Subs & Satellites - can it be done right?

[Ultrasonic]

- Finally, many are concerned about localization of the subs and 80hz seems to be the number people agree on. I also don't think the physics supports this notion. And it is the 90-120hz that I always find the most difficult to get smooth in my room and I don't think I could do it without crossing the subs higher. It is possible I am fooling myself, but I can't locate subs crossed at 120-130hz.

I'm currently using 110 Hz actually but I've not tried higher. One key factor is how rapidly the crossovers fall-off. A typical AV 80 Hz crossover uses a very low roll-off whereas I use 48 dB/Octave LR crossovers. I'm guessing you probably use fairly steep crossovers too?

 

[Bill Brown]

At this point I only use the XO in the subs, 24 db/octave. In the spirit of full disclosure, I must confess that if I listen to one in isolation, I can hear frequencies above the XO (obviously), but I think (? hope) that these are masked when the full system is playing. I can also see full system measurement effects above the XO as the level falls, so it must be taken into account. Hopefully I will have a chance to experiment with this at some point as I change my system architecture to more digital XOs (from the current passive mains and line-level subs).

Bill

 

[MikeFromNorway]

Once again Thanks to everyone the time and insight. Must say i am overwelmd by the knowledge and the effort to help!

So many different opinions raises a simple, though maybe stupid question.

Could i trust the automatization in the Neumann DSP monitor and sub control?

Wouldn’t the smoothest (easiest) thing for a newbe like me be to just go neumann kh420+3x750dsp sub, get their mic and monitor control software, let it do the Magic. Hook everything up to a minidsp shd studio, let Dirac do the fine tuning and trust the automation put into these technologies?

 

[Ultrasonic]

Could i trust the automatization in the Neumann DSP monitor and sub control?

Wouldn’t the smoothest (easiest) thing for a newbe like me be to just go neumann kh420+3x750dsp sub, get their mic and monitor control software, let it do the Magic. Hook everything up to a minidsp shd studio, let Dirac do the fine tuning and trust the automation put into these technologies?

I know nothing about the Neumann system but for information Dirac Live on an SHD Studio will do nothing to optimise time delay integration. The SHD Studio has the functionality to do so (independent of Dirac Live) but you'll need to make measurements yourself in REW to decide what they should be.

 

[Bill Brown]

Wouldn’t the smoothest (easiest) thing for a newbe like me be to just go neumann kh420+3x750dsp sub, get their mic and monitor control software, let it do the Magic. Hook everything up to a minidsp shd studio, let Dirac do the fine tuning and trust the automation put into these technologies?

Using that approach, with perhaps a little thought about where to place the subs before turning on the dsp, you will likely be enjoying amazing, near state-of-the-art reproduction in short order and avoid the rabbit-hole of the minutia, especially if you do some acoustic treatment. A straight path to enjoying the music. I have anticipatory envy of you .

Bill

 

[feynman]

Would Dirac even be necessary there?

 

[dshreter]

Would Dirac even be necessary there?

I’m not sure Neumann supports FIR filtering, just PEQ and phase. So theoretically there’s benefit to Dirac’s capabilities. Would be good for a Neumann expert to verify.

 

[sigbergaudio]

At this point I only use the XO in the subs, 24 db/octave. In the spirit of full disclosure, I must confess that if I listen to one in isolation, I can hear frequencies above the XO (obviously), but I think (? hope) that these are masked when the full system is playing. I can also see full system measurement effects above the XO as the level falls, so it must be taken into account. Hopefully I will have a chance to experiment with this at some point as I change my system architecture to more digital XOs (from the current passive mains and line-level subs).

Bill

Most people seem to be fine with a relatively high crossover without being bothered by localization. 80hz was selected by THX because at that point almost no one has a problem with localization even with very suboptimal subwoofer placement. This means most people with a more optimal placement can get away with far higher crossovers. Localization is also often due to noise/distortion in the form of mechanical noise from poor subwoofers, noise from things vibrating close to the subwoofer (activated by the vibration from the sub), etc.

