just after the threshold.  Then, when the total MGain (only down to -15dB) is exceeded, no more compression occurs.

This is not at all what happens.  Instead, the overall compression ratio is mostly determined by the number of dEQ's you put in sequence.  With 1,2,3 dEQ's you get 8, 16, or 24dB total compression within the range above the threshold (with -26dB threshold, 20Hz test and 80Hz corner frequency), and it's distributed fairly evenly from the threshold to 0dB.  Meanwhile the "compression ratio" control really only does fine tuning, changing the total compression by +/- 2dB or so, while also very slightly changing the shape of the curve of how the compression is being applied just below the threshold, but it's not a matter of using up the compression then reverting to no more compression.  In every case, there is more and more compression going all the way to 0dB, though possibly only about half as much near 0dB (because that doesn't change much with the higher ratios, only the part nearest threshold).

Without doing more testing, I'm not sure which "compression ratio" gives the nicest looking compression curve.  It does seem like the higher ratios give more compression right after the threshold, whereas they all give about the same compression later.  I think in general it's better to have less compression right below the threshold and more later, so a lower "compression ratio" than 1:5 might be preferable, but 1:5 is a good compromise because with only 2 dEQ's, with 1:2 compression ratio the total compression falls by 2.5dB, whereas there is only 0.8dB additional compression (added mostly up front) by going to 1:20 and just a trace more going to 1:100, but I don't think that extra compression is worth changing the curve to relatively more up-front compression.

Even at system level of -9dB, the bass on Bass Connection frequently exceeds -15dB at the input to my subwoofer DEQ.  So it is definitely within the compression range a lot.

I think I've read somewhere that nobody buys the DEQ 2496 just for it's compressors, but once you have it they can be useful, if you can figure out how to use them.  One nice thing about the dEQ is that frequencies above or just below the corner frequency are hardly being affected at all.

But the ideal version would have ratios that actually worked, no stupid -15dB limit so stacking not needed, and a more intuitive way to set the shape of the onset of compression, and of course little or no compression below threshold.

My listening tests suggest that using only 2 dEQ's is better sounding, but it might be worth listening to the different "compression ratios", or using 3dEQ's with a higher threshold which can only be done with 3dEQ's.

2 DEQ's puts me just below a 1:2 compression ratio overall, despite my 1:5 ratio selection.

In an hours long serious listening comparison of DDD and DD, playing Drivin' Bass by Bass Connection (which I'm now thinking is the best of these Bass Test records) it was clear that DD was far better.

Listening to DDD, I was thinking, "This is good bass, but I think DD was better."

Switching back to DD, I wasn't thinking, just feeling the bass, and minutes flew by like seconds because I was in such ecstacy.

DD has the magic, DDD doesn't.

There definitely are some room rattles with DD however.  I've been checking into them.   There does not seem to be any problem with the subs themselves.  In fact, I can crank up the level to at least -2dB on the chairside EQ, for the most part, with few issues on this record.  (Without any limiter, I can hardly play the album at -15dB.)

DDD seems to sanitize the bass, possibly even making it slightly more tuneful by emphasizing the higher harmonics, but you are much more missing the actual deep bass and how it feels.  DDD also gives you the feeling that you can just crank the level up to anything, even above 0dB, and nothing bad happens.  DD doesn't give quite that same sense of impunity, but listening isn't just about avoiding bad sounds--it's more important to be enthralled.

So I'm sticking with DD for now.  I see it as possible that a different configured set of 3 DEQ's might be better than DD if the threshold is raised, which remains a possibility for 3 DEQ's.  With only 2 DEQ's, it's clearly at the limit and I don't want to push things any farther, I need all the bass attenuation I can get from starting them both at the same threshold.

This test may not have been entirely "fair" because DDD has about 2dB more bass loss even at lower levels.  To be completely fair, I'd need to compensate for that.  But I'm already convinced that wouldn't restore the lost magic, and DD is also easier to use because there is less bass loss at lower levels too.

Speaking of bass lost, I need to check it with pink noise and such.  I plan to adjust DD to being as flat as response was before at the same fairly high level as I have used previously, and then see what it's like at others.

I'm thinking now there is some loss at 20-40 Hz at low levels, so their EQ's should be raised a bit to compensate and get flat at a reasonably high level.  But I'll keep J5 separate and clone it for an out-of-chair EQ (which may even add more dynamic EQ).  So I'll have one new EQ for serious listening, which reaches flat despite the dynamic EQ, and one for outside-chair listening.  And now is a good time to split because I'm boosting the bass just a tad more for serious listening, when for non-serious listening it doesn't need any more boost and could actually use some cuts.

Update July 14

Today I was going to just make some "definitive" measurements.  But's it's hard for me to make a measurement without immediately wanting to change something.  And I made mistakes, and took screen shots of so many spectra I couldn't remember which was which.

So I just went ahead and made what appeared to be the needed adjustments to get back the previous flatter bass.  It didn't seem to make much difference what the level was between -12dB and -6dB as far as the loss in the deep bass because of the double dEQ compressors.  So I standardized on -9dB on the chairside EQ, which is now exactly that relative to signal level because I have the Tact set to 0dB.  Previously I was using -10dB with about another -1.2dB on average with the Tact.

So this was adjusted for louder Pink Noise than I've generally used, and I generally used less because I was afraid of damaging anything, but now I have dynamic EQ protection, and louder test signal also means greater signal to noise.

For lower levels than the EQ is set for, as levels are lowered the bass should fall very slightly less than everything else, giving a slight "loudness" effect, but so slight it's hardly worth mentioning.

So basically this involved raising the bottom 4 GEQ sliders on the DEQ up about 3dB.  I first followed the advice given by my RTA app, but then backed off quite a bit.  Before I backed off, the EQ levels were as high as +8 and +9.5 dB for the 25 and 31 Hz, because there was already considerable boost there.  Watching how it affected the spectrum, I backed off on both to +7.5dB, still more boost than before, especially at 25 Hz.  At 20 Hz, I cancelled 6dB of the originally -10dB cut, leaving it at -4dB.

The new EQ sounds great in the listening spot, and rather boomy (but not intolerable) everywhere else, as intended.  It's saved as DD matching DD in the subwoofer EQ.  For the nicely limited but not re-boosted sound, which is now preferable not in the listening spot, memory J5A should be chosen to go with DD in the subwoofer EQ.

July 24,

The current dEQ is working very well.  It can almost be cranked up to 0dB with impunity on anything, just a bit less so than DDD.

I did the finishing editing on my own new record with the level at -4.5dB and all was fine.  (Though I am still worried how some things might be with simple level normalization that ignores deep bass levels.)  Then I played something Roon recommended, Conditioned Air.  Well, my album had nothing like the sustained bass power of this album...though it is already normalized.

So I remain convinced of the necessity of deep bass limiting (if not already done well enough by the speaker designer, etc) and that I've done a pretty good job.

Last time I devoted any significant time to it at all, I used both dEQ and DYN.  That was a few years ago.  Since then about all I generally did was change the threshold in the dEQ.  I didn't even know there was more than one dEQ that you could layer.  I think I'm doing much better now, and only with dEQ which is more appropriate because I'm running the sub up to 125 Hz.

I generally didn't keep the dEQ and DYN up to date with gain structure changes (in any of the myriad level controls,  of which some can be made lower while others higher achieving the same effect, etc) of which there were several important ones.  Anyway, it wasn't generally as effective as I have now, and in the early moments of the designing the new bass EQ (earlier article) I switched it off altogether and the result was much music was impossible to listen to.  Especially "bass tracks," which I often like to listen to.