I dont have any of that spiffy audio graphing software nor really any desire to spend a hundred or so dollars obtaining a sound card and mic to run it. Since i dont have an EQ to change anything so precise measurements are not that important. So i just used the realtraps test tones cd and my spl meter to measure some responses of my sub at the listening position. I have never done anything like it before and only had the info from here http://www.realtraps.com./test-cd.htm to go on. I just set my receiver to my normal listening level for movies since thats the level that really only matters to me. But the numbers i got were all over the place with huge gains and losses throughout. I did it 3 times to make sure they were all correct. These numbers are with no correction applied. Im not sure if the correction table for the RS analog meter would even apply to something like this.

15hz-87db
20hz-79db
25hz-93db
30hz-92db
35hz-95db
40hz-89db
45hz-81db
50hz-83db
55hz-86db
60hz-88db
65hz-91db
70hz-95db
75hz-94db
80hz-90db

So my questions are is it like this because of the placement in the room and the room itself??? And would buying an EQ fix this???

I dont have any of that spiffy audio graphing software nor really any desire to spend a hundred or so dollars obtaining a sound card and mic to run it. Since i dont have an EQ to change anything so precise measurements are not that important. So i just used the realtraps test tones cd and my spl meter to measure some responses of my sub at the listening position. I have never done anything like it before and only had the info from here http://www.realtraps.com./test-cd.htm to go on. I just set my receiver to my normal listening level for movies since thats the level that really only matters to me. But the numbers i got were all over the place with huge gains and losses throughout. I did it 3 times to make sure they were all correct. These numbers are with no correction applied. Im not sure if the correction table for the RS analog meter would even apply to something like this.

15hz-87db
20hz-79db
25hz-93db
30hz-92db
35hz-95db
40hz-89db
45hz-81db
50hz-83db
55hz-86db
60hz-88db
65hz-91db
70hz-95db
75hz-94db
80hz-90db

So my questions are is it like this because of the placement in the room and the room itself??? And would buying an EQ fix this???All that variability is very normal and is due to the effects of the room and, yes, it is also related to the sub's location in the room. Moving the sub around will change the room's effects and your response (FR) can be either worsened OR improved by altering the sub's placement within your room.

Yes, you would need to apply the correction factors that are appropriate for your meter to your readings for them to be "correct". And before you measure your FR you would want to make sure your sub is calibrated to the right volume level with your speakers using generic calibration tones (not individual frequencies) such as those provided by your receiver or a calibration disc. I do not know if your test CD includes these generic calibration tones or not.

There are physical ways to adjust your FR to be flatter using placement and room treatment (i.e. traps, etc.), but EQ is what people use beyond that to adjust their FR to be flatter. If a flatter FR is your goal, then you would try to achieve that using whatever combination of placement, room treatment, and EQ you have at your disposal. It CAN be improved by placement and room treatment, but if you really want it to be as flat as possible, it would very most likely require EQ.

Correct your measurements with the correction factors and post back with your corrected measurements. But, in just giving them a cursory look, they are not really all that bad. It looks like your sub may be running a bit hot. But in order to really be able to "see" what your room's FR looks like, you would need to graph your measurements.

Thankyou for your reply. I have already used the receivers auto cal setup and internal test tones to calibrate everything before i did this. I will apply the correction to the numbers here shortly just to see what it really is. But i cant move the sub since I built it to be part of a entertainment center. The sub looks this way since i just set the volume to my normal listening level and not to the 75db of pink noise through all speakers. But i like running it about 10db hot, i love bass. I just mainly want to correct the dip at 20hz. Theres a 14db dip from 25-20hz and then it climbs back to 87db at 15hz then falls from there. I would like to flatten the FR so im thinking an EQ will be the best option.

Here's a very rough hand drawn corrected graph using paint. And the corrected number's.

10hz-76.9
15hz-97.2
20hz-85.5
25hz-97.9
30hz-95.5
35hz-97.5
40hz-90.8
45hz-82.4
50hz-83.8
55hz-86.5
60hz-88.3
65hz-91.3
70hz-95.2
75hz-94.1
80hz-90.0

http://i11.photobucket.com/albums/a170/Hotroxychick4u/TestGraphListeningPosition5hzincrem.gif

Here's a very rough hand drawn corrected graph using paint.If your receiver's auto-cal includes EQ, then I think it should have done a better job with your FR.

Otherwise, those peaks and valleys are all normal. I assume that is the FR of your front speakers (if running SMALL) and the sub. You'd want to go ahead and measure it out further, well beyond your crossover setting (if running the fronts as SMALL), so that you could see how it all looks, relatively, so that you could really define whether what you are seeing is more peaks or valleys. But what you are showing is enough to see that, yes, there are peaks and valleys. With EQ pulling down peaks is considered to be much better than trying to raise valleys. So, with EQ you might, for example, pull down some peaks so that they are closer to those valleys at 20Hz and 45Hz and then raise the sub's overall level in order to get the whole curve back up. But, again, you need some more data points out further so that you can get a better picture of what's going on.

Ok that makes sense. I only did the LFE since i just built this new sub and was curious about the response. I am going to start looking into EQ's and all the necessary software and components needed to fix this. I like my bass really loud so i would definitely have to raise the sub's level after doing this to get it back to where i want it. The test tone cd i have goes up to around 300hz in 1hz increments. So i will be able to measure throughout the whole range. Are there any good inexpensive EQ's ($100-$150) that would allow me to adjust the whole frequency range not just the low's or high's only?? I am really only interested in a sub EQ, so the EQ would not have to be full range i was just asking.

Ok that makes sense. I only did the LFE since i just built this new sub and was curious about the response. I am going to start looking into EQ's and all the necessary software and components needed to fix this. I like my bass really loud so i would definitely have to raise the sub's level after doing this to get it back to where i want it. The test tone cd i have goes up to around 300hz in 1hz increments. So i will be able to measure throughout the whole range. Are there any good inexpensive EQ's ($100-$150) that would allow me to adjust the whole frequency range not just the low's or high's only?? I am really only interested in a sub EQ, so the EQ would not have to be full range i was just asking.Try this: http://www.edesignaudio.com/product_info.php?cPath=2_123&products_id=657

Is that eQ.2 better than the Behringer DSP1124P from PE???? They are the same price at $100 just curious which one is more user friendly.

