Optimizing subwoofers and integration with mains: multi sub optimizer - Page 26 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 162Likes
Reply
 
Thread Tools
post #751 of 777 Old 08-09-2017, 03:41 PM
Senior Member
 
AV_mike's Avatar
 
Join Date: May 2009
Posts: 425
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 115 Post(s)
Liked: 105
The target curve is simply a text file of the following example format:-

20.0 5.0
30.0 5.0
50.0 4.0
60.0 3.0
100.0 0.0

So it's just frequency values followed by desired levels - to match with what ever shape you are trying to achieve.
MSO can either 'join the points' with straight lines (log interpolation), or effectively draw a smooth curve between them (cubic spline).
Help that helps, Mike.
AV_mike is online now  
Sponsored Links
Advertisement
 
post #752 of 777 Old 08-09-2017, 05:42 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
Quote:
Originally Posted by Iamjcl View Post
Does anybody know how to generate a "target curve" for import into MSO?
I came up with a kind of klugey way of making fancier curves within MSO itself than can be done by hand, by exporting the response of a filter channel. I've attached a project that is almost empty (no imported measurements). The only thing I did was add a low-shelf filter of type "LF Shelf Second-Order (Alt)" and created a graph of its response. You can adjust this shape by adjusting the center frequency and LF boost in dB of this filter using the tuning feature in the Properties Window. Select the filter FL1 in the Config View, then move over to the Properties Window and click on the numeric field in the second column of the Value property of either the LF boost or center frequency parameters. This will cause a spin control to appear as in Figure 9 in the tutorial. As you manipulate the spin control, the graph will change in real time. The ballistics of this button are a little slow, but they speed up after you've held the mouse button down on the spin control for a couple of seconds.

The "center frequency" of an LF shelf of this type is the frequency for which the boost in dB is half of its maximum low-frequency value. A boost of 6 dB and a center frequency of 105 Hz corresponds roughly to the low-frequency portion of the Harman curve discussed by Todd Welti (figure 5, black curve).

If you can get a curve you like, you can export the response of this filter by clicking on the "Filters" node above FL1 in the Config View, then right-clicking and choosing "Save this Channel's Response as Target Curve". A sample target curve is also in the zip file.

If you want more control over the curve, you can delete F1 and replace it with a low-shelf filter of type "LF Shelf Variable Q Second-Order (Alt)". In this case, the Q allows you to control how gradual the transition is from low to high frequencies.

Edit: To actually achieve that curve when optimizing, you'd likely want to add exactly such a filter (such as "LF Shelf Variable Q Second-Order (Alt)") to the shared sub filters of your real configuration, as plain old PEQ isn't well suited to that function. IOW, if you're using a sub-only configuration and you use the technique above to make your target curve, adding the same kind of filter with the exact same parameter values to your shared sub filters will get you there without further ado if you've already optimized to the default flat target curve.
Attached Files
File Type: zip target_curve.zip (8.3 KB, 15 views)

Last edited by andyc56; 08-09-2017 at 06:11 PM.
andyc56 is offline  
post #753 of 777 Old 08-09-2017, 06:51 PM
Advanced Member
 
Iamjcl's Avatar
 
Join Date: Feb 2000
Location: Down South
Posts: 874
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 145 Post(s)
Liked: 32
Quote:
Originally Posted by andyc56 View Post
I came up with a kind of klugey way of making fancier curves within MSO itself than can be done by hand, by exporting the response of a filter channel. I've attached a project that is almost empty (no imported measurements). The only thing I did was add a low-shelf filter of type "LF Shelf Second-Order (Alt)" and created a graph of its response. You can adjust this shape by adjusting the center frequency and LF boost in dB of this filter using the tuning feature in the Properties Window. Select the filter FL1 in the Config View, then move over to the Properties Window and click on the numeric field in the second column of the Value property of either the LF boost or center frequency parameters. This will cause a spin control to appear as in Figure 9 in the tutorial. As you manipulate the spin control, the graph will change in real time. The ballistics of this button are a little slow, but they speed up after you've held the mouse button down on the spin control for a couple of seconds.

The "center frequency" of an LF shelf of this type is the frequency for which the boost in dB is half of its maximum low-frequency value. A boost of 6 dB and a center frequency of 105 Hz corresponds roughly to the low-frequency portion of the Harman curve discussed by Todd Welti (figure 5, black curve).

