Double Bass Array experience

[ppataki]

I thought I would share my experience about Double Bass Array (DBA) with the ASR community
DBA is nothing but placing your 2 or 4 subs in the opposite end of your room, preferably in the middle of its width (see pictures in the referenced PDF) with the rear one(s) having opposite phase
This method shall cancel out the axial room modes generated by the length of the room
I got the idea from here:
https://www.nubert.de/downloads/dba.pdf

Here Nubert used 4 subs as you can see in the pictures - in my room I just used two SVS SB-1000 but it shall work the same way

Here is the frequency response at the listening position when the two subs are in the front below my Audio Nirvana speakers (blue) vs using DBA as described above (orange)
The reference volume was 80dB, both with 1/12 octave smoothing and using Dirac Live 3

The room mode above 50Hz is fixed nicely + I got an additional 2Hz of bass extension
So I am happy to confirm that the method indeed works as expected!

Step response curves:
The slight pre-ringing is caused by the mixed-phase crossover filters that I am using but it is totally not audible
In my opinion the orange one looks better here as well since it does not fall that much into the negative domain

A few comments:
- setting the delay for the rear sub did not work well as suggested in the PDF, instead I used RTA with REW to find the optimal delay value (I kept inceasing the delay until the frequency response curve showed the least dips at the listening position - I did not care about the peaks since Dirac can cut those but it cannot fix the dips)
- I am now in the process of building my own subs with Dayton Audio Ultimax 12" drivers; I will post the measurements here in approx. 8 weeks once they are finished

I hope this helps those considering the DBA method

 

[ppataki]

As promised above I am sharing my experience with DBA using my DIY Dayton Audio based subs

First of all my plan using Ultimax subs did not work out well: I ordered one that had a manufacturing defect (mechanical noise) then ordered the second one from a different source, same defect....I returned both
I ended up with a pair of Dayton Reference RSS315HF-4 in sealed boxes
I am thrilled by these subs - excellent speed and precision!

A quick comparison showing the IR at 25Hz measured at 1cm

SVS SB-1000 (my previous subs, now for sale)

The Dayton RSS315HF-4

The difference in speed is really remarkable!
@bennybbbx for your information, you might like this

So then I set up the DBA but unfortunately my living room does not have a perfect rectangular shape but one of the corners is 'cut' + I physically could not place the two subs on the same axis
I had to settle with something like this:

Here are the results with Dirac:

This is how it looks like if I use the two subs in stereo mode (placed under my Audio Nirvana full range speakers)

You can see that both the room modes at 53Hz and 89Hz are nicely filled using DBA + I have a better bass extension
So the theory still works!

However, I need to add that I switched back to stereo mode....
Probably due to the shape of my room the way I heard the bass was a bit weird - it felt like the bass was coming from everywhere in the room vs in stereo mode when it comes from the front only
I mean in DBA it was definitely coming from the front too but still I could feel a kind of post-ringing from around/the back of my head

Probably I heard this:

Clearly there is a second peak after 35ms which is not there is stereo mode:

Also the calculated delay for the back sub did not work. My room has a length of 7.2m; the corresponding delay would be 21ms
Using 21ms resulted in clearly hearing the two subs being misaligned. I had to set the delay by ear and ended up with 16.5ms

I would be curious to see this excercise in a perfect rectangular shaped room where the subs can be placed on the same axis - probably the results would be much better

Still I hope that this helps those considering DBA

 

[bennybbbx]

when in stereo the bass sound not so in front is more a bad mix. many use stereo spreader to make wider. slow speakers produce then not so dry mixes. but they make all frequency wider. this sound not good. i think best solution for this is use a stereo spreader that reduce the stereo width below 100 hz. there is a free gstereo effect. https://www.gvst.co.uk/beta.htm

you can also use the https://www.gvst.co.uk/gmonobass.htm

 

[Kvalsvoll]

Thank you for posting this, as I believe it can be helpful for others trying to get the perfect bass by measuring and tweaking positions and dsp settings.

So I am happy to confirm that the method indeed works as expected!

Except - it did not work! Because if it worked, you would not switch back because the sound actually was not so good.

There are reasons why it did not work, several factors at play here. You are on to some of them in you own analysis.

The good news is that it is likely it is possible to get reasonable bass in your room with 2 subwoofers. You may not be able to achieve a perfect measurement, but that does not mean you can not get acceptable sound. The easiest approach is to start out like you did - 2 subwoofers up front, behind the front speakers.

Only 1 small sound source placed on the floor like this is not sufficient to create a plane wave even at low frequencies, and then the major requirement for a DBA/SBA is no longer met. Then the cancellation will not work as intended, and the subwoofer in the back will be detectable as a separate sound source. You can see that is does not work by looking at what happens in time, and also if you measure the direction of the sound field, you will see that there is no plane wave travelling through the room from front to back.