Kal, is that something you would couple with Audyssey, and if so, what would the procedure be? EQ first and then run Audyssey, or vice versa?

Kal, is that something you would couple with Audyssey, and if so, what would the procedure be? EQ first and then run Audyssey, or vice versa?With a single sub, I would not couple any other EQ with Audyssey.

I have two, but running from the a Y'd output to two amps. Just wondering if this type of EQ could be beneficial above that provided by Audyssey, which I feel has done a nice job with the contouring of the sound...just can't see what it did due to Denon not showing this information.

I have two, but running from the a Y'd output to two amps. Just wondering if this type of EQ could be beneficial above that provided by Audyssey, which I feel has done a nice job with the contouring of the sound...just can't see what it did due to Denon not showing this information.The only item one might wish to add on to Audyssey for multiple subs would be the new Audyssey-based SVS EQ which is made to handle two subs.

Thanks Kal! Not sure I would want to invest in that component, but it does sound interesting!

My receiver is not equipped with audyssey. Just an auto cal feature that sets the distance of the speakers for you. Im not even sure if it adjust any levels for you as i have always used an SPL meter and tuned it manually to 75db. My center channel is directly on top of my sub so there is no distance difference. But the auto cal set up has the sub set almost 3 feet further away. Is that normal and could that be the cause of the big drops in the FR???

My receiver is not equipped with audyssey. Just an auto cal feature that sets the distance of the speakers for you. Im not even sure if it adjust any levels for you as i have always used an SPL meter and tuned it manually to 75db. My center channel is directly on top of my sub so there is no distance difference. But the auto cal set up has the sub set almost 3 feet further away. Is that normal and could that be the cause of the big drops in the FR???Possibly due to the time delay in the active circuitry of the sub. Not uncommon.

Here's a very rough hand drawn corrected graph using paint. And the corrected number's.

10hz-76.9
15hz-97.2
20hz-85.5
25hz-97.9
30hz-95.5
35hz-97.5
40hz-90.8
45hz-82.4
50hz-83.8
55hz-86.5
60hz-88.3
65hz-91.3
70hz-95.2
75hz-94.1
80hz-90.0

http://i11.photobucket.com/albums/a170/Hotroxychick4u/TestGraphListeningPosition5hzincrem.gif


Not sure what your dip in the 20Hz range is but I think that dip in the 40Hz range is a null (a popular room null actually) and you can not do anything about that with EQing. You need to move your sub around or add more subs.

Sometimes measurements will drive you crazy instead, nulls are not exactly audible in the low frequencies meaning you can fix them but the flatter FR does not sound different. I measure lots of things but at the end of the day all that matters is how it sounds to you and as long as you know the differences you can pick whatever sound you like.

btw, I have all the different Sub EQing products. eQ.2 is much easier then the 1124p, connections are easier, settings are easier and it has a SSF dial. The only drawback to the eQ.2 is that has only two EQ bands to fix problems....most times that is okay though.

Kal, is that something you would couple with Audyssey, and if so, what would the procedure be? EQ first and then run Audyssey, or vice versa?

Run Audyssey first then manually EQ your subs with those devices if you want something different. I have Audyssey but I also use the DCX2496 for specific sub integration (I have 4 subs).

Not sure what your dip in the 20Hz range is but I think that dip in the 40Hz range is a null (a popular room null actually) and you can not do anything about that with EQing. You need to move your sub around or add more subs.

Sometimes measurements will drive you crazy instead, nulls are not exactly audible in the low frequencies meaning you can fix them but the flatter FR does not sound different. I measure lots of things but at the end of the day all that matters is how it sounds to you and as long as you know the differences you can pick whatever sound you like.

btw, I have all the different Sub EQing products. eQ.2 is much easier then the 1124p, connections are easier, settings are easier and it has a SSF dial. The only drawback to the eQ.2 is that has only two EQ bands to fix problems....most times that is okay though.


I cannot move my sub around since its integrated into my EC see pic below. Would remeasuring it with the SPL meter closer to the sub reveal if that dip in the FR is a room null. Since it was initially measured at my listening position 14feet away from the sub?? I would hate to spend the $150 on the EQ and soundcard if that will not fix it, when i could use that towards another sub.
http://i11.photobucket.com/albums/a170/Hotroxychick4u/Speakerplacment044.jpg

I already know the build well , We all bought the NHTs drivers in bulk ;)

I would build another sealed sub to handle mid range....30Hz or so to 100Hz....it will improve your overall bass response in room better then any EQ will.

Here's a very rough hand drawn corrected graph using paint. And the corrected number's.

That looks about right for a room with no bass traps. But I don't think "correction" curves are useful for many reasons. The Radio Shack SPL meter is accurate enough between 30 Hz and 1 KHz:

Comparison of Ten Measuring Microphones (http://www.realtraps.com/art_microphones.htm)

You really should be looking at bass traps to improve this rather than EQ. Bass traps raise nulls and lower peaks, and also reduce ringing. EQ can only reduce peaks.

--Ethan

That looks about right for a room with no bass traps. But I don't think "correction" curves are useful for many reasons. The Radio Shack SPL meter is accurate enough between 30 Hz and 1 KHz:

Comparison of Ten Measuring Microphones (http://www.realtraps.com/art_microphones.htm)

You really should be looking at bass traps to improve this rather than EQ. Bass traps raise nulls and lower peaks, and also reduce ringing. EQ can only reduce peaks.

--Ethan


For these frequencies he is going to need very,very thick bass trap, no?? Plus, I never read anywhere that they can actually fix room nulls. All advice previously has pointed me to the fact that only more subs and sub placement can fix a null.