If you can get a curve you like, you can export the response of this filter by clicking on the "Filters" node above FL1 in the Config View, then right-clicking and choosing "Save this Channel's Response as Target Curve". A sample target curve is also in the zip file.

If you want more control over the curve, you can delete F1 and replace it with a low-shelf filter of type "LF Shelf Variable Q Second-Order (Alt)". In this case, the Q allows you to control how gradual the transition is from low to high frequencies.

Edit: To actually achieve that curve when optimizing, you'd likely want to add exactly such a filter (such as "LF Shelf Variable Q Second-Order (Alt)") to the shared sub filters of your real configuration, as plain old PEQ isn't well suited to that function. IOW, if you're using a sub-only configuration and you use the technique above to make your target curve, adding the same kind of filter with the exact same parameter values to your shared sub filters will get you there without further ado if you've already optimized to the default flat target curve.
Excellent - very easy with that explanation. Thanks!

Still trying to find optimum settings - with MiniDSP and MSO I've now got so many parameters it is almost too much. Getting it all dialed in so it sounds "right" is my current project. Have some great looking FR graphs but actual sound is less-great so still on the learning curve. Fascinating science in all of this!
Iamjcl is offline  
 
post #754 of 777 Old 08-09-2017, 06:52 PM
Advanced Member
 
Iamjcl's Avatar
 
Join Date: Feb 2000
Location: Down South
Posts: 874
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 145 Post(s)
Liked: 32
@AV_mike , thanks - that's simple enough!
Iamjcl is offline  
post #755 of 777 Old 08-10-2017, 01:29 PM
Advanced Member
 
Iamjcl's Avatar
 
Join Date: Feb 2000
Location: Down South
Posts: 874
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 145 Post(s)
Liked: 32
Quote:
Originally Posted by Iamjcl View Post
@AV_mike , thanks - that's simple enough!
Another quick question...

Since one of the 6 subs in my optimizations has to have a delay of 0 (no delay filter channel at all), and presumably MSO tries to time the other 5 that do have delay filters assigned to match that one, would that imply that the AVR "Sub Distance" could remain the same, even when trying various MSO optimizations that are all "time aligned" to that one sub with no delay?

I ran Yamaha YPAO after an optimization and it selected a sub distance (for the group of 6, collectively) - so would I need to re-run this each time I try a new optimization (again - what seems to me to be the "reference sub", the one without a delay parameter, isn't time manipulated by MSO / MiniDSP but the others are).

Hope this makes sense.
Iamjcl is offline  
post #756 of 777 Old 08-10-2017, 07:45 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
Quote:
Originally Posted by Iamjcl View Post
Since one of the 6 subs in my optimizations has to have a delay of 0 (no delay filter channel at all), and presumably MSO tries to time the other 5 that do have delay filters assigned to match that one, would that imply that the AVR "Sub Distance" could remain the same, even when trying various MSO optimizations that are all "time aligned" to that one sub with no delay?
That's difficult to predict, so I can't give a definite answer. The sub distance will affect the integration of the main speakers and subs, but I don't know how much these different MSO configurations will differ from one another.

If you're not using MSO to optimize integration between main speakers and subs, you'll need to do some sort of manual procedure. Commercial room correction systems are notorious for getting the sub distance wrong, though I've been told that Dirac does a very good job. One way to do it with REW is to use the RTA with "Pink PN" as the source signal type to observe combined main and sub responses at the MLP. Then manually tweak the sub distance and level to optimize integration while observing the RTA display in REW. With many AVRs you have to back out of the menu one level for the distance or level change to "take".

Quote:
Originally Posted by Iamjcl View Post
I ran Yamaha YPAO after an optimization and it selected a sub distance (for the group of 6, collectively) - so would I need to re-run this each time I try a new optimization (again - what seems to me to be the "reference sub", the one without a delay parameter, isn't time manipulated by MSO / MiniDSP but the others are).
Be careful with this if you're using the LF shelf in the miniDSP to hit a target curve. If YPAO has a flat target curve, it will try to eliminate the boost provided by the LF shelf in the miniDSP.
andyc56 is offline  
post #757 of 777 Old 08-11-2017, 12:59 AM
Senior Member
 
