Skip to main content

First, start with the Wikipedia article on L-R and focus on the 2nd order L-R crossover response plot.

"Second order Linkwitz–Riley crossover (LR2, LR-2)
Second-order Linkwitz–Riley crossovers (LR2) have a 12 dB/octave (40 dB/decade) slope. They can be realized by cascading two one-pole filters, or using a Sallen Key filter topology with a Q0 value of 0.5. There is a 180° phase difference between the lowpass and highpass output of the filter, which can be corrected by inverting one signal. In loudspeakers this is usually done by reversing the polarity of one driver if the crossover is passive. For active crossovers inversion is usually done using a unity gain inverting op-amp."
https://en.wikipedia.org/wiki/Linkwitz%E2%80%93Riley_filter#Second_order_Linkwitz%E2%80%93Riley_crossover_(LR2,_LR-r2)

Note that the Butterworth filter is what is implemented in Master Fader high pass and low pass parametric functions and there is no option to cascade two of them into a Linkwitz-Riley. Attempting to use the high pass or low pass function to implement one half of a crossover results in that erroneous 3dB bump at crossover when the two halves of the signal are combined acoustically. If you have ever tried to cross your subwoofers in Master Fader without analyzing the result, you were probably disappointed in the boomy bass.

You can't get there from here. Or can you? Let's try.


response plot attribution: https://commons.wikimedia.org/wiki/User:Krishnavedala

In Master Fader, we can set a 1000Hz crossover in the user interface and ignore the trailing three zeroes to obtain a normalized frequency axis for the purposes of comparison. Then we can see what is going on with our own eyes and not need any math to figure this out.

Here is the Butterworth lowpass filter response overlaid with a Bell filter at 1KHz and the Q tuned to 0.5 for the purposes of comparison to the erroneous 2nd-order Butterworth crossover plot on Wikipedia:

The Bell filter with Q of 0.5 looks like it overlays that 3dB bump in the Wikipedia plot fairly well. All we have to do now is subtract that error from the combined response by changing the gain of the Bell filter from +3dB to -3dB.

Now we're talking. We still have issues though. Here's the combined response of these two filters with an aux fed subwoofer crossed at 75Hz to eliminate breath pops and footfalls without losing the low E on a guitar:

The 2nd order crossover is too gradual for a subwoofer that has approximately one octave of bandwidth. By the time we finish adding a 27Hz high pass to preserve the low B on a 5-string bass while sparing subwoofer power from any residual inaudible sub-bass mud that might be contaminating the mix, the subwoofer gain is down by at least 3dB across its entire bandwidth.

The bass-limited SA1530z 3-way Mackie main that the KW181 is paired with in this system still has to do more heavy lifting than it can credibly handle. Plus there is the necessity of a signal inversion at crossover to match the phase between the bass drivers through the crossover region, so the bass-limited Mackie mains are driving most of the bass in the performance from their 15" woofers out of phase with the stage. That's not good for a live performance but most especially problematic in the small venues that bar bands play where the stage volume is a substantial portion of the total mix.

Comments

CherylJosie Fri, 03/23/2018 - 15:21

Let's revert to Wikipedia and look at a 4th-order L-R crossover instead:

"Fourth order Linkwitz–Riley crossover (LR4, LR-4)Fourth-order Linkwitz–Riley crossovers (LR4) are probably today's most commonly used type of audio crossover. They are constructed by cascading two 2nd-order Butterworth filters. Their slope is 24 dB/octave (80 dB/decade). The phase difference amounts to 360°, i.e. the two drives appear in phase, albeit with a full period time delay for the low-pass section."

So can we concentrate more sub bass in the subwoofer and eliminate that phase inversion in the bass-limited mains? Here's the plots.

Whoops! We have to increase the Q to 1 because the slope of the 4th-order filter is doubled. Let's sanity check the combined response:

That looks pretty good. So what is it like as a subwoofer crossover?

That looks workable.

Here's the bass-limited crossover, for completeness, with a 17KHz lowpass to protect the tweeters and spare the audience from inaudible sizzle or feedback that might cause unwitting hearing damage:

I have not tried this method out yet. I have been using the graphic as a notch that subtracts out the 3dB error.

The advantage of using the parametric is that it has better tunability and avoids confusing the crossover with the room curve.

I hope this information comes in useful. Comments and suggestions are appreciated. Thanks everyone.

bouldersound Fri, 03/23/2018 - 15:33

When you change the gain of one band you change the crossover point and, by extension, the phase relationship. I'm not convinced that those plots account for that. Even if they do, they may not account for the phase and frequency response of the cabinets. Measurement is likely needed to integrate electronic and acoustic contributions to the end result.

CherylJosie Fri, 03/23/2018 - 20:24

bouldersound, post: 456290, member: 38959 wrote: When you change the gain of one band you change the crossover point and, by extension, the phase relationship. I'm not convinced that those plots account for that. Even if they do, they may not account for the phase and frequency response of the cabinets. Measurement is likely needed to integrate electronic and acoustic contributions to the end result.

I get skewered with that imminently valid point every time I post another crossover thread with no measurements.:oops:

I own a UMIK-1 for calibrating my home theater. A laptop and stereo wireless would help me align the PA with Room EQ Wizard or similar. That investment is not happening at this time so I have been setting up by ear and just trying to minimize the know error sources with this DIY approach and minimal gear.

Maybe I could generate a sine wave at crossover to level match the mains and subs with an app or two on my phone. If the combined response doesn't gain 6dB I know I have a problem with the phase that maybe a delay tweak can fix.

Then I could generate two additional tones to check the balance through crossover while adjusting the Q of that Bell filter. It's not the same as a sweep but it's better than nothing.

