A pretty common question, I know, and I have a reasonable idea of what to do, but I’d like some other opinions before diving in.
This is for a live band (smallish events) and DJ work. No HT, no EDM, no raga parties. (g)
I have complete digital control over eq and such. (Currently within my digital mixer, but I will be getting the EAW UX3600 soon.)
The subs have yet to be determined, but they will be a dual 12” design, either TH (similar to the Danley TH212) or a vented box @ around 6-7 cubic feet.
(I plan to build both, and see which works best for me.)
I the past, I’d just cross at about 100 hz, and then adjust the eq and amp levels by ear, but that was back when I had less control. Plus, the KF394’s may require a little extra finesse.
—————————————————
DETAILS:
The KF394 uses dual tens for the low mids, in a vented enclosure.
Unprocessed, it’s about -15 dB at 64 Hz, and it starts dropping pretty hard below about 105 Hz.
Processed, using an EAW box and their presets, the KF394 is -10 dB at 64 Hz, and doesn’t start falling off on the bottom end until about 80 Hz.
FWIW, the full specs are here:
https://eaw.com/docs/1_Current_Products/KF/Spec_Sheets/KF394_SPECS_revA.pdf
An EAW tech recommend that I cross at 80 Hz, with a 24 dB LR slope. I find this to be extremely surprising, and actually I question it rather strongly, for several reasons.
What do you guys think?
I’ll of course be using processing so they WILL be flat down to 80 Hz, but I’m concerned that this is “eq’d” response, and thus more of a strain on the drivers. Perhaps even not sonically ideal due to EQ phase issues or something.
- Plus, doesn’t such a steep slope cause phase issues all by itself?
At the same time, it has always been my experience that ANY live subwoofer sounds best when its LPF is at 100 Hz or less. Preferably 80 Hz. - But my experience is quite limited.
And yet, finally, I know that many guys run their subs much higher, even for live music. (Hence some discussions in that PPSL thread that surprised me.)
——————————————
Well, I’d be foolish not to be open to new ideas and concepts, so if this wasn’t too many words already, I sure would welcome some opinions & advice.
Thx.
This is for a live band (smallish events) and DJ work. No HT, no EDM, no raga parties. (g)
I have complete digital control over eq and such. (Currently within my digital mixer, but I will be getting the EAW UX3600 soon.)
The subs have yet to be determined, but they will be a dual 12” design, either TH (similar to the Danley TH212) or a vented box @ around 6-7 cubic feet.
(I plan to build both, and see which works best for me.)
I the past, I’d just cross at about 100 hz, and then adjust the eq and amp levels by ear, but that was back when I had less control. Plus, the KF394’s may require a little extra finesse.
—————————————————
DETAILS:
The KF394 uses dual tens for the low mids, in a vented enclosure.
Unprocessed, it’s about -15 dB at 64 Hz, and it starts dropping pretty hard below about 105 Hz.
Processed, using an EAW box and their presets, the KF394 is -10 dB at 64 Hz, and doesn’t start falling off on the bottom end until about 80 Hz.
FWIW, the full specs are here:
https://eaw.com/docs/1_Current_Products/KF/Spec_Sheets/KF394_SPECS_revA.pdf
An EAW tech recommend that I cross at 80 Hz, with a 24 dB LR slope. I find this to be extremely surprising, and actually I question it rather strongly, for several reasons.
What do you guys think?
I’ll of course be using processing so they WILL be flat down to 80 Hz, but I’m concerned that this is “eq’d” response, and thus more of a strain on the drivers. Perhaps even not sonically ideal due to EQ phase issues or something.
- Plus, doesn’t such a steep slope cause phase issues all by itself?
At the same time, it has always been my experience that ANY live subwoofer sounds best when its LPF is at 100 Hz or less. Preferably 80 Hz. - But my experience is quite limited.
And yet, finally, I know that many guys run their subs much higher, even for live music. (Hence some discussions in that PPSL thread that surprised me.)
——————————————
Well, I’d be foolish not to be open to new ideas and concepts, so if this wasn’t too many words already, I sure would welcome some opinions & advice.
Thx.
