Audiolense solutions for correction of a multi-sub array.

[jjazdk]

It was a bug fix. that enabled hulkss to high pass two subs that ran in parallel with three others.

Btw, jjazdk, did you read my examination of hulkss three subwoofer configurations above? And if so what do you think of my conclusion?

I did, yes.

It is fairly obvious, to me, why the 5 driver subwoofer setup is much more wellbehaved above 100Hz, and therefore also a superior impulse response. Wether that is relevant for a subwoofer, maybe, maybe not. It sure is impressive!

I do however think the MSO solution will often be better, in terms of "power management", i.e. not putting excessive strain on some of the subwoofers but a more equal load on all of them (of course depending how you set up the correction-ranges in MSO).

MSO gives me a really nice control of the individual subwoofers, where the current AL implementation will wrestle my subwoofers into order with no way for me to adjust that wrestling to a more sensible level (pr. subwoofer).
That is why I propose an individual level control of the subwoofers (to take into account each subwoofers output capability).

A correction option that does the following for a minimum phase system:
1. Measures and uses individual delay correction for each subwoofer
2. Lets the user account for output capability (level adjusting the individual subwoofers)
3. Makes one combined frequency correction, based on the summation of the delay-corrected subwoofers
4. Makes one level correction of the combined outputs.

A correction option that does the following for a linear phase system:
1. Measures and uses individual delay and phase (TTD) correction for each subwoofer
2. Lets the user account for output capability (level adjusting the individual subwoofers)
3. Makes one combined frequency correction, based on the summation of the delay and phase corrected subwoofers
4. Makes one level correction of the combined outputs.

I acknowledge that this is a spiral of never ending requests, for a myriad of different options, as our setups get more and more complicated.

An example on why I need the individual level control:
In my current setup my "front subwoofer" consist of 15x10" drivers, on one amplifier channel. Where as my "kitchen" subwoofer consists of 6x10" drivers, and is farther away from MLP. A measurement at MLP of the two subwoofers, driven by the same signal level from the amplifier, yields roughly 12dB more output in most of the frequency range from the front subwoofer than the kitchen subwoofer. When AL tries to correct this, the front subwoofer drivers ends up getting very little power compared to the drivers in the kitchen subwoofers. I also have a "rear subwoofer" with 7x10" drivers, when measured in MLP this has an output somewhere in between the front subwoofer and the kitchen subwoofer.
With my background (electro-acoustic engineer) it makes a lot of sense to have the ability to ensure equal load (within reason of course) for all of the drivers. In my own setup that would be easy, as I just want the same drive level for all my subwoofer drivers (28x10").

 

[hulkss]

My case with the subwoofer is probably different from what MSO is best at. MSO really does work and can do exceptionally well with a distributed group of subs used to randomize and average out room modes.

In my case I selected all "closed back" subs that individually can function as coherent monopole sources. They are all placed in a line array extending fully across the floor/wall room edge with the front main speakers. There are a enough closely spaced sources to launch a coherent cylindrical sound wave into the room. I believe this is much different than spherical waves coming from multiple subs around the room.

In my case matching timing, polarity, phase, and acoustic gain seems to make this sub line array a reality. Audiolense is really good at precision alignment of sound sources.

One thing I am curious about is that the array is "curved" when AL matches the sound arrival timing at a point. Maybe I should measure at several points along a line using speaker by speaker?

 

[HerbertWest]

My reply was to HerbertWest's setup, not yours :-)
Regarding your setup, I agree.. I have somewhat similar challenges in my own setup.

Thank you! I'll try over the weekend and confirm Probably Saturday - speaker measurement sweeps on Valentine could trigger speaker ejection out of the window by my otherwise tolerant wife.

 

[jjazdk]

In my case I selected all "closed back" subs that individually can function as coherent monopole sources. They are all placed in a line array extending fully across the floor/wall room edge with the front main speakers. There are a enough closely spaced sources to launch a coherent cylindrical sound wave into the room. I believe this is much different than spherical waves coming from multiple subs around the room.

In my case matching timing, polarity, phase, and acoustic gain seems to make this sub line array a reality. Audiolense is really good at precision alignment of sound sources.

One thing I am curious about is that the array is "curved" when AL matches the sound arrival timing at a point. Maybe I should measure at several points along a line using speaker by speaker?

When your subwoofers are individually frequency corrected and have individual delays, their summation do not perform like a line array. If you want to achieve a line array, you must feed the 4 identical subs with the exact same signal (and, disconnect the IB sub).
Not that I see any benefit in implementing a line array subwoofer in your room.

