john k... said:
You point is beside the point. You sidestepped my point by not answering my question.
Then how would YOU model a cardiod? Your comment is not much help. And I DO think that what I suggest is approximately correct. More correct would be, like I said, three points. The outer two a dipole and the inner on a monpole In the far field this combination WILL have a cardiod pattern when the monople strngth is properly set. Now how much of an error is it to move the center monopole out to one of the two dipole sources?
I hope you are not going to try and delay my demonstrating my points in a simulation by arguing the methods that I use. You tell me what you would accept.
Hi Earl,
I too feel that your proposed way of modelling cardioid is not appropriate to audio reproduction in a normal room. You cannot have constant cardioid with varying frequency where there is a fixed distance between dipoles without varying the time (phase) response to one with frequency, in order to develop that rear null. Then, using two dipoles in different positions will generate different room effects after loudspeaker energisation which are relative to each and which cannot be countered electrically before they arise .
Even though I would not choose to use JohnK's cardioid because of the necessary rear cabinet depth and wall/corner distance necessary due to the null failing circa 100Hz where significant close wall peaking can still arise, also the rear of cone loading, (does not suit my PERSONAL requirements, and my preference would probably not suit JohnK either) I am not aware of any LF audio cardioid generation system which is equally simple and useful.
I have tried something similar to JK's cardioid, and it is quite uncanny the way that the LF SPL falls away behind the loudspeaker. Thus his resistive cardioid loudspeaker cannot generate the same kind of peak and dip rear of LS to wall reflections as will a monopole, dipole or bipole.
There is no far field in most domestic rooms at LF.
Cheers ........ Graham.
I too feel that your proposed way of modelling cardioid is not appropriate to audio reproduction in a normal room. You cannot have constant cardioid with varying frequency where there is a fixed distance between dipoles without varying the time (phase) response to one with frequency, in order to develop that rear null. Then, using two dipoles in different positions will generate different room effects after loudspeaker energisation which are relative to each and which cannot be countered electrically before they arise .
Even though I would not choose to use JohnK's cardioid because of the necessary rear cabinet depth and wall/corner distance necessary due to the null failing circa 100Hz where significant close wall peaking can still arise, also the rear of cone loading, (does not suit my PERSONAL requirements, and my preference would probably not suit JohnK either) I am not aware of any LF audio cardioid generation system which is equally simple and useful.
I have tried something similar to JK's cardioid, and it is quite uncanny the way that the LF SPL falls away behind the loudspeaker. Thus his resistive cardioid loudspeaker cannot generate the same kind of peak and dip rear of LS to wall reflections as will a monopole, dipole or bipole.
There is no far field in most domestic rooms at LF.
Cheers ........ Graham.
Its trivial in a model to adjust of source strength to keep the cardiod pattern should that be necessary. Three point exactly yields a cardiod and I still don't see the problem with using two.
In fact if John tells me how to place the source, how to point it and where to put it and what size room, I'll use his suggested configuration. Then I'll compare that to the monopole (not at that same location of course as that would not be fair) and a dipole, (probably at Johns same location) and a second monopole.
In fact if John tells me how to place the source, how to point it and where to put it and what size room, I'll use his suggested configuration. Then I'll compare that to the monopole (not at that same location of course as that would not be fair) and a dipole, (probably at Johns same location) and a second monopole.
Hi Earl,
You now mention three 'point' sources to generate cardioid instead of two dipoles without explaining what they are, so I cannot comment further because you do not explain how you expect these to work.
You yourself say that additional monopoles need to be carefully placed for their additional effect to be beneficial. Any loudspeaker created from separate point sources in order to simulate a 'cardioid' response wrt to any one of those elements is going to have critical room position dependency for every element at the chosen test frequency, and those results will be irrelevent for music reproduction !
Your proposed test cannot be relevent unless you have completely independent and free standing fully integrated LS systems.
Another point I should like to mention is the incorrect thought that phase shift induced delay is not important at LF due to the wavelengths being longer than we can sense.
Where has this been proven, how and with what reference equipment ?
I have spent a fair bit of time running different LF drivers resistively in small sealed cabinets so that the resultant transduction phase could not be modified by any 'within-band' driver resonance; ie. sub-resonant so that the transduction phase would remain linear with falling frequency, say below 100Hz.