Last edited:
With live sound, driver protection and control trumps other factors so 4th order crossovers are common and are the minimum I will use, 4th order LR crossovers are well behaved in the phase domain too. But I sorta agree with you on the "forced" 80hz crossover for the KF394. Looking at the response plots alone I'd be inclined to use a higher crossover but who knows maybe it doesn't sound bad at 80hz and the little extra driver excursion this creates helps cool the drivers better.
^ Thanks, that helps a lot.
In the old days, and to the best of my limited knowledge, 4th order x-overs were considered problematic. Maybe modern DSP has made the difference?
Unless others feel differently, I'l stick with 4th order L-R for now, and try between 80 and 100 Hz.
-----------------
- But I'd still love to hear other opinions on this important topic.
In the old days, and to the best of my limited knowledge, 4th order x-overs were considered problematic. Maybe modern DSP has made the difference?
Unless others feel differently, I'l stick with 4th order L-R for now, and try between 80 and 100 Hz.
-----------------
- But I'd still love to hear other opinions on this important topic.
Can't say I have ever heard anyone say that, consider that most pro sound analog active crossover used 4th order filters.
They have managed to come up with phase neutral processing in the latest and greatest DSP processors but even with a lesser unit the average end user can get very good results.
Good to know, but within my very limited knowledge, using high-end gear my whole life (Tannoy, Turbosound, Bag End, etc) every tech or manual always recommended 12 db per octave, and with the x-over point best set at both speakers' -e dB point, if possible. Obviously, that's a very small sample of opinions, but still pretty valid.)
Well, I like the idea of 4th order, anyway, as long as there are no serious phase issues, so I'm good! Again, thanks for this.
Whether to use even or odd order crossovers in a PA application has nothing to do with the phase response of the crossover per se. It has to do with the total acoustic power response in the crossover region. That is a function of the both the crossover order and the ratio of distance between drivers to crossover wavelength. This is not simple.
Odd order crossovers deliver a constant amount of electrical power to be delivered to the drivers at all frequencies. When the drivers connected to the HP and LP filter are separated by more than half a wavelength at the crossover frequency, using an odd order crossover makes sense, because the spatial phase relationship is more or less random. On average for all listeners, the drivers are never in phase acoustically, because the path length differential varies too much compared to the wavelength at the crossover frequency. To get an average flat frequency response over the widest listening area an odd order crossover does this.
Even order crossovers deliver flat frequency response, even though the amount of power they deliver to the drivers varies throughout the crossover range. They are best used when the distance between the drivers is less than half the wavelength at the crossover frequency. Despite the fact that they don't deliver a constant amount of electrical power to the pair of drivers through the crossover range, they do give flat frequency response, because when the drivers are close to each other acoustically, the efficiency of the driver pair increases 3dB in the crossover range.
So if you crossover at 80Hz (14' wavelength), between cabinets which are 3' apart, you are much better using an even order crossover. The average frequency will be flatter. If you used an odd order crossover, you would have a 3dB bump on axis and if you average all listening positions, there would still be a hump.
Where to put the crossover frequency is a simple calculation, assuming you have accurate data. In PA use, the most important issue is usually about maximum total output. In your case, you need a graph of the subwoofer plus subwoofer amplifiers maximum SPL versus frequency. Then you need the same for the EAW cabinet and amplifier. The optimum crossover frequency is then where these two curves cross each other.
Odd order crossovers deliver a constant amount of electrical power to be delivered to the drivers at all frequencies. When the drivers connected to the HP and LP filter are separated by more than half a wavelength at the crossover frequency, using an odd order crossover makes sense, because the spatial phase relationship is more or less random. On average for all listeners, the drivers are never in phase acoustically, because the path length differential varies too much compared to the wavelength at the crossover frequency. To get an average flat frequency response over the widest listening area an odd order crossover does this.
Even order crossovers deliver flat frequency response, even though the amount of power they deliver to the drivers varies throughout the crossover range. They are best used when the distance between the drivers is less than half the wavelength at the crossover frequency. Despite the fact that they don't deliver a constant amount of electrical power to the pair of drivers through the crossover range, they do give flat frequency response, because when the drivers are close to each other acoustically, the efficiency of the driver pair increases 3dB in the crossover range.
