Speaker/Room Measurement Witchcraftery

[Elephantears]

Oh no! The problem is that for us audiphile Princesses, there is always a pea under the layer of mattresses. We just have to learn to sleep. The heightened consciousness of a period of measuring is bound to unearth a few peas.

On the other hand, oh for a monitor that gets it just right in the presence region. I could take upper mid emphasis if I had that. The REF 3A Reflektor sounded like it was getting it righ to me, but at a price.

 

[Purité Audio]

I would be tempted to sort out the 35dB +/- FR before considering loudspeaker change.
Keith

 

[Elephantears]

35dB? You have a funny way of reading other people's graphs Keith. And from someone who deliberately obscures the measurements you publish (screengrabs with the left colum missed out, smoothing applied), I think its poor form to falsify/exagerate the evidence of someone else's.

 

[Ayya Khema]

I would be tempted to sort out the 35dB +/- FR before considering loudspeaker change.
Keith

if we look at the 1khz as the reference, S-man FR is +-20db.
however, this is due to his 30hz peak and how high is sub is set compared his mains.

S-man, your subwoofer is about 10db too high. your subwoofer should be around 70db on the graph, not 80db.

as for the upper mid forwardness, I suggest using no toe in on the kef ls50. measurements on the kef ls50 off axis response and on axis show that the kef ls50 is 3 db too high between 2khz up to 6khz when listend on axis, when listened off axis of 30 degree, the rise in that same region flattened and becomes neutral. even in some case, some toe out would be beneficial with the kef ls50. id start with no toe in at all, firing straight into the room.

 

[S-Man]

As I keep saying... the "horrific" peak and trough is the room. The ear "expects" this and in combination with >95% of program material having little low or no output at that frequency it doesn't cause anthing like the trouble the graph would leave you to believe.
When I adjust the ZRB level there is a very clear sweet spot where it sounds right - like baby bear's porridge, not too lumpy and not too sloppy .

The LS50s are firing straight.
The frustrating thing with the LS50s is that they can sound amazing. However they can also sound too "in yer face". Whether this is high resolution or metallic colouration is open to interpretation. My CAOW1s do not have quite the same resolving power in the midrange, but they sound good with a much wider range of recordings... a compromise I find better for long term satisfaction.

 

[Elephantears]

As I keep saying... the "horrific" peak and trough is the room.

I don't know why people don't see this as obvious. Using REW to measure speakers in my room (14'8" x 12') I get a big 40Hz peak whatever speakers I use, even if I bring in the little speakers I use in my office, which have no bass. The difference in all the graphs I've made is between big peaks and humungous peaks (in between sounds best).

 

[Purité Audio]

35dB? You have a funny way of reading other people's graphs Keith. And from someone who deliberately obscures the measurements you publish (screengrabs with the left colum missed out, smoothing applied), I think its poor form to falsify/exagerate the evidence of someone else's.

You are quite correct, only 30dB from 20Hz to 20kHz.
S-man I meant to ask you about the infrasonic component ?
Keith

 

[Ayya Khema]

As I keep saying... the "horrific" peak and trough is the room.

so you will tell me you have a room mode that gives you a straight FR from 10hz to 35hz? of +12 db compared to 100hz? it does not make sense. even if its such a case which I highly doubt that its a room mode, then enjoy! I sure couldnt live with 12db boost from 10hz to 35hz but if you like it, thats all that matters.

can you send measurements, taken at the listening position, at the same volume, of 1- your subwoofer measured by itself 2- the ls50 with the highpass 3- the combination of ls50 and subs 4- your mains + sub out of phase. this should take 5 minutes of your time.

 

[ToTo Man]

I hope I'm not thread-hijacking here if I upload some of my own graphs instead of starting yet another thread on room measurement / sub integration. Hopefully this will not only help me but also others pursuing a similar endeavour.

This is how my speakers measure from my listening position (not my exact listening position but 25cm lower as I did not want to disturb the height of the microphone which is currently set precisely at tweeter height). Note: Ignore the big dip around 300Hz-450Hz, this disappears when the mic is raised to my correct listening height.

Comments are welcomed on my sub-70Hz EQ'd frequency response and how its balance is compared to the rest of the spectrum (it sounds just about right to my ears but I'm willing to experiment if you think it can be improved upon!).

First off this is the REW Simulation of my room with mains and subs running. I've inputted all data as accurately as possible, except for the subs' XO frequency (this should be 25Hz but the lowest the Simulation can model is 40Hz. My room is 4.2m x 3.8m x 3.25m and is on the first floor of a Victorian house with suspended wooden floors.