 

[HerbertThePervert]

The concept is widely used in german home theatres.

Here are some really interesting papers (in german of course)...

http://hannover-hardcore.de/infinity_classics/!!!/Entwicklung SBA.pdf
http://hannover-hardcore.de/infinity_classics/!!!/Dokumentation SBA-01.pdf
http://hannover-hardcore.de/infinity_classics/!!!/Alternative DBA-Anordnungen.pdf
http://hannover-hardcore.de/infinity_classics/!!!/DBA im L-Raum.pdf


You will need a rectengular or L-shaped room for it to work properly. You need more then 2 Subwoofers, atleast if you want to get the desired effect up until higher frequencies. Depending on the dimensions 4 or even 8/9 Subwoofers (per wall!) are need to get a useable DBA up until 120hz (full LFE coverage).

You can calculate the highest possible frequency :

(c=343m/s, d= distance between woofer midpoints)

You usally optimise it via decay measurements, not frequency response.

BTW:
I am using heavily dampened Single Bass Array with four subs, it is the same concept (plane wave), but the absorbers are passive instead of active like with a DBA.

 

[ppataki]

@HerbertThePervert
Your SBA sounds very interesting! Can you please share some more information about it?
Thank you

 

[HerbertThePervert]

Hi ppataki,
it is the same arrangement of subwoofers as with a DBA just at the front wall and porous (or any other non active) absorbers in the back.
In my case that is four Subwoofers each at 1/4 room height and width:
http://www.poisonnuke.de/Grafiken/1-4_Positionen.jpg

If you use two subs they will be placed at 1/4 width and 1/2 height.

The author of the papers i posted created a post on Gearslutz aswell:
https://www.gearslutz.com/board/bas...rray-massive-damping-measurements-inside.html

In the back i plastered my whole wall full of porous absorbers with 20cm depth (+20cm air space behind it, which has the same effect as 40cm porous).
I will control the room decay with slats in front of it.
Depending on the room size you might need thicker absorbers with a lower "längenspezifischer Strömungswiderstand" maybe "flow resistance?", i dont not know the exact english term, sorry. The usual rockwool stuff works until 60cm thickness, after that there is no more improvement.

You can model that here: http://www.acousticmodelling.com/porous.php

Since the frequency response is valid for a whole row you can easily eq it to the wished reponse. There will be no seat to seat variations.
Furthermore there is no sound propagation loss, it is just as loud 2 or 5 meters from the front.

I will provide some pics of my installation later.

 

[ppataki]

@HerbertThePervert Thank you, that sounds really cool!

When talking about porous absorbers, if I wanted to use natural material would combed wool do it?
For example this one: https://www.hificollective.co.uk/catalog/014-0425-jantzen-combed-natural-wool.html

I was thinking about creating a corner trap filled with this....

 

[bennybbbx]

did you know http://en.audionet.de/apps/carma/ ? this is not a full automatic room correction system but it give best results for me and its free. I find it some days ago. it can do and it is suggest to do mono measure(with all speakers enable) in the low bass range. this is what i notice too. what really happens in bass with room modes can only hear when all speakers play. impossible to correct correct with a system as dirac or other.

I hope you get not shocked that it offer only 5 eq. but when you correct as they suggest in tutorial maybe you hear big enhancements. because it is a half automatic you can correct with 5 EQ then later measure the correctet and use another 5 eq for fine tune. but i need for my worse room on left side 5 and on right side 4 and there is 1 EQ include to boost the bass at 48 hz 7 db because of weak bass. you can then the eq settings copy to DSP

Use 1/3-octave-band-smoothing for general evaluation of the frequency response∑Use 1/12- or 1/24-octave-band-smoothing to determine room resonances. Don’t try to fill notches at low frequencies by narrow band peak filters. If the sound pressure is cancelled at these frequencies by room acoustics you can’t remove this cancellation by more energy input. Only changing the speaker position or the listening position would change the result basically.∑Please compare the determined EQ-settings with different octave band smoothing to achieve the best EQ setting. ∑For optimizing the sound quality at one listening position you should perform the measurement at some different microphone positions. We suggest shifting the microphone up to 40cm in all directions based on the listening position. Please compare these measurements and try to find an EQ setting which fits best to all listening positions. ∑EQ settings usually influence subjectively other frequencies besides the desired ones; this effect is well-known by audio engineers. So EQ settings at low frequencies not only have an effect on the bass ∑Identical EQ-Settings for left and right channel usually deliver the most balanced sound image if the speakers are positioned symmetrically. In special cases (e.g. a listening room with roof pitch, asymmetric speaker placement, L-shaped ground plain) a different EQ setting for left and right channel could be better. So please check both options and decide by yourself.

http://en.audionet.de/apps/carma/

 

[ppataki]

Thank you @bennybbbx , it looks great and apparently it supports multichannel too! (which is a must-have for me)
I will try it in a few days and will come back with the results
One question: do you know if it corrects the impulse response/step response as well or only the frequency response?

