Success with FBQ2496

[OvalNut]

Hi all,

I just want to take a moment and say thank you to Sonnie, brucek, Wayne and many others here who impart their special brand of knowledge and enlighten the masses.

A couple days ago, I installed a FBQ2496 and dialed it in using REW. Fantastic results. Up to this point, I have used an ART351 1/3 octave graphic EQ for my sub, and it worked, but going with a parametric EQ really made a big difference.

The thing is though, using an EQ for a sub in general, and using a BFD in particular, involves alot of detailed decisions/questions/head scratching along the way. During my little journey along this path, this site helped immeasurably.

Anyway, shown here are some before and after graphs to give you an idea of the kind of difference the Behringer FBQ2496 can make. Can you say dynamic transient response with no overhang? I can.

The sub is a PC13 Ultra tuned to 10hz and located nearfield. The after EQ'd charts show a targeted 8db house curve. I didn't quite achieve that target, but hey, I just first hooked it up a couple days ago, still futzing with it...


Before - Frequency Response



After - Frequency Response



After - Low Frequency Waterfall



Anyways, thanks again to all the good folks here, and for Sonnie et al for putting together a VERY useful site.

Tim

[Sonnie]

You are quite welcome Tim... it is our honor and pleasure.

Looks like you got it all figured out...

[brucek]

You've done an excellent job.

Your final response is very smooth. Actually, it's even better than it looks since you've chosen a more aggressive vertical scale than we normally recommend here. Usually we recommend a vertical scale of 45dB-105dB. If you chose that scale, your end result would look even better on a graph. Good job.

I do have a comment though that may be hard to explain.

House curves are normally added to the existing target. Generally 75dB is a good target and level to calibrate REW.
Then added house curve file indicates the frequency where you would like to begin the increased level. We assign this frequency and give it a level of 0.0dB. Then we choose an end frequency and tell the file the amount of gain to use. So, if I wanted an increase of 8dB from 80Hz down to 30Hz, I would use the simple file of:

30 8.0
80 0.0

You're also allowed to interpolate any number of frequencies and values between 80 and 30 for different slopes you might want. The file I show is in its simplest form for explaination purposes..

Your graphs would indicate that you have entered a value of 0.0 dB at 30Hz and -8.0dB at 80Hz? So if I set my room level with REW to 75dB and use a housecurve file of:

30 0.0
80 -8.0

then I would end up having to be quite aggressive with my filters to lower the signal down to the target.

I may be wrong about this in your case, but that's what it looks like to me.

Here are three graphs showing the three situations of no housecurve, file with positive curve and file with negative curve.





See what I'm getting at? The negative value file lowers the entire target level.

Or, am I way off here and you simply adjusted your target manually to produce the results you obtained (which is quite possible). In the end, it will all sound the same, but my point is (if I'm correct), that you may have been overly aggressive with your filters.

bruce

[OvalNut]

Or, am I way off here and you simply adjusted your target manually to produce the results you obtained (which is quite possible). In the end, it will all sound the same, but my point is (if I'm correct), that you may have been overly aggressive with your filters.

That is exactly what I did, just lowered the target line. Also, I used the simple positive house curve of:
30 8.0
80 0.0

My intent going into this was to maximize extension as well as smooth the response and address low end ringing to the best extent possible, so I lowered the target to the point where the response started steeply dropping off around 15hz, then pulled everything down to there. Yes, there are a couple pretty big cuts in there. I don't have all the details with me here, but I think the biggest one was around -14db in the mid 20's. I'll tell you though, it sounds as great as it looks. Really remarkable.

My intent is to leave it alone like this for awhile and enjoy the results before considering further adjustments. Best laid plans, eh ....?

[brucek]

My intent is to leave it alone like this for awhile

Oh for sure. Anyway, I made my point to those that use a negative house curve. You've certainly done yours correct.

Your waterfall is pretty good too. Mine rings at just about every frequency - a horrible mess. You have a real smooth low ring above 30Hz....... Did you treat your room or are you just lucky?

brucek

[OvalNut]

I'm lucky alright, ... to have a wonderfully understanding wife who really doesn't say much about how my HT room looks.

Yea, it's pretty thoroughly bass trapped and acoustically treated using GIK products. Putting those in made a dramatic improvement in the sound, without question.


Tim

[Wayne A. Pflughaupt]

I’m with brucek – I would have reduced the sub level before equalizing for a target so low. Typically we’ve seen that wholesale cutting or wholesale boosting can lead to problems. Some of the other guys here don’t agree with me on this, but if possible I prefer a combination of boosting and cutting filters.

