Multiple Small Subs - Geddes Approach

Hi Paul,

You know I have never thought about it in terms of cycles, but more like time. The window should be about 4 ms. ( rough guess) so I would say about 4 wavelengths. 4-5 sounds about right. But I don't have any real data on that. But two seems too short and 10 wavelengths clearly far too long.

I once published a scale of gammatone filters which very accurately match the ears. I posted the impulse responses of each band. I don't remember the scale but I clearly remember how short they got, almost a a/f type thing, where a is TBD but has to match that chart.

Go to Wikipedia and you can find the filter. Clearly by five cycles you get 95% of the filter.
 
I think impulse time-alignment of sub(s) and main speakers may be more important for sound quality than an 'even' bass distribution across the room at all cost. Multiple subs will make impulse time-alignment more difficult and restrict the sweet spot. :cool:

"We spend a great deal of time designing, testing, refining speaker system. Then, as an industry, we just hand big sonic hammer in the form of a subwoofer to a customer and say 'here, you give it a try.'"

I am paraphrasing Andrew Jones discussing the foibles of handing consumers subs. That is the reason Elac uses a phone app to assist in time alignment.



When I first setup my two subs I used REW to make measurements and adjust delay on my NU1000dsp to time align the signal at the listening position. Now this is handled via Dirac.

Time alignment is THE primary function of Dirac as expressed by one of their engineers to me while at Axpona this year. Using multiple measurement locations and making one measurement per speaker per location Dirac is able to create a sonic three dimensional image of direct and reflected sound.

This allows them to create filters for time correction; EQ is just a nice add on.
 
Time alignment is THE primary function of Dirac as expressed by one of their engineers to me while at Axpona this year. Using multiple measurement locations and making one measurement per speaker per location Dirac is able to create a sonic three dimensional image of direct and reflected sound.

This allows them to create filters for time correction; EQ is just a nice add on.

Sure this works great at HF, but how does one determine the arrival time at LFs (I hope not from a broadband impulse response.), physical distance input would work. But does the arrival time make any difference at LF? I would claim not.

I am sure that Dirac has some handle on things at LF, but I doubt that its arrival time in the lead.
 
Sure this works great at HF, but how does one determine the arrival time at LFs (I hope not from a broadband impulse response.), physical distance input would work. But does the arrival time make any difference at LF? I would claim not.

I am sure that Dirac has some handle on things at LF, but I doubt that its arrival time in the lead.

Following is what I would imagine.

- Dirac is built in to the NAD
- Dirac application communicates with the NAD
- Dirac application initiates a frequency sweep via NAD thus able to measure the time from initiation to time of response (sound) reaching the microphone.
- Sound is recorded via included calibrated microphone attached to NAD.

If speaker A is 11' away, B is 14' away, and the sub is 19' away, Dirac will create appropriate delays to deliver the signal from each source coherently to the listening position.

Dirac will pickup on phase/delay issues within a crossover network on a speaker - if I recall correctly.

As the engineer said, Dirac does most of its heavy lifting in the time domain.
 
While I would prefer a large space for bass and you would prefer outdoors, we are both stuck with the fact that a small space in our homes is the only choice that we have. So make the best of it. Learn to think of Bass in a small room as unlike any other situation in sound reproduction. It takes some time to get past the "intuition" that comes from the free field experience, but it is essential to do so because otherwise you will never get the bass right.

Yep, we do have our preferences, huh? And may they all be equally blessed...

I guess, having so strongly heard what I like outdoors, it's kinda hard to care about in..
Oh well, :)
 
Since it doesn't matter, then sure stick with your way.

As you're showing me, it doesn't matter indoors.

But outdoors, ah yes, it matters ;)

My understanding of the software I use to locate distance from broadband (raw) sub response, is that it looks to determine timing where the belly of the inevitably concave phase curve lies around zero, which supposedly also produces the highest integrated impulse response.

Do you see any issues with this technique for determining timing to acoustic origin?
 
Ok, as fortuitous timing would have it,
I got to play with some beginning multiple subs trials.

The thread says 'Multiple Small Subs'...hope big subs work too:)
Have 4 double 18"s in the room to play with. Normally I use only 1, or maybe two when listening in stereo.

Anyway, I tried using 3 spread around, 1 of them underneath a mono synergy. Quickly checked/matched timing, polarity, levels at a listening zone.
All 3 low-passed steeply at 100Hz, linear phase LR 72dB/oct.

Sustained bass notes seemed to do well.
Bass transients not so well.
There was nothing I could to keep from hearing separate subs on transients like kick drum etc.....short of turning all but one down to the point of no contribution.

It's got me thinking...exactly how do we characterize the frequency content of the kick that comes out of a sub. I can watch the cone jump...a single jump...what the heck frequency is it, especially when steeply low passed?

And how does the multiple sub approach deal with such transients?
When they create a strong sense/source of localization?

