Wider soundstage for small rooms - fire your speakers backwards!

[whazzup]

Thanks to all the great content here and elsewhere, I'm gaining a better understanding how rooms affect sound (and how speakers are NOT MAGICAL ). And with the realization that reflected sound has such a big role (along with all the literature debating its absorption / diffusion etc), I was curious what happens if I got rid of the direct sound. So I flipped my speakers around.

And it actually doesn't sound too bad! I'm digging the 'bigger' sound, like my speakers are now at the walls, although yes, the highs are definitely reduced. Imaging is probably reduced too, but that's pretty hard to measure and the speaker placement is already bad, so....

First, my small untreated, not-to-scale room and a 2.1 setup (KEF 3005se + Definitive Tech sub).

Typical front facing speakers vs the flipped speakers, measured at sitting position.
Note that there's already EQ applied for up to 200hz.

The beginning of the high frequency dropoff is so precise at around 1.2khz. Or maybe it's further impacted by my monitor. My noise floor can be quite high, so I do boost my bass. After some trial and error, these are my EQ settings, with a 4db, 1.5khz high shelf filter boost for the lost high end.

The 4db boosted result (highlighted in yellow)

And this is after dropping all eq filters by -4db (from my understanding that boosting is generally not good for speaker longevity).

Sure, there are still some issues like in the 350hz region and the dip at 165hz, but the result is still very listenable compared to just the conventional front facing position. I'd like to believe now the sound is more 'airy' But the nice thing is now I have a convincing illusion (to my ears) that there's an actual band / singer right behind my monitor. In a small room. The speakers disappeared!

It has been an interesting experiment. I'm still listening to my music in this position and figuring out what to tweak along the way. Obviously for people who have their speakers right against the wall the result might not be so good.

My conclusion is it's important to understand the concepts, but also not to be too limited by what are best practices. When consuming music at least.

 

[MarsianC#]

Dr. Amar Bose smiles in heaven
Please add ETC graphs.

 

[dasdoing]

Dr. Amar Bose smiles in heaven

he is refering to Bose 901 btw
this is the backside!!!

Please add ETC graphs.

i would rather see step response / group delay

 

[whazzup]

I switched speakers to the Focal Bird satellites, with the idea that speakers with rising high end might help when they're flipped. All measurements done with the speakers flipped, mic at seated position.

@MarsianC# @dasdoing Unfortunately, my audio education hasn't been comprehensive enough to understand what ETC / step response / group delay mean, but I did the frequency sweeps. I have my crudely equalized subwoofer on when doing the tests, so please ignore anything below 100hz.

Here's the REW file as well, if anyone finds it useful.
https://app.box.com/s/gbo6e9uhj8p7ymilhzhjq1dsexkodvn2


RTA (moving mic) unequalized L/R speakers
Left speaker (highlighted in yellow) has the boosted response from the side wall

Equalized left (highlighted in yellow) / right / both channels response

Impulse response of right channel
Not quite sure if it's supposed to look like this....

RT60
157 hz is... strong resonance...?

 

[wwenze]

At this point it becomes philosophical whether you want to increase treble by 4dB because it is the nature of reflected sound to have less treble. On the other hand the woofer looks to be pretty omnidirectional up to 1.2kHz so some balancing is necessary.

Boosting does not hurt speakers, but it causes clipping in the digital domain.

 

[dasdoing]

you actualy posted ETC and step response in this graph; though ETC needs more zoom, and SR in normaly plotted in % scale.

however, the SR shows the expected: your frequencies are all arriving at diferent times (the IR window isn't even big enough to show the whole extension). that's actualy the problem of this approuch. omnidirectional frequencies will come via direct sound, the mids and highs arrive later via reflections.
I have no clue how this sounds like tough; important is that you like it

 

[whazzup]

you actualy posted ETC and step response in this graph; though ETC needs more zoom, and SR in normaly plotted in % scale.

however, the SR shows the expected: your frequencies are all arriving at diferent times (the IR window isn't even big enough to show the whole extension). that's actualy the problem of this approuch. omnidirectional frequencies will come via direct sound, the mids and highs arrive later via reflections.
I have no clue how this sounds like tough; important is that you like it

How is it that Rew can show such a clean graph but mine has overlapping elements?
https://www.roomeqwizard.com/help/help_en-GB/html/graph_impulse.html

 

[dasdoing]

How is it that Rew can show such a clean graph but mine has overlapping elements?
https://www.roomeqwizard.com/help/help_en-GB/html/graph_impulse.html

look at the bottom boxes. you overlapped 3 graphs. uncheck every box but ETC to see ETC only, for example

 

[whazzup]

look at the bottom boxes. you overlapped 3 graphs. uncheck every box but ETC to see ETC only, for example

Oh! Got it, thanks!