I have never calibrated a system in a very large room. I have read that commercial theaters use white noise for calibration instead of sweeps but I don't know why or how. So far the only 'large' venue I have mixed in is outdoors and the audience was close enough that treble attenuation through air was not noticeable.

The phase change from the Bell filter applies to both cabinets equally. I don't know if it introduces the same phase shift either side of the tuning frequency.:confused:

Charlie is now stacking his subwoofers on one side with a bass-limited main on top, and putting the other main on the floor or a table. This helps keep a 'power alley' from forming in the sub bass and it also keeps the height of the stack under control on one side of the stage so the band is more visible, but it could also make the modal response less smooth in the small spaces we play. I would have to investigate the modal response in advance with some small equipment and I'm not ready for that level of involvement nor am I convinced it matters for these gigs. Asymmetry also complicates the relative phase shift between subs and mains. Tweaking by ear is probably necessary even with measurements.:confused:

Further refinements will have to wait until I have the budget and the time to practice with new toys. I am basically done with what I can do on a theoretical level.

I don't know much about mixing in general so maybe my time would be better spent learning the subtleties of effects and EQ, how to mike instruments properly, how to configure a snake or wireless to the 8 channel recorder, or how to efficiently set up and tear down without making mistakes in the sound check. I have been meaning to write a Master Fader setup procedure for myself so I can focus better, plus maybe preparing a show in the mixer in advance of every gig instead of letting that task revert to Charlie.

I added a Zoom Q2N to my collection of gear recently. It works well in low light and has excellent depth of field, so I am grateful to the guy at Guitar Center who recommended it. Even on its narrowest field of view at approximately 80 degrees I can only place it about 15' from the stage with a 20' spread between performers. That puts it smack in the middle of the audience/dance floor, and that narrow setting also discards half of the resolution because it's a digital zoom function.

The field of view is 160 degrees on its wide setting. That's huge but I found a way to make use of it last weekend and got great video of the Shamrock Mansion birthday party.

Here's a collaborative effort from the most recent gig I mixed. Another band member merged my side-shooting Q2N with audio from his hand-held XY audio recorder (you can see his Zoom taped overhead on the right side of the video directly in front of the stage).

The Q2N certainly has its advantages for casual video recording compared to more $ophisticated equipment. I just wish it didn't break the recording up into 3.7G files. That's too small for video at a live gig where I want to keep the thing running continuously throughout.

The mix at the party was configured hurriedly by Charlie and me, with a minimalist show prepared in advance and without any of my special templates. There was minimal proximity cut on the microphones because they were not in vocal/instrumental subgroups where I usually fine-tune the proximity cut EQ. There was no house curve in the graphic either so the FOH was somewhat thin sounding on the instruments but a little heavy in the vocals with their minimal proximity cut. The crossover had that Butterworth 3dB bump because the FOH graphic was flat and I had not implemented the Bell filter in the parametric yet. I didn't really get the effects tuned well.

I resorted to tweaking everything per channel with the input EQ during the sound check and first couple of songs. I've been getting over the flu and barely made it to the gig at all. They were lucky they had anyone mixing because the backup option I encouraged Charlie to investigate didn't pan out and I was 'it'.

The mix still seems to have come out OK even without all my value-added research and development. The audio captured with that Zoom sounds fine for a house party and the audience complemented us for good sound without the ear-splitting dBs typical of amateur rock bands in private residences.

Sometimes it feels that I am wasting my technical effort on this amateur stuff. If it sounds substantially the same with or without a disciplined approach what's the point? Either I am doing it all wrong or the amateur bar/house party environment without full PA has enough acoustic warts that a few electronic glitches don't matter anyway.

I'm still doing it. It's fun and I am learning. That's all I need. With my trashed hearing and my arthritic degeneration I am not going to get hired to do sound professionally. This is the apex of my 'career' as a live sound and recording engineer in my 'retirement'. I'm making the most of it because it's all downhill from here.

bouldersound Fri, 03/23/2018 - 21:52

I don't mean to knock what you're doing, just to keep things connected to real life situations.

Think of this: The phase of the two filters shifts in opposite directions. Just above or below the crossover frequency they will be sufficiently out of phase to alter the response. Far enough from the crossover frequency there's sufficient level difference that one band will dominate, but there's an area either side of it where the levels of both bands are close enough that the phase differences cause response variations. There you get a rippled response curve. I don't see that represented in the graphs so I assume they are showing gain but not accounting for phase.

None of which really matters much once you're inside a box and the sound is taking multiple paths to the listener, as you know since you mention modal response. The room can scramble your response much more than proper setup can fix it, though it's still better to have it set properly than not if you've got the time and energy. I rarely had the time to do it "right" so I learned to use my ears and wing it. I'd walk the room and if there was something noticeably wrong I'd make some tweak during the show. My sub cables (passive) were terminated in dual banana connectors, so I could hot swap the polarity. I would actually use power alley to keep the sub energy off the walls since reflections just made for all kinds of variations in response near walls anyway.

My point is be flexible and don't get bogged down in theory. Give yourself a solid starting point with theory and planning but be prepared to let it go should it not work out. Have fun: the mix will benefit.

CherylJosie Fri, 03/23/2018 - 22:17

bouldersound, post: 456293, member: 38959 wrote: I don't mean to knock what you're doing, just to keep things connected to real life situations.

Think of this: The phase of the two filters shifts in opposite directions. Just above or below the crossover frequency they will be sufficiently out of phase to alter the response. Far enough from the crossover frequency there's sufficient level difference that one band will dominate, but there's an area either side of it where the levels of both bands are close enough that the phase differences cause response variations. There you get a rippled response curve. I don't see that represented in the graphs so I assume they are showing gain but not accounting for phase.