Where conventional EQ (2x 6dB/oct) stages were used to counter the LF roll-off, the amplitude compensation could be accurate, but the phase response led to increasingly incorrect reproduction with falling frequency, and it simply did not sound right.
I developed a circuit which increased amplitude with falling frequency whilst maintaining in/out phase constancy, with the result that the 'phase/distance' characteristics of all reproduction immediately fell back into place.
The difference in reproduction between phase linear LF and phase shifting LF is instantly recognisable, though that difference might not be appreciated unless a phase linear source is already there as reference for A-B comparison, or loudspeakers are tested beside live performance !
And why is LF phase so important ? The lower frequency harmonics which arrive in normal time re-combine incorrectly in the listening room when the fundamental has been 'shifted' in time. End of story !
The loss of LF coherence is most noticeable on live open air and large concert hall recordings, and most 'hi-fi' systems already suffer from it without the listeners realising because they have not had the chance to hear better.
Cheers ......... Graham.
You now mention three 'point' sources to generate cardioid instead of two dipoles without explaining what they are, so I cannot comment further because you do not explain how you expect these to work.
You yourself say that additional monopoles need to be carefully placed for their additional effect to be beneficial. Any loudspeaker created from separate point sources in order to simulate a 'cardioid' response wrt to any one of those elements is going to have critical room position dependency for every element at the chosen test frequency, and those results will be irrelevent for music reproduction !
Your proposed test cannot be relevent unless you have completely independent and free standing fully integrated LS systems.
Another point I should like to mention is the incorrect thought that phase shift induced delay is not important at LF due to the wavelengths being longer than we can sense.
Where has this been proven, how and with what reference equipment ?
I have spent a fair bit of time running different LF drivers resistively in small sealed cabinets so that the resultant transduction phase could not be modified by any 'within-band' driver resonance; ie. sub-resonant so that the transduction phase would remain linear with falling frequency, say below 100Hz.
Where conventional EQ (2x 6dB/oct) stages were used to counter the LF roll-off, the amplitude compensation could be accurate, but the phase response led to increasingly incorrect reproduction with falling frequency, and it simply did not sound right.
I developed a circuit which increased amplitude with falling frequency whilst maintaining in/out phase constancy, with the result that the 'phase/distance' characteristics of all reproduction immediately fell back into place.
The difference in reproduction between phase linear LF and phase shifting LF is instantly recognisable, though that difference might not be appreciated unless a phase linear source is already there as reference for A-B comparison, or loudspeakers are tested beside live performance !
And why is LF phase so important ? The lower frequency harmonics which arrive in normal time re-combine incorrectly in the listening room when the fundamental has been 'shifted' in time. End of story !
The loss of LF coherence is most noticeable on live open air and large concert hall recordings, and most 'hi-fi' systems already suffer from it without the listeners realising because they have not had the chance to hear better.
Cheers ......... Graham.
gedlee said:
Zounds! That's not what I would call "very small." Three six cubic foot subs, one of them 1/2 way up the wall. 😱 Uh, not in my living room. Guess I need a bigger house! How big are the mains - if the 3 "very small" subs are 6ft^3 each?
Sorry to hark back to this old post, but the multi-sub idea seemed kinda cool, until that came up. And the placement seems kind of restrictive, not really random at all.
OK, back to cardiods - sorry for the interruption. =)
"small" is of course relative and different for each individual and given context. considering the frequency range and the intended spl levels we speak about (home cinema), i would like to call them "too small".
Graham Maynard said:
Three monopoles in a line with one on the end out of phase from the other two will generate a cardiod. Source characteristics are linear and so superposition holds. The center point source is a monpole with a monopole polar response. The outside two are a dipole with a cos(theta) response. The sum of the two is 1+cos(theta) - a cardiod. The amplitude of the monopole usually has to be adjusted to track the dipoles pressure response, but this is not a problem.
"You yourself say that additional monopoles need to be carefully placed for their additional effect to be beneficial. "
I never said this, in fact, I tend to say the opposite, that "additional" monopoles can be placed anywhere. There are, of course, better and worse places to put them, but what source is this not true for?
panomaniac said:
As the other guy said, its all relative. In the world of first rate LF capability with 120 dB respopnse at say 30 Hz. this is very small!! In loudspeakers size does matter.