So if you crossover at 80Hz (14' wavelength), between cabinets which are 3' apart, you are much better using an even order crossover. The average frequency will be flatter. If you used an odd order crossover, you would have a 3dB bump on axis and if you average all listening positions, there would still be a hump.
Where to put the crossover frequency is a simple calculation, assuming you have accurate data. In PA use, the most important issue is usually about maximum total output. In your case, you need a graph of the subwoofer plus subwoofer amplifiers maximum SPL versus frequency. Then you need the same for the EAW cabinet and amplifier. The optimum crossover frequency is then where these two curves cross each other.
Wow, Jack, that's some serious info. THANKS.
I'm now doing research on related topics, like x-over & spacial design in an MTM cabinet, (Which is what the KF394 is) but some of the math is waaaay beyond me!
Still, you've opened up a huge new can of worms, such as " what's the perfect tops-speaker height, for a given subwoofer crossover and a given tops-speaker spacing.
Not to mention, "how much toe-in, and how does that affect the optimum height, in terms of even response."
Egads.
i'll be posting new threads on these, as separate topics, soon. I'm crossing my fingers that we can get some serious discussion / opinions.
======================================
BUT FOR NOW, I need clarification. You wrote:
"So if you crossover at 80Hz (14' wavelength), between cabinets which are 3' apart, you are much better using an even order crossover. The average frequency will be flatter. If you used an odd order crossover, you would have a 3dB bump on axis and if you average all listening positions, there would still be a hump.."
What cabinets are you referring to, @ 3' apart? the subs (center to center) the tops, or the subs center to the tops center? (height) ?
I need to know this before any followup questions.
- And thanks again, this seems to be a very important topic to go further with.
I'm now doing research on related topics, like x-over & spacial design in an MTM cabinet, (Which is what the KF394 is) but some of the math is waaaay beyond me!
Still, you've opened up a huge new can of worms, such as " what's the perfect tops-speaker height, for a given subwoofer crossover and a given tops-speaker spacing.
Not to mention, "how much toe-in, and how does that affect the optimum height, in terms of even response."
Egads.
i'll be posting new threads on these, as separate topics, soon. I'm crossing my fingers that we can get some serious discussion / opinions.
======================================
BUT FOR NOW, I need clarification. You wrote:
"So if you crossover at 80Hz (14' wavelength), between cabinets which are 3' apart, you are much better using an even order crossover. The average frequency will be flatter. If you used an odd order crossover, you would have a 3dB bump on axis and if you average all listening positions, there would still be a hump.."
What cabinets are you referring to, @ 3' apart? the subs (center to center) the tops, or the subs center to the tops center? (height) ?
I need to know this before any followup questions.
- And thanks again, this seems to be a very important topic to go further with.
In my statement that you quoted, the 3' refers to the distance between the EAW and the sub cabinet (center to center). I'll explain why later.
If you are putting cabinets on the stage high up, positioning them is simple. You always want them virtually as high in the air as you can get them. Once you get them up there, the main axis of them needs to be pointed at ear height of the farthest listener from the stage. This will result in the most constant SPL for everyone in the audience. It also minimizes sending acoustic output towards surfaces where you don't want any, such as the back wall. See John Eargles book for more information.
In a PA application, the amount of toe in that can be used is always compromised by minimizing the chances of feedback. The more toe in the mains have, the more sound gets to mics at the front of the stage.
In a residential environment, using toe in, helps stabilize the image over a wider area through time-intensity trading. This does not work in PA applications for two reasons.
1) PA cabinets tend to have flatish response over a certain lateral angle, then fall off a rock once you go past that area.
2) The Haas affect which enables time-intensity trading to work, does not scale with space. Once you are 20' off of the central axis in a 500 seat venue, the sound from the nearer speaker is getting to you 15msec earlier. No amount of increased intensity from the further away speaker could ever make the image pull back to the center.
If you are putting cabinets on the stage high up, positioning them is simple. You always want them virtually as high in the air as you can get them. Once you get them up there, the main axis of them needs to be pointed at ear height of the farthest listener from the stage. This will result in the most constant SPL for everyone in the audience. It also minimizes sending acoustic output towards surfaces where you don't want any, such as the back wall. See John Eargles book for more information.