 

[Omid]

Hulkss,
All you need now is a line array of subs behind your seating position to play at 180deg to the arriving cylindrical wave to null any reflection, lol...

 

[hulkss]

When your subwoofers are individually frequency corrected and have individual delays, their summation do not perform like a line array. If you want to achieve a line array, you must feed the 4 identical subs with the exact same signal (and, disconnect the IB sub).
Not that I see any benefit in implementing a line array subwoofer in your room.

In my case, each sub is fed from the same source signal. That source signal is modified by a filter differently for each sub so that the acoustic output is the same from each sub. The similar acoustic output is what matters. The filters in Audiolense create a virtual curved array as the corrected sound from each sub arrives at the measurement point at the same time.

Why disconnect the IB? It is a sealed back sub like the others. The back chamber just happens to be very big (the crawl space under my house). It is not a dipole.

 

[hulkss]

Hulkss,
All you need now is a line array of subs behind your seating position to play at 180deg to the arriving cylindrical wave to null any reflection, lol...

Based on what I can hear and measure I don't have a "reflection" problem.

 

[Omid]

yes, I was just kidding...

 

[hulkss]

yes, I was just kidding...

Yes, I know. That approach of suppressing a rear reflection with delayed and inverted phase speakers does work well in rooms designed for it.

 

[jjazdk]

In my case, each sub is fed from the same source signal. That source signal is modified by a filter differently for each sub so that the acoustic output is the same from each sub. The similar acoustic output is what matters. The filters in Audiolense create a virtual curved array as the corrected sound from each sub arrives at the measurement point at the same time.

Similar acoustic output in MLP, from each of the 4 subwoofers is not going to give you the behaviour of a line array subwoofer. A line array radiation pattern, in time and space, requires the same output from each subwoofer measured at the acoustic radiation point, not measured at MLP.
I am not saying your solution is bad, I am merely pointing out that you are not making a line array subwoofer.

Why disconnect the IB? It is a sealed back sub like the others. The back chamber just happens to be very big (the crawl space under my house). It is not a dipole.

Because to make a true line array subwoofer you need each source to have exactly the same output, and unless you correct each subwoofer in the extreme nearfield the IB will not have the same output (frequency and phase) as the Epiks.

Again, I am not saying your solution is bad, as a matter of fact I find it pretty amazing. I was only pointing out thast what you wrote would not create a radiation pattern that could be named a line array.

 

[jjazdk]

Yes, I know. That approach of suppressing a rear reflection with delayed and inverted phase speakers does work well in rooms designed for it.

It works remarkably well, even in less optimal rooms. I have tried it in my own room, which is far from optimal. But, the effect of the DBA is still very much there.

 

[hulkss]

Because to make a true line array subwoofer you need each source to have exactly the same output, and unless you correct each subwoofer in the extreme nearfield the IB will not have the same output (frequency and phase) as the Epiks.

I may try that:
Measure each sub in the near field, at the same distance, to create a correction for each using the same flat target. The wavelengths are 14 feet or longer so slight differences in phase will still be quite "similar". Then I can measure the overall sub array in the main listening position.

I have tested a zero latency convolver plug-in that works with AL filters. I can create the line array corrections, put them into this convolver and then measure through it with AL as part of a system measurement.

 

[jjazdk]

I may try that:
Measure each sub in the near field, at the same distance, to create a correction for each using the same flat target. The wavelengths are 14 feet or longer so slight differences in phase will still be quite "similar". Then I can measure the overall sub array in the main listening position.

I have tested a zero latency convolver plug-in that works with AL filters. I can create the line array corrections, put them into this convolver and then measure through it with AL as part of a system measurement.

That sounds interesting, I may find time to toy with that convolver myself. Good find, thanks!

Does that convolver work as a plug-in for JRiver?

 

[hulkss]

Does that convolver work as a plug-in for JRiver?

It will load, however, JRiver support for plugins is lame and still in development. Supported channel count is limited, the host sample frequency is not being passed, and the buffer sizes are not controllable and cause a lot of latency, I can't load multiple instances of the same plugin, the list goes on.....

I find that this convolver works great in Plogue Bidule as a host. I can run JRiver into my MOTU and then loopback through Bidule and use it instead of JRiver DSP to get stable performance with much less latency. Since the MOTU accepts ASIO or WASAPI I get DSP for any player and it works on Apple Mac too.

I'll see If I can get it working tonight. I just measured the subs up close with AL.

 

[hulkss]

Another subwoofer solution from the discussion above: the "True" Line Array (TLA) as pictured in the first post. This will be a quick "proof-of-concept" test.