And I just don't get where my placement SUGGESTIONS are all that restrictive. You CAN place the subs anywhere you want and the more there are the better they will work. But just like any other type of sub there are better ways to place them such that they work better. You can't put a dipole in a corner and expect it to work properly - isn't this "very restrictive"? You can't put the cardiod in a corner. I think corners are ideal places for subs - they are out of the way. Monopoles always work best when up against boundaries - Dipoles can tolerate them unless they are oriented correctly. Which is more restrictive?
Hi Earl,
Back in Post#88 you wrote;-
>> Then add one sub at a time. Adjust the gain and the LF cutoff for best overall response and try the phase switch to see the diffrence. The phase switch should be a large difference for the first sub (pick the "better" one), a lessor difference for the second and so on. Add one sub at a time. <<
Apologies. I should not have written that the placement is critical, but that their individual level and phase set-ups are.
To me that is every bit as complex.
My thoughts are that this arrangement might focus imagery at the listening position, but smear energy at other locations in the room.
Maybe this is the intention.
Okay I am beginning to follow your idea here
>> Three monopoles in a line with one on the end out of phase from the other two will generate a cardiod. Source characteristics are linear and so superposition holds. The center point source is a monpole with a monopole polar response. The outside two are a dipole with a cos(theta) response. The sum of the two is 1+cos(theta) - a cardiod. The amplitude of the monopole usually has to be adjusted to track the dipoles pressure response, but this is not a problem. <<
Line vertical or horizontal ?
However if one driver is out of phase, and the amplitude of one has to track the other two then phase changes will affect wideband coherence and directionality.
However. Theory is worthless if we cannot actually hear the imagined imperfections.
Cheers ........ Graham.
Back in Post#88 you wrote;-
>> Then add one sub at a time. Adjust the gain and the LF cutoff for best overall response and try the phase switch to see the diffrence. The phase switch should be a large difference for the first sub (pick the "better" one), a lessor difference for the second and so on. Add one sub at a time. <<
Apologies. I should not have written that the placement is critical, but that their individual level and phase set-ups are.
To me that is every bit as complex.
My thoughts are that this arrangement might focus imagery at the listening position, but smear energy at other locations in the room.
Maybe this is the intention.
Okay I am beginning to follow your idea here
>> Three monopoles in a line with one on the end out of phase from the other two will generate a cardiod. Source characteristics are linear and so superposition holds. The center point source is a monpole with a monopole polar response. The outside two are a dipole with a cos(theta) response. The sum of the two is 1+cos(theta) - a cardiod. The amplitude of the monopole usually has to be adjusted to track the dipoles pressure response, but this is not a problem. <<
Line vertical or horizontal ?
However if one driver is out of phase, and the amplitude of one has to track the other two then phase changes will affect wideband coherence and directionality.
However. Theory is worthless if we cannot actually hear the imagined imperfections.
Cheers ........ Graham.
gedlee said:
If a system is oriented towards reproducing popular music and not HT, we can target uniform max SPL down to 40Hz only, losing 6dB max SPL (for same distortion levels) for each decade under 40. That can make subs smaller. The content of most music will not be terribly hurt from such a convention. That works especially well in pro installations too. Full range, full SPL, can be reserved for large spaces of very good structural integrity, and budgets of a high ceiling.
Graham Maynard said:
You don't set the gains and phase on your subs? Whats so complex? Using a spectrum analyzer is somewhat complex, but hardly brain surgery.
There is no image at LF.
Any line, doesn't matter.
Huh?
Hi Earl,
Brain surgery ? Oh that's nice ?
Certainly not technical discussion though !
I don't have subs !
I have used and implemented subs, but I don't like them because they cannot be made to integrate phase linearly.
And this is the point I have been making, but which you appear to refute.
Subs are merely 'add-ons' to increase LF SPL.
No image at LF ?
I am trying to make sure the LF is as coherently integrated with the rest of the reproduced sound as possible, and for this to happen the LF radiation from the LS must be coherent, including LF, for otherwise the radiating waveshape is modified = distorted.
Cheers ......... Graham.
Brain surgery ? Oh that's nice ?