More subs would obviously be the best, but explaining to the wife why im building 2 more subs would be very hard if not impossible. Would using and EQ to drop the peaks and then raising the subs output slightly to compensate fix it somewhat? I dont know much about bass traps but the way my room is setup i dont really have any corners near by. I have one about 6' away but all the other are 25'-35' away. Dont bass traps go in the corners? Or can you use acoustic wall panels as bass traps?

Would using and EQ to drop the peaks and then raising the subs output slightly to compensate fix it somewhat?

sorry, EQing can not fix nulls.

I do not remember from the other thread but are you happy with its sound?

Oh yea i am happy with the sound its very impressive for the most part. Looks like i might be stuck with the nulls then unless there is another solution rather than more subs.

I never read anywhere that they can actually fix room nulls.

Okay, you can read it here: Bass traps improve nulls. :D

You can see hard proof below that bass traps not only raise nulls, but lower peaks and reduce ringing. Ringing is shown front to back, with the "mountains" coming forward over time.

--Ethan

http://www.ethanwiner.com/misc-content/lab-ringing-both.gif

It may be common sense or just plain old fact. But im still not understanding why an EQ cannot help the FR at my listening position. Could i not lower the peaks and then boost the valleys to somewhat flatten it out?? I am new to this and just thought that is what EQ's do. Im just trying to get the best sound i can with the only placement i have without adding a second and third sub at the moment. Can someone explain to me why an EQ couldnt do this? I know penngray you sort of explained it but i can hear the null around 40hz. Its like the sub cutsout around there. And as far as around 20HZ everything is shaking so violently its not as noticeable as the other.

im still not understanding why an EQ cannot help the FR at my listening position. Could i not lower the peaks and then boost the valleys to somewhat flatten it out??

Sure, but:

1) The correction will be valid for only the one cubic inch where the microphone was when you set the EQ. An inch or two away the response will likely be made worse than it was.

2) Boosting nulls eats up amplifier power very quickly. Adding 3 dB of boost requires twice as much power, and 12 dB of boost requires four times more power. Many nulls are 25 to 30 dB deep or even more!

Much more here:

Audyssey Report (http://www.realtraps.com/art_audyssey.htm)

--Ethan

Okay, you can read it here: Bass traps improve nulls. :D

You can see hard proof below that bass traps not only raise nulls, but lower peaks and reduce ringing. Ringing is shown front to back, with the "mountains" coming forward over time.

--Ethan

http://www.ethanwiner.com/misc-content/lab-ringing-both.gif

Thanks, I guess I should qualify my "fixing nulls". I do not consider that "raised null" a good fix but its a start.

btw, I already knew about that link, I spent days and days on your site while I built my HT room, your help in the Acoustical treatements threads was outstanding (along with Terry and others).

Oh yea i am happy with the sound its very impressive for the most part. Looks like i might be stuck with the nulls then unless there is another solution rather than more subs.

Sometimes we need to back away from the measurements and actually enjoy what we have built. You will drive yourself nuts search for the best FR plots even more so when you do not use treatements.

A treated room is incredible, hopefully someday you will experience what Im taking about.

You really should be looking at bass traps to improve this rather than EQ. Bass traps raise nulls and lower peaks, and also reduce ringing. EQ can only reduce peaks. Does not the Meridian system's EQ reduce ringing?

Does not the Meridian system's EQ reduce ringing?

Anti-mode 8033 also has been said to remove ringing.

I guess I should qualify my "fixing nulls". I do not consider that "raised null" a good fix but its a start.

Nulls are typically 20 to 30 dB deep or even more. Adding a dozen bass traps in a normal size room often raises those nulls by 15 dB. This is a huge improvement! It's impossible to make any domestic size room perfectly flat unless you line literally every surface with bass traps. But even a modest number of bass traps can take you from "This is terrible" to "Wow, that's great!" And that's the real point. Versus EQ which cannot help nulls much if at all.

--Ethan

Does not the Meridian system's EQ reduce ringing?

I've never tested that system, but basic physics tells us that EQ will not reduce ringing the way bass traps can. In theory it's possible to tweak an EQ to reduce ringing for a single very small point in the room. By "very small point" I mean literally 1 cubic centimeter. Certainly not for both of your ears at the same time, and definitely not for an area spanning several people as the makers of these devices claim.

--Ethan

Sometimes we need to back away from the measurements and actually enjoy what we have built. You will drive yourself nuts search for the best FR plots even more so when you do not use treatements.

A treated room is incredible, hopefully someday you will experience what Im taking about.

Words of wisdom. Yea i am pretty much SOL on room treatments this go around this being a family room and all. My next room will be a dedicated theater area with the proper treatments. This is my very first time even having anything better than a $200 Panasonic HTIB. So i am definitely enjoying it, possibly too much the wife says. But i am also addicted, i love this hobby and dont want to leave anything well enough alone. So thats why i was wanting to purchase an EQ and have something else to mess with. Aside from building more sub's which at the moment would not be a good thing. The sub sounds great it doesn't disappoint by any means so i might waste my money on an EQ for a few hours of fun. But i realize it is not a necessity and probably will not help flatten the FR.

I've never tested that system, but basic physics tells us that EQ will not reduce ringing the way bass traps can. In theory it's possible to tweak an EQ to reduce ringing for a single very small point in the room. By "very small point" I mean literally 1 cubic centimeter. Certainly not for both of your ears at the same time, and definitely not for an area spanning several people as the makers of these devices claim. I am not saying EQ reduces ringing the same as physical treatments. You implied EQ did not reduce ringing at all. But it apparently does, over a large area. http://www.meridian-audio.com/w_paper/Room_Correction_scr.pdf

I am not saying EQ reduces ringing the same as physical treatments. You implied EQ did not reduce ringing at all. But it apparently does, over a large area. http://www.meridian-audio.com/w_paper/Room_Correction_scr.pdfFWIW, this is what Dr. Earl Geddes said on the issue--


"No multiple subs does not change the modal ringing, no matter how many you have the decay rate stays the same. But EQ can't change the decay rate either, it just brings down the excitation level. Only damping can actually change the decay rate, everything else just changes the excitation level."