1. Create a 5 sub setup and use "speaker-by-speaker" to measure each sub right up close in the near field along a straight line only inches from each sub.
2. Correct all the subs to the same target.
3. Put this correction set into it's own convolver to use as a single sub in the complete system as measured & corrected with Audiolense

OK. Here are the five subs corrected to align in time and frequency. This is a frequency only correction with very little latency.

Now this revealed a limitation, the line can only be as good as it's weakest link. I could not use the deep bass of the IB sub
Not to give up easily, I'll Make a two driver sub with the IB (that's also in the line array).

Here's the line array in a convolver, hosted in Plogue Bidule, ready to be used as a single driver. I loop back through it with my MOTU 16a.
I could only figure out how to use this convolver as a diagonal matrix, so 5 in, 5 out.

Now I set-up a two driver sub with my Infinite Baffle (IB) on the bottom, the "True" Line Array (corrected as above) on top, crossed over at 30 Hz.

Because of that crossover, I will use TTD and the linear XO filters in AL (this will be a real test for sure).

Here's the simulated frequency response in the listening position:

And the step response:

Now is this worth implementing into the full system?

 

[hulkss]

After a review of all the data from the various set-ups in the previous posts, I made the following determinations. I believe this to be true for my sound system, not necessarily all systems:

1. Stop mixing the full range main speakers in with the subwoofers. Yes, they play bass, but great subwoofers they are not.
2. Do mix the LFE channel with the center channel before convolution. The upper LFE frequency content then transitions nicely from the subwoofer into the center main speaker same as left or right bass offloaded bass with the left and right main speakers. No phase cancellation problems from some bass being rerouted through crossover filters and some not.
3. Keep all the main channel bass offloading frequencies the same to avoid phase differences between rerouted bass channels. Do set-up for unique offloading filters with a 0.1 Hz difference between L, R, C channels.
4. Do not offload bass from surround speakers. Just let them play along on their own. Don't put a collection of incoherent bass signals into the main bass mix.
5. In my case: stop correcting the subwoofers individually as a multi-driver sub or separate subs.
6. Use as little correction as needed to get the job done. In my case: stop using MSO filters.

Ok, with that said, I made a simple single sub set-up. All the Epik sub and IB sub drivers (a dozen 15" drivers) playing together in parallel. I guess this was too easy for me to try earlier

Here's the measurement in the main listening position:

I made a vey simple target:

Then I generated a correction with the option "Prevent bass boost" selected:

Resulting in flat frequency response to 6 Hz.

And a decent step response (this is frequency correction only):

With no bass boosting needed (the correction is basically a low pass filter), I've got all the dynamic range the system can deliver. All analog gains were set to reach max amplifier power with the same signal.

 

[2234rew]

After a review of all the data from the various set-ups in the previous posts, I made the following determinations. I believe this to be true for my sound system, not necessarily all systems:

1. Stop mixing the full range main speakers in with the subwoofers. Yes, they play bass, but great subwoofers they are not.
2. Do mix the LFE channel with the center channel before convolution. The upper LFE frequency content then transitions nicely from the subwoofer into the center main speaker same as left or right bass offloaded bass with the left and right main speakers. No phase cancellation problems from some bass being rerouted through crossover filters and some not.
3. Keep all the main channel bass offloading frequencies the same to avoid phase differences between rerouted bass channels. Do set-up for unique offloading filters with a 0.1 Hz difference between L, R, C channels.
4. Do not offload bass from surround speakers. Just let them play along on their own. Don't put a collection of incoherent bass signals into the main bass mix.
5. In my case: stop correcting the subwoofers individually as a multi-driver sub or separate subs.
6. Use as little correction as needed to get the job done. In my case: stop using MSO filters.

Ok, with that said, I made a simple single sub set-up. All the Epik sub and IB sub drivers (a dozen 15" drivers) playing together in parallel. I guess this was too easy for me to try earlier

Here's the measurement in the main listening position:

View attachment 39330

I made a vey simple target:

View attachment 39331

Then I generated a correction with the option "Prevent bass boost" selected:

View attachment 39332

Resulting in flat frequency response to 6 Hz.

View attachment 39333

And a decent step response (this is frequency correction only):

View attachment 39334

With no bass boosting needed (the correction is basically a low pass filter), I've got all the dynamic range the system can deliver. All analog gains were set to reach max amplifier power with the same signal.

Wow, I've been catching up on the last few pages and my head hurts.

Not become it's not interesting - quite the opposite.

I'm fascinated but it seems complex. The good news is I think this is as complex as it gets, so if I can understand this thread, I'm in good shape for the future ! LOL

hulkss, can you please share screenshots of all your Audiolense setup/settings screen now.