Certainly not technical discussion though !
I don't have subs !
I have used and implemented subs, but I don't like them because they cannot be made to integrate phase linearly.
And this is the point I have been making, but which you appear to refute.
Subs are merely 'add-ons' to increase LF SPL.
No image at LF ?
I am trying to make sure the LF is as coherently integrated with the rest of the reproduced sound as possible, and for this to happen the LF radiation from the LS must be coherent, including LF, for otherwise the radiating waveshape is modified = distorted.
Cheers ......... Graham.
Graham Maynard said:
I don't quite understand why you are in this discussion if you don't use subs.
To me the idea of adding subs is not to increase the LF SPL but to smooth out the response of the Low frequencies in the modal region. The added SPL is a nice benefit, but I usually trade this off for a smaller sized sub.
The coherent wavefront idea has some merit where this is possible, but its simply not possible in a small room at LF - i.e. the modal region. The point that I am trying to make to you is that you have to view this LF region differently than you view sound reproduction in larger rooms, outdoors, or at higher frequencies. The LF case in a small room is unique among the sound fields that we find in practice and concepts that work in other fields will not necessarily work in small rooms at LF. The small room at LF is the only situation where the modes are distinct and wave propagation is fixed in direction. Nowhere else does this occur. You have to do things diferently when this happens.
gedlee said:
Ha! No argument from me. I think I've even convinced my wife of that - with the parade of large speakers I drag thru the living room.
But for most of us, at least those who have a music system in the main room, 18ft^3 total of subs is not small.
After I posted last night ( I was at work) I went in search of something about that size. Found it right away. In the office we have several of those 1/2 height file cabinets. 24x24x16. They don't' look so huge in a large space like the office, but they would take up a bit of space in most homes.
So maybe if you are used to Altec 210 cabinets, or have a huge HT, then the 24x24x18 subs might seem small. It is all relative.
I'm with Salas on this one. For music, the bottom octave is the least important. Might be nice to have, but even a good, solid 40Hz will be quite impressive.
---Bottom line, if I had a space big enough to hide 3 6ft^3 subs, I'd just go with bigger mains. A matter of personal taste.
panomaniac said:
I think that a lot of people are missing the point about the subs. Its the number and the spatial distribution that counts. Don't obsess on the size, if its too big get smaller ones. The thing that you have to realize is that the greater the number of independent sources the smoother the response variations in frequency and location will be. Use three 8" subs if thats all the bigger you want to go, but never resort to using only the mains.
And three is not some "magic" number either - two is twice as good as one!! Three is about a 50% improvement on that. AT four you will start to see diminishing returns.
I measured my subs again and they are actually 16 x 18 x 18 a bit smaller, but those are fairly high output devices. I have a good size room and I like subs for movies - makes a big difference on sound effects. Thats why they have an LFE channel!
This is a classic issue of I was just guessing at the subs size because the size of the SUBS doesn't matter its the number, but then the discussion got to be all about the size.
The size of the mains DOES matter because of the directivity.
gedlee said:
Ah, OK, gotcha, Doc. No worries then! As I don't really need 120dB at 25Hz in my present listening room, I'll go with some smaller boxes. (My neighbors will thank me). I'm keen to try to distibuted sub idea and have drivers and amps to do it with. Just a matter of running the wires.
OK, makes sense. I have the parts to do 4, but will probably stick with 3. Gotta figure out if I can get one up high.
That was me, mostly. Was just freaked out thinking about 3 big boxes.
BTW, do you like BP boxes because of the high SPL in the limited band you're using?
I'm also keen to try the cardiod bass for the mains. Need to figure out a good way to get more of a cardioid pattern from my open baffle bass. Maybe deeper wings....
On a differnt note. What do you think of Graham's objections to LF arriving at different times or out of phase. May not be a problem with the LFE track, but how about for music?
panomaniac said:
I'm pretty sure I've heard what Graham describes, then Earl posted back saying their was no evidence of those type of effects, which made me doubt myself.
As I say, I had the distributed subs bit, but what annoyed me was that I felt I could localise the subs, which really confused me at the time cause supposedly we are unable to localise under 80 hz or so. So I left that experiment and moved on.
