AVS Forum banner

861 - 880 of 2327 Posts

·
Registered
Joined
·
1,774 Posts
Given the number of subs and listening positions you're considering, I'd recommend first doing a kind of trial run, with fewer subs and maybe three listening positions. In that way, you can gain some experience without the risk of causing an error that might invalidate a large number of measurements (78 in your case, and that's without any main speaker measurements).
In fact, I'd recommend simplifying things even more when starting out. Try making just three measurements at a given listening position.

  1. Frequency response of sub A
  2. Frequency response of sub B
  3. Frequency response of subs A and B together

Import these three measurements into MSO. Configure MSO to take the sum of the first two by making them a measurement group. Compare this computed sum to the third measurement above. If they don't match, something is likely wrong with the measurements.
 

·
Registered
Joined
·
235 Posts
In fact, I'd recommend simplifying things even more when starting out. Try making just three measurements at a given listening position.
I'd also strongly suggest done a simple trial run. The first time I used MSO I didn't realize I must press the key on my iNuke dsp windows software, vs using my mouse to click to the next value, and therefore got messed up results. The next time I did a simple test run and the calculated/optimized MSO frequency response was an incredibly close match to the after REW measurment.
 
  • Like
Reactions: andyc56

·
Registered
Joined
·
44 Posts
Thank you Andy. My setup is the Dayton EMM-6 to Audio Buddy mic preamp to 3.5mm mic input on computer. Output from computer is from soundcard's 3.5mm stereo speaker jack to aux analog inputs on processor. Processor splits signal to mains and sub where I have crossover and speaker distances set. Sub output from processor is routed to symetrix sound processor for individual delay, PEQ and gain for each sub.

I read somewhere that acoustic time reference is not reliable on subs, so I believe I need to use the loopback method. Correct?

So, are saying is that MSO does not attempt to adjust delays so that sound arrives from each speaker at a listening position at the same time, but rather adjust delays so that the waves are phase aligned, recognizing that in some listening positions the sound will arrive sooner or later from different speakers?

Something like this?

MSO Delay.jpg
 

·
Registered
Joined
·
1,774 Posts
Thank you Andy. My setup is the Dayton EMM-6 to Audio Buddy mic preamp to 3.5mm mic input on computer. Output from computer is from soundcard's 3.5mm stereo speaker jack to aux analog inputs on processor. Processor splits signal to mains and sub where I have crossover and speaker distances set. Sub output from processor is routed to symetrix sound processor for individual delay, PEQ and gain for each sub.
Oh, I see. Since you're using an analog input to your processor, and the same sound card is used for both input and output, for the loopback you could get away with just a cable going from an unused sound card analog out to its line in. Then you'd choose that previously unused output as the loopback timing output in REW, and similarly for the line input.

I read somewhere that acoustic time reference is not reliable on subs, so I believe I need to use the loopback method. Correct?
This is a misunderstanding. The acoustic timing reference is established using a speaker having a sweep applied to it that starts at, I think, 5 kHz and ends at around 20 kHz. So the speaker used to reproduce this signal must not be a sub, as you need something with good response from 5 kHz to 20 kHz to reproduce it. The measurement of this sweep is converted internally by REW into an impulse response which is sharp and narrow. The peak of this response establishes "t = 0". Any accurate timing reference, be it loopback or acoustic, can be used for sub measurements. However, it cannot be established using a sub to reproduce the acoustic timing reference test signal.

That said, since you're doing an all-analog measurement, I think the logistics will be easier with a loopback timing reference rather than an acoustic one.

So, are saying is that MSO does not attempt to adjust delays so that sound arrives from each speaker at a listening position at the same time, but rather adjust delays so that the waves are phase aligned, recognizing that in some listening positions the sound will arrive sooner or later from different speakers?

Something like this?

View attachment 2307668
I think that's a good illustration. There are many confounding factors in this problem, some of which I discuss in this post.
 

·
Registered
Joined
·
1,774 Posts
My setup is the Dayton EMM-6 to Audio Buddy mic preamp to 3.5mm mic input on computer.
Oh, I missed the fact that you're using an external mic preamp earlier. It provides phantom power to the mic, but also boosts the mic's very low output to line level. Because of the mic preamp having high gain, you should not connect its output to the mic input of your computer's sound card, but to its line input. If you were to connect the mic preamp's output to the mic input of your sound card, it would overload it.
 