AV_mike's Avatar
 
Join Date: May 2009
Posts: 425
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 115 Post(s)
Liked: 105
Hi Andy, that was a short break from support duties😀
I for one am glad you're back, as I can not answer any technical queries.
With regards to house or target curves, I can only refer to my own use.
I use MSO to combine multiple subs (was three, now two) with a flat target, no custom curve. Then I run Dirac, and initially use its 'auto target', that is effectively a slope from +2.5dB @ 20Hz to -2.5dB @ 20kHz. I then use MSO again to find the best splice between subs and mains, with MSO aiming for the Dirac target (really simple text file). Finally I modify the Dirac target to my taste, the sub/main splice does not alter during this step, confirmed with REW.
Might help others, Mike.
andyc56 likes this.
AV_mike is online now  
post #758 of 777 Old 08-11-2017, 05:17 AM
Advanced Member
 
Iamjcl's Avatar
 
Join Date: Feb 2000
Location: Down South
Posts: 874
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 145 Post(s)
Liked: 32
Quote:
Originally Posted by andyc56 View Post

If you're not using MSO to optimize integration between main speakers and subs, you'll need to do some sort of manual procedure. Commercial room correction systems are notorious for getting the sub distance wrong, though I've been told that Dirac does a very good job. One way to do it with REW is to use the RTA with "Pink PN" as the source signal type to observe combined main and sub responses at the MLP. Then manually tweak the sub distance and level to optimize integration while observing the RTA display in REW. With many AVRs you have to back out of the menu one level for the distance or level change to "take".

Be careful with this if you're using the LF shelf in the miniDSP to hit a target curve. If YPAO has a flat target curve, it will try to eliminate the boost provided by the LF shelf in the miniDSP.
Thanks! - that is great info (REW / Pink PN procedure to set AVR distance accurately). I'll give this a go and see how far it is off from YPAO, and also how each sounds.

Thanks for the warning on YPAO also - I have been returning YPAO EQ on sub channels to "Through" (no eq) after YPAO runs.
Iamjcl is offline  
post #759 of 777 Old 08-14-2017, 07:03 PM
Newbie
 
Spridle1's Avatar
 
Join Date: Jul 2008
Location: Houston
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 10
I know I've read something on this before, so please point me to the page. I have an Anthem Statement processor with ARC room correction which I want to use to on frequencies from 200-2500Hz, but I have to run it from 20-2500Hz and the creection is then applies to all channels including the sub. I want to use MSO for my 6 subs. My question is do I run the built in ARC processor room correction first, and then run MSO on the corrected processor output? Also how will I optimize the crossover?
Spridle1 is offline  
post #760 of 777 Old 08-15-2017, 11:27 AM
Senior Member
 
AV_mike's Avatar
 
Join Date: May 2009
Posts: 425
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 115 Post(s)
Liked: 105
Quote:
Originally Posted by Spridle1 View Post
I know I've read something on this before, so please point me to the page. I have an Anthem Statement processor with ARC room correction which I want to use to on frequencies from 200-2500Hz, but I have to run it from 20-2500Hz and the creection is then applies to all channels including the sub. I want to use MSO for my 6 subs. My question is do I run the built in ARC processor room correction first, and then run MSO on the corrected processor output? Also how will I optimize the crossover?
I would advise performing a sub-only configuration first - to get all the subs working together. But for this you will need some means of combining them into a single sub channel - preferably with the ability to adjust gains & delays for each sub individually, as well as being able to apply some PEQ to each sub channel. After this is completed - run ARC. It should NOT have very much to do, as the single sub channel should already be "corrected". After ARC - measure the single sub channel and mains, and use MSO to find the best settings for an optimal splice through the crossover region, again using gains & delays.
For all of the above you must measuring whilst employing a timing reference (loop-back or acoustic) in order to give MSO meaningful data.
Regards, Mike.
AV_mike is online now  
post #761 of 777 Old 08-15-2017, 03:25 PM
Newbie
 
Spridle1's Avatar
 
Join Date: Jul 2008
Location: Houston
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 10
Quote:
Originally Posted by AV_mike View Post
I would advise performing a sub-only configuration first - to get all the subs working together. But for this you will need some means of combining them into a single sub channel - preferably with the ability to adjust gains & delays for each sub individually, as well as being able to apply some PEQ to each sub channel. After this is completed - run ARC. It should NOT have very much to do, as the single sub channel should already be "corrected". After ARC - measure the single sub channel and mains, and use MSO to find the best settings for an optimal splice through the crossover region, again using gains & delays.
For all of the above you must measuring whilst employing a timing reference (loop-back or acoustic) in order to give MSO meaningful data.
Regards, Mike.
Mike, I have a Symetrix Solus 8x8 processor for my subs. So you are saying 1) optimize subs only with MSO and lock in settings with the Symetrix then, 2) run ARC with mains and subs running through the Symetrix as optimized in Step 1.