None of which really matters much once you're inside a box and the sound is taking multiple paths to the listener, as you know since you mention modal response. The room can scramble your response much more than proper setup can fix it, though it's still better to have it set properly than not if you've got the time and energy. I rarely had the time to do it "right" so I learned to use my ears and wing it. I'd walk the room and if there was something noticeably wrong I'd make some tweak during the show. My sub cables (passive) were terminated in dual banana connectors, so I could hot swap the polarity. I would actually use power alley to keep the sub energy off the walls since reflections just made for all kinds of variations in response near walls anyway.

My point is be flexible and don't get bogged down in theory. Give yourself a solid starting point with theory and planning but be prepared to let it go should it not work out. Have fun: the mix will benefit.

CherylJosie Sat, 03/24/2018 - 12:47

CherylJosie, post: 456294, member: 50446 wrote:

There is no phase plot in Master Fader and I suspect you are correct about the plot on Wikipedia not accounting for nonlinear phase either.

I don't own the mixer so I don't have access to technical support. I registered for the Mackie user's forum but my account has been pending approval for a year. I don't know why they have not approved it. I don't think I need proof of purchase to join.

Without a model of the Bell filter and high/low pass/shelf I don't know what the parametric equalizer's phase does. With a Q of 1 on the Bell filter I am not sure there would be a whole lot of abrupt phase change anyway. I found some info on Wikipedia about linear phase FIR. I suspect the Mackie does it all in DSP because that's cheaper ultimately. Hopefully it is FIR linear phase but it could be analog for all I know.

I noticed a strong bass null directly front of the stage from the reflection off the front wall behind the band apparently since that null was consistent across the front. Where I was mixing by the windows opposite the subwoofers, the bass was heavy from boundary gain off the windows but not muddy, maybe because the windows are not a solid boundary. The bass was a little thin but clear where the audio recorder picked it up.

The bass was best out in the hallway left of the stage. Bass always seems to sound better outside of a small room than inside, even if it has to pass through a doorway on its way out to get to an even smaller room.

I get the part about just winging it and having fun. I can't help analyzing it. I am always trying to optimize.

Well maybe some day I will have the opportunity to measure the system. I would like to align it for an outdoor gig where there is no room boundary confounding the issues. Then at least I have a starting point to work with for future gigs. For now it's wing it and have fun.

bouldersound Sat, 03/24/2018 - 13:19

CherylJosie, post: 456295, member: 50446 wrote: Without a model of the Bell filter and high/low pass/shelf I don't know what the parametric equalizer's phase does. With a Q of 1 on the Bell filter I am not sure there would be a whole lot of abrupt phase change anyway.

Right, but phase interactions also need similar levels. A shallower slope's less drastic phase shift is mirrored by its less drastic gain change. The peaks of the ripple may be lower, but they spread across a greater frequency range. Pick your poison.

CherylJosie, post: 456295, member: 50446 wrote: I found some info on Wikipedia about linear phase FIR. I suspect the Mackie does it all in DSP because that's cheaper ultimately. Hopefully it is FIR linear phase but it could be analog for all I know.

FIR filters trade phase shift for other negative traits. They add more processing latency than IIR filters, and they can "ring" when set aggressively. DSP can do either type, but I don't think FIR can be done with analog filters. Both have their uses.

CherylJosie Sat, 03/24/2018 - 19:56

Here's a video of Red Shift. The Q2N audio is directly behind the stack of subwoofers and a main that was to the left of the audience and by the hallway. The mics are about 12" off the side wall and the recording probably sounds substantially similar to what the lead guitarist heard on stage. I had to lean the tripod forward to see the monitor on the back of the Zoom so I didn't really know what I was recording until I played it back.

CherylJosie Sat, 03/24/2018 - 20:40

I'm still trying to process that comment about using the power alley to keep bass off the walls.

The power alley is frequency-dependent. That necessitates tuning the spacing between the woofers, and the distance from the walls, so that those frequencies cancel between the woofers at the same frequencies where SBIR and modal resonances from the sidewalls are problematic.

I can't wrap my mind around the concept of doing that by ear in real time, especially if the subwoofers are large and heavy as they usually are.

In the case of a small space the subwoofers are often right next to the sidewalls so there's no tunability. The SBIR would be with the front wall behind the band because at those low frequencies the very near sidewalls just add reinforcement at all bass frequencies anyway.

This seems like a trick that works best in a larger venue with some advance planning and maybe some acoustic measurements, but I suppose one could get lucky also. One of our venues works well on the dance floor with the subwoofers placed in a stack. It just seems to fit the acoustics well that way. The subs are basically nearfield and right next to a sidewall because the stage is off center too. They get good reinforcement off that wall and it's far enough from the room penetrations that it is basically in the deepest 'horn' in the room over there.

bouldersound Sat, 03/24/2018 - 21:15

I didn't move the subs around by ear, but I'd ballpark the placement with some quick math. If the middle of the sub band is about 75Hz then the wavelength is about 15 feet. Any multiple of 15' plus half (7.5') will put 75Hz 180° out of phase at a geometric 90° from the audience. Up close it will have little effect because of relative distances from subs to listening position (3:1 rule of thumb), but at a distance it has an effect. Just remember it's shooting backward and upward just as well as forward, so reflections off the ceiling or back wall can mess with things. Of course power alley at one frequency is multi lobed at another and trending omni at yet another.

