Jump to content

BOSSOBASS Raptor system 3


Madaeel

Recommended Posts

Wow...good info. Thanks 3ll3d00d. That does seem to be fairly complex based on what you said above.

 

However, in my case, it was fairly easy in my room, and repeatable. I have my FTW21s on one phase adjustment, and have the FV15HPs on another. Leaving them in phase, audyssey couldn't remove the peak at 27hz, and my elemental designs eQ.2s couldn't remove the peak either. So prior to running audyssey, I set the phase on both pairs of subs to tame the peak (out of phase) before audyssey. I then ran Audyssey XT to flatten it out. Viola!  B)

 

Perhaps my situation is unique...don't know, but would love someone else to try.  :)

it's certainly true that I have a tendency to overcomplicate things :D

 

I think more generally I'd say that achieving a controlled phase mismatch across the passband such that FR is maintained *without* external phase eq will be as hard as setting up multisubs in the best case & much harder, if not impossible, in the worst case.

 

Do you have a loopback capable measurement system? if so then it would be interesting to see the graphs showing the phase response. The ability to use offline calculations (e.g. trace arithmetic in REW) make this sort of thing orders of magnitude easier to implement.

Link to comment
Share on other sites

  • Replies 1.3k
  • Created
  • Last Reply

Probably one reason the theater sounds so good is that it's a big enough space that the room doesn't influence the response as much.  I have a hunch that a flatter bass response gives more kick.  Why?  It's because the strongest bass transients have energy spread widely across the spectrum.  When the response is flat (and minimum phase, technically speaking) all that bass energy arrives at once, which is a very moving experience to say the least.  On the other hand, any peaks or dips in the response also cause the energy to not all arrive at once and to be smeared out in time so the instantaneous pressure is not nearly as high.

 

This is it, right there ^^^.

 

The downside of being on forum for 13 years is that you have to repeat yourself a thousand times if you don't just cruise along with the crowd and agree with conventional wisdom. I started a thread about 8 years back that stated the obvious, specifically that all transients in nature have content to DC and that a subwoofer's transient response is dictated by the extension on the low end side of bandwidth and the low pass filter on the high end of bandwidth.

 

Examples of spectrographed scenes from DVDs (pre-BR) showed content to 1 Hz and lots of posters were surprised by that as the CW then was to believe that anything below 20 Hz was unintended noise. B)

 

In that and other similar discussions over the years, I have consistently held that primarily what we hear is frequency response. A flat response is thwarted by the infinite interaction possibilities of reflections that are constructive, destructive and everything in between as well as standing waves.

 

In reality, the first wave to hit you (your ears and your body) is that which emanates directly from the subwoofer. Every other wave is reflected and most latent is that released from a standing wave, which stores energy before it releases the wave.

 

You can't correct for that latent release with EQ. If you dial a hole in the direct-radiated wave (use EQ to "pull a peak down"), then you are still getting the latency from the standing wave with a big hole in the direct radiated wave, lessening the "in time" presentation.

 

That's why the goal in my own case has always been a flat response from 3 Hz to 100 Hz, no smoothing EQ and use of placement and/or delay only to achieve as flat a response as possible. Plus or minus 3dB seems to be a reliable standard, from hearing many variations from perfectly flat.

 

In an enclosed space, I believe that is the only correlation to tactile feel. Particle velocity varies at the source (what is being called 'near field'), by mechanically what the source is, as my graphic illustrates, but equalizes at a distance (what is being called 'far field').

Link to comment
Share on other sites

That should be plenty in that room at just under 2000cuft. You'll definitely get that wobble. I'm not sure how much that one 15 is doing with ULF in free air though. Like I said it's hard to tell when playing transients.

That's awesome. That's why I said if you didn't tell me ahead of time that ULF was in those scenes it woulda scared the crap outta me. That's the

coolest effect when there's material below 10hz. I wanted to tell you I have been looking in to short throw pj's which would allow me to do a riser. :D

 

 

 

I haven't but I always say I would never use them. Even if they felt natural I just know in my head it's artificial. I could definitely use them in my room too since I'm on a slab but I get plenty of shake even without them. I know maybe if I heard and felt them I might change my mind but right now I have no desire to. You're system is ULF capable and then some Scott...you got em?

 

I do not have Crowsons but they (and some other transducers) interest me in some ways. Thank you! Yes, I do have a fairly capable system and I have been amazed at how much energy is still transmitted through by concrete floor to the couch. Holy crap! I do agree that transducers can be artificial but when set up well enough to just accentuate the bass then I feel like there is a good sense of feeling that is sometimes missing with out them. Granted I do not have any of these in my room and never had any before. I might try them out sometime though.

 

I'm going to build a floating riser in a couple weeks and put a couple left over 18's. That will be essentially free and I'll see if I like it. :D

 

 

I would like to hear what you guys think of tactile feeling from 50-500Hz, where small room effects rob us of so much SPL/SIL/Impact.  The most tactile movie presentation I ever heard was TF2 at this theatre:

 

CanaTheater9_070909.jpg

 

An old small town Texas theater that was completely renovated and THX certified.  'Punch in the chest' was a very mild way to put it.  More like 'Mule Kick to the chest'.  I have never felt more impact save for at hearing-damage-loud large outdoor concerts.  I know the response did not dig below 25Hz, but it was impressive nonetheless.  So many HTs simply cannot do this, and it may be due to room size and the inherent cancellations therein.

 

Thoughts?

 

JSS

 

Damn right. Skull crushing midbass. That's how I likes it!

 

I do wonder if we miss that clean, tight midbass because of the size of our domestic-sized rooms. Nearfield midbass as a fix?

 

 

A lot of people seem to go the other way and try to do ULF nearfield. Ehhh.

Link to comment
Share on other sites

In reality, the first wave to hit you (your ears and your body) is that which emanates directly from the subwoofer. Every other wave is reflected and most latent is that released from a standing wave, which stores energy before it releases the wave.

 

You can't correct for that latent release with EQ. If you dial a hole in the direct-radiated wave (use EQ to "pull a peak down"), then you are still getting the latency from the standing wave with a big hole in the direct radiated wave, lessening the "in time" presentation.