FWIW, this is what Dr. Earl Geddes said on the issue--


"No multiple subs does not change the modal ringing, no matter how many you have the decay rate stays the same. But EQ can't change the decay rate either, it just brings down the excitation level. Only damping can actually change the decay rate, everything else just changes the excitation level." It is certainly true that signal processing does not change the physical properties of the room, so the decay characteristics (or the frequency response aberrations) must remain unchanged. However, EQ can alter the spectral response we hear, and certain notch filters can alter the modal resonance we hear (or measure).

>>At the onset of a transient event, the filter’s output starts at full scale and falls to the steady-state equalized level; this has the useful effect of causing the room contribution to build more quickly. When the signal stops (offset) the filter produces an output that can cancel the energy stored in the mode resonance.

Furthermore, Figure 9 illustrates that although the reverberation time of a mode can be significantly reduced by this equalisation technique, the envelope of the signal coming from the loudspeaker is significantly changed from that of the original input. In a sense we are able to reduce the effect of room modal resonance by creating a deliberate distortion in the signal coming from the loudspeakers.<<

[The Loudspeaker–Room Interface – Controlling Excitation of Room Modes, Rhonda J Wilson, Michael D Capp, and J Robert Stuart, Meridian Audio Ltd]

Roger,

The room acts as a filter between the speaker and the listener, and via signal processing, one can construct an inverse filter which undoes the effects of the room filter. Such an inverse filter can reduce or even eliminate ringing, both in theory and in practice, but there are a number of caveats.

1. It only works for one listening spot. And the size of the spot is not one cubic inch, but is a function of the wavelengths involved. For subwoofer frequency modal resonances, this spot can extend over at least a full seat.

2. The simplest yet still flexible EQ filters are a set of parametric equalizer filters. This is a "minimum phase"* system, which can only completely undo both the frequency and time effects of another minimum phase system. There has been much debate over whether a room is or is not a minimum phase system. Generally, it is not. I have measured this myself, and the less damped the room, the more non-minimum phase stuff it is likely to have. But often, isolated room resonances are minimum phase, and can be cancelled in both the frequency and time domains.

3. Building a non-minimum phase equalizer poses some problems. Probably its most serious potential audio issue is that of "pre-echo" -- hearing a bit of sound artifact before it is actually supposed to happen! If the filter is too inaccurate, pre-echo can happen. If the listening position is off, pre-echo can happen. For subwoofer frequencies, the time scale of pre-echo and our perception of it is not necessarily problematic. But for the full range of audio frequencies, it is. I don't think ringing cancelation is practical for anything but subwoofer frequencies. And there is to my knowledge no existing room correction system which even attempts to do it at higher frequencies.

Regards,
Terry

* Without going into a lot of technical explanation, "minimum phase" means that for any given desired frequency shaping, the time delay through the filter at every frequency will be at its absolute minimum possible value.

The room acts as a filter between the speaker and the listener, and via signal processing, one can construct an inverse filter which undoes the effects of the room filter. Such an inverse filter can reduce or even eliminate ringing, both in theory and in practice, but there are a number of caveats.

1. It only works for one listening spot. And the size of the spot is not one cubic inch, but is a function of the wavelengths involved. For subwoofer frequency modal resonances, this spot can extend over at least a full seat. Hi Terry,
Thanks much for your post. Unless I'm missing something, you appear to be in general agreement with the Meridian position in several respects. Is that right?

I think we must be careful about the term "inverse filter" in this discussion. Meridian is explicitly avoiding that condition in their system, as that would tend to reduce the resonance to zero. Their goal is only to reduce the decay rate to that of the room's overall RT60. This distinction may prove to be important wrt your point about the effective area of coverage for an inverse filter being small. Do you agree with their data that shows that while the magnitude of the modal peaks varies quite a bit at different seating positions, the decay rates are consistent?

You said >>isolated room resonances are minimum phase, and can be cancelled in both the frequency and time domains<< and I have have no reason to doubt this. If so, and if a room's resonance is effective in disturbing the sound quality at multiple seating locations to varying degrees, as excited by the loudspeakers, and the loudspeaker outputs are modified by a notch filter that essentially "de-rings" the room's resonance to a partial degree, why would not that modification in the room's energy at that frequency not apply to every seat where the resonance was audible?

ETA: This interesting paper (http://www.davidgriesinger.com/asa05.pdf)by David Greisinger was just brought to my attention by another Forum member, which speaks to this same subject. See pages 18-20 in particular. Thanks, Sanjay!

Hi Terry,
Thanks much for your post. Unless I'm missing something, you appear to be in general agreement with the Meridian position in several respects. Is that right?

I didn't know what the "Meridian position" was, so I Googled "Meridian room correction" and got this:
http://www.meridian-audio.cz/room-correction.pdf

Is this what you are talking about? After reading it, I guess I would have to say "kinda yes."

I think we must be careful about the term "inverse filter" in this discussion. Meridian is explicitly avoiding that condition in their system, as that would tend to reduce the resonance to zero.

Yes, I see this point about the "inverse filter" terminology.

Their goal is only to reduce the decay rate to that of the room's overall RT60. This distinction may prove to be important wrt your point about the effective area of coverage for an inverse filter being small. Do you agree with their data that shows that while the magnitude of the modal peaks varies quite a bit at different seating positions, the decay rates are consistent?

The "before" and "after" waterfall plots in this paper? I don't see this very clearly, except below 100 Hz. The higher frequency slopes look much the same, though the energies of these look equalized (which of course can be done without phase correction). But it is hard to accurately judge a decay time from a waterfall plot. There are better ways to show RT60 -- an RT60 calculation, for one! :) There is an international spec for this.