Have you looked into the concept of cardioid subs? You place them longitudinally a quarter wavelength apart and delay the front sub by that time. The wave from the front sub is projected just as the wave from the back sub passes it. But the front sub's wave reaches the back sub delayed a half wavelength (1/4 wavelength from electronic delay + 1/4 wavelength from acoustic delay), creating a null toward stage.

CherylJosie Sun, 03/25/2018 - 09:22

I have looked into cardioid sub a little after posting on a couple of forums about the various iterations of my DIY crossover in Master Fader.

That is as far I took it. There’s no chance of actually using it for these gigs. With the subs run aux-fed they cross low and their center frequency is closer to 55Hz than 75Hz. They would need to be something like 5’ apart center-to-center and with the size of these subs that would make the entire array close to 8’ deep. Plus most of the bass is in the bass-limited mains anyway so the rear attenuation from doing cardioid would be minimal.

There isn’t enough room to put the subwoofers that far apart front-to-rear in any of our venues. There’s no room on or in front of the stagea to place a cabinet and we would need one in both locations. Outdoors it doesn’t matter because without boundary reinforcement the bass is thin enough with our equipment that the band can use all the bass signal they can get off the backside of the FOH as a monitor.

Eventually I will be setting up a semblance of a studio in my home now that I am moving back in. It’s not a mansion and the quarters will be cramped. I won’t need much PA for it. If I do a house party in the back yard that might benefit from cardioid sub to reduce the noise pollution that the neighbors experience. That would be a longer term investigation. So far my inclination is to put the band against the house anyway so the rear wave would be reflected out toward the neighbors anyway and using cardioid won’t address that at all. We’d have to set the band up in the garden aiming at the house. I don’t think that works with the foliage and planters back there.

At this time I am just editing and posting video. I have a few recordings backlogged from computer failures and winter sickness. One band member wants a DVD so I’m also looking into authoring software. It’s part of the plan but I’m using all free tools at this time while I invest I better recording equipment.

I did some surround recording with multiple microphones but it’s like field recording for movies and nature programs more than music production. I was just experimenting with adjustable condenser polar patterns and mid-side so I can experience what that’s like firsthand.

paulears Wed, 03/28/2018 - 10:36

I'm totally lost. That video sounds fine and much better than the typical blend of live and mixer audio people normally record, so I'm confused by the depth of interest in crossover and eq curves. My frequent work venue has a 12K house system, run usually at nowhere near it's capability, but when I switched to a digital mixer, I soon found that I really could NOT hear any of the little humps and bump tweaks I could see on the displays.

what was wrong with the sound you could get? I'm interested in what you perceived as a problem that is clearly deeply concerning you. I experimented with cardioid PA subs, and decided that it really wasn't worth doing for anything other than control reasons - and often the acts on stage would comment on the lack of bass - when out front there was plenty. Subs usually leak everywhere and on stage you get used to it, and miss it when not there.

Very often if I have to adjust a crossover with old fashioned knobs for freq/boost/cut I find myself doing it totally by ear, and often being quite surprised by the control settings for the best sound - often done with a well known music track rather than visual aids and pink noise. I'm not even sure that flipping a waveform upside down is as obvious as it perhaps should be. The two switch positions sound different, but not better/worse. In the UK, we have quite a well known sound guy who toured at some point with everyone, and he was deaf in one ear, and mixed with his stool looking left. Even with defective hearing, it can still sound good. Judged on the recording, I suspect in the room it was pretty decent.

pcrecord Wed, 03/28/2018 - 12:54

paulears, post: 456323, member: 47782 wrote: often done with a well known music track rather than visual aids and pink noise

That's what the thread inspires me too. Spectrum analysers and other measurement tools gives us a good starting point.
But to me it's followed by ear tweaking all the time.
I'm working to make a pleasing sound for human ears like mine, so there is no reason, my ears shouldn't be in the decision making process.
When I work with a line array, I get the renting team to make the setup and tune it for me. They check the phases and general balance of the kit.
So I don't mess with placement and things like that.
My only goal is to make the kit sing and sounding musical.
The team I work with gets crasy everytime I start the one song that helps me place the vocal frequencies but it's all fun because I know the results will be as expected... ;)

CherylJosie Fri, 03/30/2018 - 00:02

paulears, post: 456323, member: 47782 wrote: I'm totally lost. That video sounds fine and much better than the typical blend of live and mixer audio people normally record, so I'm confused by the depth of interest in crossover and eq curves.

I guess I can take that as a compliment. Thanks. I'll always be reaching for more. It's just who I am.

If the tweaks I put on the input EQ did audibly reduce the errors in the FOH alignment (I am pretty sure they did) they also by definition made audible and probably undesirable changes to the post fader aux channel recording mix (pretty sure that happened too).
I am approaching this from a higher theoretical/technical level than others are but I am still in a steep learning curve and my performance is inconsistent.
The crossover has been a stumbling point because L-R is not actually implemented in the architecture of the mixer. I have to tweak the implemented filters with a somewhat complicated mixture of two different EQ functions in order for the resulting combination to behave somewhat like a L-R crossover and even then it's probably compromised in the phase (definitely if the Bell filter is analog).

what was wrong with the sound you could get? I'm interested in what you perceived as a problem that is clearly deeply concerning you.

That first East Crescent recording with merged audio is better than I have gotten before. It sounded fine to me but I did note some anomalies that I would prefer to have not been there.

In the room, the bass was a little muddy and strong near the windows where the iPad was, but in front it was clear. The stage monitors were a little muddy also from lack of proximity equalization but otherwise fine. The deep null right in front of the stage was disappointing though because it took some of the punch out of the sound up close.

Part of my confusion is that despite not being 'technical' the mix was good anyway. It sounded OK in the room and on the recording. I am just wondering what gains I left on the table by being sick and unprepared as well as behind the development curve on the crossover integration. I'd like to mix consistently well, at least within the limits of my hearing.