But in the end, “theory” is often academic. As long as you’re not seeing any ill effects, then it’s not a huge deal.

My intent going into this was to maximize extension as well as smooth the response and address low end ringing to the best extent possible...

You do know the equalizer can’t reduce ringing, right?

Regards,
Wayne

[OvalNut]

You do know the equalizer can’t reduce ringing, right?

D'oh, silly me.


Tim

[JohnM]

Wayne A. Pflughaupt wrote:

You do know the equalizer can’t reduce ringing, right?

Yes it can.

[brucek]

I suspect Wayne means that equalization doesn't reduce ringing with a first order effect. Rather, reducing the stimulus (by supplying the subwoofer with less voltage at the ringing frequency through equalization), as a second order effect results in less ringing at those frequencies..... At least that's my take on it....

brucek

[Wayne A. Pflughaupt]

Something like that. Qualifying that so-called “ringing” is a more precisely long decay times of bass signals – the low-frequency equivalent of what we call “reverberation” in the upper frequencies - I don’t see how there can be an electronic solution to that.

Regards,
Wayne

[brucek]

I don’t see how there can be an electronic solution to that.

Well, the end solution is to lower the SPL level at that modal frequency with an electronic inverse filter. This lessens the amount of time it rings before disappearing...

The result can be classed as an electronic solution.

brucek

[Wayne A. Pflughaupt]

But reducing the amplitude (SPL) of the ringing frequency doesn’t really solve the problem. A high-decibel explosion scene in an action flick will put it right back up! Take another look at Ovalnut’s waterfall chart:

Notice the black marker @ 39.6 Hz, and compare it to the blue marker @ 15 Hz. Both frequencies are at the same SPL level: ~74 dB. The green marker @ 20 Hz is only 1 dB higher than 39.6 Hz, but look at the difference in signal decay times.

Do we really want to use the equalizer to fix the low frequency ringing problem? Sure, I guess cutting drastically below 30 Hz will make the waterfall chart “look” better, but it’s certainly going to destroy extension by the time it “looks” as good below 40 Hz as it does above.

Ovalnut’s bass traps are the reason his waterfall looks so good above 40 Hz. As we all know, bass traps are absorption devices. We typically recommend absorption for the upper frequencies when someone has a long signal decay problem there, too, don't we?

I’ll admit, it took me a while to wrap my head around this. The problem, with me anyway, was that the term “ringing” is a really disingenuous word to describe the problem. To me the word evokes an irritating upper frequency sound, like PA system feedback, if you will. Once I finally figured out that “ringing” is essentially the low frequency equivalent of what we call reverberation in the upper frequencies, it began to make sense. No one recommends equalization to fix an “echoey” room – the solution for that is absorption. Likewise, absorption is the only real way to properly address low-frequency “reverberation,” too.

So –
Absorption to reduce extended signal decay times, aka ringing or reverberation.
Equalization to smooth frequency response.

Regards,
Wayne

[JohnM]

It is not simply a case of altering the level at the resonances. The EQ filter, when properly matched to the resonance in both gain and bandwidth, actually changes the reverberation time at that frequency - indeed if the bandwidth is made too narrow, it will end up increasing the reverberation time, when the bandwidth match is accurate the reverberation time at the modal frequency will fall into line with the rest of the response. Remember that the action of the equaliser is to correct the response at the point of measurement - it does not and cannot globally correct the response and reverberation time throughout the room (which is something appropriate room treatments can do to a large extent) but only at the location where the mic is placed and only for the source (subwoofer) at the location it was when the measurement was made. At lower frequencies the range over which the corrections deliver worthwhile results is larger than at high frequencies, above a few hundred Hz the range becomes very small which is why this form of correction is only recommended for the low frequency range. At high frequencies the region in which the correction is valid is tiny (fractions of an inch from the measurement point) which is why electronic correction of specific resonances or features is of no value there and would typically make things worse, however broad corrections using filters with bandwidths of 1/3 octave or larger can be useful for correcting the general trends of the response at high frequencies.

For Tim's response a filter at 32Hz or so (whatever frequency accurately matches the peak there) of a cut sufficient to bring the peak down to that of the nearby frequencies should have its bandwidth adjusted, remeasuring each time, until the reverberation time is seen to be corrected. The REW estimate for the filter will usually be a good starting point from which to start tweaking the bandwidth. In some cases a small adjustment upwards or downwards in frequency might also be needed, but it is quite evident in the waterfall when the filter is correctly set. The same approach can be used for the other frequencies which show extended decay times.