Not sure I follow the next step on how to then use MSO to find best splice at crossover.
Spridle1 is offline  
post #762 of 777 Old 08-16-2017, 09:45 AM
Senior Member
 
AV_mike's Avatar
 
Join Date: May 2009
Posts: 425
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 115 Post(s)
Liked: 105
Quote:
Originally Posted by Spridle1 View Post
Mike, I have a Symetrix Solus 8x8 processor for my subs. So you are saying 1) optimize subs only with MSO and lock in settings with the Symetrix then, 2) run ARC with mains and subs running through the Symetrix as optimized in Step 1.

Not sure I follow the next step on how to then use MSO to find best splice at crossover.
Step 3 is indeed re-measuring the system - mains & the now combined single sub channel, with ARC active. You measure the mains while the sub is muted, and the sub while the mains are disconnected. So the measurements will be affected by ARC and by the crossovers configured in your Anthem. Now feed all the data into MSO, and it will show you how well the sub blends with the mains. You may be lucky and get a perfect 'splice' - but if you don't, MSO can help determine changes in relative delays & gains in order to make the transition between subs & mains as seamless as possible.
You can also try different crossover frequencies - as the usually quoted 80Hz does not work for everyone. My mains work best at 60Hz, while my surrounds are better at 90Hz. Do all the measurements you can think of, then let MSO do the hard work.
Regards, Mike.
AV_mike is online now  
post #763 of 777 Old 08-29-2017, 12:05 PM
AVS Forum Addicted Member
 
noah katz's Avatar
 
Join Date: Apr 1999
Location: Mountain View, CA USA
Posts: 21,563
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 1095 Post(s)
Liked: 380
Sounds like MSO won't work with OmniMic - right?

Quote:
Originally Posted by andyc56 View Post
If one is using a USB mic, REW is required, as its acoustic timing reference is the only way I know of to get time-synchronized measurements with that arrangement. If there is some other software that can do this using a USB mic, I"d be interested in finding out so I can add it to the documentation. If one is using an analog mic and e.g. an external USB sound device, any measurement software supporting loopback time correction (which includes REW, FuzzMeasure and others) can be used. HolmImpulse can do time-synchronized measurements without a physical hardware loopback via its time-locking feature. But according to Bill Waslo (OmniMic author), that only works if the output DAC and the input A/D clocks are locked to one another. This can't happen with a USB mic arrangement. All of this is explained in the MSO documentation.
noah katz is offline  
post #764 of 777 Old 08-29-2017, 12:14 PM
AVS Forum Special Member
 
markus767's Avatar
 
Join Date: Apr 2009
Posts: 9,873
Mentioned: 111 Post(s)
Tagged: 0 Thread(s)
Quoted: 4189 Post(s)
Liked: 1577
Quote:
Originally Posted by noah katz View Post
Sounds like MSO won't work with OmniMic - right?
If you use REW's acoustic timing reference feature it will work.

Markus

"In science, contrary evidence causes one to question a theory. In religion, contrary evidence causes one to question the evidence." - Floyd Toole
markus767 is offline  
post #765 of 777 Old 08-29-2017, 01:04 PM
AVS Forum Addicted Member
 
noah katz's Avatar
 
Join Date: Apr 1999
Location: Mountain View, CA USA
Posts: 21,563
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 1095 Post(s)
Liked: 380
I got OmniMic so I wouldn't have to deal with REW; I have limited computer skills and patience.

Last edited by noah katz; 08-29-2017 at 02:22 PM.
noah katz is offline  
post #766 of 777 Old 08-29-2017, 01:49 PM
AVS Forum Special Member
 
markus767's Avatar
 
Join Date: Apr 2009
Posts: 9,873
Mentioned: 111 Post(s)
Tagged: 0 Thread(s)
Quoted: 4189 Post(s)
Liked: 1577
Quote:
Originally Posted by noah katz View Post
I got OmniMic so I wouldn't have to deal with REW; I have limited computer skills and patience.
Sounds like MSO won't work with you
pepar likes this.