Very often if I have to adjust a crossover with old fashioned knobs for freq/boost/cut I find myself doing it totally by ear, and often being quite surprised by the control settings for the best sound - often done with a well known music track rather than visual aids and pink noise. I'm not even sure that flipping a waveform upside down is as obvious as it perhaps should be. The two switch positions sound different, but not better/worse. In the UK, we have quite a well known sound guy who toured at some point with everyone, and he was deaf in one ear, and mixed with his stool looking left. Even with defective hearing, it can still sound good. Judged on the recording, I suspect in the room it was pretty decent.

We all liked the sound. The PA is capable and sized well for the gigs. We are all getting better at this a little at a time.

I don't trust myself to align a crossover by ear, but I am developing a method to do it anyway since I don't want to drag any more equipment with me.

I decided to level match at crossover and then flip the sub phase and tune the delay for a null at crossover. I'll add delay to whichever channel causes the null with the least added delay.

I can do that as a first cut and then tweak it by ear. I would still like measurements though so I will work on setting up something. I suppose I need a wireless now, for talkback and for calibration, maybe for house music also. Wireless and a cellphone should do it. Maybe bluetooth? Or would a real wireless be required? I need line level. I have inline transformers for talkback mike if needed. I know nothing about wireless. I've only ever browsed some on the Internet.

paulears Fri, 03/30/2018 - 01:35

A world class sound designer I work with with top notch equipment in gorgeous spaces spends a lot of his time once the system is in, running around the venue with his pad or Mac, tweaking tiny eq and delay parameters so that virtually every speaker is individually aligned - BUT - his close to perfect is based on the average of the sound at numerous locations. It's clearly impossible to make the sound the same everywhere. I don't quite get your end aim. Monitors being muddy through proximity equalisation? Monitors rarely get eq'd for best sound quality, but most volume vs susceptibility to feedback. The deep null at the centre front is perfectly normal - and today I'm putting in some front fills because it's HF that's lacking and spoiling intelligibility, not the bass. Sure, the duh duh is missing in the middle, but it's fine 6m either side of the centre seat.

I just don't really understand what you quest is? Not being funny - but where do you find the time in a soundcheck to start flipping polarity and tweaking delays? The show I have tomorrow is big and loud, and they have a truck of gear, and we can't fly their PA, so it's going to be ground stacked - so the best that will happen is some curve changes on the EQ. They won't be tweaking delays, and frankly I think it's unproductive and unnecessary. I'll also bet their FOH op will get a great sound, just from controls FOH. Everything will be done by ear. Probably EQ via a walkabout. You say 'despite not being technical', yet you are being very technical. What PA have you got? Digital or analogue? If it's digits then wireless control should be a doddle - you did mention iPad, then you say you are considering wireless - which is it?

The crossover has been a stumbling point because L-R is not actually implemented in the architecture of the mixer. I have to tweak the implemented filters with a somewhat complicated mixture of two different EQ functions in order for the resulting combination to behave somewhat like a L-R crossover and even then it's probably compromised in the phase (definitely if the Bell filter is analog).

I've no idea what on earth you're doing - what crossover needs this kind of thing? If it's house install, the the installers normally put the system in, align it and then lock down the adjustments in the rack processor so people cannot make arbitrary changes. All that's needed is to eq the input to it.

You must learn to use your ears, otherwise small tweaking gets out of control.

You also mention you all liked the sound. In my very humble experience, you need to quantify who's opinion you trust the most. Anyone on stage should be discounted immediately. Like a trombone player, they have no idea whatsoever what it sounds like from out front. I pay attention to the people out front who know how to describe what they hear. Oddly, the lighting people and occasionally educated house staff who can compare and verbalise how this show compares with others. In that video clip, it could be argued that the sound quality is too good, as it exposes the performers less than good areas - tuning, for example. I'd probably spend more time assisting that element than tweaking delays on eq. Sometimes the better the sound is, the worse it can sound. There's also another feature that makes delay tweaking less pertinent - distance. You have a big stage area, so when you tweak delays, what about the direct sound path - either direct too the punter's ears or from the bass cab to the furthest open mic it leaks into. That can wreck the tiny tweaks by a huge uncoordinated delay being introduced into the system.

I admire what you're trying to do, but are you certain you're not making something that should be quite simple, over-complicated? I have far too much of my time taken up stopping feedback, controlling levels, and providing the performers with exactly what they need to perform better. The out of tune vocal BV in the clip, for example - could that be a decent singer who couldn't hear what he needed to hear to pitch properly? This is where I'd spend my time. If you are in charge of sound, and this involves monitor mixes, then you are split - and if you make a choice, then keep the artistes happy first, and then the audience get a good performance, and sound that virtually all of them won't notice. Your efforts fine tuning the PA won't be noticeable by normal people. "Hang on, does it sound better now?" always gets answered with yes - even when you have done nothing at all. Do not trust these people, they are not listening to what you are.

I look at those diagrams you posted and really doubt that in a live environment I could hear these. Grabbing a knob and sticking in a big boost or cut and sweeping it is crude and very effective for spotting what is happening. Nudging m/s on an mini parameter doesn't do the same thing.

I've never seen anyone apply home theatre audiophile techniques to live sound - the two things to me are streets apart. Home theatres are also stereo, live sound is very rarely stereo, just two channels often working virtually mono.

I'm totally unconvinced that live sound can ever be considered in the way home theatre people do, and it's understandable that when they do live sound they want to apply the same systems - but this attempt I believe to be fatally flawed and largely pointless in the quest for better live sound.

