• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Speakers and Subs - FR or XO?

oursmagenta

Active Member
Forum Donor
Joined
Jan 19, 2021
Messages
161
Likes
187
Location
France
Since I'm testing the Dirac Live Bass Management thing, I've seen in their docs that they plan to use the mains (if they have enough low-frequency output) to contribute to the overall bass part of the spectrum.

In this sense, they may go toward what you are thinking @GDK in a holistic way, optimizing (phase, FR etc ...) for all low-frequencies contribution (L + R + Subs).

One would argue that would be some form of smart cross-over (heww I hate this word smart, but I can't find another one right now), but a cross-over still ...
 
  • Like
Reactions: GDK

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
The received wisdom on ASR seems to be that you should always have some form of a cross-over between your mains and your subs. However, is this really always true?

Shall we first see why we use different size drivers and crossover/filters between the amplifier and the drive units?

An electromagnetic drive unit is a high-pass filter. It has a resonance frequency (fc) and its output will slope off at 12dB octave below the -3dB point (f3). In order to have the full audio spectrum you have to make the f3 at or above 20Hz. This requires a heavy cone and a very powerful magnet. A heavy cone cannot produce high frequencies well. You have to split the frequency to multiple bands and use two or more drive units each optimised to a limited frequency range. It can be seen by entering the parameters of the common materials and the available manufacturing techniques that a 2-way speaker with a cross over frequency of between 1-3kHz is the most optimised solution.

In order to make the low-frequency driver work up to 1-3kHz the cone size is limited. This in turns limits how low the f3 can go and/or how loud the speaker can go. However, as the musical range of recorded music is mostly above 40Hz. A speaker with an f3 of around 80Hz, placed against a wall (bookshelf) or on a corner can produce that 40Hz at an acceptable level, allowing us to hear almost the entire acoustic instrument range.

If you are happy with the above then a small 2-way speaker will satisfy you. If you are not you have two options: a 3-way speaker or the addition of a subwoofer. In the 3-way speaker the low frequency driver (woofer) and the enclosure is made larger and a second crossover is used between it and the midrange driver. However, various other factors comes into play and the sound quality gets deteriorated when the enclosure gets larger and the crossovers gets complicated. These makes a 3-way speaker much more expensive than an equivalent quality 2-way speaker. That is why using small 2-way speaker and subwoofers is sometimes preferred.

As the 2-way speaker will be high-pass filter working around 80Hz with a slope of 12dB/oct why not use a subwoofer which is optimised to work as low as 20Hz, with its own amplifier feeding the signal via low-pass active filter is that matches the 2-way speaker's inherent filter is easy? The answer to that is, why not indeed? The subwoofer and the 2-way match perfectly as a symmetrical crossover is formed. However, there is a catch: As the input signal is increased the voice coil can move outside the magnetic field. This will put it outside it's operating zone and the sound quality is dramatically lowered. This can happen even at low levels if a source with high dynamic range is played. You will have distorted peaks.

This is why it is advisable to feed a 2-way speakers via a high-pass filter. If you use a 2nd order filter set at the f3 of the 2-way speaker you can form a perfect 4th octave filter. This is where my old friend Linkwitz RIP came in, inventing the proper way to cross two drive units back in the 70s (Linkwitz-Riley filter). The subwoofer's low-pass filter should then be changed to a 4th order filter so that the cross-over is symmetrical and the 2-way speaker's and the subwoofer's outputs blend seamlessly. (Side note: I have never seen a 4th order filter on a sub, has anyone?)

There should be no argument on above. It is simple physics and elementary electronics. Richard Small wrote his thesis on how loudspeakers work half a century ago, in 1972. Unfortunately the Hi-Fi industry is still producing devices that are plainly breaking the rules and make us question the otherwise obvious.

This is a great forum. If I managed to add something to it I will be happy.
 
Last edited:

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,294
Likes
5,072
Location
Nashville
I'm at a loss, figuring what advantatge you might get by not high passing your mains. Doing so, relieves a burden on them of reproducing the lows where a number of users have pointed out, distortions can reach greater than 10 and possibly as high as 100 percent. Clearly those levels will limit overall system headroom dramatically.