Markus

"In science, contrary evidence causes one to question a theory. In religion, contrary evidence causes one to question the evidence." - Floyd Toole
markus767 is offline  
post #767 of 777 Old 09-14-2017, 02:21 PM
Member
 
genesplitter's Avatar
 
Join Date: May 2011
Location: Oakland, CA
Posts: 143
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 41
Question about iNukeDSP and the Robert Bristow-Johnson PEQ convention - How can one limit MSO to 0.10 <= Q (RBJ) <= 10 before starting the optimization? MSO property pages use the classic Q definition, which works for products like the miniDSP. The iNukeDSP also limits PEQ gain from plus/minus 15 dB.

Basically I need to constrain MSO like this for an iNukeDSP amp:

Q(RBJ) min = 0.10
Q(RBJ) max = 10.0


Here's a reference about how to find the Q(RBJ) values after optimization is complete - http://andyc.diy-audio-engineering.o...atibility.html

Attached Thumbnails
Click image for larger version

Name:	peq.png
Views:	148
Size:	12.0 KB
ID:	2282938  

U-Turn Orbit Plus turntable, Marantz SR-7009 AVR, iNuke 6000DSP, DIY LinkwitzLab speakers, MartyCube + JohnnySub, BenQ W1070, Roku stick, Plex

Last edited by genesplitter; 09-14-2017 at 02:42 PM.
genesplitter is offline  
post #768 of 777 Old 09-14-2017, 03:16 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
Quote:
Originally Posted by genesplitter View Post
Question about iNukeDSP and the Robert Bristow-Johnson PEQ convention - How can one limit MSO to 0.10 <= Q (RBJ) <= 10 before starting the optimization? MSO property pages use the classic Q definition, which works for products like the miniDSP. The iNukeDSP also limits PEQ gain from plus/minus 15 dB.

Basically I need to constrain MSO like this for an iNukeDSP amp:

Q(RBJ) min = 0.10
Q(RBJ) max = 10.0
If you constrain the classic Q to a maximum of 10, the Q(RBJ) will always be less than 10 as well, since Q(RBJ) is always less then the classic Q (equation 10 on this page). Unfortunately, this actually limits the maximum Q to something somewhat less than what the iNuke can accommodate. For instance, if a PEQ has a center frequency attenuation of 6 dB and a classic Q of 10, the Q(RBJ) would be 10 / sqrt(2) = 7.07 (that's A0=0.5 in the above mentioned equation). One of the things I was considering was to make a second type of PEQ, one that uses Q(RBJ) directly, such that if you constrain its Q to a maximum of 10, that will be equivalent to a maximum Q(RBJ) of 10 as well.

Is that something you might be interested in?

Also, there is a section regarding the two ways of specifying filter parameter limits in the documentation if you haven't yet seen it.

Edit: Regarding the minimum allowable value of Q, Welti and Devantier in "Low-Frequency Optimization Using Multiple Subwoofers" say this:

Quote:
Originally Posted by Welti and Devantier
Finally limiting the Q values of the filters to a value greater than 1 seems appropriate, since lower Q values would simply start to emulate an overall change in the subwoofer gain setting.
For this reason, I stuck with a minimum value of 1. A Q of 0.1 is a really broad filter.

Last edited by andyc56; 09-14-2017 at 03:22 PM.
andyc56 is offline  
post #769 of 777 Old 09-14-2017, 03:32 PM
Member
 
genesplitter's Avatar
 
Join Date: May 2011
Location: Oakland, CA
Posts: 143
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 41
This forum is really amazing, just last month Floyd Toole answered a question I posted in another thread, and now the author of MSO!

I suspect you workaround will converge closely to the optimal solution, and I could always re-run the optimizations bumping up classic-Q slightly each time, so I wouldn't put this as a high-priority item, but it certainly would be appreciated. Thank you for creating such a great product!

U-Turn Orbit Plus turntable, Marantz SR-7009 AVR, iNuke 6000DSP, DIY LinkwitzLab speakers, MartyCube + JohnnySub, BenQ W1070, Roku stick, Plex
genesplitter is offline  
post #770 of 777 Old 09-14-2017, 03:40 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
Quote:
Originally Posted by genesplitter View Post
This forum is really amazing, just last month Floyd Toole answered a question I posted in another thread, and now the author of MSO!
I ain't no Floyd Toole.

