Timing Measurements

[DrWho]

Early in this thread Rudy mentioned that it's hard to align signals when using a different range of frequencies....this is totally true.

 

I use the impulse plots to get an idea for the relative time arrival, but then I go to the frequency response plots to dial it in exactly. What you do is set your xovers and EQ and all that so that the plots overlayed intersect at -6dB. Then you invert the polarity of one of the drivers (doesn't matter which one), and then starting with the delay measured from the impulse plot, adjust +/- 180 degrees until you get a massive dip at the xover frequency. Then revert the polarity back to normal and it should sum perfectly to +6dB, which will be at your 0dB target.

 

It's a lot more complicated to type than to do in practice...it really only takes a few minutes.

 

 

If you want to experiment with the difference that time-arrival makes, then you can take that delay setting and then increase it by 360 degree intervals. The steady state frequency response should measure the same, but now you'll have a full wavelength (or more) of delay. Switching back and forth between the two delay settings should be straightforward for most active xovers. If the xover frequency is in the 500Hz to 2kHz range, then I like to use the sound of snares to hear the difference. If you know what a real drum snare sounds like in person, then you can very quickly tell the difference...the sound will tighten up quite a bit when it gets aligned properly. Then once hear the difference on the snare, then you start to notice a similar difference in timbre across all the instruments - it's just that it's easier to notice with transients. The more constant, or steady state the signal, then the less the time delay matters.

[Rjk1972]

I find all of this interesting. What about extreme slope crossovers? It has always been my understanding that the advantages of the extreme slope network was to eliminate comb distortions and eliminate the need to physically align your drivers. Has anyone hear compared an extreme slope network to an active network or done any comparisons between either and time aligned drivers?

[Chris A]

What about extreme slope crossovers? It has always been my understanding that the advantages of the extreme slope network was to eliminate comb distortions and eliminate the need to physically align your drivers.

 

Actually, extreme slope crossovers exacerbate the time alignment problem.  You need all-pass filters (analog or digital) or time delay (digital) to correct the phase shift between the low frequency and higher frequency outputs of the crossover networks.  Every "order" of the crossover above first order adds another 90 degrees of phase shift between the high and low outputs.  Tom Danley talks about this in his patents (Unity horn and Synergy horn patents, that is).  He boasts that he has developed "non-:named" (i.e., Butterworth, Linkwitz, Bessel, etc.) passive crossover filter topologies to correct this for his full-range horn-loaded loudspeakers.

 

 

 

Has anyone hear compared an extreme slope network to an active network or done any comparisons between either and time aligned drivers?

 

Extreme slope networks can be both active (analog or digital) or passive (analog). 

 

Yes, I've heard both "extreme slope" and "gentle slope" crossovers of both passive and active varieties.  I've also heard the effects of even very small time/phase misalignments: it's a big deal to get the phase aligned to within 1/10th of a wavelength at the center crossover frequency.  Trust me on this rule of thumb: within 1/10 wavelength alignment (i.e., 36 degrees) is required to avoid hearing the artifacts of misalignment.

 

Chris

Edited August 17, 2015 by Chris A

[Rjk1972]

If the extreme slope networks make it worse then what are the benefits of an extreme slope network?

[Bubo]

 

One day.... when my two braincells have grown to three [ when that even will happen in the first place].... then I hope to understand what you guys do with measurements....

 

I only have my wife's ears and mine, but we do use time-delay in our two-way system and it's a bit like the icing on the cake for us.

 

Much of this is done for setting loudspeaker delays automatically using Audyssey and YPAO, etc. using exactly the same techniques.  All we're doing is using the technique for single loudspeakers, and all you really need to buy is a calibrated USB microphone, a microphone stand, and an active crossover.  You probably already have a laptop or desktop computer.

 

Chris,

 

thanks for sharing your expertise.

 

In order to introduce delay, does the active crossover digitize the signal and use data buffers, then reconvert to analog?

 

On the highly regarded JBL studio monitors of old, all of the drivers appear to be lined up exactly, does this eliminate the phase issue?

[Chris A]

If the extreme slope networks make it worse then what are the benefits of an extreme slope network?