 

This is not entirely correct.  The problem with this argument is that it is ignorant of the phenomenon best described as the uncertainty principle for waves.  Time and frequency are interdependent, and EQ necessarily changes responses in both domains at the same time.  A fast transient is very localized in time, and that means that the "frequency response" of that transient is only *defined* with very low resolution.  Note, that my quotes here emphasize that this short time "frequency response" ("FR") is different from the long time/steady state frequency response (FR) that we label as just FR.  This principle is essentially why transients are always broadband.  In fact, the tighter the transient, the more broad the frequency response is.  The dirac delta impulse is precisely an impulse with a flat "FR" from 0 to infinity.

 

So what does flat frequency response have to do with ensuring the energy of a transient arrives at the same time?  Well, provided the (linear) system is minimum phase and has a FR that is perfectly flat in magnitude, the system will have a perfect response in time as well.  Any distortions in the frequency response of that minimum phase system will cause distortions in the transient response.  The flatter the FR is over a wide range, the more energy from that impulse will arrive at the same time.

 

Your suggestion that using EQ to correct for a room mode peak introduces a dip in the direct sound "FR" at that frequency is nonsense, mathematically and physically.  You simply cannot define frequency response at fine enough granularity to say there is a "hole" in the direct part of the impulse at the frequency you aimed to correct.  The math and physics constrain you from doing so.  In the same way that a peak in room FR will smear an impulse in time, the EQ filter that is used to correct that peak does the same thing.  And in fact, provided both are minimum phase, the EQ filter that ideally corrects the magnitude response of the room at a particular listening location will also completely cancel out the room reflections at that location.  In the same way that you describe the sound from the sub as arriving first directly and then reflected from other parts of the room, you can describe the EQ filter as allowing the impulse to pass through initially unchanged but then having additional ringing that follows.  Under the idealized condition I describe above this ringing perfectly cancels the ringing contributed by the room.

 

A key point here is that most (but not all, see below) EQ methods are minimum phase and can only cancel in-room responses that are also minimum phase.  What does minimum phase mean?  This concept is rather hard to understand and explain conceptually.  Essentially, *the* unique minimum phase response for a given *magnitude* frequency response (i.e., ignoring the phase component) is the response that minimizes the delay in the arrival of energy that follows the initial peak.  In audio, the term "minimum phase" almost always connotes a causal response as well, where a causal response in one in which all the energy follows the initial peak in time.  As long as the reflections that follow the initial impulse are low enough in energy and spread out enough in time, the response will be minimum phase.  Note that sound from a sub placed in a corner reflects almost immediately from the adjacent walls, but the distance involved is small enough relative to the wave-lengths that the sub produces that the reflections essentially combine with the direct sound so that they are indistinguishable from a distance.  Even a very narrow Q room mode with a 20 dB peak can be minimum phase.  What's actually happening there is that energy is arriving in time at periodic intervals and each arrival is only slightly diminished compared to the previous arrival.  The period of arrivals is the same as the period of the frequency in question, and for a sustained tone at that frequency all these arrivals constructively interfere to create a huge peak.  Being minimum phase, a minimum phase EQ filter with a dip at the same frequency and the same Q will completely cancel this out.

 

In my room, each of my four subs is minimum phase in each of my listening locations, so I can readily use minimum phase EQ filters on each sub to achieve a very flat minimum phase response across a wide listening area.  Not all combinations of sub and listener placement will yield a minimum phase response.  In my case, some sub locations *did not* yield a minimum phase response.

 

Where room response is not minimum phase, a minimum phase EQ cannot be used to correct the problem.  This rules out essentially all forms of analog signal processing.  Correction of non-minimum phase (aka mixed phase) response also requires mixed phase correction filters.  This can only be practically realized with a digital system that can store the signal in memory for some length of time in order to be combined with a later part of the signal to produce the output.  This process necessarily introduces a time delay, which is another practical constraint.  An ideal mixed-phase correction for a non-minimum phase room problem requires infinite delay (and infinite CPU power too), but frequently a good compromise can be reached that mostly corrects the problem while keeping the overall delay and CPU requirement reasonable.  Naturally there are situations where the required correction just demands too much headroom from the sub system to be achievable, and this can occur with both minimum phase and mixed phase room responses.

 

Another thing important to mention here is that EQ on a single sub cannot change the variation of response between seats.  To the extent that the variation is limited as is often the case with room modes, a single sub EQ can still help a lot.  With multiple subs running independent EQs, it's also possible to alter the variation of response between seats as I have done with my 4 subs to achieve a very tight FR (and temporal response) over a wide listening area.

 

Let me know if you're confused about something.  This is a very difficult subject to understand intuitively.  I've devoted many hours of thought experiments to develop an intuitive sense of what the math says is happening here.  I use my own software to simulate the responses of my subs with a given set of MiniDSP 2x4 filters at multiple listening locations.  Then I manually optimize the biquads (includes PEQs, shelves, and crossovers) for each sub until I get the responses I'm happiest with.  I recently extended the software to automate creation of high resolution minimum phase FIR filters to also optimize my mains responses at multiple locations.  I'm in the process of implementing mixed-phase FIR filter calculations, but this is taking me a long time because of the difficulty involved.  In all cases, my software accurately predicts what the actual responses will be at each location, both for time and frequency, except where the uncertainty is unavoidably high, i.e. in deep nulls.

 

I strongly urge you to reconsider the use of EQ.  It can make a very positive difference *if* it's done with care and proper knowledge.  For example, the importance of reserving minimum phase corrections for minimum phase problems cannot be over-emphasized.  Likewise, you cannot adequately simulate the combined response of multiple subs by merely summing the response magnitudes.  My software requires response measurements that are aligned correctly in time for its predictions to work.  For off-center seats, the peak impulse from each sub will arrive at a slightly different time, and this needs to be taken into account.