If you're having a hum problem, there's a grounding issue you need to address with better cabling, better paths for those cables (like keeping them away from power cords), or better power filtration. Solve that problem properly, and you'll have no need or desire to forego high passing your mains.
 

johnp98

Active Member
Joined
Jan 29, 2021
Messages
133
Likes
201
As the 2-way speaker will be high-pass filter working around 80Hz with a slope of 12dB/oct why not use a subwoofer which is optimised to work as low as 20Hz, with its own amplifier feeding the signal via low-pass active filter is that matches the 2-way speaker's inherent filter is easy? The answer to that is, why not indeed? The subwoofer and the 2-way match perfectly as a symmetrical crossover is formed. However, there is a catch: As the input signal is increased the voice coil can move outside the magnetic field. This will put it outside it's operating zone and the sound quality is dramatically lowered. This can happen even at low levels if a source with high dynamic range is played. You will have distorted peaks.

This is a great forum. If I managed to add something to it I will be happy.

I am wondering if a sealed design (or plugging the port) of a 2 way speaker changes anything substantial?
My understanding is that it will lower the low frequency output and then protect the cone from over excursion below the prior port tuning frequency. But does that effectively put a high pass filter on it? Or should that also be done electronically (either with a crossover, or electronically with 'bass management') when crossing over to the subs.

I guess I could see there being utility in having a larger area of overlap between the subs and the mains, as numerous low frequency sources is better for an overall flat response.

Thoughts?
 

oursmagenta

Active Member
Forum Donor
Joined
Jan 19, 2021
Messages
161
Likes
187
Location
France
Shall we first see why we use different size drivers and crossover/filters between the amplifier and the drive units?

An electromagnetic drive unit is a high-pass filter, hence has a resonance frequency (fc). Its output will slope off at 12dB octave below the -3dB point (f3). In order to have the full audio spectrum you have to make the f3 at or above 20Hz. This requires a heavy cone and a very powerful magnet. A heavy cone cannot produce high frequencies well. You have to split the frequency to multiple bands and use two or more drive units each optimised to a limited frequency range. It can be seen by entering the parameters of the common materials and the available manufacturing techniques that a 2-way speaker with a cross over frequency of between 1-3kHz is the most optimised solution. Hence, the 2-way speaker is born.

In order to have the low-frequency driver to work up to 1-3kHz the cone size is limited. This in turns limits how low the f3 can go and/or how loud the speaker can go. However, as the musical range of recorded music is mostly above 40Hz. A speaker with an f3 of around 80Hz, placed against a wall (bookshelf) or on a corner can produce that 40Hz at an acceptable level, allowing us to hear almost the entire acoustic instrument range.

If you are happy with the above then a small 2-way speaker will satisfy you. If you are not you have two options: a 3-way speaker or the addition of a subwoofer. In the 3-way speaker the low frequency driver (woofer) and the enclosure is made larger and a second crossover is used between it and the midrange driver. However, various other factors comes into play and the sound quality gets deteriorated when the enclosure gets larger and the crossovers gets complicated. These makes a 3-way speaker much more expensive than an equivalent quality 2-way speaker. That is why using small 2-way speaker and subwoofers is sometimes preferred.

As the 2-way speaker will be high-pass filter working around 80Hz with a slope of 12dB/oct why not use a subwoofer which is optimised to work as low as 20Hz, with its own amplifier feeding the signal via low-pass active filter is that matches the 2-way speaker's inherent filter is easy? The answer to that is, why not indeed? The subwoofer and the 2-way match perfectly as a symmetrical crossover is formed. However, there is a catch: As the input signal is increased the voice coil can move outside the magnetic field. This will put it outside it's operating zone and the sound quality is dramatically lowered. This can happen even at low levels if a source with high dynamic range is played. You will have distorted peaks.

This is why it is advisable to feed a 2-way speakers via a high-pass filter. If you use a 2nd order filter set at the f3 of the 2-way speaker you can form a perfect 4th octave filter. This is where my old friend Linkwitz RIP came in, inventing the proper way to cross two drive units back in the 70s (Linkwitz-Riley filter). The subwoofer's low-pass filter should then be changed to a 4th order filter so that the cross-over is symmetrical and the 2-way speaker's and the subwoofer's outputs blend seamlessly. (Side note: I have never seen a 4th order filter on a sub, has anyone?)