In a PA application, the amount of toe in that can be used is always compromised by minimizing the chances of feedback. The more toe in the mains have, the more sound gets to mics at the front of the stage.
In a residential environment, using toe in, helps stabilize the image over a wider area through time-intensity trading. This does not work in PA applications for two reasons.
1) PA cabinets tend to have flatish response over a certain lateral angle, then fall off a rock once you go past that area.
2) The Haas affect which enables time-intensity trading to work, does not scale with space. Once you are 20' off of the central axis in a 500 seat venue, the sound from the nearer speaker is getting to you 15msec earlier. No amount of increased intensity from the further away speaker could ever make the image pull back to the center.
Thanks again, Jack.
I have two immediate followup questions:
NOTE: THESE QUESTIONS ARE FOR EVERYONE. I don't expect poor Jack to carry all the weight here! (Although I sure do appreciate it.)
1: (more of a confirmation) You wrote "If you are putting cabinets on the stage high up, positioning them is simple. You always want them virtually as high in the air as you can get them. "
I love simple concepts! - But I had inferred from your first post that, when using a 3rd order x-over between subs & tops, and with them MORE than 1/4 wavelength apart, that the exact distance would still matter, in terms of even response through the x-over region.
- Actually, maybe I inferred that from my other readings about MTM design.
It probably IS true to some extent, but basically you're saying that the advantages of height (more even distance between close & far audience members) outweighs everything else, correct?
-----------------
2: About tops spacing and toe-in: I never thought about the Haas effect before, in terms of two different direct paths. (Only in terms of studio control room design, where you need to create a reflection-free zone, and diffuse anything within the Haas window. If memory serves, that's actually around 11.2 feet, but it's been a long time...)
So this is interesting. Would it perhaps then be best, with only a pair of point-source speakers, to space them CLOSER than 11 feet? (esp if tryng for some semblence of stera perception.) I usually do, anyway, but it's a fun question.
The toe-in issue still alludes me. - But in a small setup, you are often inside the Haas window, so your reason #1 for it being problematic might not apply.
----------------
And what about LCR arrangements, with all three speakers pointing straight out? Would this give better HF coverage, or introduce more phase smear than it's worth?
I have two immediate followup questions:
NOTE: THESE QUESTIONS ARE FOR EVERYONE. I don't expect poor Jack to carry all the weight here! (Although I sure do appreciate it.)
1: (more of a confirmation) You wrote "If you are putting cabinets on the stage high up, positioning them is simple. You always want them virtually as high in the air as you can get them. "
I love simple concepts! - But I had inferred from your first post that, when using a 3rd order x-over between subs & tops, and with them MORE than 1/4 wavelength apart, that the exact distance would still matter, in terms of even response through the x-over region.
- Actually, maybe I inferred that from my other readings about MTM design.
It probably IS true to some extent, but basically you're saying that the advantages of height (more even distance between close & far audience members) outweighs everything else, correct?
-----------------
2: About tops spacing and toe-in: I never thought about the Haas effect before, in terms of two different direct paths. (Only in terms of studio control room design, where you need to create a reflection-free zone, and diffuse anything within the Haas window. If memory serves, that's actually around 11.2 feet, but it's been a long time...)
So this is interesting. Would it perhaps then be best, with only a pair of point-source speakers, to space them CLOSER than 11 feet? (esp if tryng for some semblence of stera perception.) I usually do, anyway, but it's a fun question.
The toe-in issue still alludes me. - But in a small setup, you are often inside the Haas window, so your reason #1 for it being problematic might not apply.
----------------
And what about LCR arrangements, with all three speakers pointing straight out? Would this give better HF coverage, or introduce more phase smear than it's worth?
1) As with everything in acoustics, there are tradeoffs.
To get the best overall SPL coverage for the entire audience you want to fly the satellite speakers very high and angle them down towards the back. This only works because of the R^2 law and because the speaker directivity is high at mid and upper frequencies.
To minimize comb filtering between the satellites and subs, you want them close together. However since the subs aren't flown and because you want them clustered in the middle of the stage for most even bass response, these two things are incompatible to some degree.
You have to decide which is more important. No matter what I would fly the satellite speakers up high as long as they are angled down the correct amount. The bass integration is of secondary importance.