It's obviously easier to hear it if nothing else is playing, and also it's easy to think you can hear it because after all you know it's there.

 

[sigbergaudio]

That could certainly be true

And definitely nothing I write is the gospel.

On the other hand, according to some experts that I respect, my approach isn't too far afield. Though perhaps the concepts are not well-known here and so seem "out of the box."

I don't want to clutter the thread, but if curious, one could start here:

http://www.gedlee.com/Papers/multiple%20subs.pdf

Multiple Subwoofer after Geddes

For those that want to go deeper only, there is a long thread on the topic. You can see the approach over many years, responses to the many criticisms I see here, etc., weigh the ideas and see what you think:

Multiple Small Subs - Geddes Approach

After reading about the AudioKinesis "Swarm" sub system and the Geddes paper on loudspeakers, I am fairly convinced that 3-4 randomly placed small subs are well worth trying in both 2-channel and HT rooms. Geddes specifically recommends multiple and different (not matching) subs to force more...

www.diyaudio.com

Bill

I am familiar with Geddes. I have not read the diyaudio.com thread, thanks for the tip - will definitely dive into that. His approach (at least as initially described) largely ignores the last decade (couple of decades?) of improvement in DSP, which allows us to not just even out frequencies by adding subwoofers, but also by actively employing EQ, which changes the game a bit. Also, as most other things, it's a matter of compromise. The benefit of highpassing the speakers (in addition to smoother integration in many cases) is significant increase in capacity in the midrange, since the speaker now aren't forced to play deep bass, they can play louder without the woofers bottoming out. The result is lower distortion and cleaner playback. In a 2-way system this is also the same driver that plays all the way up into the midrange. Good and even bass is important, but midbass is also important, and often neglected.

 

[Sancus]

I don't want to clutter the thread, but if curious, one could start here:

http://www.gedlee.com/Papers/multiple%20subs.pdf Multiple Subwoofer after Geddes

Here's the problem with Geddes. His baseline research on how modes work is great and correct. His specific advice on how to integrate subwoofers is a combination of nonsense and outdated at this point. Here are the reasons:

1) Most of the stuff you'll see from him in videos are based on him, a PhD in acoustics, optimizing specific setups(his own, clients, etc). So his advice to run full range mains is literally for his specific setup with extremely large mains. If you don't have woofers as big as most people's subwoofers, then it's bunk. If you do then yeah sure go ahead, but...

2) He contradicts himself on various points in different videos and papers with how he suggests to set delays and what to EQ and what not to EQ among other things. This is not really a criticism of his research. It's more the result of the fact that he said different things at different times regarding different setups. He never kept one source updated with all of his detailed, final conclusions and advice. Instead there are snippets from different stages of his thoughts in different contexts.

3) His explanations are very vague, because he wanted people to send him measurements so he could run software on them and then provide the settings. Any information about how that software worked was never disclosed, and his papers certainly do not contain enough information to even guess at the algorithms used. I'll quote from "Setting Up Multiple Subwoofers."

The other point that I want to make is that the use of filters and EQ in a multiple sub situation is going to be highly non-intuitive.This is because the summation of all the subs with the mains is hyper phase dependent and with gain changes there are also phase changes. It is not at all uncommon to find that lowering the level of some sub in a certain band will actually raise the level of that band and visa-versa. Basically the complex summation of the mains and all the subs is not something that can be readily predicted except by a massive calculation in a computer.It would be impossible for a human to do such a calculation in their head, for example.

And later on...

Based on a spatial average of the listening area (do not use a single point – please) adjust the available parameters of the first sub to achieve the best possible transition from this sub to the mains.Depending on the number of available degrees-of-freedom this could be an arduous task.(That’s why I developed a computer program to do it!)But heck, DIY’s and audiophile’s will do anything for the best possible sound.Right!?

Geddes quite literally tells you to do something that he describes as a "highly non-intuitive, arduous task." This is something HE considered too difficult to do by hand, which is why he wrote software for it. And then made jokes about how audiophiles would think they could do it manually. And keep in mind his level of education. To suggest we random people setting up home theatres can manually adjust multiple subwoofers to perform a Geddes optimization and end up with a result anywhere near optimal is absurd.

