R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 [2]  All   Go Down

Author Topic: 2 subs  (Read 8443 times)

seriousfun

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 329
Re: 2 subs
« Reply #15 on: August 13, 2008, 05:23:56 PM »

OTR-jkl wrote on Tue, 12 August 2008 21:01

...

How far apart should the subs be from each other (the room is about 12' wide)? How close to the wall should they be? How far from the corners? Is there some formula for figuring a starting point for this? Do they need to be time-aligned with the mains (by physically aligning the drivers)?

...


You can't predict the distance between the two subs. You have to experiment with placement, measuring and listening carefully. There is no real formula.

Sounds eminate from the subwoofer cabinet essentially omnidirectionally, and reflect from adjacent surfaces. It can have one adjacent surface (the floor), two (the floor and one wall), or three (the floor and two walls - a corner). Each additional reflecting surface will increase gain from the subwoofer, and each will probably engage more ringing at different frequencies. In some rooms, the placement with the fewest adjacent surfaces gives you the least-colored sound, and in other rooms the corner gives you the most gain with relatively uncolored sound.

Every sound travels at the speed of sound  Razz (well, the front of the compression, or rarefaction, moves at that speed), and a subwoofer needs to be in polarity with the main speakers as much as any speaker. Moving the subwoofer so it is perfectly time-aligned with the main speakers - the same distance from the subwoofer voicecoil to the listener as from a tweeter voicecoil - is an admirable goal, but that repositioning in many rooms will result in the excitment of different room modes, with more ringing at different frequencies - compromises often must be made, and flat frequency response should be the first goal.






Logged
doug osborne | my day job

OTR-jkl

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 869
Re: 2 subs
« Reply #16 on: August 13, 2008, 10:09:39 PM »

Until today, the mains have been forward of the subs somewhat (closer to the listening position) so I moved them back so that they are aligned with the subs and now about 1' or so from the front wall. Wow!....what a difference!! I'm now able to hear really low freqs quite clearly w/o any big build-ups of any bass freqs. It seems quite even actually (at least to my ear; an RTA may show otherwise).

For fun, I re-mastered a few songs from a recent project and was able to quickly make some adjustments that made big improvements. The translation is bigger, more open and the bottom is cleaner, tighter and yet more full.

I shouldn't be so surprised, but acoustics is voodoo. Sure makes a world of difference when you can hear stuff (as Bob O says over and over and over again....). I may still need some traps in the corners but its sounding really good so far. Its like, I don't want to touch anything cause its the best I've heard yet....
Logged
J Lowes ยท OTR Mastering
Professional Audio Production for Life
www.ShoutLife.com/OTRMastering

Sam Lord

  • Full Member
  • ***
  • Offline Offline
  • Posts: 139
Re: 2 subs
« Reply #17 on: August 14, 2008, 12:35:46 AM »

One sub can *never* allow time coherence, period.  Put the subs and every other driver in correct time alignment with the ear position.  That usually requires setting higher-freq drivers farther from the ears, depending on the crossover and the rise times of the drivers.  That's a big reason many tweeters now are recessed a half-inch or so into flared mounting plates.  Manufacturers should provide correct ear-to-driver distance for any 3-or-more-driver rigid loudspeaker, because *one and only one* distance is correct.  What, yours don't?**  Big surprise...

Unless you can adjust drivers singly or in pairs for time alignment, there will be *only one* sweet spot or, sadly, none at all.  You can do it yourself most quickly by using single-miked percussion samples, listening to one speaker channel only.  Adjust woofer-to-midrange first, then mid-to-tweeter.  Do separate subs vs. woofers last.  Pray that the design *has* a coherent point.

I urge that the coherence be gotten first, then the nodes be tackled within the time coherence constraint.  You can still move the speakers around, just keep the driver-ear distances correct wrt one another.  For example, tweeter-mid ear-distance offset = +x inches, mid-bass ear-distance offset = +y inches, and bass-sub ear-distance offset = +z inches.  All rooms ring, and you may have to compromise a little time accuracy.  But unless your room is just too small or square, you *can* maintain timing with acceptable nodes.  

** If you push them they might spill the beans.
Logged

bslobodian

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 51
Re: 2 subs
« Reply #18 on: August 16, 2008, 11:17:26 AM »

Bob Olhsson wrote on Wed, 13 August 2008 12:15

I screwed around with subs for twenty years. What finally worked for me translation-wise was placing the subs right next to the speakers, treating it as a full range speaker and dealing with the low frequency room issues.