Quote:
Originally Posted by genesplitter View Post
I suspect you workaround will converge closely to the optimal solution, and I could always re-run the optimizations bumping up classic-Q slightly each time, so I wouldn't put this as a high-priority item, but it certainly would be appreciated. Thank you for creating such a great product!
Thanks for the kind words.
andyc56 is offline  
post #771 of 777 Old 09-14-2017, 03:54 PM
Deep Sea Sound
 
dgage's Avatar
 
Join Date: Dec 2003
Location: Knoxville, TN
Posts: 3,431
Mentioned: 183 Post(s)
Tagged: 0 Thread(s)
Quoted: 1663 Post(s)
Liked: 1480
Quote:
Originally Posted by andyc56 View Post
I ain't no Floyd Toole.
We still love you anyway though. . And we appreciate all of your efforts in creating such a great tool as MSO and supporting it as much as you have. You're awesome man!
sdurani, pepar, arsMatrix and 3 others like this.

David Gage
Deep Sea Sound
"You don't listen to our subs, you EXPERIENCE them!"
dgage is offline  
post #772 of 777 Old 09-15-2017, 05:12 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
Thanks for the kind words guys.

Version 1.23 is now up. It adds a new "Parametric EQ (RBJ)" filter type, which uses the RBJ Q directly and allows the usage of the full range of available Q for the iNuke DSP amps. While I was in there adding filters, I also added the missing first-order LPF and HPF. The filter context menus in the Config View have been changed a bit for improved ease of use (hopefully).

@genesplitter
dgage likes this.
andyc56 is offline  
post #773 of 777 Old 09-16-2017, 03:58 PM
Member
 
genesplitter's Avatar
 
Join Date: May 2011
Location: Oakland, CA
Posts: 143
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 41
Thanks Andy, last night I simply re-ran the optimizer until Q(RBJ) was 10 or less. Your new filter is greatly appreaciated. Ultimately I was not successful in getting the calculated optimization to match my after, measured response. I initially thought I was using too many PEW filters (the iNuke6000DSP supports 8 per channel), so this morning I did a test by creating the simplest possible MSO model, using only 2 subs, MLP only, with gain and delay but no PEQ.

Here are my MSO settings, with only Pos 1 (MLP) and REW .frd of my center, sub1, and su2:





Before Optimization:



MSO - Before Optimization (calculated). Notice the extremely close match to the REW before-optimization.



Next is the MSO - After optimization run. Theoretically looks good, especially considering no PEQ's yet:


Unfortunately my actual measurement did not match up:


Because the before MSO graph is so accurate, the after optimization should be equally accurate, therefore I am missing something. Some things I tried:

1. Don't move the microphone
2. Swap channels A and B in the iNuke, in case I got sub1 and sub2 swapped.
3. Copy iNuke values back into MSO to update the optimized graph. The iNuke won't accept all MSO values exactly as provided, but changes into the closest available (discrete) value.
4. Tried "wait for timing reference" checked and unchecked. Since I measured the center channel, I used the right channel as the timing reference.

I used a 16 foot usb extention cable + a good quality 6 foot usb cable, which is slightly over the maximum allowed usb length. I don't thing this would subtly effect my CSO calibrated microphone readings and the before MSO graph was spot on.

Any tips on what direction to investigate?
Attached Thumbnails
Click image for larger version

Name:	data_view.png
Views:	94
Size:	17.5 KB
ID:	2283850   Click image for larger version

Name:	config_view.png
Views:	95
Size:	24.9 KB
ID:	2283852   Click image for larger version

Name:	Original - before optimization.png
Views:	12
Size:	51.5 KB
ID:	2283854   Click image for larger version

Name:	MSO - Optimized (calculated) Response.png
Views:	92
Size:	17.8 KB
ID:	2283856   Click image for larger version

Name:	MSO - Before (calculated) Response.png
Views:	92
Size:	18.1 KB
ID:	2283864  

Click image for larger version

Name:	REW - After (optimized) response.png
Views:	92
Size:	53.1 KB
ID:	2283868   Click image for larger version

Name:	REW - before optimization.png
Views:	93
Size:	50.7 KB
ID:	2283872  

U-Turn Orbit Plus turntable, Marantz SR-7009 AVR, iNuke 6000DSP, DIY LinkwitzLab speakers, MartyCube + JohnnySub, BenQ W1070, Roku stick, Plex