 

If you are referring to passive (analog) extreme-slope crossovers then you'll need to add all-pass filters to correct for the added phase shift, or you need to move the drivers in the loudspeaker so that they are within 1/10th wavelength of being time-aligned while using the new crossover network.  If you have an analog active crossover, then you'll need to use all-pass filters to correct for the time misalignments ) assuming that the analog all-pass filters have enough phase adjustment range (they won't have enough adjustment range for certain loudspeakers, see below).

 

If you have a digital active crossover, you can simply use time delay or all-pass filters to correct for any time misalignment introduced by extreme slope crossover filters if using typical IIR filters and the inherent time misalignment of the drivers/horns, or you can use custom FIR filters to correct both the amplitude and phase of the crossover filters, if your digital crossover has FIR filter capability. 

 

Once you complete the task of using extreme slope crossovers by correcting their added time misalignments plus the inherent time misalignments due to the drivers not being time aligned, you'll have loudspeakers that sound much more phase coherent--much cleaner.

 

For instance, all stock Klipsch Heritage loudspeakers (Klipschorn, La Scala, Belle, Cornwall, Heresy) have time-misalignment issues of 6-8 wavelengths between the midrange and the tweeter, and 1-3 wavelengths between the bass bin and midrange horn-driver.  One of the reasons why some people prefer gentle slope networks in these loudspeakers is that the gentle slope networks tend to mask the time misalignment problems over more than one octave.  This means that the loudspeakers exhibit frequency-dependent main lobe phase issues from ~1 to 8 kHz.  Once you actually correct the root time alignment issues and use steeper slope crossover filters, they sound like you've got new loudspeakers, e.g., time aligning a Belle, time aligning Klipschorns.

 

I recommend using your ears to hear the difference, not your imagination.

 

Chris

Edited August 19, 2015 by Chris A

[colterphoto1]

sit waaaaay closer to the woofer? 

[Chris A]

That won't correct the midrange-tweeter misalignment of 6-8 wavelengths, or the bass bin-midrange misalignment of 1-3 wavelengths, but you might find a place where you're sitting on a multiple of 360 degrees phase shift between the bass bin and the midrange horn/driver, so that while playing music without transients you can't hear the effects of multiple wavelength delays at that crossover.

 

Of course, that sort of music would be a little dull to listen to over time.

 

Chris

Edited August 18, 2015 by Chris A

[Rjk1972]

If the extreme slope networks make it worse then what are the benefits of an extreme slope network?

If you are referring to an passive (analog) extreme-slope crossovers then you'll need to add all-pass filters to correct for the added phase shift, or you need to move the drivers in the loudspeaker so that they are within 1/10th wavelength of being time-aligned while using the new crossover network. If you have an analog active crossover, then you'll need to use all-pass filters to correct for the time misalignments ) assuming that the analog all-pass filters have enough phase adjustment range (they won't have enough adjustment range for certain loudspeakers, see below).

If you have a digital active crossover, you can simply use time delay or all-pass filters to correct for any time misalignment introduced by extreme slope crossover filters if using typical IIR filters and the inherent time misalignment of the drivers/horns, or you can use custom FIR filters to correct both the amplitude and phase of the crossover filters, if your digital crossover has FIR filter capability.

Once you complete the task of using extreme slope crossovers by correcting their added time misalignments plus the inherent time misalignments due to the drivers not being time aligned, you'll have loudspeakers that sound much more phase coherent--much cleaner.

For instance, all stock Klipsch Heritage loudspeakers (Klipschorn, La Scala, Belle, Cornwall, Heresy) have time-misalignment issues of 6-8 wavelengths between the midrange and the tweeter, and 1-3 wavelengths between the bass bin and midrange horn-driver. One of the reasons why some people prefer gentle slope networks in these loudspeakers is that the gentle slope networks tend to mask the time misalignment problems over more than one octave. This means that the loudspeakers exhibit frequency-dependent main lobe phase issues from ~1 to 8 kHz. Once you actually correct the root time alignment issues and use steeper slope crossover filters, they sound like you've got new loudspeakers, e.g., time aligning a Belle, time aligning Klipschorns.

