Concept of time delay?

[D-MAN]

Well, all I can say to all is breakout the LS plans and a ruler. I'm right, simple as that.

Also, measure from the slot opening in the throat, not from the dust cap, cause that ain't part of the horn itself, kids.

Here - I'll help you guys.

DM

[D-MAN]

Since we are on the subject of the LS, here's what PWK said about the time delay issues...

Bruce, I agree that the time-delay issue would measure the overall pathlength FROM the diaghragm, but that wasn't what I was considering the horn pathlength, per se. I still won't go with 3 feet to the diaghragm but its probably about 32+ inches.

This attachment is from the original LS paper.

DM

[Guest " "]

Can I add my 2 cents here. I would like to throw this on the table for comment.

Base on the below chart, We are struggling to get to 2 msec. Looks like we are at 1.8 msec under the kindest use of the tape measure.

Based on an earlier chart, it would take a 3.2 msec delay at 500hz or below before it would be at the Threshold of Audibility. We would need the path lenth in the LaScala bin to be 4ft to get to 3.2 msec.

So sounds like this is less of an issue on the LaScala, Cornwall, and I 'll throw Belle into the mix as well so we can get to the meat of the question.

Now what about the k-horn and/or Jubilee? If the answer is yes, then I guess the case for active xovers has been made. Unless someone has away to do time-delay or group-delay compensation on a per driver basis using passive xovers.

Then there is a question of return on investment. How much group or time delay is needed to justify the investment on active xovers and additional amps, recognizing that at 500hz it would take 3.2 msec or more to be at the Threshold of Audibility

Also, it turns out, this is even less of an issue for folks that are using 6db per octave crossovers. Do we have a pure 6db per octave xover for the heritage line?

[D-MAN]

My opinion is that the LS and/or Belle are well below the 3ft "tap-dance" limit of audibility.
Also, from what PWK said from the LS paper, he was looking at it the same way as I was (considering an 18" long exponential midrange horn offset, + 1 foot, about 28" overall path length). [oops, that's 30, isn't it? but I am assuming an 18" midrange horn - how long is it really?]

The Khorn and Jubilee are not, due to longer path lengths, therefore, if one wants to go for an active delay, the "big" long horns are the logical candidates.

I think that worrying about the LS having noticable timing issues is as PWK would say...

DM

[steamer]

I am trying to understand how to implement delay in my system.

I hope I have worded this correct.

Greg

I thought that was when the wife says "Not tonight......"

LOL...

[steamer]

Well damn.I get home from work and the thread I started is loaded with info.Some very interesting reading here.Some of it seems a bit over my head at this time.

SUNNYSAL wrote:

At the same time, another audio consultant, Ed Long, the inventor of the term "time align" started working on absolute polarity. He found that some music and speech was very polarized, and easily heard. Other stuff was so multimiked, both in and out of polarity, that resolution was difficult, if possible at all.

Ed Long and Ron Wickersham are the inventors of the BagEnd sound.My sub system is made by BagEnd and one of the principles involved in this system includes time alignment with the high freqencies by omitting the low pass filter found on most subs resulting in minimal delay with the higher freq..I am very impressed with the bass and how well it integrates with the mains.I have never heard bass this good.Due to this expierience I am exploring the possibilities for time alignment in the rest of the system.Maybe there wont be a noticable difference maybe there will be????If as mentioned in previous posts other effects of a misalignment can be improved then it might be worth it.Hell there are many active crossovers out there that include the delay function and very reasonably priced.I got my Rane for $75 and could probably unload it for the same then get a Behringer for $130.A reasonable upgrade to play around with time delay.

Thanks for everyones input.

Greg

[sfogg]

"Also, it turns out, this is even less of an issue for folks that are using 6db per octave crossovers."

More of an issue because there is more overlap between drivers.

Because

of the path length differences the drivers can't really be in phase at

the crossover point. If you fed the two drivers directly from an amp.

(no crossover induced phase shift) the midrange would in effect be

around 127 degrees out of phase with the woofer at the crossover

frequency.

As such you get comb filtering through the

crossover region even dead on axis which is audible. If one can't hear the comb filtering they don't know what to listen for.

Make the crossover *sharp* and the audibility of the comb filtering

(which is caused by the misaligned drivers) goes way way down.

Shawn

[Guest " "]

"Also, it turns out, this is even less of an issue for folks that are using 6db per octave crossovers."

More of an issue because there is more overlap between drivers.

Because

of the path length differences the drivers can't really be in phase at

the crossover point. If you fed the two drivers directly from an amp.

(no crossover induced phase shift) the midrange would in effect be

around 127 degrees out of phase with the woofer at the crossover

frequency.

As such you get comb filtering through the

crossover region even dead on axis which is audible. If one can't hear the comb filtering they don't know what to listen for.

Make the crossover *sharp* and the audibility of the comb filtering

(which is caused by the misaligned drivers) goes way way down.