Last edited by genesplitter; 09-16-2017 at 04:09 PM.
genesplitter is offline  
post #774 of 777 Old 09-16-2017, 04:37 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
That's odd, especially considering there's only gains and delays, and only one position, and the computed and measured sum look close before the addition of gains and delays.. I can think of a few things:
  1. Did you configure MSO in Tools, Application Options, Hardware for "System uses AVR or pre/pro with sub out?"
  2. If the answer to (1) is "yes", did you change the AVR sub distance and gain as specified in the MSO filter report after optimization? MSO tries to do as much as it can in the AVR to minimize gain and delay adjustments in the per-sub DSP.
  3. Did you measure with the delay and gain settings on the iNuke set to zero? The MSO delays and gains are changes from the as-measured state, so it can be confusing if you don't measure with a fresh baseline.

Edit: Before making AVR gain or delay changes, record the AVR sub trim gain and sub distance that were used in the measurement, as you'll need them later if you don't re-measure.
Edit 2: Also, are there any negative delays allowed for the delay blocks? If so, you must configure MSO for your AVR as mentioned above, as negative delays can only be accomplished by sub distance increases in the AVR (a reduction in delay from the as-measured state).

Last edited by andyc56; 09-16-2017 at 04:52 PM.
andyc56 is offline  
post #775 of 777 Old 09-17-2017, 06:05 PM
Member
 
genesplitter's Avatar
 
Join Date: May 2011
Location: Oakland, CA
Posts: 143
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 41
I started fresh this morning and verified that MSO simulation graphs match REW measurements - they matched extremely well. I think the problem was the iNuke windows software is finicky and doesn't save settings when you think they did. Early on I had to redo REW measurments for Pos 2 to 5 because my iNuke settings were not zero'd out as I expected.

I was having so much fun I ended up optimizing 5 positions using delay, gain, and 3 PEQs. I then re-measured the optimized positions to confirm they actually match the MSO simulation (they did).

Here is an early test confirming MSO and REW matched with some simple filters. It's scary how close they are.







Here is where I left things. 5 optimzied positions, using 2 DIY Martysubs and iNukeDSP amp. The before response is attached if you want to see it. The MSO optimized is much better, particularly the MLP.



I have a few additional questions:

1) I removed the avr from the sim by setting min delay = 0 and max delay = 40 ms. Am I correct to assume this is not the same as min delay = -20 and max delay = 20? I initially wanted to remove the avr to keep things simple, but I'm ready to add it back in if needed.

2) I ran an MSO sim with 8 PEQs per channel to max out the iNuke, but there was only a small improvement compared to 3 PEQs. I assume trying to optimize 5 positions needs more than 2 subs. Is there a rule? maybe N subs for N positions? I'm going to add a 3rd, IB sub into this setup over the Christmas break. I'll read up on how to model dissimilar subs later, but MSO should really help to combine ported and sealed/IB subs.

2) I have a 2nd system in the living room with 4 JBL 550p subs in the 4 corners. Unfortunately they are self-powered so I need a separate DSP. Is the miniDSP what most use? MSO help files warn not to buy the unbalanced miniDSP 2x4. Is there a preference between the balanced 2x4 or unbalanced 2x4 HD? My Marantz avr and jbl subs use RCA but I assume the balanced phoenix connectors can work with RCA too.

Oh, and thanks again Andy for creating such amazing software. I can't imaging using multiple subs without using a tool to optimize their combined response.
Attached Thumbnails
Click image for larger version

Name:	REW - optimized response, gain, delay 2 PEQ.png
Views:	72
Size:	41.2 KB
ID:	2284306   Click image for larger version

Name:	MSO- pos 1-5 optimized.png
Views:	69
Size:	31.1 KB
ID:	2284314   Click image for larger version

Name:	MSO - optimized response, gain, delay 2 PEQ.png
Views:	70
Size:	15.6 KB
ID:	2284340   Click image for larger version

Name:	Before Optimizations, Positions 1-5.png
Views:	9
Size:	126.0 KB
ID:	2284378  
andyc56 likes this.

