A bit of a Red Herring, that. The simulation used pink noise for the undelayed element and the same pink noise for the delayed (rear) element, but the sumation is anything but balanced. The delay causes a continuous phase shift that gives rise comb filtering. Flat energy in reflections doesn't make them more pleasant or less audible.
David S.
Thanks for those, very noticeable differences. A bit hollow sounding, like in a live room.
But I'll agree with others that these are sort of a worse case scenario. Some good diffusion on the wall behind the speakers would not sound nearly so obvious. Retain the energy of the reflections, but keep them from interfering directly.
Maybe if I have the time, I can do some of this in CARA.
Hi,
This test does not represent any realistic case since most of the dipoles are not dipole above about 2 kHz, and they are not intended to be, thus they don't radiate backwards any more than a usual box speaker would do.
If you have effort, please provide similar example but low pass filter the reflection with 1-2kHz 2nd or 4th order.
- Elias
This test does not represent any realistic case since most of the dipoles are not dipole above about 2 kHz, and they are not intended to be, thus they don't radiate backwards any more than a usual box speaker would do.
If you have effort, please provide similar example but low pass filter the reflection with 1-2kHz 2nd or 4th order.
- Elias
This is very similar to what I do, and what I recommend to others. In the constant directivity cornerhorn design, it is accomplished by blending the midrange and woofer in the 100Hz to 300Hz range. The woofer low-pass rolls it off above that. The midrange runs pretty low so they can be blended, sort of like a 2.5-way aproach. When using the matched-directivity two-way speakers, which are usually put on stands, I employ what I call "flanking subs". They are woofers placed a couple feet away, below and usually behind the mains. These are blended to smooth the floor bounce and front wall bounce notches. They're also part of the multisub approach, which serves to smooth lower frequency room modes.
More about these approaches below:
I'm not sure about "most dipole" since all electrostatic and planer magnetics would be dipole for the full range. What is the point of having "the dipole benefit" for lower frequencies only?
I'll do you a filtered reflection tonight.
David S.
I'll concede that absorption or diffusion might get it below a threshold of perception. At the same time I wonder how many owners set up appropriate treatment for that perticular wall bounce?
The original point was that dipoles sent energy in an unfortunate direction (necessitating treatment, lets say) while distinctly not sending it in the desirable directions.
David S.
I have to agree. Any time I have speakers positioned in front of the short wall of a room (firing along the length of the room) below a certain room size (less than about 4 metres wide relative to the listening orientation) I find myself in constant battle with the early side wall reflections when trying to optimize speaker placement and/or room acoustics. It's a major drawback of this room orientation unless the room is quite large - there is a fundamental compromise between getting the speakers wide enough to give good stereo imaging, but still far enough away from the side walls that becomes impossible below a certain room width.
Yes, it does make the sound field somewhat wider and it can make some recordings (generally live ones) sound more interesting, however I would argue that this is an artificial effect, not a desirable or accurate effect for playback in general - after all it has the acoustic characteristics of your room, not that of the recording.
The other down sides are it considerably worsens L-R image placement accuracy, worsens image shift with lateral movements of the listener (as the phasing between the direct path and wall reflections change, altering the comb filtering effect) and makes the exact distance from the side walls of both speakers very critical - if they're not symmetrically placed (and either is less than 1 metre from the wall) the reflection delay on one side is different from the other which is very detrimental for imaging stability.
It also introduces considerable tonal colouration (even a speaker with flat off axis response at the incident angle will cause audible comb filtering in the lower midrange) unless you can get the speaker at least ~0.8 - 1 metre away from the side wall, which is not always possible in a smaller room in that orientation.
Lots of negatives and only a subjective increase in "spaciousness" which works for a few recordings but makes most sound worse. No thanks, I'll take a relatively damped room and a lack of early side wall reflections any day.
For a smaller room (narrowest horizontal axis less than 4 metres) I'm now convinced that placing the speakers on a long wall and firing them across the shorter axis is better - it lets you get them well away from the side walls, and the detrimental effects of early side wall reflections - it's immediately obvious how much less sensitive to lateral positioning the speakers are - there is very little shift in tonal balance through the midrange and treble, allowing you to concentrate on optimizing imaging and/or bass response.