·
Registered
Joined
·
44 Posts
Oh, I see. Since you're using an analog input to your processor, and the same sound card is used for both input and output, for the loopback you could get away with just a cable going from an unused sound card analog out to its line in. Then you'd choose that previously unused output as the loopback timing output in REW, and similarly for the line input.

Here is diagram of connections. Would this be correct for a loopback timing reference?

MSO Measurement Set Up.jpg

In REW Preferences->Analysis under the Impulse Response Calculations section is where I would set timing reference option to loopback. In the Help for the loopback, it says:

"If a loopback is selected the reference channel signal must be looped back from output to input on the soundcard and measurements will be relative to the loopback timing, usually this means measurements will have a time delay that corresponds to the time it takes sound to travel from the speaker being measured to the microphone."

Not sure I understand how REW will determine the wiring configuration used. For instance, in my diagram how will REW know the left channel is looped back from output to input on the soundcard?
 

·
Registered
Joined
·
1,774 Posts
Here is diagram of connections. Would this be correct for a loopback timing reference?

View attachment 2307728
Looks good. Remember to do a loopback calibration of the sound card itself. Check the card's low-frequency response. Sometimes built-in sound cards can have questionable performance.

In REW Preferences->Analysis under the Impulse Response Calculations section is where I would set timing reference option to loopback. In the Help for the loopback, it says:

"If a loopback is selected the reference channel signal must be looped back from output to input on the soundcard and measurements will be relative to the loopback timing, usually this means measurements will have a time delay that corresponds to the time it takes sound to travel from the speaker being measured to the microphone."

Not sure I understand how REW will determine the wiring configuration used. For instance, in my diagram how will REW know the left channel is looped back from output to input on the soundcard?
That's a John Mulcahy question. I agree that when using Java drivers, it's confusing. Nowhere does it say in the Soundcard tab of the REW Preferences dialog what the loopback channel is. I guess if it can only deal with two channels, it can assume when you specify the output and input channels, the loopback can be assumed to be "the other channel". Ask John though.

Here's what my setup looks like when choosing loopback and Java drivers.



With ASIO drivers, the situation is much more understandable.



If you have problems, you might try the universal ASIO4ALL. It provides an ASIO software layer for Windows audio drivers.
 

Attachments

·
Registered
Joined
·
44 Posts
I guess if it can only deal with two channels, it can assume when you specify the output and input channels, the loopback can be assumed to be "the other channel".
Andy I will play around with it this weekend. I'll measure mains and 1 sub from 2 distances using no time reference and then with a loopback time reference so that you can look at the output and confirm my measurement method is correct.
 

·
Registered
Joined
·
11 Posts
Hi all,
first post in this thread. First off, let me say a heartfelt thank you to Andyc for making such a program freely available! Rather amazing thing to do, actually.

I have a short question: What experience to people have with using MSO with two subs? I see on the main MSO page that four subs are recommended, but that anecdotal evidence (from one reviewer) suggests it can work well with two subs also. I'm asking since Harman's research suggests that two subs that are placed at the mid-points of opposing walls work almost as well as four subs. But does the MSO software allow using the two additional subs in a way that provides even more benefits than what Harman's research suggested?

Asking since I'm considering doing the multisub thing to even out the bass in my room, and I don't know whether I should aim for two or four... :)
 

·
Deep Sea Sound
Joined
·
3,639 Posts
To me the number of subs in a room is one half of the equation and the MSO is the other half.

Multiple subs addresses room modes with the more subs you have, the more even the bass is in the room. Now just because you have multiple subs in a room and you have the opportunity to get more even bass doesn't mean you will as you then have to integrate them well. This is where the MSO is fantastic because it takes the POTENTIAL benefits of multiple subs and lets them REALIZE their potential in giving the most even and "best" bass possible with your system and your room.

Think of it like a chorus. If you just have one or two people, it isn't much of a chorus but if you have more, you COULD have a chorus. Only with everyone in perfect harmony will the chorus sound amazing like it should.

Regarding your two subs vs 4. Perfect placement gets you better but every room is different so a blanket statement of midwalls might or might not be best. Without measuring the room with acoustical software, you won't know for sure. In general, 2 subs placed midwall AND integrated well will sound better than 1 sub. In general, 4 subs placed optimally in the room (corners or midwalls) will sound more even than 2 subs but again, integration is key and percentage of improvement is key (larger improvement likely as you go from 1 to 2 to more). Beyond 4 is a diminishing return as for most rooms the extra cost and complexity might only result in minimal improvement. But in your case, I'd go with your 2 subs and give it a shot as it likely will get you most of the way compared to someone who has 2 subs but aren't placed effectively and especially compared to one sub in a room.
 