Ditto here (except for the 20 years, I would modestly put on a good 10 years or so). I have Klein & Hummel 0300D with 0800 subs and the alignment procedure makes it quite clear that the integration of the subs with the satellites have a lot to do with phase at the crossover point.
There's also a very good paper about multiple subs that you will find here: http://www.harman.com/wp/pdf/multsubs.pdf

Bernard Slobodian
Logged

Samc

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1393
Re: 2 subs
« Reply #19 on: August 21, 2008, 03:52:37 AM »

Sam Lord wrote on Thu, 14 August 2008 05:35

That usually requires setting higher-freq drivers farther from the ears, depending on the crossover and the rise times of the drivers.  That's a big reason many tweeters now are recessed a half-inch or so into flared mounting plates.  Manufacturers should provide correct ear-to-driver distance for any 3-or-more-driver rigid loudspeaker, because *one and only one* distance is correct.  What, yours don't?**  Big surprise...

Care to explain this?

Quote:

Unless you can adjust drivers singly or in pairs for time alignment, there will be *only one* sweet spot or, sadly, none at all.  You can do it yourself most quickly by using single-miked percussion samples, listening to one speaker channel only.  Adjust woofer-to-midrange first, then mid-to-tweeter.  Do separate subs vs. woofers last.  Pray that the design *has* a coherent point.

And this?.....I understand the principle of coherence etc, but I'm not sure I understand what you're trying to say here.
Logged
Sam Clayton

Samc

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1393
Re: 2 subs
« Reply #20 on: August 21, 2008, 04:06:19 AM »

Bob Olhsson wrote on Wed, 13 August 2008 17:15

What finally worked for me translation-wise was placing the subs right next to the speakers, treating it as a full range speaker and dealing with the low frequency room issues.

Yeah, keeping the sound source as a single point/system works and makes the most sense to me, I never understood the principle of treating them as two different systems and separating the sound source.
Logged
Sam Clayton

Adam Dempsey

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 546
Re: 2 subs
« Reply #21 on: August 21, 2008, 07:29:03 AM »

seriousfun wrote on Thu, 14 August 2008 07:23

Every sound travels at the speed of sound  Razz

But only if it has no direction, otherwise we're talking velocity. At sea level. In a vacuum. Unless the speaker/source is travelling at the speed of light... Razz
Logged
Adam Dempsey
Jack the Bear's Deluxe Mastering
Facebook | twitter | MySpace

Add presence, or subtract absence?

seriousfun

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 329
Re: 2 subs
« Reply #22 on: August 21, 2008, 05:22:08 PM »

Samc wrote on Thu, 21 August 2008 00:52

Sam Lord wrote on Thu, 14 August 2008 05:35

...Manufacturers should provide correct ear-to-driver distance for any 3-or-more-driver rigid loudspeaker, because *one and only one* distance is correct.  What, yours don't?**  Big surprise...

Care to explain this?

Quote:

...  Adjust woofer-to-midrange first, then mid-to-tweeter.  Do separate subs vs. woofers last.  Pray that the design *has* a coherent point.

And this?.....I understand the principle of coherence etc, but I'm not sure I understand what you're trying to say here.


Quote:

...keeping the sound source as a single point/system works and makes the most sense to me, I never understood the principle of treating them as two different systems and separating the sound source.


The front of the air compression leaves the speaker, compressing and rarefacting the air between the speaker/transducer and your ear/transducer. Whether it is a low-frequency or high-frequency sound, all should leave the speaker at the same exact time. These take the same amount of time to reach your ears. The frequency is determined by the number of times the air is compressed and rarefacted, and loudness is determined by the amplitude of this action.

(This is not one continuous wave - it is like electricity where an electron does not travel from one end of a wire to the next but one electron transfers energy to the next electron and so-on)

Let's say that the tweeter is exciting the air from the back of its cone or dome, and a mid driver is sending lower-frequency sounds from the back of its cone. Unless these two drivers are physically aligned, common sounds will be smeared.

This would seem to say that multiple-concentric drivers or full-range planar speakers would be the perfect solution, but each of those designs has its own strengths and weaknesses.

If the woofer is two feet below the mid driver, sounds common to both of them will be smeared. If the woofer is in a separate box five feet away, common sounds will be smeared worse.

This would seem to say that digital delays could fix the smearing, but that has its own strengths and weaknesses.