You said >>isolated room resonances are minimum phase, and can be cancelled in both the frequency and time domains<< and I have have no reason to doubt this. If so, and if a room's resonance is effective in disturbing the sound quality at multiple seating locations to varying degrees, as excited by the loudspeakers, and the loudspeaker outputs are modified by a notch filter that essentially "de-rings" the room's resonance to a partial degree, why would not that modification in the room's energy at that frequency not apply to every seat where the resonance was audible?
Yes, it would, absolutely. But the problem is that one seating location may have resonances which were not taken into consideration or not set at the right levels compared to those for another seating location. One could in theory treat the modes for all seating positions, provided you had all their measurements and were not stuck with minimum phase filters. But you better nail 'em exactly, and get every one. This is not easy. They get very dense and overlapping with increasing frequency. I don't think the filter notches shown in the "Graphical editing of filter characteristics" display shown in the paper are anywhere near enough.

Regards,
Terry

You implied EQ did not reduce ringing at all. But it apparently does, over a large area. http://www.meridian-audio.com/w_paper/Room_Correction_scr.pdf

No, it really doesn't. See my Audyssey Report (http://www.realtraps.com/art_audyssey.htm), which debunks similar claims by a vendor of a similar system. Then let's look closer at the graphs on page 11 of the Meridian brochure.

I see a few peaks that look to have less ringing in the After graph, but if you look at the slopes you can see they are the same - meaning the ringing is the same - but they fall through the "floor" of the graph faster only because the peaks are lower in volume! Also, what we really need to see as proof of reduced ringing over a usably large area is waterfalls measured 6 inches away, 12 inches away, and so forth. I assume this pair of waterfalls is at the same location.

For comparison, the graphs below show the ringing reduction from adding a bunch of bass traps. No contest. Not even close.

--Ethan

http://www.realtraps.com/eq-traps/r+d_fc_empty.gif

http://www.realtraps.com/eq-traps/r+d_fc_17_traps.gif

Ethan and I agree on something. Alert the media! :D

the size of the spot is not one cubic inch, but is a function of the wavelengths involved. For subwoofer frequency modal resonances, this spot can extend over at least a full seat.

In an open space the wavelengths will indeed dictate the "size" of EQ correction. But in an enclosed room the size you'll find in most homes, there are so many reflection paths that the response can change dramatically at very low frequencies over a span of only a few inches. The same multiple paths also make it all but impossible to create an inverse function electronically using a parametric EQ.

The graph below shows the response in the room for two locations four inches apart. This is the same room you visited at my company's factory a few years ago. The explanatory text below that is from my article, A common-sense explanation of audiophile beliefs (http://www.ethanwiner.com/believe.html):

http://www.ethanwiner.com/believe-lf.gif

Conventional wisdom holds that the bass response in a room cannot change much over small distances because the wavelengths are very long. (A 40 Hz sound wave is more than 28 feet long.) Yet you can see in Figure 1 above that the peak at 42 Hz varies by 3 dB for these two nearby locations, and there's still a 1 dB difference even as low as 27 Hz. The reason the frequency response changes so much even at low frequencies is because many reflections, each having different time and phase delays, combine in different amounts at each point in the room. In small rooms the reflections are strong because the reflecting boundaries are all nearby, so that further increases the contribution from each reflection. Also, nulls tend to occupy a relatively narrow physical space, which is why the nulls on either side of the 92 Hz marker have very different depths. Indeed, the null at 71 Hz in one location becomes a peak at the other.

--Ethan

Ethan and I agree on something. Alert the media! :D

Uh oh, did I just post something bad above? :D

Ethan and I agree on something. Alert the media! :D
But not for long. :D I don't agree on the frequency issue. An FFT frequency response typically lumps direct, early reflected, and modal (reverberant) sound all together. Lots of cancellation can occur for the first two at low frequencies, but not the third.

I didn't know what the "Meridian position" was, so I Googled "Meridian room correction" and got this:
http://www.meridian-audio.cz/room-correction.pdf

Is this what you are talking about? After reading it, I guess I would have to say "kinda yes."
The paper I cited was listed in my post 37. I cannot link to it as it's an AES paper, available from the AES website.

The "before" and "after" waterfall plots in this paper? I don't see this very clearly, except below 100 Hz. Probably better to have a look at the Griesinger paper I referenced in post 39. It has clearer plots.

No, it really doesn't. See my Audyssey Report (http://www.realtraps.com/art_audyssey.htm), which debunks similar claims by a vendor of a similar system. Then let's look closer at the graphs on page 11 of the Meridian brochure. The failure of Audyssey to reduce ringing doesn't mean Meridian's and Lexicon's systems cannot. Check out the Greisinger paper waterfall plots (http://www.davidgriesinger.com/asa05.pdf), figs 14 vs 15--they are easiest to see the effect.

The paper I cited was listed in my post 37. I cannot link to it as it's an AES paper, available from the AES website.

Probably better to have a look at the Griesinger paper I referenced in post 39. It has clearer plots.

Thanks, Roger. I have both of these papers. Will look at them tomorrow.

Regards,
Terry

I read both the Griesinger paper and the Meridian paper, and unequivocally, the modal ringing was reduces through EQ. Neither attempted to EQ any modes above 100 Hz, and I think that's a good idea. This is where residential-sized room modes get quite dense and overlapping, and as I mentioned, it is difficult to tease out all of them with the necessary accuracy. The current state of the art simply doesn't allow this, at least not yet.

Thinking of the practical consequences, the two lowest bass octaves, the subwoofer range of 20 to 80 Hz, respond quite well to electronic ringing reduction. This is fortunate, because even a large number of ordinary broadband bass absorbers do little down here (though I have built tuned bass absorbers, generally behemoths, which do). For the upper bass, we have commercial and homemade broadband bass absorbers, which work very nicely to control ringing in this range to the desired decay rate.

Regards,
Terry

The failure of Audyssey to reduce ringing doesn't mean Meridian's and Lexicon's systems cannot.

Agreed. To Meridian's credit, as Terry pointed out, at least they don't try to apply EQ above 100 Hz.

Check out the Greisinger paper waterfall plots (http://www.davidgriesinger.com/asa05.pdf), figs 14 vs 15--they are easiest to see the effect.