The current updated, modern approach to this is Multi-Sub Optimizer. You can certainly use Geddes' advice about placement with that software if you like, and that advice is still valid.

 

[sword]

Multiple subs placed on the sides of the room can fill bass room modes in a way that full range speakers in their proper position can't. Full range speakers with multiple subs and room treatment would be the ultimate.... if you can afford it, have the room, and have a very understanding partner.

Subs should be located where they work best, within the limits of the room arrangement, WAF, etc. The sides of the room may not be best, and should not be a general rule.

 

[Filio45]

Subs should be located where they work best, within the limits of the room arrangement, WAF, etc. The sides of the room may not be best, and should not be a general rule.

Such as directly under stereo speakers...or is that generally a lazy choice of location? Just curious

 

[tvrgeek]

I have to agree with Sigberg regarding not needing a full octave below the crossover frequency. If you are doing a 24 dB/octave crossover, and the signal is obviously -6 dB at the crossover point, then an octave lower it is ~30 dB lower (I know log scale math doesn't exactly work this way). An octave below crossover the attenuation of the sub is less than -0.1 dB. It doesn't matter if you add in a -30 dB signal or a -100 dB signal to the -0.1 dB, you end up with a flat signal. In fact, a brickwall cutoff a half octave below crossover still results is <0.1 dB attenuation. A bookshelf that is flat to mid-to-low 60 Hz range is fine to cross at 80 Hz.

The biggest problem with crossing small monitors is not where they roll off, it is the level they can drive before they get crazy distortion. A lot of 5.25 inch monitors have 5%+ THD at >100 Hz near loud-ish but possible listening levels - and those speakers would be really hard to cross over.

What do you consider "loud-ish" ? That does seem to vary as what I think is loud others may consider background.
I agree, a 5 1/2 is marginal for a full orchestra live performance levels, where a 6 1/2 will do better, but it depends a lot on the driver. The new Purifi as an example with higher X-Max than most. I cross my 6 1/2's @ 70 and they play plenty loud for me. Their F3 is around 50. The 5 1/2's I just built as a new center are crossed @ 80, but there are two of them. I tuned the box @ 80 so the combination is a 4th order acoustic. My sub has a 4th order on it. The drivers are capable of solid 50 Hz if in a larger ported enclosure.

Most AVR's cross second order, not fourth. Most sub plates are second order. I model my speakers and target less than half X-Max as the distortion climbs up as excursion goes up. A bit conservatives. Some suggest 75% is a safe target. With DSP, you can do as you please.

What really kills some sat/sub setups is not having a high pass on the sats. Then you will drive them too hard and get the distortion Turtle mentioned.

As far as set up, for music, I have always found the best place for two subs is under the sats. Easiest to manage the phase and if the sats are in a decent place away from things, it blends well. That does raise the question why not to just buy a monkey coffin in the first place. Good old 10-4-1 full range. Hard to beat actually.

 

[MikeFromNorway]

What do you consider "loud-ish" ? That does seem to vary as what I think is loud others may consider background.
I agree, a 5 1/2 is marginal for a full orchestra live performance levels, where a 6 1/2 will do better, but it depends a lot on the driver. The new Purifi as an example with higher X-Max than most. I cross my 6 1/2's @ 70 and they play plenty loud for me. Their F3 is around 50. The 5 1/2's I just built as a new center are crossed @ 80, but there are two of them. I tuned the box @ 80 so the combination is a 4th order acoustic. My sub has a 4th order on it. The drivers are capable of solid 50 Hz if in a larger ported enclosure.

Most AVR's cross second order, not fourth. Most sub plates are second order. I model my speakers and target less than half X-Max as the distortion climbs up as excursion goes up. A bit conservatives. Some suggest 75% is a safe target. With DSP, you can do as you please.

What really kills some sat/sub setups is not having a high pass on the sats. Then you will drive them too hard and get the distortion Turtle mentioned.