You do generally have a BIG room mode at a higher frequency (50Hz in my room) in this orientation due to the short speaker-listener room axis, but if you accurately correct this with parametric EQ the overall result can be better in every way, smoother mid/upper bass, no side wall colouration in the midrange, sufficient angular separation of the speakers even in a small room, and a good direct/reflected ratio that makes a room even with a moderately high RT60 sound better. (IMHO)
Personally I don't think they're needed, but it depends on what you want - do you want to be transported to the recording venue, or do you want to transport the musicians into your living room ? For the former they're detrimental, for the latter they're required. For me the former is what I want - I love the sensation of closing my eyes and feeling like I'm in a different, convincing acoustic space. I want to be at the event, not have the event in my listening room.
Agreed again. Some types of reflections (depending on direction, delay, etc) are less detrimental than others. I'm of the school of thought that all room reflections are detrimental in some way and to some degree. On certain types of music reflections can "add" something that may be perceived as musically pleasing, but that's very recording specific.
There's certainly a few different approaches that will work. I've always liked a 12" woofer very low to the floor crossing over at about 250 or 300Hz, to a midrange driver that is much higher - about 80cm or so.
That seems to take care of it very well, however with a wide baffle you can run into problems with too much gain as you near 300Hz where both the baffle and floor are giving gain, making it difficult to roll the woofer off properly with a passive network.
You then either need to narrow the baffle a bit (but you can only go so far with a 12" woofer...) or lift the woofer up off the floor just a bit so that the floor gain starts to go away as the baffle gain kicks in, which is my favoured solution.
I did some tests years ago with a 12" woofer on a 50cm wide baffle where I tried it at several heights and measured the response flatness from bass up to 300Hz and did find a specific height that gave a very balanced response with no significant holes at the listening position, nor the excessive top end rise of the lowest height, but I can't remember what height it was and don't seem to have written it down anywhere.
What really surprised me in this recent test was the fact that the woofer is unbaffled, and yet at 3 metres it's still able to increase the response at 100-150 Hz by nearly 10dB - that goes to show just how severe the floor induced notch is. I really didn't think it would have much effect without at least a minimal baffle.
Last edited:
There you go again . . . false assumptions leading to false conclusions. Those are not “the reflections” . . . you have simply created a false model that has little-to-nothing to do with reality, and then drawn false conclusions from it.
And again . . . you did not do a “reasonable simulation”. The “only other variable” is not “the attenuation of the back bounce”. Your model does not come even close to what is “realistic” regarding the dispersion and reflection of back wave energy, or what it “sounds like”.
I share gainphile’s experience in this . . . I have had very good monopoles (Thiel 3.6) and very good dipoles (Linkwitz ORION) properly set up in the same room and the dipoles present at least as good and with most recordings a better “sonic image”, and overall a more natural and accurate reproduction. The same applies to the Maggies I had years ago, and the contemporary monopoles I compared them to.
When your “theory” and my “practice” so consistently diverge the only conclusion (for me) has to be that either your theory is substantially incomplete (and thus wrong) or it’s flat out wrong, period. Either way it will not produce useful results for anyone . . .
Back to the topic of one curve that (ideally) would be similar to one's ear impression, I added some parameters in a software of mine.
Loading Impulse response, you can create various graphs. I added parameters to a so-called "psychoacoustic response" :
you can choose high frequency gating time, various smoothings (1/3 to 1/6, ERB,...).
So those interested can easy try combining parameters.
Note that it works only on windows and you need Octave (equivalent to mathlabs).
Download is here : http://www.ohl.to/audio/downloads/align2.zip
Loading Impulse response, you can create various graphs. I added parameters to a so-called "psychoacoustic response" :
you can choose high frequency gating time, various smoothings (1/3 to 1/6, ERB,...).
So those interested can easy try combining parameters.
Note that it works only on windows and you need Octave (equivalent to mathlabs).
Download is here : http://www.ohl.to/audio/downloads/align2.zip
I actually have a fair background in acoustical modeling including some very high power concert hall simulation with CATT acoustics, a ray tracing program. The theory is pretty simple. Sound eminates from a speaker in all directions based on the 3D polar pattern (a function of frequency). At every boundary it traverses the energy is reduced by the absorption coefficient (alpha) and perhaps scattered. The energy carries on until it hits the next boundary and is agian reflected, absorbed and scattered. Energy drops due to 1/R distance loss. Send out enough rays and you can simulate a complex concert hall and come up with all the standard measurements of RT60, RT30, G, EDTG, lateral fractions, etc.