To sum it up - no speaker design is perfect, and placing a subwoofer directly below a main speaker might not be better than placing it elsewhere.

Now - the other big problem with reproducing the bottom two octaves in a typical small room is: the room. Whether the woofer is concentric with the tweeter or five feet away, the room will happily screw up the bass frequency response - ringing due to reflections causing dips and peaks.

If you can find a place in the room to place a subwoofer where it returns flat frequency response sound to the listener, you've won most of the war already. Nine times out of ten, if you predetermine a subwoofer location, even if based on time coherence, the subwoofer will end up where it can't play flat.

IMO play the bottom two octaves from a speaker that's designed to do that job. And embrace the room, don't fight it - put that speaker where it plays flat frequency response.
Logged
doug osborne | my day job

Samc

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1393
Re: 2 subs
« Reply #23 on: August 21, 2008, 10:54:55 PM »

seriousfun wrote on Thu, 21 August 2008 22:22


If the woofer is two feet below the mid driver, sounds common to both of them will be smeared. If the woofer is in a separate box five feet away, common sounds will be smeared worse.

If this is true......

Quote:

To sum it up - no speaker design is perfect, and placing a subwoofer directly below a main speaker might not be better than placing it elsewhere.

How can this also be true?...... this situation (you describe) has nothing to do with the inherent problems of loudspeaker design, it is about correcting low frequency issues in the room.  This would also seem to be setting up perfect conditions for group-delay issues.


I'll quote Bob O again...Please note what he wrote about the low frequency issues:

Bob Olhsson wrote on Wed, 13 August 2008 17:15

I screwed around with subs for twenty years. What finally worked for me translation-wise was placing the subs right next to the speakers, treating it as a full range speaker and dealing with the low frequency room issues.

Logged
Sam Clayton

Sam Lord

  • Full Member
  • ***
  • Offline Offline
  • Posts: 139
Re: 2 subs
« Reply #24 on: August 22, 2008, 12:00:41 AM »

Samc wrote on Thu, 21 August 2008 03:52

Sam Lord wrote on Thu, 14 August 2008 05:35

That usually requires setting higher-freq drivers farther from the ears, depending on the crossover and the rise times of the drivers.  That's a big reason many tweeters now are recessed a half-inch or so into flared mounting plates.  Manufacturers should provide correct ear-to-driver distance for any 3-or-more-driver rigid loudspeaker, because *one and only one* distance is correct.  What, yours don't?**  Big surprise...

Care to explain this?


Samc wrote on Thu, 21 August 2008 03:52

Quote:

Unless you can adjust drivers singly or in pairs for time alignment, there will be *only one* sweet spot or, sadly, none at all.  You can do it yourself most quickly by using single-miked percussion samples, listening to one speaker channel only.  Adjust woofer-to-midrange first, then mid-to-tweeter.  Do separate subs vs. woofers last.  Pray that the design *has* a coherent point.

And this?.....I understand the principle of coherence etc, but I'm not sure I understand what you're trying to say here.


Let's look at the single-channel, 2-dimensional model (a side view) for a rigid 3-way speaker with a fixed crossover and vertically-spaced drivers.  The geometry of three sources, fixed in space wrt one another a which radiate sound to a specific source, in this case your ear, yields a certain result.  Assume that the midrange distance will be a little shorter than the tweeter distance, and the bass distance still shorter.  The critical point is that those differences are are *fixed* for a given speaker type.  I call those differences "absolute offsets," e.g. Otm and Otb.  Let's say you're a speaker designer and you've chosen your drivers and crossovers and recorded those absolute offsets.  You then draw a box with some slope of front baffle with three points on it to represent the infinite-distance acoustic center of those drivers.  Now draw the ear-to-tweeter line and the other two.  Is there any ear-to-tweeter distance and speaker angle at which the lines converge at the ear?  If yes, you have found your *only* time-coherent listening position for that speaker.  Now, if just one driver on the speaker can be adjusted to move forward or back, *and* the other two drivers can be tilted, you will have a large area in which your speakers can be placed wrt your ear yet still achieve time coherence.  The model for stereo simply adds the vertical dimension and uses the equation for a sphere, x^2+y^2+z^2=R^2.  My spreadsheet models for our adjusting speakers treated each speaker independently, having inputs for the distance between a given listener's ears.  Any deviation in floor flatness has to be accounted for.