Shawn

Important to note, since we are now measuring in degrees, as opposed to msec's, we may have introduced another phenomena.

But back to your conclusions, Intuitively one would think so. But higher order xovers introduce phase shifting in multiple leaps. This adds another pheonima all together. Using EV vintage crossovers as an example (they were great at providing the data) 6db per octave is around 135 degree shift and 12 db per octage is 270 degree shift based. 4th and 6th order get even more hairy. Swap polarity of drivers at 12db in some cases, unswap at 4th order xovers, and re-swap at 6th order xovers.

Effects of comb filtering, this is yet another phenomena all together? Time-alignment?

I think we informally suggested that the Threshold of Audibility for time or group delay at any of the given frequencies were not an issue for the Cornwall, Belle, LaScala, and is an issue for the K-horn and Jubilee.

Should we circle back and re-review.

Is this more like a function of the electronics rather than pathway lenth and driver positioning?

Would this other phenomena change our assestment? If so, any examples.

[3dzapper]

This is a very interesting discussion but has IMHO little relevelance to audibility in a musical reproduction sense v live It is not something to lose sleep over. if it was, Dr Bruce Edgar would never rest. His Titan system's 6' horn typically is used several feet behind the salad bowl squakers. A far greater time differential than that of a Khorn or Jubilee.

The crossover at 400Hz xover within the primary musical and vocal range and the constriction of the throat in the K-400 to improve upper frequency response has more impact in my mind.

In a live setting, say a symphany, there can be as much as a hundred feet between two instruments. The bass instruments are located far behind the strings. In an amplified stage setting, likely horns are used for the upper registers and direct radiators for the lower with the players amps further behind. Thus we as listeners are conditioned to accept these delays as "normal" and they don't register as "distortion" in our psycics.

Too, the use of a subwoofer with LaScalas, etc, will introduce far more time smear than the LaScala itself. Yet, the addition of the lower registers is pleasant to our brains.

Is the tap dance test a valid evaluation of the capabilities of the human auditory system? Yes. Is it relivant???

Rick

[sfogg]

"Intuitively one would think so."

And audibly that is the case too. Been there done that.

"But higher order xovers introduce phase shifting in multiple leaps."

True, but the two devices are overlapping each other less and less the sharper you go. And as such the possible frequency range where combing artifacts could be heard gets smaller and smaller.

"I think we informally suggested that the Threshold of Audibility for time or group delay at any of the given frequencies were not an issue for the Cornwall, Belle, LaScala,"

And again... if people can't hear the effects of the different path lengths in a LaScala they simply don't know what to listen for.

Go play pink noise on a single LaScala. Stand 4' or so away from it on axis. Squat down slowly then stand back up. Listen to how the pink noise changes... that is from the comb filtering between the two changing as you alter the relative path lengths between the drivers to your ears by moving in the vertical plane.

Do the same thing with *sharp* filters in the LaScala and the pink noise stays very consistent sounding until you get outside of the dispersion of the horns.

Shawn

[Deang]

Squat down slowly then stand back up. Listen to how the pink noise changes... that is from the comb filtering between the two changing as you alter the relative path lengths between the drivers to your ears by moving in the vertical plane.

If I sit in my chair and just listen to music, I don't notice any of that. However, with the ESN's, sitting in my chair 12' away -- there were times I thought I could hear the sound moving between the drivers, a somewhat phasey quality -- even though I wasn't moving. Something was happening I'd never heard before, and whatever it was -- I noticed it. Also, and this is just my opinion -- speakers using first order filters sound more natural to me, and I think they do a much better job of relaying transient information. I just got done reviewing JC's thread on his DBB CornScala project. He went through a lot of networks, and after all of the dust settled -- he ended up preferring 6db/oct, 6dB/oct, and 18dB on the tweeter. I found it interesting that he reached the same conclusion I did after my merry-go-round. I think people are just sensitive to different artifacts in the sound, and though something like the effects of comb filterng are irritating to you -- it apparently doesn't bother someone like me at all, and it might even be possible that the effects contribute in such a way that makes the music sound right to me. Al always says, "It may sound good, but that doesn't mean it's right." I always respond, "Yes it does."

[sfogg]

Dean,

"I think people are just sensitive to different artifacts in the sound"

Agreed.

I was just pointing out that the sound of comb filtering is readily audible if someone wants to listen for it. To claim different path lengths have no effects on the sound is wrong and it is easy to demonstrate.

How much it bothers someone is a different discussion.

"- there were times I thought I could hear the sound moving between the drivers, a somewhat phasey quality -- even though I wasn't moving."

I have never noticed that and that was something I was concerned about at first with them too so I ran some frequency sweeps and such.

"and I think they do a much better job of relaying transient information."

Different strokes for different folks I guess. I tend to think the opposite. The smearing I hear with first order and horns make sharp transients blurred IMO.

Shawn

[DrWho]

Since we are on the subject of the LS, here's what PWK said about the time delay issues...

...in 1963.