As far as set up, for music, I have always found the best place for two subs is under the sats. Easiest to manage the phase and if the sats are in a decent place away from things, it blends well. That does raise the question why not to just buy a monkey coffin in the first place. Good old 10-4-1 full range. Hard to beat actually.

The simplest answer which has already been mentioned in this thread is that full range speakers require space. Being able to put a pair of quality sats as the Neumann KH420 in their recommended distance from the back wall which is less then 20 cm us a hughe pluss in livingroom integration with kids and Animals. Neumann recommends placement less then 80cm to the back wall measured from the front baffle, since the speaker is 65cm deep its practicly «on» the front wall if you tow them 30 degrees.

Using a sub under each speaker is the plan. Since they also recommend Max 2m between subs if used as an array i figured getting 3, putting one in between the sub+speaker towers and use it as a table for the minidsp shd studio control as a good solution. Both for sound snd waf.

Please correct me if i’m wrong. These recommendations are taken from the user manuals.

 

[Bill Brown]

Most people seem to be fine with a relatively high crossover without being bothered by localization. 80hz was selected by THX because at that point almost no one has a problem with localization even with very suboptimal subwoofer placement. This means most people with a more optimal placement can get away with far higher crossovers. Localization is also often due to noise/distortion in the form of mechanical noise from poor subwoofers, noise from things vibrating close to the subwoofer (activated by the vibration from the sub), etc.

It's obviously easier to hear it if nothing else is playing, and also it's easy to think you can hear it because after all you know it's there.

Great post. I agree. Last night I tried walking to the subs and listening while the full system was playing. While looking at them, I couldn't tell they were making sound. The bass still seemed to be "everywhere." Certainly if they make mechanical sounds they are of higher frequency and can be located. When I sat back down, acoustic or electric basses, orchestral bass drums were located correctly in the soundfield (? the ear locking onto their higher frequency components), while deep bass seemed to originate in the center, then filling up the room.

Bill

 

[Bill Brown]

His approach (at least as initially described) largely ignores the last decade (couple of decades?) of improvement in DSP, which allows us to not just even out frequencies by adding subwoofers, but also by actively employing EQ, which changes the game a bit. Also, as most other things, it's a matter of compromise. The benefit of highpassing the speakers (in addition to smoother integration in many cases) is significant increase in capacity in the midrange, since the speaker now aren't forced to play deep bass, they can play louder without the woofers bottoming out. The result is lower distortion and cleaner playback. In a 2-way system this is also the same driver that plays all the way up into the midrange. Good and even bass is important, but midbass is also important, and often neglected.

Definitely. His work dates back decades, and the diyaudio.com thread dates from 2008, so you see significant evolution/changes in his approach. Rather than looking for a cookbook, I learned from the back and forth, weighing the ideas, supporting scientific explanations, etc.

And yes, high-passing small mains would be required to fully benefit from the increased dynamic capabilities of adding multiple subs. He uses 15" pro woofers, previously vented, now sealed, and we know that pro woofers in relatively small sealed enclosures will lead to a fairly high F3.

My mains have a vented 12" woofer (I have thought a lot about plugging the port if only to see the effects of the lower slope/different phase characteristics, but haven't experimented with it yet). With four subs, I have a low shelf to bring the bass into balance. Combined with parametric eq at my stubborn 35Hz and 78Hz nodes, the mains' woofers are receiving less signal. Last night I played music at a very high level and they moved much less than I was expecting.

"Good and even bass is important, but midbass is also important, and often neglected."

I think so too. I have found that getting < 100hz is fairly straightforward. It is 100-200hz that is the struggle. I have been in a (probably ridiculous, or maybe just insane, per my family, as they listen to seemingly endless signal sweeps) experimental phase with mains placement and sub adjustment over the last week. Thankfully after last night I am back to "leave it alone and listen to music, dummy."