 

Edit: a few clarifications and typo fixes

Link to comment
Share on other sites

Your suggestion that using EQ to correct for a room mode peak introduces a dip in the direct sound "FR" at that frequency is nonsense, mathematically and physically. You simply cannot define frequency response at fine enough granularity to say there is a "hole" in the direct part of the impulse at the frequency you aimed to correct. The math and physics constrain you from doing so. In the same way that a peak in room FR will smear an impulse in time, the EQ filter that is used to correct that peak does the same thing. And in fact, provided both are minimum phase, the EQ filter that ideally corrects the magnitude response of the room at a particular listening location will also completely cancel out the room reflections at that location. In the same way that you describe the sound from the sub as arriving first directly and then reflected from other parts of the room, you can describe the EQ filter as allowing the impulse to pass through initially unchanged but then having additional ringing that follows. Under the idealized condition I describe above this ringing perfectly cancels the ringing contributed by the room.

 

You are ignoring your hearing here. Does the auditory system have the capability to resolve the difference between steady state and transient response at low frequency in a small room?

 

As I understand it, this is not currently known. It is certainly true that the steady state behaviour at LF can vary from the transient response hence it is not unreasonable to suppose your hearing can detect this. On the other hand, your hearing didn't evolve to hear ULF in a room so it is also reasonable to suppose your capability here is v v poor.

Link to comment
Share on other sites

Let me know if you're confused about something.

Er, most of it? :blink::lol:  lol

 

This makes my head hurt!!

 

 

Having read and attempted to get my head round that, am I correct in my understanding that, assuming a single burst of sound for the purposes of my example...

- a minimum phase response effectively means that the burst has time to finish before the room effects come in?

whereas...

- a mixed phase response means that the room effects begin before the burst has time to finish?

 

If this is correct, am I therefore correct in thinking that...:

- the minimum phase response is easier to deal with because you just need to time-align the wavefronts from all the speakers, so they arrive at the MLP at the same time, and any room effects lingering after the burst has finished can be relatively easily removed by the use of sound-/reflection-absorbing room treatments?

whereas...

- a mixed phase response is more difficult to deal with because you need to measure the room response, record its effects across the bandwidth of interest via electronic EQ operating in both the frequency- and time-domain, and then implement a 'negative'/'reverse' of the effects via EQ so it acts on the burst itself, altering the burst's waveform to cancel out room effects as they happen?  (Which is why power/displacement limitations will determine the level of mixed phase EQ that can be applied, because if you design a system to deal with just a 7.1 worst case scenario disc, for example, you are not allowing any system headroom to deal with the room effects as they happen?)

 

 

If placing a subwoofer in the corner of a room removes the 'rear / side wave bounce' you'd get (and have to deal with) if it was located several feet into the room / away from the walls, you are therefore (to a degree) ensuring you have a mainly minimum phase response to deal with, rather than a mixed phase response?

 

If so, I wonder if that's the reason Steinway Lyngdorf use a 'boundary woofer' solution in their big systems, with the subwoofers set into the front corners where possible.  (It also appears to be the Bosso-recommended solution! :D)  Corner placement of subs therefore would seem to be the best option for the majority of rooms, if I'm correct?  in that you get 'reinforced' and therefore more output in-room, as well as better phase response?

 

 

I hope that's broadly correct, if not then I'm giving up and going to bed, my brain hurts from work and now trying to understand this... lol

 

Thank you for the detailed post, either way! :)

Link to comment
Share on other sites

Probably one reason the theater sounds so good is that it's a big enough space that the room doesn't influence the response as much.  I have a hunch that a flatter bass response gives more kick.  Why?  It's because the strongest bass transients have energy spread widely across the spectrum.  When the response is flat (and minimum phase, technically speaking) all that bass energy arrives at once, which is a very moving experience to say the least.  On the other hand, any peaks or dips in the response also cause the energy to not all arrive at once and to be smeared out in time so the instantaneous pressure is not nearly as high.

Now that makes sense. That's why even with my subs close it was not any better regardless of how much PVL was there, or not since we can't measure it, since my FR was a roller coaster. Which is what I assumed.

 

I think more generally I'd say that achieving a controlled phase mismatch across the passband such that FR is maintained *without* external phase eq will be as hard as setting up multisubs in the best case & much harder, if not impossible, in the worst case.

 

So IOW just stick to nearfield or transducers. :rolleyes:

In an enclosed space, I believe that is the only correlation to tactile feel. Particle velocity varies at the source (what is being called 'near field'), by mechanically what the source is, as my graphic illustrates, but equalizes at a distance (what is being called 'far field').

Agreed and great graphic.

 

I do not have Crowsons but they (and some other transducers) interest me in some ways. Thank you! Yes, I do have a fairly capable system and I have been amazed at how much energy is still transmitted through by concrete floor to the couch. Holy crap! I do agree that transducers can be artificial but when set up well enough to just accentuate the bass then I feel like there is a good sense of feeling that is sometimes missing with out them. Granted I do not have any of these in my room and never had any before. I might try them out sometime though.

 

I'm going to build a floating riser in a couple weeks and put a couple left over 18's. That will be essentially free and I'll see if I like it. :D

 

 

 

Damn right. Skull crushing midbass. That's how I likes it!

 

I do wonder if we miss that clean, tight midbass because of the size of our domestic-sized rooms. Nearfield midbass as a fix?

 

 

A lot of people seem to go the other way and try to do ULF nearfield. Ehhh.

Man you lucky bastard! That riser should give you all the shake you're missing. "Like it" would prolly be an understatement haha. With my stupid bulkhead right above me even if I used a ST projector I don't think I could do a riser. A riser or a suspended floor is all you need for below 10hz. Every other frequency range I have no problem feeling in my room either, and in spades.

 

I think mid bass near field is much better then ULF. I felt it more but not enough to compromise the rest of my FR. My brother has the eD A7s-450 firing into his couch and though it only digs to 25hz or so it feels amazing. I'd imagine midbass modules near field would be like that Texas theater Max was at. :o

 

I like that idea...

Link to comment
Share on other sites

You are ignoring your hearing here. Does the auditory system have the capability to resolve the difference between steady state and transient response at low frequency in a small room?