Again, the waterfall plots do not show what happens elsewhere in the room, and this is the key. Before I'll be convinced that EQ can reduce ringing by a useful amount, I need to see what happens not only at the original microphone position, but also at other locations a foot or more away.

Also, just to pick a bit :D I find this naïve:

Actually there is a dropped ceiling composed of 80% absorptive tiles, with the actual ceiling height of 15’. Because of the very high vertical absorption, all modes in the vertical direction are suppressed in this room.

It seems the author believes that typical ceiling tiles can "suppress" all modes for the 15 foot vertical dimension. That would be quite a feat for ceiling tiles to absorb 100 percent at 37 and 78 Hz!

--Ethan

But not for long. :D

Actually, I'm sure you and I agree on most of this stuff. And where we disagree it may be semantics more than anything else.

An FFT frequency response typically lumps direct, early reflected, and modal (reverberant) sound all together. Lots of cancellation can occur for the first two at low frequencies, but not the third.

I understand and agree with your semantic point :D but the proof is in the graph I posted above. It doesn't really matter what "class" of reflections cause the response to change so drastically over small distances at very low frequencies. That the modal contribution is not the main culprit is irrelevant. The fact is the response does change quite a lot only 4 inches away. So any EQ applied to fix the response at one ear will surely make things worse at the other ear.

--Ethan

I understand and agree with your semantic point :D but the proof is in the graph I posted above. It doesn't really matter what "class" of reflections cause the response to change so drastically over small distances at very low frequencies. That the modal contribution is not the main culprit is irrelevant. The fact is the response does change quite a lot only 4 inches away. So any EQ applied to fix the response at one ear will surely make things worse at the other ear.

--Ethan
Nope, it's not just a semantic point. We are talking about modal ringing, right? So why do you all of a sudden bring in a measurement which includes two other components?

If you stay on topic, then we can discuss spatial variation of modal ringing reduction.

- Terry

Also, just to pick a bit :D I find this naïve:

Actually there is a dropped ceiling composed of 80% absorptive tiles, with the actual ceiling height of 15’. Because of the very high vertical absorption, all modes in the vertical direction are suppressed in this room.

It seems the author believes that typical ceiling tiles can "suppress" all modes for the 15 foot vertical dimension. That would be quite a feat for ceiling tiles to absorb 100 percent at 37 and 78 Hz!

--Ethan
I can think of many words to describe David Greisinger, but "naïve" is not one of them! The Harmon Specialty Group room has a 10 foot drop ceiling with said absorptive tiles, but 5 feet above the tiles to the actual hard ceiling. This is certainly enough space in which to suppress all vertical room modes. :eek:

OK, so much for the ad hominem criticism of David. I am bracing for your ad hominem criticism of me. :D:cool:

I am bracing for your ad hominem criticism of me. :D:cool:

Not gonna happen my friend. http://www.ethanwiner.com/misc-content/bigthumb.gif

I'm still waiting for someone - anyone! - to post a graph showing ringing being reduced by EQ at locations other than where the microphone was when the EQ was adjusted.

--Ethan

I'm still waiting for someone - anyone! - to post a graph showing ringing being reduced by EQ at locations other than where the microphone was when the EQ was adjusted.

Figures 21 through 23 in the paper that Roger cited in post #37. I know you're not going to want to spend the $20 to get it from the AES :D, but I have reservations about posting copyrighted material on the web. Don't know if just these figures constitute "fair use." If this was some old historical material, I'd be more inclined, but I don't want to be sued by the AES or have them throw me out of their club!

- Terry

This certainly turned into a nice debate. Well you guys went way over my head. I am not even qualified to read these post lol. But im going to try and steer it back to my original question. It seems that a combination of bass traps and an EQ should do some good. I obviously am not wanting to go all out on bass traps and have been researching several DIY options for the price aspect. But i have decided to go with a combo of wall hanging panels and couple cylinder bass traps. I have read a few write-ups on placement options but am still somewhat confused on functional placement options. I am thinking of hanging 2 large panels on either side of the Entertainment Center behind the mains. A cylinder trap in the left corner and then two cylinder traps mid wall on both side walls with my surround speakers resting on top of these. So they would replace my current speaker stands. See the photos below, i am not working with a perfectly square area and not many corners anywhere near the LP.
http://i11.photobucket.com/albums/a170/Hotroxychick4u/100_1057.jpg
http://i11.photobucket.com/albums/a170/Hotroxychick4u/100_1053.jpg
http://i11.photobucket.com/albums/a170/Hotroxychick4u/Speakerplacment051.jpg
http://i11.photobucket.com/albums/a170/Hotroxychick4u/Speakerplacment052.jpg
http://i11.photobucket.com/albums/a170/Hotroxychick4u/Speakerplacment050.jpg

Figures 21 through 23 in the paper that Roger cited in post #37. I know you're not going to want to spend the $20 to get it from the AES :D

LOL, the last time this came up was about the audibility of nulls, and I spent the $20 only to be proven (to my satisfaction) that I was right all along. I recall we exchanged a few emails about that.

In case I never sent it to you or posted it here, this article was the result of that $20 investment:

Audibility of Narrow-Band EQ (http://www.realtraps.com/art_narrow.htm)

--Ethan

Edit: I see that I have that AES paper, so I'll read it again and report back later.

Figures 21 through 23 in the paper that Roger cited in post #37.

Terry, I read that article twice now and I'm still not sure what exactly they are showing. As best I can tell, they did not apply EQ based on a measurement at the prime seat, then measure the decay times elsewhere. As I read it, they instead did simulations.

As far as I can tell, all of the "proof of concept" graphs through Figure 11 are not real room measurements. What room has RT60 times of exactly 1.4 seconds at all of the first four modes?