U-Turn Orbit Plus turntable, Marantz SR-7009 AVR, iNuke 6000DSP, DIY LinkwitzLab speakers, MartyCube + JohnnySub, BenQ W1070, Roku stick, Plex

Last edited by genesplitter; 09-17-2017 at 06:25 PM.
genesplitter is offline  
post #776 of 777 Old 09-17-2017, 07:08 PM
AVS Forum Special Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 1,174
Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 391
Quote:
Originally Posted by genesplitter View Post
I have a few additional questions:

1) I removed the avr from the sim by setting min delay = 0 and max delay = 40 ms. Am I correct to assume this is not the same as min delay = -20 and max delay = 20? I initially wanted to remove the avr to keep things simple, but I'm ready to add it back in if needed.
They are not the same thing, that's correct. It's unlikely you'd need anything close to 40 msec delay, unless the sub distance were way off in the AVR. One thing you can do is the "Rearrange delays" command in the "Subwoofer channels" node of the Config View after doing an optimization. For N sub channels, this will make an equivalent configuration with one shared delay and N-1 individual delays. These individual delays can then be constrained to be positive, matching the available physical delays in the amps or DSP box. But the shared delay represents the AVR sub distance, so it can be allowed to have a negative value, corresponding to increasing the sub distance relative to the measured condition. I've found it makes sense to start with all individual delays first, then doing an optimization. Doing a "Rearrange delays" afterwards determines which sub channel will end up with zero delay (as the maximum number of delay blocks is N). If you do this, sometimes the delay limits are changed as this is sometimes the only way to get an equivalent configuration, so keep an eye on that.

Quote:
Originally Posted by genesplitter View Post
2) I ran an MSO sim with 8 PEQs per channel to max out the iNuke, but there was only a small improvement compared to 3 PEQs. I assume trying to optimize 5 positions needs more than 2 subs. Is there a rule? maybe N subs for N positions? I'm going to add a 3rd, IB sub into this setup over the Christmas break. I'll read up on how to model dissimilar subs later, but MSO should really help to combine ported and sealed/IB subs.
That's exactly right about the number of subs. I don't have rigorous mathematical proof, but there is for another sub optimization method, the matrix solution method with FIRs used in the old JBL BassQ. This is described in section 5 of Welti and Devantier (equation (8)). Their matrix equation shows a 2x2 matrix, but it can be generalized to a matrix with M rows and N colums. The number of rows in the matrix is equal to the number of positions measured. The number of columns is equal to the number of subs with independent filter control. When these are equal and the matrix is non-singular, there is a solution. When the number of positions is greater than the number of subs, a least-squares solution is possible using the pseudo-inverse, but the more rows (measurement positions), the larger the mean-squared error. That's linear least-squares. MSO is nonlinear least-squares using an optimizer, but the basic idea is similar. Based on this, I'd say, "make the number of listening positions at least as large as the number of subs, but if it's made too large, the mean-squared error will increase."

Quote:
Originally Posted by genesplitter View Post
2) I have a 2nd system in the living room with 4 JBL 550p subs in the 4 corners. Unfortunately they are self-powered so I need a separate DSP. Is the miniDSP what most use? MSO help files warn not to buy the unbalanced miniDSP 2x4. Is there a preference between the balanced 2x4 or unbalanced 2x4 HD? My Marantz avr and jbl subs use RCA but I assume the balanced phoenix connectors can work with RCA too.
The only thing I don't like about the balanced miniDSP 2x4 is the maximum delay of 7.5 msec, which is limiting. It means more fiddling with MSO.

Quote:
Originally Posted by genesplitter View Post
Oh, and thanks again Andy for creating such amazing software. I can't imaging using multiple subs without using a tool to optimize their combined response.
You're welcome! Thanks for presenting your data. I was getting a bit worried that you found a bug. It would have been a pretty serious one. Also, @rumpeli deserves the credit for finding out that the iNuke uses the RBJ convention for Q.
andyc56 is offline  
post #777 of 777 Old 09-17-2017, 07:36 PM
Deep Sea Sound
 
dgage's Avatar
 
Join Date: Dec 2003
Location: Knoxville, TN
Posts: 3,431
Mentioned: 183 Post(s)
Tagged: 0 Thread(s)
Quoted: 1663 Post(s)
Liked: 1480
Quote:
Originally Posted by genesplitter View Post
Here is an early test confirming MSO and REW matched with some simple filters. It's scary how close they are.
That is some impressive software with how closely those two match. Amazing!

David Gage
Deep Sea Sound
"You don't listen to our subs, you EXPERIENCE them!"
dgage is offline  
Sponsored Links
Advertisement
 
Reply Subwoofers, Bass, and Transducers

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off