CherylJosie Fri, 03/30/2018 - 19:59

paulears, post: 456360, member: 47782 wrote: where do you find the time in a soundcheck to start flipping polarity and tweaking delays?

The delay of the crossover has never been measured or even tweaked by ear for alignment that I know of. The delay adjustment on both frequency bands of the FOH has historically been set to 0 at every gig.

I admire what you're trying to do, but are you certain you're not making something that should be quite simple, over-complicated?

I am certain that I over-complicate things as a rule out of habit. Usually I pare down to essentials once I have conducted my 'design of experiments'.

I see your point though. We usually stack the mains on the subs. The cabs probably add little delay of their own and it probably matches fairly well when they are stacked even though the tech is different generation and from different manufacturers too. I just don't know though without measuring at least once.

What PA have you got?

1x DL1608
Master Fader 3.2.1
my iPad Air 2 and iPhone 6S plus
the drummer's iPad v3? (it's his PA).
1x wireless router
2x Mackie SA1530z
2x QSC KW181
(floor monitors vary between 8" and 15" depending on the gig, up to 3 aux channels/speakers)

The wireless audio I mentioned would be to send a calibration signal to the mixer on stage while I measure at FOH. The DL1608 does not implement wireless audio, only wireless control.

what crossover needs this kind of thing?

We are implementing the sub crossover exclusively in the DL1608 to save money, space, and setup time. There is no automated alignment and there is not even a purpose-built crossover, so I have to create an approximation and align it manually.

My impression is that the lack of phase alignment is making the sound check more difficult by warping the frequency response and possibly causing variability in the bass around the venue due to beaming same as a cardioid sub does, but unintentionally in the vertical dimension rather than deliberately in the front-rear dimension.

I would feel much better about leaving the delays at 0 if we had a system of matched subs and mains that were designed to work well together through their internal crossovers. Then at least I would know that the cabs are not skewing the phase. With a mismatched system all bets are off IMO but honestly I don't know how important it is. You could be right and I am just making this overly complicated. I suppose measuring once or twice would prove it one way or the other.

There's also another feature that makes delay tweaking less pertinent - distance. You have a big stage area, so when you tweak delays, what about the direct sound path - either direct too the punter's ears or from the bass cab to the furthest open mic it leaks into. That can wreck the tiny tweaks by a huge uncoordinated delay being introduced into the system

I never considered the effect of delayed feedback on the phase, or that it might vary by cab when they are physically separated from each other.:confused: I suppose it can be mitigated but not eliminated. I had already considered the asymmetry of having both subs stacked on one side of the stage and just shrugged.

In a small venue like we amateurs play, the early reflections off room boundaries are also an issue for the crossover integration. I've run into that already when tuning my home theater. I suppose that could be another reason why a methodical measurement approach to crossover delay might have uncertain value for live sound. Home theaters tune for a limited seating arrangement and the acoustic environment is more controlled, but even the reflections off my modular sofa I sit in cause huge variations of tens of dB's through crossover. I confirmed this by removing the sofa and re-measuring.

I suppose these error sources could really muck up the measurement and alignment despite my best intentions. That is why I prefer to measure the system outdoors and far from buildings.

The out of tune vocal BV in the clip, for example - could that be a decent singer who couldn't hear what he needed to hear to pitch properly?

Almost certainly. I've had singers out of tune before and it's usually because they couldn't hear themselves with the other issue being respiratory illness. It's more of a problem when the instruments are not in the PA because then the stage volume is way high. The singers start straining their voices and go off tune, almost always tending flat. Hot monitors definitely help in that case but it's tough making sure the feedback is under control.

In that video I didn't set up the monitors. I was sick/slow and the band handled it themselves. My role with the monitors is usually limited to EQing out feedback and giving them tweaks if they ask, but the drummer typically sets up the monitors because he is on stage anyway and can hear it himself. I try to keep off the stage so I'm not in their way. They are much more agile than I am.

keep the artistes happy first, and then the audience get a good performance, and sound that virtually all of them won't notice.

OK, sounds like a plan. Noted.

Thanks for your detailed response. Very kind of you.

paulears Sat, 03/31/2018 - 01:32

Not sure if any of his stuff is easy to find, but Andrew Bishop, who designed many of the speaker systems Carlsbro used for years, eventually left that company when they drifted into cleverness and science and left behind what sound really is. He started up again, making his own loudspeaker drivers, and his quest was for that old, instantly spottable 'British' sound, designed out of modern kit. I've been tempted by some of his current kit. Other firms, Sherman Audio, for example, give quite blunt instruction on how to stack their kit. Two cabinets, side by side works, two cabinets one on top of the other doesn't.

As for measurement - unless you have some very expensive software and the correct hardware (Ease, for example) - how can you measure this, and what worries me is why you'd actually want to with a simple system - I think you mentioned it's tops and subs. If you are trying to calculate and achieve phase coherence in a multi-box line array, then I see the need, but with a smaller system in a smaller room - if you determined that the subs were 20cm too far forward, would you hear it. They PA we tour with is fine for up to maybe 1500 seat venues, and touring it is a real pain. Many bands nowadays leave their FOH system on the vehicle, bring in their FOH and their monitors, because the turns like consistency and then provide a L+R to the house system, and they mix happily with an unknown system - just with a bit of an EQ curve. This is really common now, as venues have systems that suit the space. At the venue I run, we always offer ours. Some decline, and bring in boxes and boxes of posh PA, others look at the long push, play a CD and say - yep, that's fine. Nobody in the past ten years has ever pulled out analysis equipment. Some try to set eq with pink noise and a mic and flatten the room, but oddly, this guarantees nothing in practice.