Below are my actual measurements.

Graph 1: Red = left speaker with no sub and no parametric EQ, Blue = left speaker with left sub and no parametric EQ, Green = left speaker with left sub plus parametric EQ.

Graph 2: Red = right speaker with no sub and no parametric EQ, Blue = right speaker with right sub and no parametric EQ, Green = right speaker with right sub plus parametric EQ.

Graph 3: Red = Stereo speakers with stereo subs and no parametric EQ, Blue = Stereo speakers with stereo subs plus parametric EQ.

Graph 4: Red = Min sub XO setting (25Hz @ 24dB/oct), Green = Max sub XO setting (120Hz @ 24dB/oct), Blue = Mid XO setting (??? Hz @ 24dB/oct).

As I run my main speakers full range with no high-pass filter, I have set the subs to the lowest XO setting possible (25Hz @ 24dB/oct). This minimises the overlap, minimises the additional excitement of my 43Hz room mode, and therefore minimises the amount of parametric EQ I need to apply to balance the overall system response (the parametric EQ is system wide so affects both the subs and the mains).

I wonder if I need to set my BK XXLS400 subs louder to raise the level from 20Hz to 40Hz? I'm still fairly new to subs so don't know how much they can be 'cranked' before bottoming out (on a scale of 7 o'clock being minimum gain and 5 o'clock being maximum gain, I've currently got them set to 2 o'clock). With respect to the above graph showing the effect of the different crossover points, perhaps a better solution is to raise the subs' XO frequency and then apply a deeper notch filter on the parametric EQ at 43Hz?

 

[S-Man]

I hope I'm not thread-hijacking here if I upload some of my own graphs instead of starting yet another thread on room measurement / sub integration. Hopefully this will not only help me but also others pursuing a similar endeavour.

Not at all. You are most welcome.
I think somebody else's measurements will add a lot of extra info to the debate.

S-man I meant to ask you about the infrasonic component ?
Keith

The ZRBs are set to a 1st order rolloff at 10Hz, so they would still have a significant ouptut at VLF if there was any signal there to amplify.
The reasone for doing this is to minimie group delay at audible frequencies - see post #75. I suspect there is also some rolloff of the 2i2 headphone output included in the results.

 

[MattSPL]

Not sure if I missed it or not in the thread, but did you measure just your mains to see if you have that same 40hz dip, or do they not go that low?

What kind of music do you listen to as I find lots of music has 20 and 30hz (some even sub 10hz) content and definitely plenty of content in the 35 - 55hz range.

 

[S-Man]

Take a look at post #76. That's the LS50s run full range with the port open.

 

[S-Man]

Oh no! The problem is that for us audiphile Princesses, there is always a pea under the layer of mattresses. We just have to learn to sleep. The heightened consciousness of a period of measuring is bound to unearth a few peas.

I would suggest it's not a pea but something more metallic like a ball bearing

 

[March Audio]

Hi S-Man

Thanks for sending the REW files through, they are very interesting. Just to say I have not looked at the rest of this thread, so excuse me if I go over old ground. I have had a very quick look at the data and for the moment I will just make a few comments on the things that stand out to me, knowing nothing about your room/set up.

OK the frequency response shows modes at about 30, 60 and 90 Hz suggesting a major room dimension is about 5.7m. That cancellation dip is pretty nasty.

The waterfall confirms those modes are pretty nasty, the 30 and 60Hz are ringing for a very long time. The scale on the waterfall is 1 second! Suggests there to be very little in the way of low frequency absorption.

The RT60 plot confirms this showing the 30Hz mode taking 1.1 seconds to fall 60dB. However above the schroeder frequency its well behaved and in the target area I would aim for, suggesting at the higher frequencies appropriate diffusion and absorption. The room shouldnt sound too "live". Its OK and normal for the low frequency decay time to rise over the higher frequencies but not that much !

The impulse response shows a strong reflection at about 3ms which is at about 1 m distance. Any hard objects that close to the speakers? IIRC you should be aiming to have all reflections -20dB down within the first 10 ms.

So on face value I would say certain music that excites these modes is going to sound boomy and unclear at those low frequencies.

What is the sub XO frequency?

What are the room dimensions?

 

[S-Man]

Interesting. Thanks for all the analysis!

Length = 5.65m
Width = 4.25m
Height = 2.3m

L speaker is 60 cm from front wall and 100cm from the side wall (cause of the reflection you note?)
R speaker is 60 cm from front wall and 90cm from the side wall (cause of the reflection you note?)
Listening position is 63% of the length from the front wall (which is most likely the cause of the cancellation)

The back wall has a rectangular bay window about 50cm deep (included in the 5.65m). The curtains were open for the measurements but are usually closed for evening listening sessions.
Furnishing is about normal for a UK room, 2 sofas, 1 uphostered chair, a few bit of wooden furniture and carpets.