There should be no argument on above. It is simple physics and elementary electronics. Richard Small wrote his thesis on how loudspeakers work half a century ago, in 1972. Unfortunately the Hi-Fi industry is still producing devices that are plainly breaking the rules and make us question the otherwise obvious.

This is a great forum. If I managed to add something to it I will be happy.

Pretty interesting post.

Now you could have a setup for which there is a broad overlap region (FR wise) where the subs and the mains can output decent energy/spl with controlled distortion.

If you have that, there is no hard reason that an algorithm couldn't find the right balance between the L, R and subs (not in the form of a typical crossover but more in terms of a general time-domain filter) to even out the overall response.
 
Last edited:

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
I am wondering if a sealed design (or plugging the port) of a 2 way speaker changes anything substantial?
My understanding is that it will lower the low frequency output and then protect the cone from over excursion below the prior port tuning frequency. But does that effectively put a high pass filter on it? Or should that also be done electronically (either with a crossover, or electronically with 'bass management') when crossing over to the subs.
A ported speaker exhibits even worse behaviour than a closed box speaker when it comes to cone being out of control below f3. The port will reduce the cone excursion around the port resonance (different to f3) but it will not help below the vent resonance.

I guess I could see there being utility in having a larger area of overlap between the subs and the mains, as numerous low frequency sources is better for an overall flat response.
The overlap as you call it can only be one way, via matching high & low pass filters i.e. a cross over. Anything else will cause frequency and phase anomalies. A cross-over is not something you tune by ear!
 
Last edited:

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
.. there is no hard reason that an algorithm couldn't find the right balance between the L, R and subs (not in the form of a typical crossover but more in terms of a general time-domain filter) to even out the overall response.

There is no reason, full-stop. It is filter theory. Anyone who studied electrical/electronic engineering knows it. There are loads of even free programs around that can calculate the basics. There are also specific programs that calculate the entire speaker system. This one is the best and has been around since the early 90s.
 

oursmagenta

Active Member
Forum Donor
Joined
Jan 19, 2021
Messages
161
Likes
187
Location
France
A ported speaker exhibits even worse behaviour than a closed box speaker when it comes to cone being out of control below f3. The port will reduce the cone excursion at and and around the port resonance (different to f3) but it will not help below the vent resonance.


The overlap as you call it can only be one way, via matching high * low pass filters i.e. a cross over. Anything else will cause frequency and phase anomalies. A cross-over is not something you tune by ear!

Not by ear, but algorithmically I'd say yes. One way to look at it very crudely would be to use the mains as subs in their distortion controlled region, leveraging the fact that L, R, and subs are at different room locations.
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Not by ear, but algorithmically I'd say yes. One way to look at it very crudely would be to use the mains as subs in their distortion controlled region.
I don't follow you. Please elaborate.
 

oursmagenta

Active Member
Forum Donor
Joined
Jan 19, 2021
Messages
161
Likes
187
Location
France
I don't follow you. Please elaborate.
Imagine your 2 mains can have a controlled distortion (under 1%) down to 40hz. Now you have a sub that do 16 to 150hz.
If the setup would benefit from having a sub at the L and R location to excite the right room modes and fill-in a dip at 60hz perceived at the listening location. Then instead of crossing at say 40hz (because past that you loose the controlled distortion of the mains), you could use the 3 speakers to help excite correctly the room to fill in the dip.
With a basic crossover the speakers cannot help each other in the bass region to fill-in dips. It's either the mains (together) or the sub(s), never all of them hand to hand together.

N.B: of course finding the right eq + phase setup for all the speakers optimizing the response for one or multiple listening position would be very hard (or near impossible) by ear or even by trial and error with measurements. That's why I mentioned algorithmically.
 
Last edited:

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Imagine your 2 mains can have a controlled distortion (under 1%) down to 40hz. Now you have a sub that do 16 to 150hz.
If the setup would benefit from having a sub at the L and R location to excite the right room modes and fill-in a dip at 60hz perceived at the listening location. Then instead of crossing at say 40hz (because past that you loose the controlled distortion of the mains), you could use the 3 speakers to help excite correctly the room to fill in the dip.
With a basic crossover the speakers cannot help each other in the bass region to fill-in dips. It's either the mains (together) or the sub(s), never all of them hand to hand together.