I am nervous about showing images of how it went, but I am pleased with my 100-200hz results, so will show them and they are perhaps illustrative.....please ignore if too much! Or maybe pictures are worth a thousand words? Above 300 hz there are SBIR effects and multiples of room modes. I have convinced my wife to do some radical acoustics work (after I finish the back porch project....), mainly the ceiling and rear wall (currently just treating first reflection points), so this will get better. Right now I am just using PEQs in my RME DAC to fine tune. First two subs are under the mains, fed L/R signals, other two subs fed mono. All 1/48 octave smoothing, 50db scale with target room curve I was shooting for:

Mains:

Add sub under left main (blue):

Add sub under right main (orange):

Clearly, because they are co-located with the mains, their effect on modes isn't huge, though there is some "moving around" and a bit of filling in the nulls.

Add Left rear sub (green):

Add fourth sub, R rear. Fascinatingly, the response was good, but there was one deep null at around 120hz remaining. I thought hmmm...let me try moving it a bit. After 3 or 4 moves I ended up with this (pinkish?):

Mains only and mains with 4 subs, no EQ:

After a bass shelf starting at approx. 85hz (relieving the mains to some degree as described above and 2 bands of parametric EQ in the RME (35 and 78Hz):

Sounds pretty good to me at this point. You will notice I don't have huge extension (subs are sealed 10" JBL studio units), but there is good dynamics and "power." I did have a bit of a learning curve to be able to do this in fairly short order, and I am NOT suggesting the OP (or anyone, really) takes this route. It is only for illustrative purposes. With his proposed equipment he will have an easier go of it with their built-in capabilities. I don't know how I could have smoothed > 100hz without higher crossovers.

Now I am praying that when I hit "reply" my formatting will stay and this won't look like a jumbled mess......

Bill

 

[levimax]

The simplest answer which has already been mentioned in this thread is that full range speakers require space. Being able to put a pair of quality sats as the Neumann KH420 in their recommended distance from the back wall which is less then 20 cm us a hughe pluss in livingroom integration with kids and Animals. Neumann recommends placement less then 80cm to the back wall measured from the front baffle, since the speaker is 65cm deep its practicly «on» the front wall if you tow them 30 degrees.

Using a sub under each speaker is the plan. Since they also recommend Max 2m between subs if used as an array i figured getting 3, putting one in between the sub+speaker towers and use it as a table for the minidsp shd studio control as a good solution. Both for sound snd waf.

Please correct me if i’m wrong. These recommendations are taken from the user manuals.

Using subs under satelite speakers is just like having full range speakers... good LF extension but room modes get excited. In most cases bass response will be quite uneven and EQ will only do so much. While "easy" I think you are losing a lot of the benefits of subs i.e. using strategic locations and DSP to fill in the room modes.

 

[sigbergaudio]

@Bill Brown That's certainly a great result, looks like a lot of work but well invested.

I'm curious, based on your target curve it apperas 80dB is your "flat" level, indicating that your total lift here is around 5dB, and at 100hz you're only 1-2dB up. That looks like a pretty lean target.

Suggestion the next time you feel like spending a week experimenting: Start your lift at 300hz instead of 200hz, and aim for maximum lift (+5dB in your case) at around 100hz, and then flat the rest of the way.

So something along the lines of the target drawn in orange below. Would be interesting to hear how you find that.

 

[Bill Brown]

Thank you. And definitely worth a try. Thanks. And yes, I am fairly new to room curves in the bass, initially basing it on a best guess from this:

But experimentation is certainly in order. Such a critical area, 100-200hz.....lots of energy there in music.

Bill

 

[Ultrasonic]

Such as directly under stereo speakers...or is that generally a lazy choice of location? Just curious

One of the key arguments for the use of subwoofers rather than simply having a pair of very large main speakers is that subwoofers can be located independently of the drivers that cover the higher frequency range. The locations of the latter are limited by the need to provide the stereo image whereas a subwoofer (or two, or more) can go anywhere you want. The chances of the locations under the main speakers being optimal is essentially zero, which is not to say that they can never be made to work (just as very large main speakers can, and more so the larger the room).

One way to get an appreciation of how different subwoofer locations can affect the bass response at the listening position is to have a play with the Room Simulator in REW.