In the case we are discussing of a dipole speaker in front of a smooth drywall surface the model is as simple as stated. Alpha is fairly low, the off axis rearward angle isn't far off of on axis, hence the polar won't have dropped much (but it could be determined and accounted for) and the rear wall bounce has farther to go so there is extra 1/R drop. I combined those as a 5dB loss, primarily due to some distance assumptions. This may be worse case, in that no purpose applied absorption or significant diffusion is assumed. Beyond that it is exactly the same modeling rational that, say, the Soren Bech study uses.
Please let me know what I am overlooking that has you so up-in-arms. Why am I making "false assumptions leading to false conclusions" or why it is a "false model that has little-to-nothing to do with reality"? Be specific, if you know more about the subject then I, then please share your thoughts on what a realistic model would be.
David S.
Last edited:
When people understands this they usually have a breaktrough in reproduction of high quality sound in small rooms.
Low freqs (below 1kHz) is exactly where dipoles should be used. I never recommend dipoles for treble.
- Elias
Low freqs (below 1kHz) is exactly where dipoles should be used. I never recommend dipoles for treble.
- Elias
You are "overlooking" the demonstrable fact that "prediction" based on your "theory" produces false results. For a start. There's more than one "reflection" from the front wall in my listening room, and none so pronounced as your "-5dB". Averaged together they are uncorrelated, and don't produce noticable comb filtering or "timbral changes" (to quote another of your claims). Maybe in a cinder-block-and-drywall dungeon they would, but not in any of the real-world rooms where I've listened to either my or other people's dipoles . . .
Can't answer that . . . armchair psychiatry is not my gig. Sounds like a "personal problem", though . . .
rather it is a problem of the industry, it's an old problem, well known:
When the only tool you have is a hammer everything starts to look like a nail. The more sophisticated the tool the graver the risk that its use can seriously skew priorities
Since we were discussing dipoles and I was voiceing a concern about extra full range energy that would bounce off the front wall, while others were saying "so what", I thought I'd do a reasonable simulation of it. Dipoles 5 ft in front of a wall were suggested so that would be a reflection delay of about11 ms. Rear output is out of phase (which sounded like it lowered all the resonance pitches when I did that in the simulation).
The only other variable would be the attenuation of the back bounce which is a combination of extra distance, actual radiation angle and wall reflectivity. I think the 5dB reduced reflection is realistic for a dipole in front of a non-absorbtive wall.
Cool Edit Pro
David S.
So your sound files are the sum of direct sound plus all reflections lumped together in a single channel? This is not what humans perceive. It would make things easier if such a simple model would be true but unfortunately sound perception is still a mystery. We know very little.
Some people even make money on selling this comb filter myth. See this video at 4:30. While it's true that comb filtering is a problem when recording sounds, two ears and a brain behave differently.
False results? For the scenario I describe I still assume the results are correct and haven't heard anything dissuasive. If you get multiple front wall bounces behind your speaker then you must have a different scenario than a single flat wall. Since you've measured the impulse response of your system, please post a copy so that we can see the actual reflections.
I didn't say my model exactly matches your particular case. I stand by it being reasonable and realistic. You may not like the implications but that doesn't mean you can trash the model without valid reason.
No cinder block required, no dungeon, just one smooth drywall wall, as found in any home in America.
"Timbral changes" was fairly obvious in listening to the sim.
Now, you can't turn the reflection on and off in your living room as I have in the simulation. Maybe you've just gotten used to the sound?
David S.
Yes, guess so. Sort of like the folks who learn to do on-axis frequency response measurements, and suddenly decide that "flat-on-axis" is all there is to it . . .
The problem is that Dave stepping on some "hallowed beliefs". Nothing that he has said is false or unrealistic, its simply a reasonable simplification for demonstrative purposes.
Its amazing how rude people can get when you question their belief system.
Ok, jump on me for awhile, I started this anti-dipole stuff!
Its amazing how rude people can get when you question their belief system.
Ok, jump on me for awhile, I started this anti-dipole stuff!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.