I recommend using your ears to hear the difference, not your imagination.

Chris

Thanks for the answers.

Any recommendations for a crossover that can accomplish this for a pair of Cornwalls? Time aligning the drivers really doesn't seem like an option.

[Rjk1972]

If the extreme slope networks make it worse then what are the benefits of an extreme slope network?

If you are referring to an passive (analog) extreme-slope crossovers then you'll need to add all-pass filters to correct for the added phase shift, or you need to move the drivers in the loudspeaker so that they are within 1/10th wavelength of being time-aligned while using the new crossover network. If you have an analog active crossover, then you'll need to use all-pass filters to correct for the time misalignments ) assuming that the analog all-pass filters have enough phase adjustment range (they won't have enough adjustment range for certain loudspeakers, see below).

If you have a digital active crossover, you can simply use time delay or all-pass filters to correct for any time misalignment introduced by extreme slope crossover filters if using typical IIR filters and the inherent time misalignment of the drivers/horns, or you can use custom FIR filters to correct both the amplitude and phase of the crossover filters, if your digital crossover has FIR filter capability.

Once you complete the task of using extreme slope crossovers by correcting their added time misalignments plus the inherent time misalignments due to the drivers not being time aligned, you'll have loudspeakers that sound much more phase coherent--much cleaner.

For instance, all stock Klipsch Heritage loudspeakers (Klipschorn, La Scala, Belle, Cornwall, Heresy) have time-misalignment issues of 6-8 wavelengths between the midrange and the tweeter, and 1-3 wavelengths between the bass bin and midrange horn-driver. One of the reasons why some people prefer gentle slope networks in these loudspeakers is that the gentle slope networks tend to mask the time misalignment problems over more than one octave. This means that the loudspeakers exhibit frequency-dependent main lobe phase issues from ~1 to 8 kHz. Once you actually correct the root time alignment issues and use steeper slope crossover filters, they sound like you've got new loudspeakers, e.g., time aligning a Belle, time aligning Klipschorns.

I recommend using your ears to hear the difference, not your imagination.

Chris

Thanks for the answers.

Any recommendations for a crossover that can accomplish this for a pair of Cornwalls? Time aligning the drivers really doesn't seem like an option.

[Chris A]

Time aligning the drivers really doesn't seem like an option.

Because...?

 

You can time align the midrange-tweeter easily by removing the K-77 tweeters from the cabinet and placing them on top of the loudspeaker, centered at the back of the cabinet, approximately aligned with the midrange driver (K-55) below it.

 

Moving the tweeter fore and aft on top of the cabinet, you will hear it come into focus when the time alignment occurs.  From there, you can listen and determine whether or not it's worth your time to place the tweeters in small baffles, like Bruce did with his replacement tweeters (top picture).  It's this midrange-tweeter crossover time misalignment that has the greatest effect on the sound of the loudspeaker. 

 

While I actually do not recommend removing the midrange horn/driver from the cabinet due to the benefits of clamping the midrange horn mouth to the front baffle to eliminate ringing, I don't see any reason why you couldn't simply rest the tweeter on top of the cabinet, like Michael (Thaddeus) did in the bottom picture, but aligned with the midrange driver below it.  This actually has more benefits in terms of eliminating the "looking through the baffle" effect of having the tweeter mounted to the rear side of the front baffle thus improving its polars.  You can also turn the tweeter vertically to get slightly better horizontal coverage performance (...in fact ElectroVoice's recommended orientation).  Simply put something absorbent on top of the cabinet to absorb the reflections off the top of the cabinet, and you're good to go.

 

Edited August 18, 2015 by Chris A

[mark1101]

 

What about extreme slope crossovers? It has always been my understanding that the advantages of the extreme slope network was to eliminate comb distortions and eliminate the need to physically align your drivers.