2) I think this depends a lot on the size of the venue. In a smaller venue and especially a narrower one, you can get away with positioning the main speaker closer together. In a larger or wider venue, you can't do this because the SPL will be too low for listeners on the sides. In most indoor venues, you will get the most even SPL coverage by toeing the speakers out, not in.
To get the best overall SPL coverage for the entire audience you want to fly the satellite speakers very high and angle them down towards the back. This only works because of the R^2 law and because the speaker directivity is high at mid and upper frequencies.
To minimize comb filtering between the satellites and subs, you want them close together. However since the subs aren't flown and because you want them clustered in the middle of the stage for most even bass response, these two things are incompatible to some degree.
You have to decide which is more important. No matter what I would fly the satellite speakers up high as long as they are angled down the correct amount. The bass integration is of secondary importance.
2) I think this depends a lot on the size of the venue. In a smaller venue and especially a narrower one, you can get away with positioning the main speaker closer together. In a larger or wider venue, you can't do this because the SPL will be too low for listeners on the sides. In most indoor venues, you will get the most even SPL coverage by toeing the speakers out, not in.
Jack, I'm still trying to figure out how to apply your information.
I don't yet have the EAW processor, so I'm a bit limited in my computer's DSP capabilities (I can only do third order right now) I'd like to better understand what you wrote. so:
quote: "Even order crossovers deliver flat frequency response...... They are best used when the distance between the drivers is less than half the wavelength at the crossover frequency..... because when the drivers are close to each other acoustically, the efficiency of the driver pair increases 3dB in the crossover range."
Forgetting specific distance for a moment, why would you want that 3 db bump? Aren't we supposed to find the x-over point where both speakers (ideally) are 3 db down, in free-space?
- And isn't that easily corrected with DSP, hence being perhaps not the most important consideration? (I'm just asking)
-----------
quote: "So if you crossover at 80Hz (14' wavelength), between cabinets which are 3' apart, you are much better using an even order crossover. The average frequency will be flatter. If you used an odd order crossover, you would have a 3dB bump on axis and if you average all listening positions, there would still be a hump."
Are you talking about two speakers (sub & top) that have enough frequency range that they overlap, and there IS no "common -3db" sweetspot?
(Also, could that 3 dB hump not actually be a dip?)
------------
And finally, in a typical indoor gig, the tops will be near the ceiling, which is usually 8 - 10', so the top driver is at 7 - 9' max. The sub center will be about 1' or less. The tops will be out towards the sides t some extent, and in my case the subs clustered in center.
That make the typic subs-to-tops distance maybe 8' minimum to 11 - 12" maximum. I assume that if I cross at 80 - 100 hz, meaning always outside of the 1/2 wavelength mark, you are suggesting a third-order crossover.
What puzzles me is why? Is there a sonic advantage to the 18 dB slope? "All other things being equal" (which they are not) it seems to e that the best slope is the steepest slope you can get away with, specifically to minimize comb filtering. So, why specifically would you choose third order when the spacing allows for either choice?
I don't yet have the EAW processor, so I'm a bit limited in my computer's DSP capabilities (I can only do third order right now) I'd like to better understand what you wrote. so:
quote: "Even order crossovers deliver flat frequency response...... They are best used when the distance between the drivers is less than half the wavelength at the crossover frequency..... because when the drivers are close to each other acoustically, the efficiency of the driver pair increases 3dB in the crossover range."
Forgetting specific distance for a moment, why would you want that 3 db bump? Aren't we supposed to find the x-over point where both speakers (ideally) are 3 db down, in free-space?
- And isn't that easily corrected with DSP, hence being perhaps not the most important consideration? (I'm just asking)
-----------
quote: "So if you crossover at 80Hz (14' wavelength), between cabinets which are 3' apart, you are much better using an even order crossover. The average frequency will be flatter. If you used an odd order crossover, you would have a 3dB bump on axis and if you average all listening positions, there would still be a hump."
Are you talking about two speakers (sub & top) that have enough frequency range that they overlap, and there IS no "common -3db" sweetspot?
(Also, could that 3 dB hump not actually be a dip?)