 

As I understand it, this is not currently known. It is certainly true that the steady state behaviour at LF can vary from the transient response hence it is not unreasonable to suppose your hearing can detect this. On the other hand, your hearing didn't evolve to hear ULF in a room so it is also reasonable to suppose your capability here is v v poor.

 

Hearing is very important, and I'm ignoring it on purpose to avoid complicating my explanation.  Generally speaking, if you can achieve an ideally flat minimum phase frequency response, then you can basically ignore hearing.  Presumably, the mixer monitored and adjusted the mix in a room of similar quality, so you will hear what he/she hears.  Where psychoacoustics is more important is where compromise is necessary (which is always the case, in practice).  With psychoacoustics, one can weight the correction to address more audible issues at the expense of issues that may be less audible.  Of course, this too is easier said than done.  All the psychoacoustic models we have model hearing very crudely at best.  Despite the number of commercial solutions that advertise use of "psychoacoustic modeling" in their algorithms, the best sounding correction methods likely minimize the use of such models because where they are wrong, they can actually cause more harm than good.  Still, it's a good idea to adjust the target curve to sound best instead of assuming "flat is always best".  Likewise, it is a well known psychoacoustic fact that we are more sensitive to pre-ringing (ringing that happens before the impulse peak) than post-ringing, and this is important when doing mixed phase corrections whose errors can potentially contribute pre-ringing to the response.

 

As for whether the auditory system can resolve differences between transient and steady response: of course it can!  Can you notice the difference between a pulse centered at 50 Hz and a continous tone centered at 50 Hz?  Of course you can.  They are completely different sounds.  Supposing there's a high Q room resonance centered at 50 Hz too, how does that effect things?  Well, the transient won't hit quite as hard, and it will additionally have a decaying tail around 50 Hz that gives it kind of a boom sound instead of just a solid hit.  The continuous tone will sound exaggeratedly loud especially in comparison to the adjacent tones.

 

To really understand what we hear, it is most helpful to look at both steady response and responses using different time windows.  Another very useful tool is the spectrogram which can give a sort of visual "best of both worlds" picture of how the energy is distributed in time and frequency.  In the same way that the spectrogram of a soundtrack (such as that generated by a tool like SpecLab) gives us a lot of information about what's going on in it, the spectrogram of a room impulse response gives us a very good picture of how the energy from the system in the room is distributed across time and frequency.

 

Of course when using the spectrogram, you should be aware of the fact that it's actually an "over-complete" representation of what's going on.  The fact that you can generate a spectrogram with ultra-fine resolution in both time and frequency does not mean you are able to get around the uncertainty principle.  For example, one thing you should never see in a spectrogram is a strong bass hit, with all its energy concentrated between 40 and 60 Hz that appears and disappears in the span of 3 milliseconds.  Even though a full wave at 40 Hz is only 2.5 millisecond long, the effective frequency resolution for events of that duration is not enough to clearly distinguish between frequencies in the 40-60 Hz range.  Instead, that quick hit will necessarily occupy a wider band along the frequency axis.

Link to comment
Share on other sites

it's certainly true that I have a tendency to overcomplicate things :D

 

I think more generally I'd say that achieving a controlled phase mismatch across the passband such that FR is maintained *without* external phase eq will be as hard as setting up multisubs in the best case & much harder, if not impossible, in the worst case.

Agree. Without EQ, it would be very difficult.

 

Do you have a loopback capable measurement system? if so then it would be interesting to see the graphs showing the phase response. The ability to use offline calculations (e.g. trace arithmetic in REW) make this sort of thing orders of magnitude easier to implement.

 

Not sure?...I use Omnimic.

Link to comment
Share on other sites

I haven't been paying a lot of attention over the past months so I need to ask:

 

Dom, have you tried the 21" driver only near field vs far field and if so, what was the result?

 

dbaf3de078d8dab6604a2cd3c24185db.png

 

 

This is it, right there ^^^.

 

The downside of being on forum for 13 years is that you have to repeat yourself a thousand times if you don't just cruise along with the crowd and agree with conventional wisdom. I started a thread about 8 years back that stated the obvious, specifically that all transients in nature have content to DC and that a subwoofer's transient response is dictated by the extension on the low end side of bandwidth and the low pass filter on the high end of bandwidth.

 

Examples of spectrographed scenes from DVDs (pre-BR) showed content to 1 Hz and lots of posters were surprised by that as the CW then was to believe that anything below 20 Hz was unintended noise. B)

 

In that and other similar discussions over the years, I have consistently held that primarily what we hear is frequency response. A flat response is thwarted by the infinite interaction possibilities of reflections that are constructive, destructive and everything in between as well as standing waves.

 

In reality, the first wave to hit you (your ears and your body) is that which emanates directly from the subwoofer. Every other wave is reflected and most latent is that released from a standing wave, which stores energy before it releases the wave.

 

You can't correct for that latent release with EQ. If you dial a hole in the direct-radiated wave (use EQ to "pull a peak down"), then you are still getting the latency from the standing wave with a big hole in the direct radiated wave, lessening the "in time" presentation.

 

That's why the goal in my own case has always been a flat response from 3 Hz to 100 Hz, no smoothing EQ and use of placement and/or delay only to achieve as flat a response as possible. Plus or minus 3dB seems to be a reliable standard, from hearing many variations from perfectly flat.

 

In an enclosed space, I believe that is the only correlation to tactile feel. Particle velocity varies at the source (what is being called 'near field'), by mechanically what the source is, as my graphic illustrates, but equalizes at a distance (what is being called 'far field').

This makes sense to me as well, and what coolrda and Okv has also observed. Excellent time alignment also increases tactile feedback.

 

Your graphic is a good one....question is, which is most effective at PVL?  B)  Need the microflown...

Link to comment
Share on other sites

SME,

 

Do you EQ each sub independently prior to summation? I found that EQ'ing each of my sub clusters yielded no benefit compared to a global EQ solution after delays adjusyed for the near field cluster to time-align it.