N.B: of course finding the right eq + phase setup for all the speakers optimizing the response for one or multiple listening position would be very hard (or near impossible) by ear or even by trial and error with measurements. That's why I mentioned algorithmically.
From purely electrical point of view the cross-over frequency should be set at 40Hz for the sub and for the main speakers. Anything else will cause the combined signal to be irregular.

However, at 40Hz the room modes have so much power that the relatively minor signal anomalies caused by miss-matched filters will be irrelevant. Anything lower than around 70Hz, where the wavelength of the sound is 5m, the room modes will be the main effect on frequency response. Unless you have a room that is considerably larger than 5m, the lowest crossover frequency you should use is around 70Hz. On smaller rooms there is no practical and repeatable way to calculate (even measure) the resulting frequency response. Intelligent DSP based filters like Dirac will do a much better job.
 
OP
GDK

GDK

Addicted to Fun and Learning
Forum Donor
Joined
Aug 8, 2020
Messages
560
Likes
1,525
Location
Toronto
I'm at a loss, figuring what advantatge you might get by not high passing your mains. Doing so, relieves a burden on them of reproducing the lows where a number of users have pointed out, distortions can reach greater than 10 and possibly as high as 100 percent. Clearly those levels will limit overall system headroom dramatically.

If you're having a hum problem, there's a grounding issue you need to address with better cabling, better paths for those cables (like keeping them away from power cords), or better power filtration. Solve that problem properly, and you'll have no need or desire to forego high passing your mains.
Well, solely in my case it is simply because I am not suffering from significant LF distortion or cancellations between my mains and subs. So I don’t think it will offer me anything. Therefore, there is not really any reason to worry about the noise issue from the cross-over (which is a noticeable hiss, so I don’t believe it’s a ground loop).

I do think there may be some debate around what level of LF distortion is noticeable. March Audio started a thread on this point, but IIRC it got side-tracked and then, well, we know what happened next. :eek:
 

oursmagenta

Active Member
Forum Donor
Joined
Jan 19, 2021
Messages
161
Likes
187
Location
France
From purely electrical point of view the cross-over frequency should be set at 40Hz for the sub and for the main speakers. Anything else will cause the combined signal to be irregular.

However, at 40Hz the room modes have so much power that the relatively minor signal anomalies caused by miss-matched filters will be irrelevant. Anything lower than around 70Hz, where the wavelength of the sound is 5m, the room modes will be the main effect on frequency response. Unless you have a room that is considerably larger than 5m, the lowest crossover frequency you should use is around 70Hz. On smaller rooms there is no practical and repeatable way to calculate (even measure) the resulting frequency response. Intelligent DSP based filters like Dirac will do a much better job.

Not sure where you are getting at. I will try another angle.

There is a good amount of research work on how multiple subs can even out the bass spatial distribution and FR.

If you have L and R that can act as subs you can count them as virtual subs (you should definitely reduce their output past their distortion controlled region).

To match the multi sub framework you could even really construct virtual subs. Like having L and R virtually crossed at Xhz resp. Yhz, and then having L-Sub and R-Sub (virtual subs crossed at Xhz resp. Yhz) the real sub crossed also at Zhz. And feed the multisub optimization framework with this virtual setup: L, R, L-sub, R-sub, sub(s).

The last paragraph is really for the sake of the argument, full optimization shouldn't need virtual crossover at all.
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Not sure where you are getting at. I will try another angle.

There is a good amount of research work on how multiple subs can even out the bass spatial distribution and FR.

I was trying to explain the electrical response of speakers, not their interaction with the room i.e. acoustics. You are talking about the acoustics. As I said earlier when it comes to low frequency response of a room, the room modes will make much more effect on the frequency response then the electrical filters. The thread is about filters, hence I have only covered that.
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
As the 2-way speaker will be high-pass filter working around 80Hz with a slope of 12dB/oct why not use a subwoofer which is optimised to work as low as 20Hz, with its own amplifier feeding the signal via low-pass active filter is that matches the 2-way speaker's inherent filter is easy? The answer to that is, why not indeed? The subwoofer and the 2-way match perfectly as a symmetrical crossover is formed. However, there is a catch: As the input signal is increased the voice coil can move outside the magnetic field. This will put it outside it's operating zone and the sound quality is dramatically lowered. This can happen even at low levels if a source with high dynamic range is played. You will have distorted peaks.