The speakers sound much better at my friend's house. He has a much larger room and the speakers are much further from the walls.However the null can still be heard at the listening position approx 2/3 down the room. What he doesn't get though is the 30Hz mode. Although this is audible in my room, it's not that unpleasant and not that many tracks excite it.

Not sure what I can do to improve the acoustics. The speaker positions have been carefully tuned by ear (and sig gen), pulling them further out helps but is not domestically acceptable. I also tried corner and wall loading for the subs (ala F Toole) but this is subjectively not as good, even though it meaures better. I suspect this is due to messing up the time alignment

 

[Purité Audio]

Play with your seating position in REWs ‘simulation’, inputting your room’s dimensions here suggest that sitting nearer the rear wall would really help with the cancellation.
You can use a Mini DSP processor to integrate your subs, including any delay and use it’s ‘plug-in’ to reduce your peaks.
Keith

 

[Sibbers]

For me, getting the speakers as close to the rear wall helped by moving the speaker interference higher up the range and allowing me to treat with panels on the rear wall. Also sitting forward helped, but thats not comfortable. For a long time I really wanted more space for my speakers, but less became the best compromise for me. It's a dark art and moving the speakers closer to the wall felt like sacrilege.

 

[Purité Audio]

The speakers should be close to the wall, otherwise you introduce cancellations.
Keith

 

[March Audio]

Interesting. Thanks for all the analysis!

Length = 5.65m
Width = 4.25m
Height = 2.3m

L speaker is 60 cm from front wall and 100cm from the side wall (cause of the reflection you note?)
R speaker is 60 cm from front wall and 90cm from the side wall (cause of the reflection you note?)
Listening position is 63% of the length from the front wall (which is most likely the cause of the cancellation)

The back wall has a rectangular bay window about 50cm deep (included in the 5.65m). The curtains were open for the measurements but are usually closed for evening listening sessions.
Furnishing is about normal for a UK room, 2 sofas, 1 uphostered chair, a few bit of wooden furniture and carpets.

The speakers sound much better at my friend's house. He has a much larger room and the speakers are much further from the walls.However the null can still be heard at the listening position approx 2/3 down the room. What he doesn't get though is the 30Hz mode. Although this is audible in my room, it's not that unpleasant and not that many tracks excite it.

Not sure what I can do to improve the acoustics. The speaker positions have been carefully tuned by ear (and sig gen), pulling them further out helps but is not domestically acceptable. I also tried corner and wall loading for the subs (ala F Toole) but this is subjectively not as good, even though it meaures better. I suspect this is due to messing up the time alignment

As Keith says putting the speakers closer to the wall will help the cancellation, also adjusting the listening position if possible.

Treatments are not really going to help with the 30Hz mode, too low for anything domestically acceptable. They can help with the 60 hz I used Vicoustic products to good effect. Their varibass might work for this although I haven't personally tried it.

https://www.vicoustic.com/product/vari-bass

https://cdn.starwebserver.se/shops/vicousticbeta/files/vari-bass.pdf?_=1489405340

Otherwise if treatments are not domestically acceptable,

I use 4 subs which was very effective at minimising mode problems and improving bass evenness around the room.

As Keith suggests minidsp processor on the sub to EQ the modes. It should give a significant improvement to the tightness of the bass. I wouldn't try to flatten the modes completely, you will probably perceive that as too anemic. Dirac also works well.

Oh regarding that reflection you will need absorption and diffusion at that reflection point on the wall. Again I used Vicoustic panels.

https://www.vicoustic.com/product/flexi-wood-a50

https://www.vicoustic.com/product/cinema-round-premium

 

[S-Man]

I have tried the speakers closer to the wall. The result is less clarity in the midrange. The CAOW1s have full BSC and I think this messes up the balance if they are too close to the wall.
I can sit further back, but prefer the subjective balance at the usual listening position.

I don't want to use DSP on the bass because I don't want to introduce any delays. The speakers are effectively a 3-way with correct time alignment from mid to bass.

Time accuracy matters more than amplitude accuracy at these low frequencies, in my experience.

Even at the room mode frequency the system doesn't sound boomy, it just sounds a bit louder than it ought to at the very bottom end.
As a point of reference, a pair of Kii3s sounded significantly less extended, tight and tuneful in the bass than my speakers (in the same room/position).