 

JSS

Yes.  I EQ independently prior to summation.  I also do so with an eye to the combined response at multiple listening locations using my simulation software.  It's a bit tedious to do this manually, but it yields a better result than I could otherwise.

 

There's one thing I should add.  If your subs are placed symmetrically (like one in each front corner) in a (mostly) symmetric room, then running separate EQ for each sub may not be as beneficial.  In math and nature, symmetry often does stuff like that.  On the other hand, if you broke the symmetry by moving only one of the subs to a new location, then you *might* be able to get a better result with independent EQ.  I have no idea for sure.  You'd have to try it.

Link to comment
Share on other sites

I would like to hear what you guys think of tactile feeling from 50-500Hz, where small room effects rob us of so much SPL/SIL/Impact.  The most tactile movie presentation I ever heard was TF2 at this theatre:

 

CanaTheater9_070909.jpg

 

An old small town Texas theater that was completely renovated and THX certified.  'Punch in the chest' was a very mild way to put it.  More like 'Mule Kick to the chest'.  I have never felt more impact save for at hearing-damage-loud large outdoor concerts.  I know the response did not dig below 25Hz, but it was impressive nonetheless.  So many HTs simply cannot do this, and it may be due to room size and the inherent cancellations therein.

 

Thoughts?

 

JSS

Here are my thoughts...

 

Like you had suggested, the larger space plays a big role.  Assuming the space has ample Sound Power and has been optimized, SIL is increased for two reasons I know of:

 

  1. Time alignment - The large space allows for better time alignment of the longer waves so that the original waves impact all at the same time increasing SIL
  2. Sound Power or 'exertion' is greater relative to a smaller room - Because of the larger space, there is less overall room gain. Since there is little room gain, the drivers are working harder when compared to smaller spaces. More exertion of the drivers mean more overall PVL being produced and more overall SIL from the room's perspective. Put the same system in a smaller room and they won't have to work as hard because the room will benefit from room gain to achieve the target SPL.
  3. So while listeners in smaller rooms benefit from the proximity to the source system, increasing SIL; listeners in larger rooms benefit from more overall Sound Power or 'exertion', increasing SIL. The question is which one is better at being more tactile? Perhaps it varies by frequency in each scenario? I dunno...
Link to comment
Share on other sites

You can't correct for that latent release with EQ. If you dial a hole in the direct-radiated wave (use EQ to "pull a peak down"), then you are still getting the latency from the standing wave with a big hole in the direct radiated wave, lessening the "in time" presentation.

-This is pretty straightforward and makes sense. 

 

Your suggestion that using EQ to correct for a room mode peak introduces a dip in the direct sound "FR" at that frequency is nonsense, mathematically and physically.

-I don't get where you're coming from with this.  Pulling out a section of bandwidth in the signal with a filter reduces the voltage out of the device for whatever frequencies the filter is set for.  How is that not creating a dip in the direct sound for those frequencies affected by the filter? 

Link to comment
Share on other sites

This is not entirely correct.  The problem with this argument is that it is ignorant of the phenomenon best described as the uncertainty principle for waves.  Time and frequency are interdependent, and EQ necessarily changes responses in both domains at the same time.  A fast transient is very localized in time, and that means that the "frequency response" of that transient is only *defined* with very low resolution.  Note, that my quotes here emphasize that this short time "frequency response" ("FR") is different from the long time/steady state frequency response (FR) that we label as just FR.  This principle is essentially why transients are always broadband.  In fact, the tighter the transient, the more broad the frequency response is.  The dirac delta impulse is precisely an impulse with a flat "FR" from 0 to infinity.

 

So what does flat frequency response have to do with ensuring the energy of a transient arrives at the same time?  Well, provided the (linear) system is minimum phase and has a FR that is perfectly flat in magnitude, the system will have a perfect response in time as well.  Any distortions in the frequency response of that minimum phase system will cause distortions in the transient response.  The flatter the FR is over a wide range, the more energy from that impulse will arrive at the same time.

 

Your suggestion that using EQ to correct for a room mode peak introduces a dip in the direct sound "FR" at that frequency is nonsense, mathematically and physically.  You simply cannot define frequency response at fine enough granularity to say there is a "hole" in the direct part of the impulse at the frequency you aimed to correct.  The math and physics constrain you from doing so.  In the same way that a peak in room FR will smear an impulse in time, the EQ filter that is used to correct that peak does the same thing.  And in fact, provided both are minimum phase, the EQ filter that ideally corrects the magnitude response of the room at a particular listening location will also completely cancel out the room reflections at that location.  In the same way that you describe the sound from the sub as arriving first directly and then reflected from other parts of the room, you can describe the EQ filter as allowing the impulse to pass through initially unchanged but then having additional ringing that follows.  Under the idealized condition I describe above this ringing perfectly cancels the ringing contributed by the room.

 

A key point here is that most (but not all, see below) EQ methods are minimum phase and can only cancel in-room responses that are also minimum phase.  What does minimum phase mean?  This concept is rather hard to understand and explain conceptually.  Essentially, *the* unique minimum phase response for a given *magnitude* frequency response (i.e., ignoring the phase component) is the response that minimizes the delay in the arrival of energy that follows the initial peak.  In audio, the term "minimum phase" almost always connotes a causal response as well, where a causal response in one in which all the energy follows the initial peak in time.  As long as the reflections that follow the initial impulse are low enough in energy and spread out enough in time, the response will be minimum phase.  Note that sound from a sub placed in a corner reflects almost immediately from the adjacent walls, but the distance involved is small enough relative to the wave-lengths that the sub produces that the reflections essentially combine with the direct sound so that they are indistinguishable from a distance.  Even a very narrow Q room mode with a 20 dB peak can be minimum phase.  What's actually happening there is that energy is arriving in time at periodic intervals and each arrival is only slightly diminished compared to the previous arrival.  The period of arrivals is the same as the period of the frequency in question, and for a sustained tone at that frequency all these arrivals constructively interfere to create a huge peak.  Being minimum phase, a minimum phase EQ filter with a dip at the same frequency and the same Q will completely cancel this out.