Ya, I do lose the sharp imaging definitely. But it's a cool effect, and amazing that I can use Rew to visualise that. Will continue to experiment

 

[ElNino]

I've played around with this a little in the past in an oddly-shaped room. If you're going to do this and your speakers have rear ports, I recommend trying using the port bungs (or rolled up socks) to plug the ports. Midrange leakage from the ports tends to be uneven, and if the ports are pointed towards your listening position, I found it contributed to long-term fatigue.

 

[whazzup]

I've played around with this a little in the past in an oddly-shaped room. If you're going to do this and your speakers have rear ports, I recommend trying using the port bungs (or rolled up socks) to plug the ports. Midrange leakage from the ports tends to be uneven, and if the ports are pointed towards your listening position, I found it contributed to long-term fatigue.

Thanks! That brings up another question: port plugging could apply to the sub as well right?

 

[ElNino]

Thanks! That brings up another question: port plugging could apply to the sub as well right?

It could, but I have no experience there -- my sub is sealed. Since subs are not run full range, I think it's less of a potential issue with subs, although if you're crossing high it might make sense to orient the sub so that the port is not facing your listening position.

 

[whazzup]

At this point it becomes philosophical whether you want to increase treble by 4dB because it is the nature of reflected sound to have less treble. On the other hand the woofer looks to be pretty omnidirectional up to 1.2kHz so some balancing is necessary.

Yes, I do notice the treble energy is not really dissipated even when flipped, probably the room is too small.

Boosting does not hurt speakers, but it causes clipping in the digital domain.

But clipping in the digital domain, I suppose it 'harms' the fidelity of the final signal, but shouldn't affect the speaker hardware then?
Guess was thinking increasing the eq to say, +12db, will result in harmful distortion as the driver tries to vibrate physically in a forced manner...

 

[dasdoing]

Oh! Got it, thanks!

Ya, I do lose the sharp imaging definitely. But it's a cool effect, and amazing that I can use Rew to visualise that. Will continue to experiment

go to the group delay tab, generate minimum phase, and checkbox the excess group delay plot. this will show you the delays you system is producing

 

[whazzup]

go to the group delay tab, generate minimum phase, and checkbox the excess group delay plot. this will show you the delays you system is producing

damn....it's a serious time sink...so much to learn, so little time...

 

[whazzup]

A comparison between R + sub channels, front facing and flipped. Actually the normal front facing speaker doesn't measure too different from the flipped speaker? So possible conclusion is most problems are caused by room?
https://app.box.com/s/p3yy46bekri6xsm7zxkkgfxnug6hnfkd

Flipped R speaker highlighted in yellow

 

[whazzup]

Following in the footsteps of bose 901, did a mix and match with a pair of Dali Menuet and flipped Focal Birds. And I must say I like the big sounding results in my dingy little room.

Click here to Download the REW file

A photo of how it looks on my desk....please ignore the leftmost speaker and no, it's not meant to be a conventional setup. Where's the fun in that??

Using the A/B inputs from my singular amp to drive the 2 pairs of speakers, so can only eq both speakers (on a channel) at once. Light purple graph is the summation of the L/R/sub channels.

EQ for the LR channels were a simple affair though.
Only used a low shelf @~300hz and a high pass @90hz, to knock down the rising low ends + my L channel's side wall gain + kill the sub 100hz stuff. Can see the original and equalized graphs below.

 

[tomtoo]

Cool! Nice posibilitys with the new technologis. Dsp the front and back firing speaker independent would be the next step. Measure and listening. Bose had dreamed of this possibilitys! And the most importend, have fun!

 

[whazzup]

Cool! Nice posibilitys with the new technologis. Dsp the front and back firing speaker independent would be the next step. Measure and listening. Bose had dreamed of this possibilitys! And the most importend, have fun!

Unlikely for now as I'll need to replace my current stereo amp. Haha... maybe will find a good deal in the future.