• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Bass arriving out of phase at the listening position

tmuikku

Member
Joined
May 27, 2022
Messages
73
Likes
72
Hi, tried to look info on this line of thought^, how sound source "feeds" or triggers a mode, but didn 't find anything to feed imagination :) I tried to reason over the issue and got to a line of thought similar to what theSuede wrote. When mode gets going thats it, and if either speaker is not in sync with it the situation happens as in this thread, direct sound of other speaker is trumped by the mode initiated by perhaps the other speaker and fed by both. When setup and room is symmetrical both sound sources feed modes similarly and such issue is absent? There would be peaks and nulls but either speaker response at listening spot would be similar to when both playing simultaneously.

If this is true then fixing, or at least normalizing the situation, would require identifying which mode it is and placing speakers symmetrical to the mode? This might make other issues. Well, anyway.

What do you think is happening?
 

theSuede

New Member
Joined
Sep 24, 2022
Messages
4
Likes
1
@daftcombo: Then you have a 40Hz resonance between the listening room and the general "outer space" outside that corridor.... That resonance could be either acoustical or mechanical (structural). No matter which, any brick wall filter can/will sound strange in some circumstances. Sorry, can't help you there. If you have the processing to do it, and you aren't dependent on group delay, try a linear phase FIR filter? A 48 or 96dB pole filter will have extreme group delay changes in the crossover region.
 

theSuede

New Member
Joined
Sep 24, 2022
Messages
4
Likes
1
@tmuikku: It's not that complicated really, a mode is roughly a planar wavefront, and it always has a major dimension and direction in space. A primary mode is between two surfaces, higher order modes can be between two corners. That flat planar wave is why it's a "mode", at least the way I think about it.
The phase lag of the mode will basically be 2x the distance between the source and the major first reflection of the mode. If the mode is weak compared to the source, then the summed phase will be close to the source phase, if the mode is strong compared to the source, the sum will be closer to the mode phase. The phase (group delay) of the source will depend on your distance from the source, the group delay of the mode will depend on where along the major axis of the mode you are - two different things.

The basics of the math behing the Klippel NFS is rougly evolved from that, find a complex matrix of delay values that when convolved with the measurement results from lots of different positions separate direct sound and reflected sound. It's a lot worse in error margins (due to numerical instabilities) compared to a real anechoic chamber, but more than good enough for almost everything we do.
 

gnarly

Senior Member
Joined
Jun 15, 2021
Messages
484
Likes
575
16 thousand taps, 44.1 kHz
Thx. That's what i use too, although at 48kHz, so I have slightly less frequency resolution than you.
16k taps is the limit that my processor allows (qsys).
Are you using PC convolution?

fwiw, i think a linear-phase high pass for subs is the wrong way to go. It's clearly a source of pre-ring in measurements, and i really think i hear wonky stuff sometimes down low.
So, i'm a believer in sub's high pass being IIR. For a while, I tried embedding an IIR high-pass into the sub's otherwise linear phase FIR file. That helped, but I think mainly due to the IIR high-pass simply having to be low order.
I also came to the realization that yes, I could imbed an IIR high-pass in the FIR filter, but it would still only have the freq resolution of the FIR filter, and not the resolution of a true IIR high-pass.
Nowadays, my subs FIR filter has no hpf, and I use an IIR high- pass before line signal goes to the FIR filter. I like this the best, because it is very easy to play with a separate IIR hpf in real-time to combat boominess in over the top recordings, and I get the theoretically infinite freq resolution of IIR. (and without any pre-ring potential).

Hope that all made sense...
 
OP
edechamps

edechamps

Addicted to Fun and Learning
Forum Donor
Joined
Nov 21, 2018
Messages
870
Likes
3,331
Location
London, United Kingdom
@NTK I am glad to report that, after fixing the absorption coefficient for wall impedance, the simulator provides results that are a significantly better match against reality:

fl.png
fr.png


It's clearly not perfect - for example the simulator completely missed the 115 Hz dip on the right channel, which might be related to the wardrobe next to the speaker. But to me this is already quite impressive given how naive the model is (same 0.2 absorption coefficient for all boundaries, no imaginary part for the acoustic wall impedance, etc). With a little luck, it might get even closer to ground truth with a couple of low-hanging fruit improvements to the way the boundaries are modelled.

But here's the real cherry on the cake: the simulator is now able to predict the phase issue that prompted this very thread!

sum.png


This is absolutely wonderful. I am steadily getting closer to the point where I can simulate various setups and produce results that are reliable enough to be useful.

The Jupyter notebook that produced the above is attached.
 

Attachments

  • roomsim.zip
    824.2 KB · Views: 15

Flyers_Fan

New Member
Joined
Nov 20, 2021
Messages
1
Likes
0
Any updates? Because I’m having kinda the same problem and not sure it is a room issue. I suspect that it has something to do with electricity. So it messes up with audio gear. Room response is just a result of that.
 
Top Bottom