 

In my room, each of my four subs is minimum phase in each of my listening locations, so I can readily use minimum phase EQ filters on each sub to achieve a very flat minimum phase response across a wide listening area.  Not all combinations of sub and listener placement will yield a minimum phase response.  In my case, some sub locations *did not* yield a minimum phase response.

 

Where room response is not minimum phase, a minimum phase EQ cannot be used to correct the problem.  This rules out essentially all forms of analog signal processing.  Correction of non-minimum phase (aka mixed phase) response also requires mixed phase correction filters.  This can only be practically realized with a digital system that can store the signal in memory for some length of time in order to be combined with a later part of the signal to produce the output.  This process necessarily introduces a time delay, which is another practical constraint.  An ideal mixed-phase correction for a non-minimum phase room problem requires infinite delay (and infinite CPU power too), but frequently a good compromise can be reached that mostly corrects the problem while keeping the overall delay and CPU requirement reasonable.  Naturally there are situations where the required correction just demands too much headroom from the sub system to be achievable, and this can occur with both minimum phase and mixed phase room responses.

 

Another thing important to mention here is that EQ on a single sub cannot change the variation of response between seats.  To the extent that the variation is limited as is often the case with room modes, a single sub EQ can still help a lot.  With multiple subs running independent EQs, it's also possible to alter the variation of response between seats as I have done with my 4 subs to achieve a very tight FR (and temporal response) over a wide listening area.

 

Let me know if you're confused about something.  This is a very difficult subject to understand intuitively.  I've devoted many hours of thought experiments to develop an intuitive sense of what the math says is happening here.  I use my own software to simulate the responses of my subs with a given set of MiniDSP 2x4 filters at multiple listening locations.  Then I manually optimize the biquads (includes PEQs, shelves, and crossovers) for each sub until I get the responses I'm happiest with.  I recently extended the software to automate creation of high resolution minimum phase FIR filters to also optimize my mains responses at multiple locations.  I'm in the process of implementing mixed-phase FIR filter calculations, but this is taking me a long time because of the difficulty involved.  In all cases, my software accurately predicts what the actual responses will be at each location, both for time and frequency, except where the uncertainty is unavoidably high, i.e. in deep nulls.

 

I strongly urge you to reconsider the use of EQ.  It can make a very positive difference *if* it's done with care and proper knowledge.  For example, the importance of reserving minimum phase corrections for minimum phase problems cannot be over-emphasized.  Likewise, you cannot adequately simulate the combined response of multiple subs by merely summing the response magnitudes.  My software requires response measurements that are aligned correctly in time for its predictions to work.  For off-center seats, the peak impulse from each sub will arrive at a slightly different time, and this needs to be taken into account.

 

Edit: a few clarifications and typo fixes

 

 

Everything I post is from experience unless otherwise noted, none of it being nonsense. ;)

 

You cannot EQ a standing wave out of existence. Rooms are "mixed" phase, not minimum phase. The concept of minimum phase is hard to convey because it doesn't exist in our rooms. In the case of standing waves, sub and/or seat placement beats EQ. "Spatial averaging EQ" is a wet dream in a multi-channel listening environment.

 

Maybe you can post some data? Pictures are easier for me. :)

Link to comment
Share on other sites

Having read and attempted to get my head round that, am I correct in my understanding that, assuming a single burst of sound for the purposes of my example...

- a minimum phase response effectively means that the burst has time to finish before the room effects come in?

whereas...

- a mixed phase response means that the room effects begin before the burst has time to finish?

 

No, this is not right.  My long post was meant to clarify where and how EQ can be very helpful.  It was not meant to explain why certain locations work better for subs.

 

 

If this is correct, am I therefore correct in thinking that...:

- the minimum phase response is easier to deal with because you just need to time-align the wavefronts from all the speakers, so they arrive at the MLP at the same time, and any room effects lingering after the burst has finished can be relatively easily removed by the use of sound-/reflection-absorbing room treatments?

whereas...

- a mixed phase response is more difficult to deal with because you need to measure the room response, record its effects across the bandwidth of interest via electronic EQ operating in both the frequency- and time-domain, and then implement a 'negative'/'reverse' of the effects via EQ so it acts on the burst itself, altering the burst's waveform to cancel out room effects as they happen?  (Which is why power/displacement limitations will determine the level of mixed phase EQ that can be applied, because if you design a system to deal with just a 7.1 worst case scenario disc, for example, you are not allowing any system headroom to deal with the room effects as they happen?)

 

No.  The minimum phase part of the response is easier to deal with because you can effectively cancel it out at one location (if you have enough headroom, of course) using minimum phase filters.  Many digital and almost all analog filters are minimum phase.  In audio, engineers define minimum phase to imply a causal response as well.  A causal response is one that depends only on the signal now and in the past.  Most PEQ solutions including those implemented in MiniDSP products are minimum phase.  A nice thing about using minimum phase filters to correct a minimum phase response is that you don't need to worry about correcting the phase separately.  Any improvements to the magnitude response will improve the temporal response also.  Also, the final response is also minimum phase, which essentially guarantees that you won't introduce pre-ringing.

 

Correction of mixed phase aspects of the response is more difficult because it requires filters whose output depends upon the signal in the future.  Since it's not possible to go back in time, a mixed phase filter is practically implemented on a digital device by storing input samples in memory until the "future" samples are available.  This means that the filter introduces a constant delay.

 

Power and displacement will limit EQ possibilities regardless of whether the response is minimum or mixed phase.  There's no free lunch here.  If the frequency response is 10 dB too low at a certain frequency and you want to correct that, then you'd better have 10 dB headroom in your system to do so.  Where there are deep nulls, you'll never be able to get a good correction.

 

If placing a subwoofer in the corner of a room removes the 'rear / side wave bounce' you'd get (and have to deal with) if it was located several feet into the room / away from the walls, you are therefore (to a degree) ensuring you have a mainly minimum phase response to deal with, rather than a mixed phase response?