This is why it is advisable to feed a 2-way speakers via a high-pass filter. If you use a 2nd order filter set at the f3 of the 2-way speaker you can form a perfect 4th octave filter. This is where my old friend Linkwitz RIP came in, inventing the proper way to cross two drive units back in the 70s (Linkwitz-Riley filter). The subwoofer's low-pass filter should then be changed to a 4th order filter so that the cross-over is symmetrical and the 2-way speaker's and the subwoofer's outputs blend seamlessly. (Side note: I have never seen a 4th order filter on a sub, has anyone?)

Thank you for posting this, I see people ignore that a 2/3-way will have it's own high pass behavior on the low end when implementing crossovers all the time. Presumably this is the case because unless you DIY and/or take measurements you will not understand the high pass behavior, easier to be ignorant and set the LPF and HPF with same the slope / f3 and move on.

I am a big fan of the approach you describe, combining a sealed 2/3 way with a 2nd order HPF to achieve 4th order high pass acoustic behavior and then use a 4th order LPF on the sub. A miniDSP is really helpful here as you can use a Linkwitz Transform to fine tune the acoustic high pass acoustic behavior prior to applying an electronic HPF to achieve the exact crossover point / alignment you want.

Michael
 

oursmagenta

Active Member
Forum Donor
Joined
Jan 19, 2021
Messages
161
Likes
187
Location
France
On the other side, if you are running your mains full range then you will have multiple bass sources that could (should?) help with room correction.

@sarumbear, OP was a bit broader than just asking a to cross or not to cross question.
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Thank you for posting this, I see people ignore that a 2/3-way will have it's own high pass behavior on the low end when implementing crossovers all the time. Presumably this is the case because unless you DIY and/or take measurements you will not understand the high pass behavior, easier to be ignorant and set the LPF and HPF with same the slope / f3 and move on.

I don't think they ignore, they simply do not know. It is almost impossible to build and run a petrol engine without knowing how it works but it is perfectly possible to build a speaker that sounds. I think that is why DIY loudspeaker builders don't feel the need to learn. Whereas an engine mechanic has to at least go to a course.

What I summarised above is not difficult to learn. There are lots of resources. However, most DIYers prefer to rely on bits of information, which is like hearsay and try to make sense of it by making assumptions. In basic science there is often one way of solving a problem and after 50 years of analysis it is pretty well documented.
 

Chrispy

Master Contributor
Forum Donor
Joined
Feb 7, 2020
Messages
7,751
Likes
5,910
Location
PNW
I don't think they ignore, they simply do not know. It is almost impossible to build and run a petrol engine without knowing how it works but it is perfectly possible to build a speaker that sounds. I think that is why DIY loudspeaker builders don't feel the need to learn. Whereas an engine mechanic has to at least go to a course.

What I summarised above is not difficult to learn. There are lots of resources. However, most DIYers prefer to rely on bits of information, which is like hearsay and try to make sense of it by making assumptions. In basic science there is often one way of solving a problem and after 50 years of analysis it is pretty well documented.

How did you come to these specific conclusions about diy folk? Experience?
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
How did you come to these specific conclusions about diy folk? Experience?

Yeah, I do not really understand the comments on DIY folks (although I am admittedly biased being a DIYer myself :)). In my experience DIYers are always trying to learn something new and know/understand much more than typical "audiophile" consumers.

My comments regarding not thinking about the main acoustic high pass were made with a typical consumer in mind trying to set up a sub/main crossover with commercial products.

Michael
 

nerdoldnerdith

Senior Member
Joined
Sep 10, 2020
Messages
495
Likes
690
Location
Chicago
Try using a Driverack PA2 instead of the MiniDSP. It has balanced XLR inputs and outputs, which will allow you to get rid of that annoying ground loop.
 
Top Bottom