------------
And finally, in a typical indoor gig, the tops will be near the ceiling, which is usually 8 - 10', so the top driver is at 7 - 9' max. The sub center will be about 1' or less. The tops will be out towards the sides t some extent, and in my case the subs clustered in center.
That make the typic subs-to-tops distance maybe 8' minimum to 11 - 12" maximum. I assume that if I cross at 80 - 100 hz, meaning always outside of the 1/2 wavelength mark, you are suggesting a third-order crossover.
What puzzles me is why? Is there a sonic advantage to the 18 dB slope? "All other things being equal" (which they are not) it seems to e that the best slope is the steepest slope you can get away with, specifically to minimize comb filtering. So, why specifically would you choose third order when the spacing allows for either choice?
In all of my comments above, where ever I said even order crossovers, I should have said LR (Linkwitz-Riley) even order.
Section 1:
If you have a HP and LP filter which are in phase at the crossover frequency and each is -6dB down, they will sum to a flat frequency response. This is the case with an LR filter. The Q of each filter is 0.5.
If you have a HP and LP filter which each have a Q of 0.707 (Butterworth), they will each be -3dB at the crossover frequency and they will sum to +3dB.
See the graph at the link below for a comparison of them.
Linkwitz–Riley filter - Wikipedia
When you select an even order filter in most DSP boxes, it is going to give you an LR filter. If you select and odd order filter, it is going to give you a Butterworth.
In the situation above, when there is significant comb filtering between the sub and satellite due to how far apart they are placed, compared to the wavelength at the crossover frequency, no amount of eq placed before the filters will make the average frequency response flat if you are using LR filters. By average frequency response flat, I mean the frequency response measured over a large listening window (physical area). This is the most important criteria in sound reinforcement as the field is mostly reflective at these frequencies.
By having a 3dB hump from an odd order crossover, the system will have flatter averaged response over a wider listening area. Why this is true, requires a lot of math. Because the environment is so reflective at these frequencies, you are not going to hear the 3dB hump much at all, even if sitting directly on axis.
Section 2:
I'm not talking about subs and sats that have any overlap.
Section 3:
If your DSP box has an option for 48dB/oct at 80Hz, I would use this. As you say it will reduce comb filtering in the crossover region. The problem is that with such high slopes at a low frequency, there may be too much latency through the LP filter. This creates problems in live sound applications. Because of the potential for latency problems, at low frequencies they generally don't use very high slope filters. Given the choice between 18dB/Oct and a 24dB/Oct filters at 80Hz in this case, the odd order choice will have flatter overall response over the listening window.
Section 1:
If you have a HP and LP filter which are in phase at the crossover frequency and each is -6dB down, they will sum to a flat frequency response. This is the case with an LR filter. The Q of each filter is 0.5.
If you have a HP and LP filter which each have a Q of 0.707 (Butterworth), they will each be -3dB at the crossover frequency and they will sum to +3dB.
See the graph at the link below for a comparison of them.
Linkwitz–Riley filter - Wikipedia
When you select an even order filter in most DSP boxes, it is going to give you an LR filter. If you select and odd order filter, it is going to give you a Butterworth.
In the situation above, when there is significant comb filtering between the sub and satellite due to how far apart they are placed, compared to the wavelength at the crossover frequency, no amount of eq placed before the filters will make the average frequency response flat if you are using LR filters. By average frequency response flat, I mean the frequency response measured over a large listening window (physical area). This is the most important criteria in sound reinforcement as the field is mostly reflective at these frequencies.
By having a 3dB hump from an odd order crossover, the system will have flatter averaged response over a wider listening area. Why this is true, requires a lot of math. Because the environment is so reflective at these frequencies, you are not going to hear the 3dB hump much at all, even if sitting directly on axis.
Section 2:
I'm not talking about subs and sats that have any overlap.
Section 3:
If your DSP box has an option for 48dB/oct at 80Hz, I would use this. As you say it will reduce comb filtering in the crossover region. The problem is that with such high slopes at a low frequency, there may be too much latency through the LP filter. This creates problems in live sound applications. Because of the potential for latency problems, at low frequencies they generally don't use very high slope filters. Given the choice between 18dB/Oct and a 24dB/Oct filters at 80Hz in this case, the odd order choice will have flatter overall response over the listening window.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.