 

Placing the subwoofer in the corner doesn't remove the rear / side wave bounce at all.  It's just that the bounce happens so soon with respect to the period of the waves that the direct sound and bounce basically combine into a single wavefront.  I believe the big advantage to corner placement is that this places the sub as far away as possible from the *other* boundaries in the room.  The farther away the other boundaries are, the more the sound can expand and therefore diminish in level before it is reflected back to the listener.  The lower the reflections are, the smoother the frequency response will be and the better chance that it will stay minimum phase and be correctable with typical PEQ solutions.  Of course, corner placements can excite room modes more severely, making the uncorrected response worse, but these problems are often easier to fix with EQ when subs are in the corner.  I know this isn't the argument in favor for corner placement that Dave would give you.  ;)

Link to comment
Share on other sites

Now that makes sense. That's why even with my subs close it was not any better regardless of how much PVL was there, or not since we can't measure it, since my FR was a roller coaster. Which is what I assumed.

The closer you get to the sub, the more the response looks like the anechoic (or ground plane) response of the sub, to a point.  Things start getting weird again when you are inches away from a large driver, but in any case, this trend mostly holds true.  The problem is that even a very close sub placement might not be *close enough* for the sub response to "win" against the room's at all frequencies.  In my room, I get better mid and upper bass response with subs placed on each side of my sofa.  However, at low frequencies, the room response overwhelms the direct sub response and makes the sound worse than (both in frequency and time) than if the subs are placed against my rather long front wall.  That's a big part of why I'm using a bi-amped configuration with pairs of subs in both locations.

Link to comment
Share on other sites

Everything I post is from experience unless otherwise noted, none of it being nonsense. ;)

 

You cannot EQ a standing wave out of existence. Rooms are "mixed" phase, not minimum phase. The concept of minimum phase is hard to convey because it doesn't exist in our rooms. In the case of standing waves, sub and/or seat placement beats EQ. "Spatial averaging EQ" is a wet dream in a multi-channel listening environment.

 

Maybe you can post some data? Pictures are easier for me. :)

 

I don't doubt your extensive experience, but this is an area I have studied fairly carefully and extensively from a theoretical viewpoint.  I also have hands-on experience applying this knowledge to my own system, and have verified the results both with measurements and subjective evaluation.  :)

 

It is true that you cannot EQ a standing wave out of existence, but a lot of times that is unnecessary.  You care about what the response is where you actually listen.  If the response exhibits a similar peak across multiple seats, then it is possible to improve the response at all those seats.  If the peak is minimum phase as is often the case, then a minimum phase EQ correction can improve both the time and frequency response at those locations.  Listeners located at those locations will not hear the adverse effects of the standing wave, even though it's still present.  Even outside of those locations, the response may be improved some overall, even if not as much.

 

It is also true that room responses are typically mixed phase.  However, most rooms exhibit minimum phase behavior for at least part of the frequency range.  In these regions, minimum phase corrections can be very useful.  If a sub is placed in a beneficial location, its response at the listening position(s) may be almost completely minimum phase and very amenable to minimum phase EQ correction.  If you haven't done so already, you should use a program like REW to measure the sub impulse responses at a few listening locations.  REW can decompose the response into the minimum phase response having the same magnitude FR and the so-called "excess phase" portion, which is all-pass in nature (i.e., flat response in frequency but not time).  This allows you to see what response features are essentially minimum phase and which are not.  You may be surprised to see that most if not all your bass response is minimum phase.

 

As you say, sub and listener placement are more important than EQ, but often one or both is constrained by the room design or for other reasons.  In some cases, listener placement may also be constrained by performance requirements for the rest of the spectrum.  Even ideal sub and listener placements are usually not enough to get a flat response.  EQ can improve the response beyond what is otherwise possible with placements alone.

 

Your statement that "Spatial averaging EQ is a wet dream" is (pardon the pun) prematurely dismissive.  Spatial averaging is indeed an unreliable automated room correction approach for several reasons.  For manual corrections, it can be useful as long as the end result is verified.  Either way, spatial averaging basically solves the wrong math problem.  There are many other multi-seat correction approaches possible that yield much better results.

 

Hmm pictures?  What kind of pictures would you like to see?  I'll start by posting REW captures of spectrograms of my LFE output at MLP using window sizes 25, 50, 100, and 200 ms.  All of these are derived from the same response.  The shorter windows provide better time resolution at the expense of frequency resolution.  In aggregate, they give a rough idea of what a LFE impulse sounds like in my best seat:

 

Edit: It helps if I show the attached files in the post.

 

post-1549-0-69794100-1428051683_thumb.png

post-1549-0-36149800-1428051686_thumb.png

post-1549-0-62105500-1428051691_thumb.png

post-1549-0-13204300-1428051695_thumb.png

 

Take note that the y-axis indicates milliseconds and the color ranges across 40 dB.  Note that almost all the energy is concentrated near the start.  The time of arrival of energy is delayed slightly as frequency decreases.  This is due to a combination of things: the 4th order Linkwitz Riley crossover at 50 Hz for bi-amped sub pairs, the port tuning, and the protective HPF on the deep bass subs.  These contributions are non-minimum phase, so even the relatively flat magnitude response I have (not shown) is not enough to fix this temporal problem, but it is relatively minor.  Either way, I hope to repair much if not most of it with my first deployment of mixed phase filters.

 

If you want to see them, maybe tomorrow I could post spectra of each sub at the MLP without EQ to show what a huge difference EQ can make.

Link to comment
Share on other sites

@Madaeel, your observations from the nearfield experiment are interesting, and matching my own experiences.

 

At very low freqs, the velocity potenial may not be that important, or it is simply that the velocity we are able to create simply is not enough.

 

Interesting discussion, and many good points made in this thread, and so well hidden I nearly missed it all.

 

I have posted on different places about particle velocity, tactile feel and bass, may be we could create a thread on this subject here on data-bass - where there is interest, knowledge and you don't need to waste energy in discussing "you can't hear below 20hz" or "what is this nonsense about bass you can feel".