Now you know a big reason why designers use active crossovers.  You can use them to adjust the signals for phase (always adding delay--no going backwards in time!) and get time coherence with a flat, vertical baffle.  Or  you can set them to default to converge *downwards* to your ears, as ATC does.  They expect most folks now to soffit-mount.  Thomas  Barefoot has his MM27's sub drivers mounted on the sides, so he has to slow the tweeter some and the mids a bit more, I expect.  My ATC20 passives (fixed, 2-way example) converge on a line normal to a point about at the top of the woofer.  There are more details, such as the fact that a driver's acoustic center varies some with distance.  A woofer converges to roughly the surround inner edge with distance at infinity, for example.  Does that help?  Sorry, it's much easier with drawings which I no longer have...

Hope this helps, best, Sam
Logged

Samc

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1393
Re: 2 subs
« Reply #25 on: August 22, 2008, 10:56:18 AM »

Sam Lord wrote on Fri, 22 August 2008 05:00


The critical point is that those differences are are *fixed* for a given speaker type.  I call those differences "absolute offsets," e.g. Otm and Otb.  Let's say you're a speaker designer and you've chosen your drivers and crossovers and recorded those absolute offsets.  You then draw a box with some slope of front baffle with three points on it to represent the infinite-distance acoustic center of those drivers.  Now draw the ear-to-tweeter line and the other two.  Is there any ear-to-tweeter distance and speaker angle at which the lines converge at the ear?  If yes, you have found your *only* time-coherent listening position for that speaker.  Now, if just one driver on the speaker can be adjusted to move forward or back, *and* the other two drivers can be tilted, you will have a large area in which your speakers can be placed wrt your ear yet still achieve time coherence.  The model for stereo simply adds the vertical dimension and uses the equation for a sphere, x^2+y^2+z^2=R^2.  My spreadsheet models for our adjusting speakers treated each speaker independently, having inputs for the distance between a given listener's ears.  Any deviation in floor flatness has to be accounted for.

I understand the principle you are trying to explain, although I'm not sure I agree with some of the details.  

After all there are many good sounding systems that do not have the voice coils of all the components perfectly 'lined-up', and there are some loudspeaker designers that do not consider this an absolute requirement.  I also know that there is a margin of 'misalignment' that is beyond human hearing, the misalignment must result in group delay that is beyond the threshold of hearing and my google search turned up this little tidbit:

"Given that the minimum audible group delay is claimed to be 1ms at 2kHz, that amounts to a physical driver displacement of 345mm - assuming the velocity of sound to be 345m/s (22
Logged
Sam Clayton

Sam Lord

  • Full Member
  • ***
  • Offline Offline
  • Posts: 139
Re: 2 subs
« Reply #26 on: August 22, 2008, 11:51:59 PM »

Samc wrote on Fri, 22 August 2008 10:56

Sam Lord wrote on Fri, 22 August 2008 05:00


The critical point is that those differences are are *fixed* for a given speaker type.  I call those differences "absolute offsets," e.g. Otm and Otb.  Let's say you're a speaker designer and you've chosen your drivers and crossovers and recorded those absolute offsets.  You then draw a box with some slope of front baffle with three points on it to represent the infinite-distance acoustic center of those drivers.  Now draw the ear-to-tweeter line and the other two.  Is there any ear-to-tweeter distance and speaker angle at which the lines converge at the ear?  If yes, you have found your *only* time-coherent listening position for that speaker.  Now, if just one driver on the speaker can be adjusted to move forward or back, *and* the other two drivers can be tilted, you will have a large area in which your speakers can be placed wrt your ear yet still achieve time coherence.  The model for stereo simply adds the vertical dimension and uses the equation for a sphere, x^2+y^2+z^2=R^2.  My spreadsheet models for our adjusting speakers treated each speaker independently, having inputs for the distance between a given listener's ears.  Any deviation in floor flatness has to be accounted for.

I understand the principle you are trying to explain, although I'm not sure I agree with some of the details.  

After all there are many good sounding systems that do not have the voice coils of all the components perfectly 'lined-up', and there are some loudspeaker designers that do not consider this an absolute requirement.  I also know that there is a margin of 'misalignment' that is beyond human hearing, the misalignment must result in group delay that is beyond the threshold of hearing and my google search turned up this little tidbit:

"Given that the minimum audible group delay is claimed to be 1ms at 2kHz, that amounts to a physical driver displacement of 345mm - assuming the velocity of sound to be 345m/s (22
Logged

seriousfun

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 329
Re: 2 subs
« Reply #27 on: August 23, 2008, 08:53:25 PM »

Samc wrote on Thu, 21 August 2008 19:54

seriousfun wrote on Thu, 21 August 2008 22:22


If the woofer is two feet below the mid driver, sounds common to both of them will be smeared. If the woofer is in a separate box five feet away, common sounds will be smeared worse.