Likewise, the following text from page 12 does not seem to describe a real-world test, but was another simulation with filters applied to recorded signals after the fact. Versus having the filters inserted in the playback path and each location measurement shown as it was captured:

To check that the method used is spatially robust the measurements used for Figure 6 were also filtered. The pre-filter was calculated from the preferred seating position, position 1 in Figure 4, but applied to measurements taken at all four positions. Decay curves with and without pre-filtering at 32Hz, 60Hz and 92Hz are shown in Figure 21, 22 and 23 respectively. One can see a general improvement in the control of the decay time at all four positions.

The whole thing is written in such a dense style that a lot of what they are saying is not clear. At least to this feeble mind. :D

Anything you can clarify about my above points is most welcome.

--Ethan

Terry, I read that article twice now and I'm still not sure what exactly they are showing.
Yea, Ethan read a scientific journal paper!:D

As best I can tell, they did not apply EQ based on a measurement at the prime seat, then measure the decay times elsewhere.

They claim to have EQed for just position 1 out of the four microphone positions widely scattered in the room, as shown in figure 4. The dominant room modes were first identified using analysis from all four microphones.

As I read it, they instead did simulations.

It seems that the only simulations were done in section 4, to see the effect of their EQ notch filters on synthetic tone bursts.

As far as I can tell, all of the "proof of concept" graphs through Figure 11 are not real room measurements. What room has RT60 times of exactly 1.4 seconds at all of the first four modes?

The authors claim that these are real room measurements, from their 5.5m x 5.2m x 2.4m room.

Likewise, the following text from page 12 does not seem to describe a real-world test, but was another simulation with filters applied to recorded signals after the fact. Versus having the filters inserted in the playback path and each location measurement shown as it was captured:


To check that the method used is spatially robust the measurements used for Figure 6 were also filtered. The pre-filter was calculated from the preferred seating position, position 1 in Figure 4, but applied to measurements taken at all four positions. Decay curves with and without pre-filtering at 32Hz, 60Hz and 92Hz are shown in Figure 21, 22 and 23 respectively. One can see a general improvement in the control of the decay time at all four positions.


Ah, I see the issue you are raising. It didn't register at first. They simply convolved their filter with the real room measurements they made from all four listening positions.

For any linear system (and a room's acoustical response is fully linear up to about 130 dB), the results from such a convolution will be identical to what you would measure if you did the filtering "live." This is not a simulation, but a mathematical convenience.

It is well known to anybody doing nuts and bolts filter processing, so they didn't bother to explain it. Perhaps Roger or someone else "skilled in the Art," as the patent office says, can back me up here. :)

Anything you can clarify about my above points is most welcome.

--Ethan
Hope the above helps.

- Terry

Terry, I read that article twice now and I'm still not sure what exactly they are showing. As best I can tell, they did not apply EQ based on a measurement at the prime seat, then measure the decay times elsewhere. But that's in fact exactly what they did. More below.

As far as I can tell, all of the "proof of concept" graphs through Figure 11 are not real room measurements. What room has RT60 times of exactly 1.4 seconds at all of the first four modes? Correct, several of the figures are simulations (however, Figs 5 and 6 are actual measurements). They are there to help illustrate the problems:
>>The following set of figures demonstrates some of these problems by showing the response of a resonance, or a sum of resonances, to a tone burst. For simulations using a single resonance the response of the resonance was scaled to give no gain at the centre frequency.<<

Likewise, the following text from page 12 does not seem to describe a real-world test, but was another simulation with filters applied to recorded signals after the fact. Versus having the filters inserted in the playback path and each location measurement shown as it was captured:I can see how you can read it that way. I think the sentence that is most troubling is this one: >>The pre-filter was calculated from the preferred seating position, position 1 in Figure 4, but applied to measurements taken at all four positions.<< If they were actually simulating the result, they would have said >>The pre-filter was calculated from the preferred seating position, position 1 in Figure 4, but applied to the measurements already taken at all four positions.<<

What they are saying is they created a prefilter based on the measurement made in the prime seat, and rather than making new/different prefilters for each of the other seats, they used the same prefilter for the other seats/measurements. The wording could have been better (those Brits!:D) but it's exactly the data you are looking for.

I can see how you can read it that way. I think the sentence that is most troubling is this one: >>The pre-filter was calculated from the preferred seating position, position 1 in Figure 4, but applied to measurements taken at all four positions.<< If they were actually simulating the result, they would have said >>The pre-filter was calculated from the preferred seating position, position 1 in Figure 4, but applied to the measurements already taken at all four positions.<<

Interesting, Roger! We interpreted this text differently. I chose the second meaning, which would have been more precisely stated if they used words as you gave in the second case, of course.

But even if they did what I originally thought, the experimental results would have been the same. If you have impulse response A for the pre-filter and impulse response B for speaker + room (specific to desired mic position, of course), it doesn't matter if you convolve A and B together in the computer, or allow system B to convolve impulse response A "directly" via live playback.

- Terry

Well, I'm glad I'm not the only one who found the text less than clear. :D

Me, I won't believe it until I see it in person. :eek:

--Ethan

Me, I won't believe it until I see it in person. :eek:

Why did I somehow suspect this conclusion?
:D

Why did I somehow suspect this conclusion?
:D

As we all know, science progresses when someone has a theory and does a test, then others duplicate the test and get the same results. There are a huge number of people with a direct commercial interest in EQ being able to reduce ringing. Yet to my knowledge, not one vendor of "room correction" products has ever shown proof that their products reduce ringing, even though all of them make that claim.

A few years ago I emailed Audyssey asking (very nicely!) for any graph and data etc they had to support that claim. I also asked for clarification on THIS (http://www.audyssey.com/graphs/graph1.html) page of their web site - room size, microphone position, and so forth. I didn't even get the courtesy of a confirmation they received my email. And their site still has nothing to back up their claims. Indeed, when I tested (http://www.realtraps.com/art_audyssey.htm) the Audyssey system it clearly did not reduce ringing. In fact, at some frequencies the ringing got worse!

