In room low frequency response

[johnk...]

I have posted a new web page on in room response of different woofer types which address both amplitude and transiend woofer response. It's a little complicated (modal analysis) but if you read through it I think many of you will get the idea of just how bad in room bass rersponse really is and why there isn't any simple slolutions. The aritcle may be found here

[johnpastuck]

Hi John,
Very interesting. I just made some long cables for my subs. I'll give it a try.

[johnk...]

I'm working on a little Excel fronted code that will allow users to play a round with monopole, dipole and cardioid woofers. Hope to get it done in a few days. It's will be a free toy. It's a lot more advanced than most of the room response tools I have seen. It will only handle rectangular rooms but I have tested it against SoundEasy's FEM code.

[djarchow]

John,

I assume this is an expansion of your work you were discussing with Earl Geddes. I will look forward to the spreadsheet as all those equations gave me a bit of a headache despite my digging out a half dozen physics and math texts from the physics degree. Did I really know this stuff back then? But hey I did find my old college bartenders guide...now there's something I remember how to do.

Regards,

Dennis

[markk]

John,

I assume this is an expansion of your work you were discussing with Earl Geddes. I will look forward to the spreadsheet as all those equations gave me a bit of a headache despite my digging out a half dozen physics and math texts from the physics degree. Did I really know this stuff back then? But hey I did find my old college bartenders guide...now there's something I remember how to do.

Regards,

Dennis

Exactly! I just printed it out and will digest it slowly as I have time.

[mzisserson]

Very nice. Thak you... Gret reading, may take several time to sink in, but much appriciated!

[Dirk]

When I grow up, I don't think I want to be John K. I just want to understand half the math involved. Thanks for the inclusion of the graphs. I "get" those.

From the article:

If so, it would seem to create a dilemma; a source which excites the most modes should be least sensitive to position yielding a frequency and transient response which may be more consistent over a wider range of listening and source positions while, on the other hand, a source which excites fewer modes could potentially yield more accurate response for specific source and listener location, but would likely show greater variation with changes in position.

I think this is a key takeaway, but I'll admit I'm a little fuzzy on its implications. I know that all of the analysis was for a single woofer. What about situations where there is more than one source? If there are multuple woofers, situated as to excite evenly as many modes as possible, is that desirable?

[wingnut]

What about situations where there is more than one source? If there are multuple woofers, situated as to excite evenly as many modes as possible, is that desirable?

check out this Welti/Toole interview:

http://www.onhometheater.com/features/20040801.htm

[Dirk]

check out this Welti/Toole interview:

http://www.onhometheater.com/features/20040801.htm

Yeah, I read the underlying white paper a few years ago and recently rediscovered it. The white paper (and interview) address frequency response, but don't really go into transient response, etc etc.

I guess that's the crux of the matter for me: what should I really be optimizing for?

My listening area has the sofa at about the halfway mark in a long, irregular room. I've got a couple obvious room modes that don't vary with seating location. One is obviously the floor/ceiling axial and I've EQ'd that one out (that one band made a HUGE improvement). The other modes seem to vary disturbingly, with massive suck-outs, some appearing & disappearing within a 6" space. There's no EQ for that and the sub is already int he best possible location.

In order to try and "swamp" the room mode problem, I've been considering a "table" style box directly behind my sofa containing (4) 8" woofers. My plan was to excite as many dominant room modes as possible, but based on that one paragraph I originally quoted, I'm not sure if that's what I should be shooting for.

[johnk...]

I'll try to response as best I can. First, not being much of an expert web page developer I really don't know what I can do to change how the stuff prints. I'm just too much of a dummy on that aspect of web publishing. If It's just too wide I'll see what I can do. I just use the tool Yahoo gives me. Sorry.

Second, yes, this is an extension of some of the discussion with Earl G. over at DIY. I imagine he might disagree with some of it. He tends to see things his way. That's ok with me but I think the math is solid.

It certainly does seem to be a dilemma on how to really get the best response. And sitting exactly in the center (side to side) doesn't help things either because all the odd number axial modes in the side to side direction will have a pressure node there. Fortunately few of use really have rectangular rooms.

The upcoming spread sheet will have a section for single woofer (monopole, cardioid, and dipole) evaluation and a section for up to four woofer, but all monopoles. Maybe some day I'll upgrade it to multiple dipoles and cardioids and even mixed types.

Anyway, thanks for the feed back., but remember the spreadsheet isn't going to be definitive. Just another play toy.

By the way, I tried to write it up so what you don't really have to understand the math to understand the concepts. How close did I come?

[philiparcario]

JOHN this is a nice tool. My home theater does not have the right shape, but my 2 channel music room will work. I will spend some time checking out the site. Thank you Phil

[donradick]

I'll try to response as best I can. First, not being much of an expert web page developer I really don't know what I can do to change how the stuff prints. I'm just too much of a dummy on that aspect of web publishing. If It's just too wide I'll see what I can do. I just use the tool Yahoo gives me. Sorry.