Quite a bit later, I built some subs using four peerless 10s I had laying around, and had them next to the mains. It was good, cept something about the bass just didn't sound quite right, and I couldn't put my finger on it.
I was using a DCX to eq the subs, and then I vaguely recalled somewhere that the DCX has thrutime of a millisecond or so, so I moved the subs forward from the mains a foot and that fixed the problem I was sensing.
Of course that led me to think back to when I could localise the distributed sub setup, and the lightbulb went off and wondered if it was not so much I could localise by hearing, but could do so from the different arrival times of the 'wave'. As I say, instead of a colloective 'thump' so to speak, it was ever so slightly a succession of thump thump thump, especially as the mains were still giving out reasonable output as well.
Even later on I went back to the distributed sub type thing, but in that case I utilised the ability of the DCX to seperately delay each channel individually (ie each sub in this case) - by tape measure and guesstimate - and I'm pretty certain that I do hear the difference when the bass is time aligned so to speak.
I'm very conscious that it could be I'm fooling myself (esp as Earle 'nixed' the idea), boy now I know the hunted haunted feeling a dyed in the wool subjectivist gets when someone challenges their ability to hear the difference between power cords!! ha ha, but unlike the dyed in the wool subjectivist I'd be more than happy to participate in a scientific test to see if I can indeed hear what I think I can!!
panomaniac said:
I know you weren't asking me but I'm going to comment anyways. The mental picture that's used when talking about things like arrival times is one of free space propogation. At LF in rooms the modes are discrete and sparse, so this picture doesn't really make sense. Instead of imagining a wave propogating from the subwoofer to the listener (what happens in free space) a more appropriate picture is to think of the sub as driving a system of discrete oscillators (the room's modes). The "objection" raised is based upon a viewpoint that is inconsistent with reality, so one really cannot argue against it on those terms.
Now if you want to talk in terms of time-domain behavior, then I would contend the best result comes from making the frequency domain as flat and smooth as possible, which means multiple subs located around the room.
localising / localizing
Remember that there are two steps here
a) can you localise the subs?
-lets accept for the moment that you could
b) if so why?
-possibly not due to delays
I have had subs that I could localise and others that I cannot, even with approximately the same frequency range, time delays etc. (I use a ~55 Hz crossover, and that should be low enough to make localisation impossible)
The reason seems to be various non-linear / upconversion effects. The difference between the subs is easily revealed by playing swept sines - the ones I could localise have easily audible port noises, noises from airflow in spaces around the voicecoil, distortion peaks and other unwanted output at various frequencies (probably also when distortion harmonics hit port resonances, and the like).
Remember that the ear is somewhat more sensitive above the sub range than below, and it is possibly less surprising that relatively quiet subs are wanted. I think if you hear these noises unmasked for even a few percent of the music, the brain remembers where the sub is most of the time (particularly if you are concerned about it).
I think it is quite difficult to make a reflex sub quiet enough, at least a new design (see PD1550 tapped horn in the subwoofer forum) is very much better. (Though I do use an old 100l ported one to fill in the biggest holes, and digital correction to remove the delays - music playing only.)
Ken
terry j said:
Remember that there are two steps here
a) can you localise the subs?
-lets accept for the moment that you could
b) if so why?
-possibly not due to delays
I have had subs that I could localise and others that I cannot, even with approximately the same frequency range, time delays etc. (I use a ~55 Hz crossover, and that should be low enough to make localisation impossible)
The reason seems to be various non-linear / upconversion effects. The difference between the subs is easily revealed by playing swept sines - the ones I could localise have easily audible port noises, noises from airflow in spaces around the voicecoil, distortion peaks and other unwanted output at various frequencies (probably also when distortion harmonics hit port resonances, and the like).
Remember that the ear is somewhat more sensitive above the sub range than below, and it is possibly less surprising that relatively quiet subs are wanted. I think if you hear these noises unmasked for even a few percent of the music, the brain remembers where the sub is most of the time (particularly if you are concerned about it).
I think it is quite difficult to make a reflex sub quiet enough, at least a new design (see PD1550 tapped horn in the subwoofer forum) is very much better. (Though I do use an old 100l ported one to fill in the biggest holes, and digital correction to remove the delays - music playing only.)
Ken
- Status
- Not open for further replies.