There is no 'best' sound, just the quest for the most appropriate one for the venue and people on stage. I like in-ears, so does our drummer - the keys and guitar like speakers, loud speakers. The 'best' sound for me rarely happens because those two spoil it, by spilling into everything and in my ears, I lose clarity, and often find the keys going all phasey when it leaks into multiple mics I have up in my IEM mix. Nothing I can do with it. Our sound man complains that he needs a negative fader for the guitar - so often the overall volume is set by the guitar amp direct, and he just brings everyone up in the mix to sit with it. The PA mix has no guitar in it at all - this kind of thing would wreck your approach, and you'd hate it.

Clearly - what works for you fits perfectly, but I'd just encourage you to stop chasing your tail. If all the delays are wrong, and the filter curves have little weirdnesses in them, but it sounds brilliant, why spend any time building the lily, and probably breaking it in the process?

pcrecord Sat, 03/31/2018 - 06:32

CherylJosie, post: 456370, member: 50446 wrote: The delay of the crossover has never been measured or even tweaked by ear for alignment that I know of. The delay adjustment on both frequency bands of the FOH has historically been set to 0 at every gig.

I agree with Paulear, the sound can't be the same everywhere unless the assistance uses headphones... (got a friend who does it for special events.)

About the delay ; the best way, IF you don't want to do the math, is to choose the target best sounding spot, (usually the near FOH mix position) and do a null test.
On the line array and subs, play a frequency that both can handle, then flip the polarity and listen to the results.
At this point you can adjust the delay until you get the most cancellation, save and flip the polarity again.

dvdhawk Sat, 03/31/2018 - 11:36

CherylJosie, post: 456288, member: 50446 wrote: First, start with the Wikipedia article on L-R and focus on the 2nd order L-R crossover response plot.

"Second order Linkwitz–Riley crossover (LR2, LR-2)
Second-order Linkwitz–Riley crossovers (LR2) have a 12 dB/octave (40 dB/decade) slope......

Or, you could refer to the very first response you got to your original post on the subject last March and read the Rane whitepaper on the topic and accept the fact, as Boswell rightly points out, that the right tool for the job is an active crossover, and that an EQ acting as a pseudo crossover will always have inherent issues. If you're getting results that make everyone happy (the band, the audience, the employers) you can move on to other things you can do to finesse the mix. I personally often use the aux-fed sub approach he describes when I'm using a system like you're describing with perfectly good results. It took 5 minutes to configure, and hasn't required any more than a minor adjustment (done to taste by ear) from one job to the next.

And it's worth noting, again, that you don't even need a separate active crossover. The self-powered subs (both the Mackie and QSC) have all the crossover functions you would need built-in. Your configuration (powered sub + powered fullrange) is the exact job they were designed to do.

I definitely admire your perseverance, your methodical, engineer's approach, and seeming fascination with this line of work. We all share that to a large extent, but as far as I can see, this particular horse is dead. Since last year you've upgraded the sub cabinets and found an alternate method of stacking them that works for you in certain venues. That's all good, we live and learn. (Although you would lose style points with me for putting a speaker on a table.) Now I think it's time to borrow a page from the discipline of efficiency engineering and ask, "is there something else I could be focusing on that would allow me to do my job better and yield more significant improvements to the finished product (the show)?"

My 2¢.

Boswell Sun, 04/01/2018 - 11:53

Dave (dvdhawk) gives a good summing up. I would add only that my years of experimenting with live sound crossovers ended up with active 24db/octave Linkwitz-Riley (L-R) crossovers ahead of the power amps and an experimentally-determined offset between the subs and the mids. All the cabinets I used already have the mids and tweeters aligned. I then made sure I set them up the same each time.

The method I devised for the offset was as follows. In an acoustically dry room, I put the sub on a stand lined up 150mm from the edge, and a SDC microphone on the axis of the speaker a few feet from it (that distance doesn't matter unduly). I played a sinewave from a generator set to the crossover frequency (e.g. 120Hz) and recorded a couple of minutes of the sinewave direct from the generator and also from the microphone on separate tracks. I then replaced the sub with the mid/top cabinet with its front in exactly the same plane as the sub was. The microphone was adjusted in height only so that it was on the axis of the mid-range speaker. I played the sinewave recording at the same time as recording the mid-range speaker, so the two live speaker tracks were the same amplitude and nominally in phase. I displayed the difference waveform between them and moved the mid/top cabinet backwards or forwards to minimise the amplitude. In this way, the time alignment was as near exact as I could get it, so I wrote down the distance needed. I repeated it for all the sub-mid combinations I use.

When I was setting up for one gig with some keen novice audio engineering students to help, I told them to use the L-R crossovers to feed the mid and sub power amps. When I came to listen to the sound check, there was something strangely wrong with the bass. I found they had wired the left sub to the right amp output and vice versa. "You said to use a left-right crossover" was the excuse.

CherylJosie Tue, 04/03/2018 - 10:14

Once I stop ROTFL I'll carefully consider the latest responses.

You guys are the greatest. Thanks so much for sharing your experiences. There's no way I would come up with all data this on my own while mixing in bars.

I liked the idea of recording the reference sine signal and the acoustic sine signal at crossover and matching the delay between the two cabs with physical positioning rather than electrical delay. Cleaner, if it works, and 100% repeatable.

I appreciate confirmation that flipping phase and testing for a null is a valid (or at least prior art tested if not valid) way to set approximate delays for live sound. I didn't know if that had been tried before or not. My measurements of home theater crossover in a very small apartment demonstrated to me that even the plywood frame of the sofa introduces multiple large amplitude excursions near crossover from the phase disturbance of very early reflections off the sofa combined with very early reflections off all the close room boundaries (especially the floor and front wall) in the bass frequency range where absorption and re-direction of those reflections is minimally effective at taming them.