[terry j]

JohnM wrote:

For Tim's response a filter at 32Hz or so (whatever frequency accurately matches the peak there) of a cut sufficient to bring the peak down to that of the nearby frequencies should have its bandwidth adjusted, remeasuring each time, until the reverberation time is seen to be corrected. The REW estimate for the filter will usually be a good starting point from which to start tweaking the bandwidth. In some cases a small adjustment upwards or downwards in frequency might also be needed, but it is quite evident in the waterfall when the filter is correctly set. The same approach can be used for the other frequencies which show extended decay times.

Thanks so much for the detailed reply John, and even more so for the fine-tuning tweak you've just given us. I can see a bit of back and forth between tabs now ha ha, but it is perfect timing for my new sub install.

[Wayne A. Pflughaupt]

JohnM wrote:

It is not simply a case of altering the level at the resonances. The EQ filter, when properly matched to the resonance in both gain and bandwidth, actually changes the reverberation time at that frequency.

The REW estimate for the filter will usually be a good starting point from which to start tweaking the bandwidth. In some cases a small adjustment upwards or downwards in frequency might also be needed, but it is quite evident in the waterfall when the filter is correctly set.

Okay John, you’ve motivated me to do a little experiment I’ve been mulling for quite a while, to see if any ringing improvements via EQ will indeed disappear once the system volume is ratcheted back up. Going into this experiment I’m assuming they will.

To start, I bypassed my BFD to get this baseline response and waterfall. Keep your eye on the 42 Hz marker in all these charts. As you can see, I have a peak at ~42 Hz. The dotted line shows predicted response with a 42 Hz filter from the BFD, cut –6 dB with a 5/60 bandwidth setting (i.e., 1/12-octave).

Baseline Response (Showing Predicted Action of 42 Hz Filter)



Waterfall of Baseline Graph

Applying the abovementioned filter resulted in response pretty close to what REW predicted.

Equalized Response with 42 Hz Filter

Here are two of the waterfalls generated from several sweeps. The upper chart shows ringing at 42 Hz to be reduced much more than the lower. I wasn’t able to account for this, and I wasn’t able to reproduce it. A few subsequent sweeps resulted in waterfalls virtually identical to the lower chart, so we’ll just use that as our reference.

Waterfall Response with 42 Hz Filter Applied

Putting the baseline and equalized graphs side by side, showing only the pertinent information, for easy comparison:

Baseline (L) vs. Equalized (R)

Predictably, the equalized graph shows reduced ringing at 42 Hz.

To qualify my “any ringing improvements via EQ will disappear once the system volume is turned back up” theory, I took an SPL reading of a 42 Hz sine wave with the equalizer turned off, and got a reading of 90 dB. After applying the filter seen above, I re-adjusted the receiver volume control so that it once again read 90 dB at 42 Hz. This would put our 42 Hz point of interest back up to the level we saw in the first waterfall graph. Here’s what I got.

Waterfall Response with Re-adjusted SPL Level

Putting the baseline and equalized / re-adjusted graphs side by side, showing only the pertinent information, for easy comparison:

Baseline (L) vs. Equalized / SPL Re-adjusted (R)

Conclusion
The graphs tell the tale: We can plainly see that the equalizer did reduce the ringing (aka decay time) at 42 Hz. Some of the improvement disappeared when the original SPL level at 42 Hz was restored, but the ringing did not return to is previous level. So, I have to relent that John is right: The equalizer can reduce ringing. Technically, at least.

Practically, the improvement is not that significant, because (1) ringing was improved only slightly once the original SPL level was restored, and (2) the improvement is limited to a more-or-less specific frequency. And as we all know, the improvement shown more than likely only applies to this specific measuring location.

So in the end, as far as ringing is concerned, the equalizer is a poor substitute for absorption, which will affect a wide range of low frequencies at (I assume) most any location in the room. Much as I love them, I've yet to see an equalizer deliver the tight decay times we see with Ovalnut's waterfall graph. Indeed, comparing his graph to mine, it looks like the bass traps are having a significant effect even below 30 Hz, even though they're not "supposed to."

Ovalnut’s Treated Waterfall (L) vs. My Untreated Waterfall (right)

So - I’m standing my previous assertion:

• Absorption to reduce extended signal decay times, aka ringing or reverberation.
• Equalization to smooth frequency response.

Once again, John, thanks for getting me motivated to do this little experiment!