By the way, I tried to write it up so what you don't really have to understand the math to understand the concepts. How close did I come?

I'm no web expert either, despite solid IT knowledge in a number of domains.
I'll PM you with some suggestions for the web format.

Uh, wish I could give you a big thumbs up on the math, but my expertise stops
at algebra and quadratic equations and such. Always wanted to crack one
of the calculus books on my shelf, but I'd really rather build and listen to
speakers. I can definitely grok your analysis, graphs, and conclusions though.

Dirty Harry: "A man's got to know his limitations"

-Don

[Deward Hastings]

It is difficult to separate room effects from speaker behavior at low frequencies. When comparing dipole radiators and monopole radiators for example first reflection from the front wall will cancel for monopole and add for dipole when the wall is ¼ wave distant. At other frequencies peaks and nulls reverse. Dipoles excite fewer room modes because of their directional radiation pattern, and they excite the “on axis” mode differently from a similarly placed monopole radiator. Relatively modest changes in position can have significant effect on mode excitation, and that in turn effects perceived frequency response and perceived transient response, since an excited room mode (resonance) extends the duration of an impulse. Exciting many room modes in a “live” or reverberant room can make bass response seem louder (except at the unavoidable nulls) . . . it also makes it seem more blurred, like the difference between a piano played with and without the sustain pedal. For accurate bass reproduction it would seem best to avoid exciting room modes as much as possible.

There are a number of computer models of room response on the web which purport to offer guidance on speaker placement, listener placement, room treatment and even furniture placement for best results. How accurate or effective they are is open to question . . . what is clear from all of them is that the room is a critical factor in bass response. There are quite a number of free response modelers,

http://www.rspu.ryazan.ru/~foboss/hi...c/room060d.xls

for example, which is an Excel spreadsheet into which you load room dimension, speaker placement and listener location and which outputs a graphic of the room and a frequency response curve for both monopole and dipole speakers, and commercial

http://www.cara.de/ENU/CARA/index.html

There are plenty of others to be found. It will be interesting to see how the results of the “room060d” spreadsheet compare with John K’s.

Some elements of the differing performance of dipole and monopole bass radiators are discussed here.

http://www.linkwitzlab.com/frontiers_4.htm

and this on the effect of (sub)woofer offset, a subject that often comes up in discussions of subwoofer “integration”.

http://www.linkwitzlab.com/frontiers_5.htm

[johnk...]

I pretty much agree with Deward except I would caution about thinking of low frequency response in terms of reflections of the wall and resulting cancellations. These are concepts of traveling waves and are more applicable (though not totally) at higher frequencies where geometric acoustic (ray tracing) is more applicable, or for the case where there is only a wall behind the source. For example consider the this image



It shows a dipole source centered 1/4 wave length from a wall at a frequency corresponding to a standing wave with wave length twice the room length. While the dipole sources are of inverted phase, so is the amplitude of the standing wave at the positions of the dipole sources. The result is that the two dipole sources contribute equally to the excitation of this mode and whether or not there is a null in the response at this frequency will only be dependent on the position of the listener. Ignoring other modes, for a listener at A there would be a null. For a listener at B there would not be a null. On the other hand, if the source were a monopole positioned positioned at the 1/4 wave point this mode whould not be excited and it would matter where the listener is sitting. There would be no contribution to the in room response from this mode anywhere in the room.

The problem is very complicated. Real sources aren't point source and their physical size means they will never really be at a node. Wall conductance and admittance can be frequency as well as amplityude dependent, and so on...

[jkrutke]

Interesting work, John. A lot of it is over my head, but I'll be paying attention to the spreadsheet you come up with. About 6 years ago, I came into this spreadsheet written by "Yavuz Aksan" that calculated room response from a monopole or dipole. It looked nice and appeared to work but it never really correlated with my real world findings measured ungated with a mic at the listening position, even in a perfect, sealed rectangle room. Have you seen this spreadsheet? I think it was from the old FRD page.

But yeah, I agree in-room bass response is troublesome. I generally just rely on a mic and relocation until I find something that sucks the least.

[johnk...]

Interesting work, John. A lot of it is over my head, but I'll be paying attention to the spreadsheet you come up with. About 6 years ago, I came into this spreadsheet written by "Yavuz Aksan" that calculated room response from a monopole or dipole. It looked nice and appeared to work but it never really correlated with my real world findings measured ungated with a mic at the listening position, even in a perfect, sealed rectangle room. Have you seen this spreadsheet? I think it was from the old FRD page.

But yeah, I agree in-room bass response is troublesome. I generally just rely on a mic and relocation until I find something that sucks the least.

Yes, I have it. It is the room060d spread sheet that Deward was referring to. I even have a modified version of it for cardioids.

I don't think any of these codes is going to do much to match reality. I put together my code because I know what in it and it's a good tool to play what if games with, particularly with regards to how different woofers behave. I'm more interested in understanding behavior than actually predicting real room response. The problem with real rooms is that there are just too many details to get right. Just considering wall, are they reflective? What is the variation with frequency? Are the reactive (do they move)? If so is it compliance dominated motion or a mass dominated motions? How does it vary with frequency? Door? windows? Hall ways?