If this is true......

Quote:

To sum it up - no speaker design is perfect, and placing a subwoofer directly below a main speaker might not be better than placing it elsewhere.

How can this also be true?...... this situation (you describe) has nothing to do with the inherent problems of loudspeaker design, it is about correcting low frequency issues in the room.  This would also seem to be setting up perfect conditions for group-delay issues.
...



Not contradictory. Moving a woofer far away from a tweeter might always smear common sounds, but the effects of arbitrarily placing a subwoofer below the speakers because the idea seems right to you, resulting in horrid frequency response, *might* be worse. I've placed essentially full-range speakers in rooms where they had to sit in order to present a proper image, but the room made their bass response horrid; adding a properly integrated subwoofer, and placing it where it delivered flat response, would have been a good solution (yes, with potential, predictable compromises everywhere). No speaker design is perfect.

A car might have wonderfully precise steering, but if it's not fast enough to drive on a highway, it's still not a very good car (stupid analogy alert).


Logged
doug osborne | my day job

Samc

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1393
Re: 2 subs
« Reply #28 on: August 24, 2008, 06:19:09 AM »

seriousfun wrote on Sun, 24 August 2008 01:53

Not contradictory. Moving a woofer far away from a tweeter might always smear common sounds, but the effects of arbitrarily placing a subwoofer below the speakers because the idea seems right to you, resulting in horrid frequency response, *might* be worse.

Doug, I would like to suggest that there is nothing "arbitrary" about placing the subwoofer close to the rest of the system to preserve coherency; science dictates it.

Quote:

I've placed essentially full-range speakers in rooms where they had to sit in order to present a proper image, but the room made their bass response horrid;

It would seem that the acoustics of the room was out of whack and the ROOM needed to be treated.

Quote:

adding a properly integrated subwoofer, and placing it where it delivered flat response, would have been a good solution (yes, with potential, predictable compromises everywhere).

Are you suggesting that splitting up the loudspeaker system, and placing the components around the room as a viable means to treating acoustic anomalies in a room?  Why not just treat the room, wouldn't that result in a more predictable situation?
Logged
Sam Clayton

Samc

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1393
Re: 2 subs
« Reply #29 on: August 24, 2008, 07:18:17 AM »

Sam Lord wrote on Sat, 23 August 2008 04:51

Hi Sam, thanks for a fine discussion.  

Thank you for the education.....

Quote:

That Googled group delay figure isn't close to correct.  Think, 1msec at 2kHz is 2 full cycles!!  No, if your 2k crossover point between mid and tweeter is off by as little as 1/16", or about 3.5 degrees out of phase, you can positively hear it.  Of course you need to be close to begin with, let's say well under 20 degrees of phase or around 0.25", otherwise you run the risk of missing a whole cycle when making adjustments to get coherent wavelaunch.  I haven't read AES papers on this, though one recent paper has postulated a minimum audibility threshold for interaural (stereo) time offset of around 2usec, which is about .026 inches distance for sound.  That's a very different parameter, but I think it bolsters the argument for tight inter-driver time alignment.
 
Lets assume that this is correct... I don't dispute your argument by the way, I'm trying to figure which way is up on this issue, anyway, how do you explain the many good sounding systems that do not abide by this theory?  Has anyone ever actually done any scientific tests on this topic?

Quote:

I spoke with Billy Woodman of ATC at the 123rd AES, and he thought my 1/16" was a little tight, but certainly believed that 1/8" or 10 degrees mid-to-tweeter phase error was audible.  Roger Quested, for all his accomplishments, doesn't get the importance of this.

I could hardly find two people who agreed on this subject too and hence my question about scientific tests.

Quote:

I would only add that tailoring amps for frequency ranges gets you almost nothing.  You can use poorer components on sub amps, but internal amp bandwidth really needs to cover the whole audible range to sound good.  

My "tailoring" statement refers only to choosing appropriate amplification type and power for each frequency band.


Quote:

These speakers are now about 65 grand, and IMHO not bettered by any.

What loudspeakers are these pray tell.
Logged
Sam Clayton
Pages: 1 [2]  All   Go Up