Multi-Sub Optimizer (MSO): Lessons Learned, Tips & Tricks

[Willem]

Thanks very much. This is a lot to chew on and it confirms that there is a challenge with levels and headroom, particularly with unequal subs, but not only in that case. For me, I do not have particularly large subs, and the room is large, so I cannot afford to waste headroom. Fortunately all three are sealed. The most powerful of my subs, the SB2000, also sits in a corner, so should do the heavy lifting. The others are mostly there for smoothness.
So I guess the first thing I should do is establish the 5% distortion level, and use that when measuring frequency response for each of them in every listening position.

 

[DaveBoswell]

you can check out the spreadsheet in this link below as well as a great source of info. Look for the CEA 2010 at 2m numbers in 31.5-50Hz range for your various subs and use that for relative limits. It's not super precise as the limit is not a hard and fast line but a range that requires some judgement and testing. I think a reasonable place to start is the CEA 2010 limit around 40 or 50Hz. The B&W sub is in the list the KEF 10 is there, not the 8, but suggest you use the 10 limits minus 1 or 2 dB, and Audioholics has published test results for the SVS sub in the other link below. That should be sufficient to get the limit matching inputs for the relative sub gains (link below as well).

https://www.audiosciencereview.com/forum/index.php?threads/subwoofer-comparison.20494/

SVS PB-2000 and SB-2000 Subwoofers Review

Review: The SVS ported PB-2000 and sealed SB-2000 subwoofers sport custom 12" drivers, a new 500 watt amp with DSP, performance below 20Hz, and a price-tag less than $800. The SVS 2000 series subs.

www.audioholics.com

 

[Willem]

Thanks, but I am still a bit puzzled. At 40 Hz I would have something like 100 dB, 105 dB and 110 dB for my three subs. However, how do I set that on the subs? Two of them just have a volume knob to turn, and the B&W leaves me even more puzzled because I have to set its sensitivity/volume by connecting it to a computer.
I take it these levels represent the maximum these subs can reproduce more or less cleanly, i.e. we are setting them up for more or less equal strain, i.e. without compression or the dsp limiters kicking in.

 

[Kervel]

However, how do I set that on the subs?

Just measure each sub with REW, and adjust physical volume button by hand until the biggest one is 5db more than the second, and that one 5 dB more than the third

 

[DaveBoswell]

agreed - what @Kervel said, however in a perfect world you are measuring each sub in the extreme nearfield (i.e. mic less than 1 inch from driver) with subs in the same place in the room, or at min in the extreme nearfield in the sub's normal position. Goal with the measurement is to minimize the impact of room resonances to the sub's amplitude response. Use pink generator in REW noise for the measurement.

 

[Willem]

In the meantime I have done some REW measurements, but those leave me puzzled. If I play the entire ensemble of main speakers and subs I get a reasonaly flat response (apart from peaks and dips), so sub levels are probably more or less OK. All filters etc on the ADI-2 are off, main speakers are high passed around 80 Hz and sub woofer low pass filters are disabled as far as possible. Since I am using a UMIK-1 I have to play the main speakers for a timing signal when I measure the individual subs at the various listening positions. The measurements show a dramatic fall off from 80 Hz down to about 40 Hz, below which there is a plateau some 30 dB below the level of the main speakers and below the level when all subs are playing at the same time. That seems rather more than I would have expected from leaving out two of the three subs, so I wondered if I am making a measurement mistake somewhere. But if not, how can MSO do its magic if the strong signal from the main speakers is included?

 

[JStewart]

how can MSO do its magic if the strong signal from the main speakers is included?

It can’t.

Disconnect one speaker. Send sweep to disconnected speaker. Sub will play.

Send timing signal to connected speaker.

 

[Willem]

Thanks. So time to figure out how I can send the REW sweep to one channel and the timing signal to the other.

 

[Willem]

Found it. Time to redo the whole measurement series....

 

[Keith_W]

I managed to get MSO to import my measurements and create a config. However, I need to export the config so that I can load it up in Acourate to integrate it with my mains. Is there a way to get MSO to export the config as a .WAV file? MSO can only export to JRiver JSON, MiniDSP biquad, and EqualizerAPO biquad text files only.

 

[3ll3d00d]

I managed to get MSO to import my measurements and create a config. However, I need to export the config so that I can load it up in Acourate to integrate it with my mains. Is there a way to get MSO to export the config as a .WAV file? MSO can only export to JRiver JSON, MiniDSP biquad, and EqualizerAPO biquad text files only.

There isn't you can either enter them manually into acourate and then use amplitude sum to create the aggregate filter or you can import them into something, export as freq response then import the text file into acourate similar to how a mic cal file is imported. The jriver Json is actually done via beqd (an app I wrote) so potentially I could add a wav export to that

 

[Keith_W]

You are Andy, the author of this piece of software? Brilliant work, and thank you for doing this! Unfortunately I haven't managed to hear or measure the results yet, all I can see is a very tasty looking result that I would love to integrate into my system. When you say "import them into something" ... what "something" are you talking about? I can't import it into REW (or at least, I haven't figured out how to do that yet). I know that REW can import text files - should I choose MiniDSP biquad or EqualizerAPO biquad? Otherwise, it looks like I will have to manually enter the coefficients and hope that I did not mess it up somewhere, because doing this exercise repeatedly could easily become quite tiresome.

 

[3ll3d00d]

You are Andy, the author of this piece of software

No, the app that handles the JSON export and can load into jriver is mine.