Regards,
Wayne

[JohnM]

Did you try altering the centre frequency or bandwidth at all Wayne? It is often easier to see the exact frequency of a resonance by looking at where the peak is centred later on in the waterfall, the initial response (which is used to estimate the filter frequency) can have its peak slightly offset from the mode's true centre frequency due to the effect of the response at nearby frequencies, on the 42Hz peak it looks like the true centre might be 42.2Hz or event slightly higher - even a small offset in centre frequency can have a significant effect on the decay. Once you're sure the centre frequency is right, try tweaking the bandwidth up and down and observe the effect on the decay in the waterfall. Thanks for taking the time to investigate.

[OvalNut]

Wow, thank you all for your extended feedback and lesson in the contributing variables surrounding decay times.

To throw a bit more red meat on the fire for you perusal, attached below are the results of some changes I made over the last day or so. Essentially, I really liked what I was hearing in the upper bass range, and don't want it overpowered by the mid-low bass, but I also found that HT effects were lacking a bit, so after some consideration, I went with the following less aggressive, yet still low end weighted, house curve that still provides good emphasis for HT big screen special effects and mood setting ambient bass.

28 5.0
63 0.0

This yielded the following results which seem to be doing the trick from what I’m hearing. It’s interesting too that with this curve, while I am still getting great impact with movies, I have lowered my overall sub level to +4db hot as measured using Avia. The increased clarity is most notable.

Tim

[Wayne A. Pflughaupt]

Thanks for the kind words, John!

Did you try altering the centre frequency or bandwidth at all Wayne?

Only “on paper” (i.e., virtually, with REW) with the first graph. You mentioned that it was important to match the filter with the peak, so I used the BFD setting that got me a cut that “looked like” the peak, if that makes sense. For instance, other frequency centers made the cut look “lopsided.” Don’t forget, you’re kinda limited in the filter tweaking you can do with the BFD because the center frequency and bandwidth controls are fixed steps, not continuously variable like with an analog parametric.

Regards,
Wayne

[Wayne A. Pflughaupt]

OvalNut wrote:

To throw a bit more red meat on the fire for you perusal...
Essentially, I really liked what I was hearing in the upper bass range, and don't want it overpowered by the mid-low bass, but I also found that HT effects were lacking a bit, so after some consideration, I went with the following less aggressive, yet still low end weighted, house curve that still provides good emphasis for HT big screen special effects and mood setting ambient bass.

Hee hee – what “red meat,” other than yet another confirmation for the shelved house curve!

Regards,
Wayne

[JohnM]

Wayne A. Pflughaupt wrote:

Don’t forget, you’re kinda limited in the filter tweaking you can do with the BFD because the center frequency and bandwidth controls are fixed steps, not continuously variable like with an analog parametric.

Agreed, but it is still worth trying the next step up or down, and it is particularly important to try other bandwidth settings. The overall shape of the initial response is the combination of many resonances large and small, so the filter predictions based on that overall shape can benefit from tweaking based on the measured effects in the waterfall. A less even initial corrected response (due to tweaks to the filters) may have better control of decay time. It is also common to find that correcting a large peak reveals other nearby smaller peaks in the waterfall and the ideal solution has extra filters to deal with those, but the filter adjustment process then becomes a manual process based on the results of the waterfall plots.

I really don't get how you are thinking...below the Schroeder frequency in a room we can only speak of decay time and how would possibly a parametric EQ lower the decay time besides that when you lower a peak lets say 6-7dB you don't "treat" the ringing you only lower that to...the only thing that happens is that the rooms doesn't respond as much at a lower level than it did before?

no a dsp/eq is not the solution to room problems...sure if you play your music in a room with lots of problems between 10-100hz then a dsp can be useful...

but i don't think that a flat frequency response is the most important thing...a well treated room with a reverberation time of 0.2-0.3sec is more important in my world..after that it can be interesting to add an EQ/DSP if it is needed to pick the biggest peaks...

but i prefer large helmholtz that not only evens the frequency response but really kills the decay time to..

[JohnM]

Highend.nu wrote:

I really don't get how you are thinking...below the Schroeder frequency in a room we can only speak of decay time and how would possibly a parametric EQ lower the decay time besides that when you lower a peak lets say 6-7dB you don't "treat" the ringing you only lower that to...the only thing that happens is that the rooms doesn't respond as much at a lower level than it did before?

You speak of the EQ filter as if its effects were only in the frequency domain, which they are not. The EQ filter, like the modal resonances of the room, has a time response. The combined effect of the filter and the room's resonance at the measurement point is to address both the excess amplitude of the initial response due to the mode's gain and the excess decay time due to the mode's decay time. To better see the time domain behaviour of the EQ filters just connect the EQ in a loopback and look at waterfall plots of various EQ filters with different gains/cuts/bandwidths.