So I can look at that in parallel with your recent summary, and piece together the puzzle slowly in my mind.

Appreciated

 

[hulkss]

Wow, I've been catching up on the last few pages and my head hurts.

Not become it's not interesting - quite the opposite.
I'm fascinated but it seems complex. The good news is I think this is as complex as it gets, so if I can understand this thread, I'm in good shape for the future ! LOL
hulkss, can you please share screenshots of all your Audiolense setup/settings screen now.
So I can look at that in parallel with your recent summary, and piece together the puzzle slowly in my mind.
Appreciated

I'll start another thread with the settings I'm working for a TTD correction.

 

[2234rew]

Keep all the main channel bass offloading frequencies the same to avoid phase differences between rerouted bass channels. Do set-up for unique offloading filters with a 0.1 Hz difference between L, R, C channels.

Hi @juicehifi

Regarding quoted, is having different frequencies of 0.1Hz only required for this re-routing of bass?

Or do I need to have all XO points for multi-way DSP speakers having 0.1 Hz differences between L and R channel XO points?

 

[juicehifi]

Only for re-routing

 

[2234rew]

Hi @hulkss

Are you using MSO still currently?

Can you explain why you need this and why Audiolense XO can't do whatever MSO is doing?

cc @juicehifi

 

[hulkss]

Hi @hulkss

Are you using MSO still currently?

Can you explain why you need this and why Audiolense XO can't do whatever MSO is doing?

cc @juicehifi

I am not using MSO at the moment but I will again when I get back into doing multi-seat correction.

From the MSO website: MSO optimizes the flatness of the combined frequency responses of main loudspeakers and multiple subwoofers at multiple listening positions simultaneously. Although it's possible to use it for identical equalization (EQ) of all subs at once, it does not depend on a global EQ approach like commercial "room correction" systems do. Such systems cannot reduce the seat-to-seat variation of the bass response. Instead, MSO performs optimization of EQ, gain and delay parameters individually for each subwoofer. This individual EQ optimization for each sub causes the variation of the bass response from seat to seat to be reduced, and the bass response at each listening position to be flattened.

MSO can be used to perform tasks that include:

• Optimizing subwoofer integration with main speakers through delay, gain and EQ adjustments
• Reducing seat-to-seat frequency response variation of the combined sub outputs through optimized EQ of each sub individually
• Achieving optimum subwoofer time alignment via individual sub delay adjustments
• Improving response flatness of the combined sub outputs at all listening positions
• Creating, specifying and achieving your desired target frequency response curve to shape the bass response to taste

 

[2234rew]

I am not using MSO at the moment but I will again when I get back into doing multi-seat correction.

From the MSO website: MSO optimizes the flatness of the combined frequency responses of main loudspeakers and multiple subwoofers at multiple listening positions simultaneously. Although it's possible to use it for identical equalization (EQ) of all subs at once, it does not depend on a global EQ approach like commercial "room correction" systems do. Such systems cannot reduce the seat-to-seat variation of the bass response. Instead, MSO performs optimization of EQ, gain and delay parameters individually for each subwoofer. This individual EQ optimization for each sub causes the variation of the bass response from seat to seat to be reduced, and the bass response at each listening position to be flattened.

MSO can be used to perform tasks that include:

• Optimizing subwoofer integration with main speakers through delay, gain and EQ adjustments
• Reducing seat-to-seat frequency response variation of the combined sub outputs through optimized EQ of each sub individually
• Achieving optimum subwoofer time alignment via individual sub delay adjustments
• Improving response flatness of the combined sub outputs at all listening positions
• Creating, specifying and achieving your desired target frequency response curve to shape the bass response to taste

Thanks, I saw all that but specifically for you, what did it offer that Audiolense could not do ?

 

[hulkss]

Audiolense could not achieve the bass target response in four different seats at the same time.

 

[2234rew]

Audiolense could not achieve the bass target response in four different seats at the same time.

Thanks, when I saw post #19 with your measurements and simulations in this thread I thought you had achieved what you wanted using only Audiolense (without MSO involved).

Audiolense solutions for correction of a multi-sub array.

I am using Audiolense (AL) to correct a large subwoofer array. In this case, the array has seven subwoofers of three totally different cabinet designs aligned across a 22-foot-wide sound stage. There is a Klipschorn in each of the left and right corners, an Infinite Baffle (IB) sub in the middle...

www.avnirvana.com

If I understood correctly, that was with Audiolense treating the 5 subs as seperate drivers, Audiolense correcting each of the 5 subs individually. As would be the 'normal' approach I guess.

But obviously in the end, Audiolense alone didn't perform 'best' ?