REW has similar phase reference measurement functionality built-in using a separate speaker channel to transmit an acoustic timing reference sweep before transmitting the measurement sweep (so I am told, with conflicting opinions on whether that function actually exists or works), but I've never looked for let alone tried it. Obviously in such scenario the reference sweep is contaminated by room acoustics and estimating absolute delay off that swept sine derived impulse with acoustic contamination on each channel is problematic, but it's better than nothing.

REW also has a way to do the reference sweep electrically to characterize the computer sound card and that method is documented in the help manual. I've never tried that either because I am using a UMIK-1 USB mike rather than a sound card and that method does not apply.

Some people have insisted that there is no way to use any reference sweep with the UMIK-1 acoustic or otherwise because it has its own internal A/D with its own clock and the resulting phase is uncorrelated between separate measurements no matter what mitigation reference signal is applied. I think that is probably technically accurate but misleading interpretation, but I just don't have the expertise to state conclusively one way or the other. For a brief sine sweep and acoustic reference I don't see that it matters if the transmit and receive clocks drift a little from each other. The acoustic phases are not going to go completely askew in a short time frame from a little clock drift. This is audio, not video. A little bit of multi-megahertz clock skew won't change the measured phase much after being divided down to audio sampling frequencies IMO.

I think for the time being I will leave the delays set at 0, use the revised DIY improvised L-R crossover in Master Fader that I outlined here in this thread, tune the FOH alignment by ear with some house music, and focus more on the monitors and mixing until I have the opportunity to measure the FOH system at my leisure in a controlled environment.

I guess if/when I get the chance I might try measuring a single sub/single main stack for symmetrical setups. I'll try to use positioning rather than delay to align the phase, if at all possible, since that is the repeatable way to do it.

I can try a dual sub/single main stack for the asymmetrical setup, where I just pad the subs down by 6dB for the level measurement and ignore the missing single orphan top cab. I'll assume the timing is OK in the center of the venue with the orphan cab contributing from the other end of the stage, and pad the subs back up by 3dB in the venue over my reference measurement. I'm not sure how the mutual coupling etc. additive works in small modal environment where the separated cabs are close enough to each other and their boundaries too for some mutual coupling at some frequencies, but I am sure that a little bit of graphic EQ will fix it if there's a problem.

My 'theoretical' assumption is that a reference measurement of one top cab and two subs boosts the subs too high by 6dB with mutual coupling, whereas in the venue the additional top cab adds 3dB of non-mutual-coupled signal to the mains so I only have to pad the stacked subs down by 3dB to normalize the level back to my measurement environment.

I know, I know, I'm being too analytical and ignoring the complex errors of live sound that make my simplistic dogmatic technical approach seem ridiculous. Noted.

I'll let you know how the dead horse is doing if/once I have borrowed the cabs from Charlie long enough to measure them, and stop torturing you with any more of these measurement-deficient crossover threads. Now that I'm moving back to my home I'll maybe set the system up in my back yard and do the measurement far from any buildings (I can get at least 30' from the nearest fence or house because it's a corner lot). I cannot measure the cabs in a 1 bedroom apartment! but maybe I can hostess a house party this summer and finalize the crossover in a controlled environment at the same time.

Unfortunately, with the Mackies and QSC mixed, there is not crossover for the top cabs built in. The Mackie top cab crossover is 100% integrated into the Mackie sub that was sold. The Mackie SA1530z top cab has no built-in crossover at all and the QSC sub only has its own half of the internal crossover. Charlie doesn't care anyway. He's going to keep using Master Fader with aux fed sub and the 75-80Hz crossover implemented in the DL1608 no matter what. It works for him, it works for me, we don't care about external crossover that just adds wiring and subtracts space in the van in exchange for theoretical electrical benefits that we cannot realize in the acoustics of a small venue anyway.

Horse is dead. Thanks for stripping it to the bones with the acid wash of your combined professional experience. That's what I came for!

paulears Sat, 04/07/2018 - 00:54

I have to keep reminding myself that lots of the rules for near field monitoring fall completely apart in a practical PA, where you have a combination of polar patterns, depending on the frequency. Omni for the subs, vaguely cardioid for the mids and torch like beams for the HF. Putting front fills in for the first few rows, and maybe delays for the rear all add up to major issues with time. Phase coherence as a concept is an interesting education wander in the park, but some of the errors we get practically cannot be 'tweaked out' and compromise is the name of the game. It does occur to me that maybe I should spend more time on the crossover settings - especially where the bass is provided by combinations of different speakers. With 18" subs, and then some cabs with 15" drivers for doing bass AND mids, there is some crossover in the crossover? Physically, the cones are separated in the front to back plane, so it would be possible to delay one to match the other, and worst case at least prevent nulling - but isn't this exactly how cardioid subs actually work? One 0f my PAs has a digital crossover to manage the bi-amping plus subs and the other uses a cheaper analogue device for the same amp system, and I'm not sure I can hear the difference.

Big spaces play strange tricks. If you use stereo subs, not for stereo, but just because the crossover has left and right outputs and dual channel amps, you can accidentally swap left and right and not notice, and from the back recently I could detect something not quite right and it turned out somebody had swapped the HF left and right inputs over when changing a dead amp. During the fault finding, pulling the left desk fader down removed one channel from the PA as expected, but from the back it was nearly the left one, but walking the 40m to the stage revealed it was the right! If the soundscape in the auditorium can be this ill-defined and confused, no wonder these tweaks to alignment physical and electronic make little sense.