·
Registered
Joined
·
11 Posts
To me the number of subs in a room is one half of the equation and the MSO is the other half.

Multiple subs addresses room modes with the more subs you have, the more even the bass is in the room. Now just because you have multiple subs in a room and you have the opportunity to get more even bass doesn't mean you will as you then have to integrate them well. This is where the MSO is fantastic because it takes the POTENTIAL benefits of multiple subs and lets them REALIZE their potential in giving the most even and "best" bass possible with your system and your room.

Think of it like a chorus. If you just have one or two people, it isn't much of a chorus but if you have more, you COULD have a chorus. Only with everyone in perfect harmony will the chorus sound amazing like it should.

Regarding your two subs vs 4. Perfect placement gets you better but every room is different so a blanket statement of midwalls might or might not be best. Without measuring the room with acoustical software, you won't know for sure. In general, 2 subs placed midwall AND integrated well will sound better than 1 sub. In general, 4 subs placed optimally in the room (corners or midwalls) will sound more even than 2 subs but again, integration is key and percentage of improvement is key (larger improvement likely as you go from 1 to 2 to more). Beyond 4 is a diminishing return as for most rooms the extra cost and complexity might only result in minimal improvement. But in your case, I'd go with your 2 subs and give it a shot as it likely will get you most of the way compared to someone who has 2 subs but aren't placed effectively and especially compared to one sub in a room.
Thank you so much! Very useful advice!
 

·
Registered
Joined
·
235 Posts
...I don't know whether I should aim for two or four... :)
Upstairs I have 4 subs and downstairs I have 2 subs. 2 subs still works great. Here are my before and after graphs with 2 Martysubs. Both are powered by a Behringer iNUKE amp. One is placed in the front-right corner and another placed along the left-midwall. I placed them at those locations for no other reason then there was space and it was convenient at the time. I've never tried other locations because my results are pretty good already and I'll likely add a 3rd sub somewhere at the back of the room when I get some free time to build another sub. I already have a spare iNuke amp sitting in a box.


2 subs, front-right corner and left mid-wall, Before MSO:



Before MSO, the room "boomed" at around 55 Hz. The MLP had no bass at 80 Hz, but Pos 3 had good bass at 80 Hz. Probably caused by the mains and sub cancelling each other at the 80 Hz crossover frequency. This is hard to fix manually because making the MLP better would make Pos 3 worse.


2 subs, front-right corner and left mid-wall, After MSO:



I weighted the MLP the highest at 1.0, Pos 2 and 3 at 0.8, Positions 4/5 at 0.5 weight. The improvement at the MLP using 2 subs is massive. Notice how even the worst optimized position, Pos 4, has better frequency response than even the best of the original positions.
 

Attachments

·
Registered
Joined
·
1,774 Posts
I weighted the MLP the highest at 1.0, Pos 2 and 3 at 0.8, Positions 4/5 at 0.5 weight. The improvement at the MLP using 2 subs is massive. Notice how even the worst optimized position, Pos 4, has better frequency response than even the best of the original positions.
Wow, nice results! Looks like I should update the documentation regarding MSO usage with two subs. Is it okay if I use your graphs in the MSO documentation?
 

·
Registered
Joined
·
11 Posts
Indeed, those are impressive results, Genesplitter!

Btw, I read a fascinating paper this morning on subjective preferences with regards to different systems for bass control: http://www.aes.org/e-lib/browse.cfm?elib=16324
(as a service to humanity I'm uploading the pdf here http://docdro.id/HJbAnjR )

The interesting finding is that a system with two subwoofers on opposing midwalls that are actively used to cancel out modes - "single source to sink" - actually is preferred over a very elaborate and intricate CABS system. And both single source to sink and CABS is significantly preferred over the rest. They also find that decay time is more significant for predicting preferences than flat frequency response in itself.

As I understand it, MSO is employing some of the same approach as the "source to sink" approach? And it will often reduce decay time as well as flatten the frequency response? Getting even more excited about trying it out now.
 