Are big subs causing the problem here?
Damn I hope not cause I really like strong clean transients :D
Very beginning attempt, pls don't shoot me for the fledgling observation..
 
It's impossible to explain what you perceive.

But here is what I know. If you took your same test and looked at the impulse response you would find that you could not tell when one sub arrived from any other sub.

And what you did is far from what I suggest. I do not use a single LP on the subs and without measurements you really have no idea what you have.
 
As Earl says, odd if you are hearing transients localized. Might be distortion in your subs producing notes that are high enough to be localizable. Lots of distortion in subs, even the best (except for Kipschorns and motional feedback).

But some of us who have experimented with time delay for subs note a tiny improvement in something that could be called transient response when the time delay is corrected. Maybe you are hearing the difference when some sub note arrives later than other sub notes.

Somewhere I posted my results with 14 sub positions in my room. Major differences. You need to know the FR for positions in order to choose the positions that help the most together or you might make it worse.

No disrespect to Mark100, but nobody's ear is good enough to analyze bass FRs. For some music, a sub with a peak at 50 Hz might sound more bass than one with very low extension but flat. Lots of folks mistake big boost for bass guitar at 80 Hz for superior low bass.

B.
 
Last edited:
As Earl says, odd if you are hearing transients localized. Might be distortion in your subs producing notes that are high enough to be localizable. Lots of distortion in subs, even the best (except for Kipschorns and motional feedback).

But some of us who have experimented with time delay for subs note a tiny improvement in something that could be called transient response when the time delay is corrected. Maybe you are hearing the difference when some sub note arrives later than other sub notes.

Somewhere I posted my results with 14 sub positions in my room. Major differences. You need to know the FR for positions in order to choose the positions that help the most together or you might make it worse.

No disrespect to Mark100, but nobody's ear is good enough to analyze bass FRs. For some music, a sub with a peak at 50 Hz might sound more bass than one with very low extension but flat. Lots of folks mistake big boost for bass guitar at 80 Hz for superior low bass.

B.

I agree, no one could analyze bass by ear for a 50Hz peak.

I'm strictly talking about major transients like a kick drum or such....where the single pulse of the sub is part of the transient response.
They are the only time I can clearly hear separate sources....even after setting timings equal at listening position.

And I really doubt distortion is an issue. Using very clean bms 18n862's at levels far below where they begin to distort. Check out this CEA-2010 test I made of them and a Labhorn. Labhorn CEA-2010 distortion

Earl, what's your measurement procedure?
I'm guessing RTA with some time alignment, as I can't see how a transfer from multiple sources could be useful at all. ??
 
...I'm strictly talking about major transients like a kick drum or such....where the single pulse of the sub is part of the transient response.
They are the only time I can clearly hear separate sources....even after setting timings equal at listening position....
Given Fletcher-Munson perception, even modest of distortion at 50 Hz would be present and audible in the sub output at 150 Hz. There may be a peak in the kind of drum strike you are talking about that produces audible localizable tones.

But mostly, a kick drum is nothing like anybody would call a transient. Snare drum and drums with wood sticks have transients to be savoured. And best of all, the cowbell. (Trying to remember if I downloaded my cowbell recording?) If you can play a good cowbell (with sticks or on the cow), you have good transient performance.

Although arm-chair theorists might think otherwise, folks with X0 (sharp slope) as high as 130 Hz generally report no sub localization on music. I have no doubt there would be localization if you turned the mids off. And of course, localization is absolutely perfect all the time when not tested blind.
 
You can never have too much cowbell....

RTA is from the eighties i would say :)
Why wouldn't a transfer function suffice?

There are multiple possible methods, they boil down to:
- just run all subs at unity gain and only apply global EQ
- EQ each sub individually, and then apply global EQ
- use mso (i don't know of an alternative) to measure each sub's individual response and model the sum (and EQ/delay/gain/other filters applied to each sub as you see fit). This last approach labor intensive but is especially useful for multiple positions optimization.
 
Earl, what's your measurement procedure?
I'm guessing RTA with some time alignment, as I can't see how a transfer from multiple sources could be useful at all. ??

For LF only steady state has any meaning. So I just do steady state measurements at a few locations up to about 200 Hz. In my case I prefer to use an FFT analyzer for steady state - I use SpectraPlus.

If you mean a transfer function, that is precisely what I use and the "transfer from multiple sources" is exactly what's required. No time alignment is used, it's unnecessary.

Seems to me you are still thinking NM not QM. What you hear is the superposition of all sources.
 
- EQ each sub individually, and then apply global EQ
This is what I do.

- use mso (i don't know of an alternative) to measure each sub's individual response and model the sum (and EQ/delay/gain/other filters applied to each sub as you see fit). This last approach labor intensive but is especially useful for multiple positions optimization.

I tried MSO and it just didn't work for me. I just did EQ in real time and found a better solution that way with much less effort. The idea of MSO is good, but I think the implementation was lacking.