 

[3ll3d00d]

I will have to manually enter the coefficients

You can enter them as filters instead rather than raw coefficients

 

[Keith_W]

OK, after a few days of difficulties I finally managed to get an output from MSO that I have confidence with. The input looked like this:

And the output looked like this:

I have exported the output from MSO into EqualizerAPO text format. Here are a few lines:

Channel: L
# FL10
Delay: 16.27 ms

Channel: R
# FL10
Delay: 16.27 ms

Channel: SL
# FL9
Preamp: -7.32 dB
# FL21
Filter: ON PK Fc 146.86 Hz Gain -19.48 dB Q 25.0000
# FL22
Filter: ON PK Fc 47.88 Hz Gain -12.16 dB Q 25.0000
# FL23
Filter: ON PK Fc 116.57 Hz Gain -28.00 dB Q 22.8788

Now, the question is what to do with the output. This is the scenario: in my current configuration, I have a stereo pair of speakers set up as 8 way active. The crossover is generated by Acourate, and the filters are convolved by JRiver. I have my subwoofers low passed at 80Hz, and the woofers band passed at 80-500Hz.

When my measurements were taken, I set up the crossover so that the subwoofer had no low pass filter - i.e. the entire frequency output of the sub was measured. I did the same for the main speakers, but this time I low passed the woofers at 500Hz. My woofers are capable of remaining reasonably flat down to 40Hz. So, as far as MSO is concerned, it thinks that I will be running my main speakers uncorrected, and it only gave me recommendations for correcting the sub.

As you can see MSO thinks the subs are delayed 16.27 ms with respect to the main speakers. This is easy enough, I can dial in a 16.27 ms delay for my main speakers in my config.

The real question is how I should set up the system from here on. Should I:

1. Run the woofer with a low pass filter only (500Hz downwards) AND low pass the sub at 80Hz? The disadvantage of doing this is that I am giving the woofer more work to do - meaning, more excursion, less linearity, more heating of the voice coil, more amplifier power needed, etc. etc. But the unknown for me is whether MSO is also using the woofer output to linearize the bass.

2. Run the woofer with a low pass filter only (500Hz downwards) AND not low pass the sub (i.e. let it run full range)?

2. Go back to low passing the sub at 80Hz, and band pass the woofers between 80-500Hz? If I do this, what do I do with MSO's recommendations for freqs above the low pass of 80Hz? Ignore them? Input those values into the woofer?

 

[kiwifi]

Now, the question is what to do with the output. This is the scenario: in my current configuration, I have a stereo pair of speakers set up as 8 way active. The crossover is generated by Acourate, and the filters are convolved by JRiver. I have my subwoofers low passed at 80Hz, and the woofers band passed at 80-500Hz.

Crossover the sub to the woofers at 80Hz as before. Sub = 80Hz LPF, woofers 80Hz HPF.
The low pass filter on the sub is not a brick wall, so you should apply all the MSO filters to the sub, even those above 80Hz. I usually correct up to two times the crossover frequency on the sub.

 

[3ll3d00d]

Now, the question is what to do with the output.

The result doesn't look great to me tbh and you are using a relatively enormous number of filters (and only peq too) which seems a little unusual. You should decide on how to use the output before you simulate so you know what to measure too.

I have a stereo pair of speakers set up as 8 way active

assume you mean stereo input with a pair of subs configured to produce mono output and 3 way mains on top?

 

[Keith_W]

Crossover the sub to the woofers at 80Hz as before. Sub = 80Hz LPF, woofers 80Hz HPF.
The low pass filter on the sub is not a brick wall, so you should apply all the MSO filters to the sub, even those above 80Hz. I usually correct up to two times the crossover frequency on the sub.

Got it! Thank you. Nice and simple!

The result doesn't look great to me tbh and you are using a relatively enormous number of filters (and only peq too) which seems a little unusual. You should decide on how to use the output before you simulate so you know what to measure too.

What would you suggest to get a better result? To my eyes the result looks infinitely better than the input. Does your raw input look as awful as mine?

assume you mean stereo input with a pair of subs configured to produce mono output and 3 way mains on top?

Sorry, I should be clear. I have a pair of subs, but they are configured as stereo and not mono. And yes, a pair of three way mains on top.

 

[3ll3d00d]

I have a pair of subs, but they are configured as stereo and not mono. And yes, a pair of three way mains on top.

I don't understand how you intend to use MSO on a straightforward 4 way stereo setup. Are you just trying to optimise the combined response of the each sub and woofer? Yet mso does not support fir filters so how can it be aware of the crossover?

 

[Keith_W]

I don't understand how you intend to use MSO on a straightforward 4 way stereo setup. Are you just trying to optimise the combined response of the each sub and woofer? Yet mso does not support fir filters so how can it be aware of the crossover?

To be honest I am not sure myself. I don't know if Acourate will allow me to create the IIR filter, somehow convert it to a FIR filter, and then integrate it with the rest of my crossover. There are so many potential dead-ends and unanswered questions that I have that I decided to tackle the roadblocks one at a time as I encounter them

What I want is an even bass response at the listening position. If I can't get MSO to work, then the next step would be to try Acourate's virtual double bass array (which I don't like the sound of, because it involves using geometry instead of measurement). If that falls through, then the last step would be installing bass traps or Helmholtz resonators - which I am not keen to do because of the size and because it is a general purpose living room and not a dedicated listening room. Acourate can deal with bass peaks but not nulls, so I was quite interested when I read that MSO has a way to deal with the nulls.

In general I try to exhaust all the free options first before opening my wallet. MSO is free, and sounded promising, so it was my first port of call.