Link to comment
Share on other sites

 

I don't doubt your extensive experience, but this is an area I have studied fairly carefully and extensively from a theoretical viewpoint.  I also have hands-on experience applying this knowledge to my own system, and have verified the results both with measurements and subjective evaluation.  :)

 

It is true that you cannot EQ a standing wave out of existence, but a lot of times that is unnecessary.  You care about what the response is where you actually listen.  If the response exhibits a similar peak across multiple seats, then it is possible to improve the response at all those seats.  If the peak is minimum phase as is often the case, then a minimum phase EQ correction can improve both the time and frequency response at those locations.  Listeners located at those locations will not hear the adverse effects of the standing wave, even though it's still present.  Even outside of those locations, the response may be improved some overall, even if not as much.

 

It is also true that room responses are typically mixed phase.  However, most rooms exhibit minimum phase behavior for at least part of the frequency range.  In these regions, minimum phase corrections can be very useful.  If a sub is placed in a beneficial location, its response at the listening position(s) may be almost completely minimum phase and very amenable to minimum phase EQ correction.  If you haven't done so already, you should use a program like REW to measure the sub impulse responses at a few listening locations.  REW can decompose the response into the minimum phase response having the same magnitude FR and the so-called "excess phase" portion, which is all-pass in nature (i.e., flat response in frequency but not time).  This allows you to see what response features are essentially minimum phase and which are not.  You may be surprised to see that most if not all your bass response is minimum phase.

 

As you say, sub and listener placement are more important than EQ, but often one or both is constrained by the room design or for other reasons.  In some cases, listener placement may also be constrained by performance requirements for the rest of the spectrum.  Even ideal sub and listener placements are usually not enough to get a flat response.  EQ can improve the response beyond what is otherwise possible with placements alone.

 

Your statement that "Spatial averaging EQ is a wet dream" is (pardon the pun) prematurely dismissive.  Spatial averaging is indeed an unreliable automated room correction approach for several reasons.  For manual corrections, it can be useful as long as the end result is verified.  Either way, spatial averaging basically solves the wrong math problem.  There are many other multi-seat correction approaches possible that yield much better results.

 

Hmm pictures?  What kind of pictures would you like to see?  I'll start by posting REW captures of spectrograms of my LFE output at MLP using window sizes 25, 50, 100, and 200 ms.  All of these are derived from the same response.  The shorter windows provide better time resolution at the expense of frequency resolution.  In aggregate, they give a rough idea of what a LFE impulse sounds like in my best seat:

 

Edit: It helps if I show the attached files in the post.

 

attachicon.giflfe-spec-25ms.png

attachicon.giflfe-spec-50ms.png

attachicon.giflfe-spec-100ms.png

attachicon.giflfe-spec-200ms.png

 

Take note that the y-axis indicates milliseconds and the color ranges across 40 dB.  Note that almost all the energy is concentrated near the start.  The time of arrival of energy is delayed slightly as frequency decreases.  This is due to a combination of things: the 4th order Linkwitz Riley crossover at 50 Hz for bi-amped sub pairs, the port tuning, and the protective HPF on the deep bass subs.  These contributions are non-minimum phase, so even the relatively flat magnitude response I have (not shown) is not enough to fix this temporal problem, but it is relatively minor.  Either way, I hope to repair much if not most of it with my first deployment of mixed phase filters.

 

If you want to see them, maybe tomorrow I could post spectra of each sub at the MLP without EQ to show what a huge difference EQ can make.

 

 

I'm interested in seeing the responses of each sub and what correction and delays you applied to each sub to get good results.  

 

JSS

Link to comment
Share on other sites

As for whether the auditory system can resolve differences between transient and steady response: of course it can!  Can you notice the difference between a pulse centered at 50 Hz and a continous tone centered at 50 Hz?  Of course you can.  They are completely different sounds.  Supposing there's a high Q room resonance centered at 50 Hz too, how does that effect things?  Well, the transient won't hit quite as hard, and it will additionally have a decaying tail around 50 Hz that gives it kind of a boom sound instead of just a solid hit.  The continuous tone will sound exaggeratedly loud especially in comparison to the adjacent tones.

 

that's not the correct comparison as we're not talking about the difference between a pulse and a tone, we're talking about the difference between a pulse that has a certain magnitude and that then decays to 0 over a given time period vs a pulse that has a lower initial magnitude but rapidly increases to that same magnitude and then decays to 0. The question then is whether or not your auditory system has the sensitivity & temporal resolution to detect that reduced initial magnitude. 

 

Take 40Hz as an example, you'll get a 1st axial mode here with a dimension of 4.2m & a full cycle at 40Hz is 25ms long. Assume the sub is in a corner so the 1st reflection is really off the wall behind you, by the time you've "heard" a full cycle at 40Hz, the leading edge of wave has already travelled the length of the room two more times hence that increasing magnitude caused by the mode is already happening before you've even heard the entire wavelength. 

 

We also know that human hearing can be modelled as a series of filters (e.g. the gamma tone filter bank model) and that those filters get increasingly long as frequency reduces. This means temporal resolution of our hearing reduces as frequency reduces while the frequency range covered by each filter gets smaller and smaller (in linear terms anyway)

 

The combination of these two physical facts (a mode acts fast relative to the period, your hearing has poor temporal resolution at LF) make me quite sceptical that we can distinguish that lower initial magnitude produced by the use of EQ to cut a mode. 

 

When I get time I'm going to try and play with this anyway by comparing correction using an IIR filter (traditional PEQ) vs a VBA based approach (same sub will emit a delayed, inverse signal to provide the cancellation) vs a physical active bass trap (same VBA sort of approach but the cancelling signal reproduced by a different sub). I can do the 1st two easily enough, the 3rd one needs me to build a box for a driver I have handy. 

Link to comment
Share on other sites

I'm interested in seeing the responses of each sub and what correction and delays you applied to each sub to get good results.  

 

JSS

 

 

+1^^^

 

As well as the final response. Looks like a 20 Hz tuned ported system?

 

I believe that on time delivery is important but is most important an octave above and below cross (which is why including the mains in the measurements is a must, IMO). Below that, the decay time tends to swamp the start time.

 

I'm interested in the main seat only. Responses at seats outside the sound field are not relevant to critical listening discussions.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...