 

Actually, extreme slope crossovers exacerbate the time alignment problem.  You need all-pass filters (analog or digital) or time delay (digital) to correct the phase shift between the low frequency and higher frequency outputs of the crossover networks.  Every "order" of the crossover above first order adds another 90 degrees of phase shift between the high and low outputs.  Tom Danley talks about this in his patents (Unity horn and Synergy horn patents, that is).  He boasts that he has developed "non-:named" (i.e., Butterworth, Linkwitz, Bessel, etc.) passive crossover filter topologies to correct this for his full-range horn-loaded loudspeakers.

 

 

 

Has anyone hear compared an extreme slope network to an active network or done any comparisons between either and time aligned drivers?

 

Extreme slope networks can be both active (analog or digital) or passive (analog). 

 

Yes, I've heard both "extreme slope" and "gentle slope" crossovers of both passive and active varieties.  I've also heard the effects of even very small time/phase misalignments: it's a big deal to get the phase aligned to within 1/10th of a wavelength at the center crossover frequency.  Trust me on this rule of thumb: within 1/10 wavelength alignment (i.e., 36 degrees) is required to avoid hearing the artifacts of misalignment.

 

Chris

 

Chris,

 

How do you know that your setup is time aligned any distance from the point in space you dial it in for?

[Chris A]

Trick question?

[Don Richard]

If the drivers are stacked vertically and they are separated farther than 1/4 wavelength, adjusting delays will time align the system at only one vertical point in space. The greater the separation distance, the smaller the sweet spot. Coaxial speakers or aperture summation speakers (Danley) do not have this issue, and can achieve coherent summation over a wide frequency range, and throughout the speaker's coverage pattern.

Edited August 18, 2015 by Don Richard

[mark1101]

Yeah.........Don makes a point that we don't discuss much.  No Chris........not a trick question..........but the answer is trick sort of.

 

In a room time alignment makes a difference.  Absolutely.  I have proven it to myself that it does.  But my opinion is that the precision is overridden to a degree by reflected sound.  You time align to a "point" in space.  Move your head 1 inch and it's off.  Too many reflections and it's off.

 

a 6Khz wave is 2.25" approximately.  A popular crossover point.  Like I said.......move your hear an inch and your off by a wide margin.,,,,,,,,or suffer too many reflections and it simply doesn't work.

 

So what do you do?  Coaxials?  Yeah, that works.  I have those with passives.....excellent.

 

What else can we do?  Serious question...........reduce reflections?, pick the point of time alignment carefully, do the best we can and eliminate gross mis-alignments (I do this)?  I don't know the answer.  Just wondering what you guys think about it.

[IB Slammin]

"Ok what exactly am i looking at in your diagram? We are talking about khorns here."

 

What part of

 

"Digital delay is the only fix."

 

Didn't you understand?

 

What part of "If the horns are deeper than the woofer, then an all-pass delay will fix it." and "I've done the above (balanced) with dual 12's"

 

Didn't you understand?

But hey boss,..

Couldn't time alignment be achieved through cabinet design? '

tc

Edit: Or steep slopes.....just saying that I have heard actives suck the life out of music

Edited August 19, 2015 by IB Slammin

[IB Slammin]

Yeah.........Don makes a point that we don't discuss much.  No Chris........not a trick question..........but the answer is trick sort of.

 

In a room time alignment makes a difference.  Absolutely.  I have proven it to myself that it does.  But my opinion is that the precision is overridden to a degree by reflected sound.  You time align to a "point" in space.  Move your head 1 inch and it's off.  Too many reflections and it's off.

 

a 6Khz wave is 2.25" approximately.  A popular crossover point.  Like I said.......move your hear an inch and your off by a wide margin.,,,,,,,,or suffer too many reflections and it simply doesn't work.

 

So what do you do?  Coaxials?  Yeah, that works.  I have those with passives.....excellent.

 

What else can we do?  Serious question...........reduce reflections?, pick the point of time alignment carefully, do the best we can and eliminate gross mis-alignments (I do this)?  I don't know the answer.  Just wondering what you guys think about it.

Mark,

To my ear your 684 subs are dialed in perfectly with the MCM grand. (a monster system in the home environment)

I think that you hit the nail on the head on this one. We move step by step to achieve a goal.