It makes no sense that EQ can reduce ringing for a usefully large area. When I discussed this with my math and DSP expert friend Bill Eppler (who you met at RealTraps) he confirmed that with all such DSP code, the more correction that is applied, the smaller the improved area becomes. So at this point I don't feel the burden of proof has been met, especially since the test and methods in that AES article are so ambiguously described.

I would be thrilled if a small unobtrusive device could really improve room acoustics at bass frequencies! This is the holy grail for acoustic treatment makers, and if EQ really could reduce ringing I'd be all over it immediately and have my own version for sale.

--Ethan

It makes no sense that EQ can reduce ringing for a usefully large area. When I discussed this with my math and DSP expert friend Bill Eppler (who you met at RealTraps) he confirmed that with all such DSP code, the more correction that is applied, the smaller the improved area becomes.

Give Bill the paper to read.

So at this point I don't feel the burden of proof has been met, especially since the test and methods in that AES article are so ambiguously described.

Both possible interpretations to the ambiguity -- that of doing the actual room convolution vs. doing the room convolution in software -- must yield the same result. This is basic linear filtering theory.

- Terry

It makes no sense that EQ can reduce ringing for a usefully large area.

You might also ask Bill to read the paper "Analysis of Traditional and Reverberation-Reducing Methods of Room Equalization" by Louis D. Fielder, J. Audio Eng. Soc., Vol. 51, No. 1/2, 2003. It analyzes the spatial tolerances of dereverberation via EQ. I wouldn't advise you to read it yourself, as it has a lot more math than the Wilson et al. paper.

The author is not optimistic for EQ room reverberation correction, but he is interested in the entire audible range. However, there is a clear relation with size of correctible area and frequency. He defines an "error rejection" which quantifies this EQ error. Figure 16 in this paper gives the relation between error rejection and frequency. It shows a zero error at approximately 1/7 of the wavelength of the sound being corrected.

So for the range of 20 to 80 Hz, the 1/7 wavelength starts at 8 feet and drops down to 2 feet. This is the spatial range over which we can expect accurate room correction. Less accurate correction can of course extend over a larger area.

- Terry

You might also ask Bill to read the paper "Analysis of Traditional and Reverberation-Reducing Methods of Room Equalization" by Louis D. Fielder, J. Audio Eng. Soc., Vol. 51, No. 1/2, 2003.

Okay, Bill is an AES member and I know he gets all the literature.

As for the sort of physical (not theoretical) proof I expect, I have a great idea: Why don't you and I get together and test this ourselves? You could EQ the speakers in a smallish room, such as the "lab" room at RealTraps, and then we'll measure the reduction in ringing at various locations. Oh wait, we already did that (http://www.realtraps.com/eq-traps.htm). :D

It shows a zero error at approximately 1/7 of the wavelength of the sound being corrected.

This takes us back to my Post #43 in this thread where I explained:

In an open space the wavelengths will indeed dictate the "size" of EQ correction. But in an enclosed room the size you'll find in most homes, there are so many reflection paths that the response can change dramatically at very low frequencies over a span of only a few inches. The same multiple paths also make it all but impossible to create an inverse function electronically using a parametric EQ.

I believe this also relates to "minimum phase" which will not hold when there are so many reflection paths. This is why I'm so adamant that I will accept only hard proof in the form of waterfall plots, such as those I used to debunk (http://www.realtraps.com/art_audyssey.htm) Audyssey's claims.

--Ethan

As for the sort of physical (not theoretical) proof I expect, I have a great idea: Why don't you and I get together and test this ourselves? You could EQ the speakers in a smallish room, such as the "lab" room at RealTraps, and then we'll measure the reduction in ringing at various locations. Oh wait, we already did that. :D

Yup, and when I said I wanted to restrict it to the subwoofer range where I would normally confine EQ and you said "no way," I caved. Big mistake! :D

I know, but even below 100 Hz the ringing was not reduced, only the level. The three waterfalls in the section Added March 8, 2006 show this pretty clearly. But I'll agree with you in advance that when setting an EQ manually, it's probably not possible to match what a computer algorithm could do. None the less, the theory of "minimum phase" still goes out the window due to all the extra reflections in every smallish room. With four walls and a ceiling you have three poles, and that's just for the (0) modes.

--Ethan

None the less, the theory of "minimum phase" still goes out the window due to all the extra reflections in every smallish room. With four walls and a ceiling you have three poles, and that's just for the (0) modes.

I don't know what this means. :confused:

First, I misspoke above. I meant to say three modes = six poles, because each filter band requires two poles.

A single electronic filter (the EQ) can "counter" a single acoustic filter (the room) if it matches exactly what the room is doing. In this case the EQ is set to be the reverse of the room. This is the theory by which an EQ can reduce ringing, and it's valid. But a room is not a single filter! A room is much more complex for two reasons. Assuming a plain rectangle room:

1) There are three dimensions.

2) Each dimension has multiple resonances at harmonically related intervals.

So to counter the effects of the room's filter you'd need three electronic filters, one for each dimension, plus each filter needs multiple poles to account for the harmonic series of resonances. Two poles are needed for each harmonic. So if we consider only frequencies below 100 Hz in a room, say, 25 by 19 by 8 feet, there are 8 resonant frequencies, which means 16 poles are needed. And that's just for mode frequencies, ignoring the multiple reflection paths that create additional peaks and nulls (though not more ringing frequencies).

It gets even more complicated when you consider the effect of the speaker and listener positions in the room. Most of the room's filter poles contribute something less than 1.0 to the response at any given location. As soon as you move the listening position even a little, the complex relationship changes, and with it changes the response.

This is why the "1/7th wavelength" you mentioned earlier does not hold in an enclosed room. There are many paths by which a single frequency from one speaker can arrive at any given location, and all the waves from all of those paths combine to yield a single response at any given cubic centimeter in the room. This is why Figure 1 in my "Believe" (http://www.ethanwiner.com/believe.html) article shows such a large change over only four inches even at very low frequencies. And this is why no DSP can truly counter the room's skewed response and ringing for an area large enough to encompass multiple seats. It's not even possible to fix things for both ears at the same time.

--Ethan