I will say that I have tested my code against the FEM part of Sound Easy for rectangular room and it yields very good agreement in those cases. If I want to look at odd shapes I’ll use SoundEasy, but it can take ½ a day or more just to compute the response for one source/listening position. if you are only interested in the response below 100 Hz or so the code I put together takes a few second. If you want to get the picture up to the 1 KHz upper limit of the code then it takes a few minutes because you have to consider more modes.

But like I said, view it like a toy.

[Deward Hastings]

It looked nice and appeared to work but it never really correlated with my real world findings measured ungated with a mic at the listening position, even in a perfect, sealed rectangle room.

I think that's the problem we all encounter with these models . . . they look nice but don't seem to correlate to real expericnce with enough reliability to have predictive value. Not least of the problems is that they all seem to assume steady-state sine waves as the excitation. It is easy, for example, to find the center-of-the-room null with a steady state tone . . . frightening, even how the signal comes and goes with a relatively slight movement of the mic (or one's head). A step function or impulse measured at the same location does not, however, come up null. So when we're measuring (or modeling) room modes with sine waves below the fundamental of the lowest playing orchestral instrument just what are we looking at, and what is its significance? Of above, for that matter . . .

I tried doing a room response measurement using a ribbon (velocity) microphone a couple years ago (essentially no response below 30 Hz. unfortunately) to see if I could see the pressure/velocity difference . . . all I got was more confusing results still not correlated to much of anything.

And then there's the additional minor problem of room variance. I listen in a lathe-and-plaster room with lots of loose fitting casement windows . . . significntly transparent to low bass. A room of identical dimension built out of concrete would measure, and sound, quite different. I note that the pictures SL displays of his listening room (large expanse of glass door and lots of space behind his listening positions) might be expected to have limited low frequency reflection . . . still he hears a difference between omni and dipole radiators in that room (although probably not as great as it would be in a more reflective environment).

I think we're a long way from having models of this behavior that tell us anything of much use . . . and perhaps a long way from understanding it. Once agan I come away from it marveling that any of our sound reproduction gimicry works at all . . .

[johnk...]

I think we're a long way from having models of this behavior that tell us anything of much use . . . and perhaps a long way from understanding it. Once agan I come away from it marveling that any of our sound reproduction gimicry works at all . . .

I am editing this a little.
I agree that the models don't tell us what we want to know, but I do believe what is going on is well understood. It's really that we don't have the necessary data to plug into tha analysis. I don't think the problem is one of steady state analysis though. Please note this transient figure from my web page



This is the transient for the 1-d case of the monopole eqed to the woofers free field response. In this case the model correctly predictes a minimum phase frequency response and EQ yields what would be the expected pulse response, except for the delay which is consistent with the source to mic distance. FFT analysis still holds between the time and frequency domain provided the room/mic/source system is basically linear and time invariant.

If we were to build a truly rectangular room and then took the time to characterize the complex admittance of the walls, floor and ceiling; and modeled the source as a source of finite size, I'm confident that the theory would agree with measurement. But it’s the old catch 22. We just don't have a good handle on what the correct surface admittances are before hand, particularly at low frequency. When we move into the higher frequency range we enter an area where things can be handled more on a statistically basis and it’s a little easier to deal with.

[Andy_G]

If we were to build a truly rectangular room and then took the time to characterize the complex admittance of the walls, floor and ceiling; and modeled the source as a source of finite size, I'm confident that the theory would agree with measurement. But it’s the old catch 22. We just don't have a good handle on what the correct surface admittances are before hand, particularly at low frequency. When we move into the higher frequency range we enter an area where things can be handled more on a statistically basis and it’s a little easier to deal with.

You would also have to take into account the transmission of sound within the structure, we all know wooden floors do different bass than concrete floors, and also consider that we pick up low frequency with a fair amount of tactile sense from the rest of our body and any furniture we might be sitting in.

I'm glad you are doing it, have fun !!

[Deward Hastings]

This is the transient for the 1-d case of the monopole eqed to the woofers free field response. In this case the model correctly predictes a minimum phase frequency response and EQ yields what would be the expected pulse response, except for the delay which is consistent with the source to mic distance. FFT analysis still holds between the time and frequency domain provided the room/mic/source system is basically linear and time invariant.

The problem doesn't have to do with what comes out of the speaker . . . that's fairly well understood for monopoles and dipoles alike. The problem is what happens once the sound is "loose in the room", and how we hear the difference between an impulse (the initial part of which a mic can "see" too, along with subsequent reflections) and the position dependent peaks and nulls (the "room modes") that form with a steady state tone.

The impulse graph you show indicates no reflections, no room resonance, no "modes" at all. Would that we all were so fortunate in our listening environments . . . or maybe not. There seems to be general agreement that some reflection is good at higher frequencise . . . it is my speculation that the goodness goes away as we get down into the bass, or into regions where wavelength and room size become similar. But I can't prove it . . .