To me the last step is to tune the room. I must admit that I am a selfish listener and could care less about dispersion. I don't walk around the room. My wife and I sit in a 5'wide sweet spot for dedicated listening. Anything less should be heard in the car...

too much ???? smile

tc

[djk]

"But hey boss,..Couldn't time alignment be achieved through cabinet design? tc"

 

Sure, but how is that going to help the Klipschorn?

"Edit: Or steep slopes.....just saying that I have heard actives suck the life out of music"

 

ES networks can sound really funny, I don't care for their sound.

 

Most commercial electronic crossovers have a lot of room for improvement, they use $0.06 opamps (really!), and polarized electrolytics in the signal path. I used to build my own designs, now I just modify off-the-shelf designs. Learn to solder.

[Chris A]

It helps me if you first ask what your real concerns and questions are, then I can have a chance to address them, one by one. 

 

Mark, in your case there are some things that I see are probably important:

 

1) looking at a picture of your main setup, I'd say that you have a bit of a "vertical separation issue" that most people who frequent here don't have (in terms of magnitude).  If you think about reducing your setup down to a single crossover at a low enough frequency (like 400-450 Hz) then we are working with essentially a point source.  What I'm talking about is the two-way Jubilee (or substituting the MWM bins for the KPT-KHJ-LF Jubilee bins).  It is coherent in-room because there are no high-frequency crossovers that are at 2.25" wavelengths and the vertical separation issue basically disappears.  I believe that you can fill in the rest of this story from our discussions with Roy. 

The KPT-305s don't have adjustment capability vertically, but you can place a small riser--like a 2x4 block--under the back rubber feet of the cabinets to tilt them forward.  The K-402s, of course, have adjust capability built-in, so you can infinitely adjust these to point at your preferred listening position, thus lessening the effects of vertical separation (if you look at the Wilson series of loudspeakers, they are doing the same thing - arranging the vertically separated series of drivers in an arc that points at the listener).  The only thing left to worry about is the K-510 HF on top.  Personally, I'd remove it and use one driver in the K-402s to cover the range from the KPT-305s upwards.  I believe that this is the area that is causing the greatest contributions to "vertical separation" that you have described.

 

2) We talked about the level of early reflections in your room, right around your loudspeaker stacks.  One of the things that you've done well is that you don't really have ceiling reflections--this is good considering the height of your loudspeakers.  I'd recommend pulling your stacks forward about a foot and pushing your subs against the front wall (assuming that they're not already touching the front wall).  [in fact, I'd recommend placing your subs into the corners-radiating into the corners, and placing your stacks in front.  Then you get the advantages of corner loading with the subs and it will tend to better mask the HF harmonics and modulation distortion emitted from the direct-radiating woofers.]  I'd also recommend putting down a bit more carpet right up to the mouths of the MWM bins and your direct-radiating subs. Also, as we've discussed, I'd recommend placing a column of 2'x2' acoustic absorption pads on the side walls next to the K-402 mouth exits.  I've found that early reflections are killers of imaging, as we've discussed.

 

3) I recommend listening near the back wall of the room (which you probably already do) due to the horizontal and vertical distribution of the radiating elements of your array to get that array of elements to coalesce into two coherent point source locations. 

 

4) Then you can dial in everything in terms of time alignment to reap the full advantages.

 

 

 

...just saying that I have heard actives suck the life out of music

 

"Just saying" means that you don't wish to discuss it, but that you wish to say it.  Okay, I won't discuss it, because like anything else, there's good equipment and not-so-good equipment that can be used as the subject of discussion.

 

Chris

Edited August 19, 2015 by Chris A

[mark1101]

The way I time aligned what you show in that picture is with a tape measure.  Believe it or not.  It is all time aligned to the front of the cabinets, not to a point in space.  It may not be perfect but I'm certain is very very close.  I have 3 systems in that room and the other two do not even come close to the clarity of the MCM.

 

As far as the vertical height separation.  I consider that a big advantage.  The system sounds huge and separated.  Without experiencing it I really can't describe it other than to say everything else sounds small to me, and the MCM sounds closest to what I hear when I go to a concert and listen to a very large sound reinforcement system.  This was my goal when I built it, and it does very well in the room.