 

[bennybbbx]

when you change the EQ knobs then the measured FR, phase or groupdelay change in realtime. try only correct the strongest peaks for bass in mono measure and EQ and compare with your dirac correction if the carma do more clarity and better transients on your speakers too.
It is up to you to use a EQ and insert the settings. You can compare yourself with a phase linear EQ or not what sounds better. if you do not play realtime music then you can also do low frequency phaselinear. then all delay around 30 ms. dirac VST delay 6 ms. so it is not full phase linear EQ.

I think the phase is not so important of a speaker when you can not change the delay of tweeter to bass/mid. Did you only have wide band speaker or do you have a more way system to test too ?. there can check if the phase is good with moscilator from melda produktion. play the sine tone in note steps on 1 speaker and hear how it sounds. when delay is good then all notes around crossover frequncy (c6 is 1.047 khz, c7 2.093 khz)c8 = 4186) should locate from the speaker. if delay between tweeter to bass/mid is worse the sound seem come from other positions and not from speaker or is less or more louder or silenter. then can change the delay and i find a delay that work good on all tones on hearing positions on kali + rbt35 sr ribbon tweeter. kali tweeter sound on some notes very worse. it depend on hearing position too. so i think a good speaker should be able to change the delay between systems.

 

[bennybbbx]

Thank you @bennybbbx , it looks great and apparently it supports multichannel too! (which is a must-have for me)
I will try it in a few days and will come back with the results
One question: do you know if it corrects the impulse response/step response as well or only the frequency response?

here can see wy automatic correction fail. i correct room node but after measure with the EQ corrections there happen a large gap after the 2. room node. there happen some things in rooms that can not calculate. see in screenshots. the darker lines show the Fr that is measure without EQ. they brighter lines show how it should look after correction. and the 2. image is the measure with that eq settings in dsp. much diffrence. this is also wenn do dual mono measure

 

[audio2design]

@ppataki ,


In your comment about speed, I noted that the SVS sub appears to go much deep than the Dayton before rolling off. I could be mistaken, but wouldn't that impact how you measured speed?

 

[ppataki]

Hi @audio2design
You are right, but then I used a low shelf to compensate for that and the Dayton was still much faster:

This is the FR of the SVS vs Dayton

The Dayton indeed has a lower Q

Let's see what happens if I compensate the Dayton with a low shelf filter
Now this is the FR of the SVS vs the compensated Dayton:

Let's see what happens with the speed of the Dayton:

But let's see again the SVS

38ms vs 60ms to reach down to 20% decay

 

[audio2design]

@ppataki, the SVS has a ton of energy <20Hz compared to the Dayton and that will not settle out quick, but may be inaudible (but you can feel it).

The rise time of the Dayton looks a lot slower than the SVS. The Dayton may have a high pass which means the effective excitation at the driver is not the same.

 

[ppataki]

The rise time from 10% to 100% for the SVS is 13.5ms, for the Dayton it is 16.6ms - way less of a difference vs in the decay time
How can you tell from the measurement that the SVS has way more energy?
Since I have both I can tell by ear also that the SVS sounds muddy and slow compared to the Dayton, hence I replaced them

 

[Astrozombie]

I've always been curious about the differences, on the high-end you have Dual subs on opposite ends in one enclosure, dual woofers in one enclosure, then there's always stacking them on top of each other. Though it'll be a while before I can afford something like that with 15'' or 18''

 

[Flaesh]

Probably due to the shape of my room the way I heard the bass was a bit weird - it felt like the bass was coming from everywhere in the room vs in stereo mode when it comes from the front only
I mean in DBA it was definitely coming from the front too but still I could feel a kind of post-ringing from around/the back of my head

May be due to the room shape.. Or unperfect tuning.. Normal DBA poduce definitely frontal bass.
To be precise, your setup (and any with subs on the floor) is\was not DBA\CABS.

 

[Cbdb2]

@HerbertThePervert Thank you, that sounds really cool!

When talking about porous absorbers, if I wanted to use natural material would combed wool do it?
For example this one: https://www.hificollective.co.uk/catalog/014-0425-jantzen-combed-natural-wool.html

I was thinking about creating a corner trap filled with this....

Rockwool is better, its heavier a lot denser and cheaper.

https://www.homedepot.ca/product/rockwool-safe-n-sound-16-inch-o-c-for-2x4-wood-studs/1000421954

A lot cheaper, $2 a pound instead of $15.

 

[Razorhelm]

for those who have tried this what spl level are the rear subs compared to the front ones? i assume you have to turn them down a little.