·
Wireless member
Joined
·
27,116 Posts
Perfectly understandable. When I added super chunk bass traps, a second collocated sub pair and (just) Audyssey the main benefit of the cleaning up of the room resonances was the unveiling of detail up and down the frequency range. Only after lifting that veil from the content did I begin to fiddle with a the flatness of my room’s bass response.

Jeff
 

·
Registered
Joined
·
1,774 Posts
Btw, I read a fascinating paper this morning on subjective preferences with regards to different systems for bass control: http://www.aes.org/e-lib/browse.cfm?elib=16324
(as a service to humanity I'm uploading the pdf here http://docdro.id/HJbAnjR )
That paper has been claimed by one particular person on the ASR forum as being the best research on the subject out there. I don't share that view. Here are Floyd Toole's comments about the paper from section 8.3 of the latest (third) edition of his book.



Their rationale for doing the 1/3 octave smoothing before EQ of the single-sub system strongly suggests that the authors don't understand the concept of minimum phase at all. Here's what they say:

The application of drastic filter parameters using high Q-factor and gain settings is not advisable. High gains may drive the loudspeaker into non linear behavior; and very high Q-factor filters have a long decay artifact in the time domain that may also be noticeable, oddly enough, as a resonance! As this type of equalization is applied to the signal before it is reproduced by the loudspeaker, the first wavefront, as it passes through the listening position and before it gets modified by the room response, will contain these artifacts, which may be audible and degrade the perceived quality.
It's established linear system theory that if you take a frequency response that has the minimum-phase property, and has, say, a large, narrow peak, and correct it with a minimum-phase filter, that filter will fix the time domain behavior as well. That is true even though the filter by itself may ring like crazy. Of course, this only applies for the listening position at which the response was measured. The heuristic explanation about the first wavefront above is simply wrong, although their objection to large boosts is well-founded. The REW manual shows how to determine whether the response has the minimum-phase property in the frequency region of interest.

Also, it's not clear to me how they can isolate the time domain behavior as they claim to have done, since the systems they compare have very different magnitude responses. The C.A.B.S. system response in Figure 8 rises about 30 dB from about 92 Hz to about 27 Hz, while the equalized single-sub system is flattish. I'm not saying that the time domain isn't important. It is, but I don't see how their testing removes the very large confounding factor of the strongly different magnitude responses of the systems compared.

The interesting finding is that a system with two subwoofers on opposing midwalls that are actively used to cancel out modes - "single source to sink" - actually is preferred over a very elaborate and intricate CABS system. And both single source to sink and CABS is significantly preferred over the rest. They also find that decay time is more significant for predicting preferences than flat frequency response in itself.

As I understand it, MSO is employing some of the same approach as the "source to sink" approach? And it will often reduce decay time as well as flatten the frequency response? Getting even more excited about trying it out now.
It's not really the same thing. Check out Todd Welti's post, "Comparison of Double Bass Array to Sound Field Management" for an introduction. MSO is similar in concept to SFM, but very different in implementation. C.A.B.S. and DBA use mode cancellation, while SFM and MSO are mode manipulation techniques. The former two arrange subs on the front and back walls so the wavefronts emanating from them approximate plane waves, with polarity reversed on the rear subs. The delay of the rear subs is adjusted to obtain, in effect, an active absorber. Its success over a large area of a room depends on the plane wave approximation. In the two-sub "source to sink" approach, you have spherical waves from each of the single subs, so the opportunities for cancellation are much poorer than for the plane wave approach. I've expressed strong skepticism about this technique in the past, but who knows, maybe it works better than I think it can. At any rate, MSO cannot insert polarity inversions at will during the optimization process, as the type of optimization algorithm it uses is tripped up by sudden, discrete changes in parameter values such as a polarity inversion. It's not capable of "teasing out" solutions involving inversions in that way. I don't recommend using such inversions with MSO unless you're specifically using a C.A.B.S. or DBA system, but who knows, I could be wrong on that.
 

·
Registered
Joined
·
1,774 Posts
Is it not the case that MSO may eliminate a particular modal peak or null?
That's true, but it's not necessarily the same thing as canceling out the effect of the mode (or many modes) altogether. I guess one could say that mode cancellation is a specific case of mode manipulation that seeks to eliminate the effect of the mode altogether, as might be done with DBA (which seeks to cancel not just one mode, but a whole family of them over a wide spatial region).
 
861 - 880 of 2327 Posts
Top