PWK's views changed greatly throughout his entire life and though it wasn't published Jim Hunter tells me PWK in the later years was all for time-aligment and felt it was the biggest downfall to hornloaded speakers. He never did anything about it earlier on because the tools simply didn't exist to correct for it and he felt other factors (like efficiency) were way more important to address. His earlier writings were more of an attempt to justify the compromises he made.

I've never understood why people try so hard to justify the misalignment - yet will spend way more money on fancy wire and swap amps left and right. On top of it all there are even more advantages to active crossovers than just the time alignment - it's just entirely a better approach on so many levels. [*-)]

[djk]

Some thoughts, in no specific order:

If any of you had actually heard the difference that a time delay makes in a horn system, you wouldn't be having this discussion. I have.

A Cornwall is under the Blauert & Laws criterea, but the time smear is plainly audible in comparison to a system that is time off-set corrected. On music.

Can you live with it?

PWK had no choice, but I do. I can hear the difference and I don't have to live with it. If you don't want to deal with it, you can just stick your head in the sand.

"he ended up preferring 6db/oct, 6dB/oct, and 18dB"

Which is what I recommended to you a couple of years ago. The 18dB on the tweeter high-pass reduces the effects of comb filtering between the midrange and the tweeter. A Butterworth (ALK) sounds better than a Chebychev (AA). It helps to use a swamping resistor around 30 ohms or so, similar to what is in a Forte network.

Some people prefer the larger peaks and dips that you get with a 6dB, it sounds more 'lively'. That's fine. It costs so little to try that I would encourage people to hear for themselves before making a decision based on any discussion.

"there were times I thought I could hear the sound moving between the drivers, a somewhat phasey quality "

The steeper the slope of the crossover, the closer the levels must be matched. 6dB is more forgiving in that regard, it has to be because you can't adjust the levels in fine enough steps with any autoformer.

"Behringer for $130"

Not the most transparent sounding piece on the market, but will work for a Cornwall type speaker, but not a LaScala or Klipschorn type speaker. Will not fix the tweeter. A DCX2496 can, they are back in production again.

"by omitting the low pass filter found on most subs resulting in minimal delay with the higher freq"

Pure marketing BS. The group delay is strictly a function of the frequency response of the filters. The BagEnd subs have a two-pole low-pass filter (they call it an integrator) that causes a 12dB/oct low pass in the transfer function at about 80hz in most of their models.

Air path length on a LaScala. If you take a piece of string and tape it to the dustcap of the woofer, and run it straight down the middle of the horn all the way to the mouth, it will be 36" ±1/4". The acoustic center will actually be behind this point by a bit.

[colterphoto1]

So then I'll just leave things pretty much as they are then, all things considered? []

And thanks Rick, that orchestra allusion really makes us toss all the time-alignment theories out the window, doen'st it? Is it possible that the brain adjusts for the differences, then ignores them? That would be one thing Doc could not measure.

Michael

[steamer]

djk you wrote:

"by omitting the low pass filter found on most subs resulting in minimal delay with the higher freq"

Pure marketing BS. The group delay is strictly a function of the frequency response of the filters. The BagEnd subs have a two-pole low-pass filter (they call it an integrator) that causes a 12dB/oct low pass in the transfer function at about 80hz in most of their models.

I was under the impression they use a dual integrator that have minimal delay charactoristics as compared to a normal lowpass filter.Also that the frequencies they operate thier subs at are 180 degrees out of phase with the higher frequencies passed to the mains.They then reverse the phase at the outputs to the mains to bring everything back into phase.

Does an integrator circut have different charachteristics than say a typical low pass filter?

Greg

[Deang]

Thanks Dennis/Shawn

Who, I don't think anyone is really trying to "justify" the issue so much as just trying to figure out if it's worth hassling with. Sounds like it is, but I'm not sure if I can live with the sound from all the solid grate devices in the signal path, and other options resulting in the level of sound I'm used to are too damn expensive.

[3dzapper]

"And thanks Rick, that orchestra allusion really makes us toss all the time-alignment theories out the window, doen'st it? Is it possible that the brain adjusts for the differences, then ignores them? That would be one thing Doc could not measure. "

That was my point Micheal. The brain is a wonderful thing. A sewage plant stinks to most of us, but a person who works in that plant does not notice the odor.

Sometimes, we get ourselves in trouble or cause more trouble for ourselves attempting to find "prefection" Perfection to mere mortals like us is unobtainable.

DJK, who's knowledge and experience I respect greatly, may well be correct, hearing a time corrected Khorn may be a relevation. But, by adding elements wouldn't one compromise the beauty of a clean sounding zero feedback pre-amp and SET or PP amp with the addition of many small resistors and op-amps?

Like the good Dr's motto, life and audio is a game of compromises.

Rick

[sfogg]

Rick,

"In a live setting, say a symphany, there can be as much as a hundred feet between two instruments. The bass instruments are located far behind the strings."

That isn't exactly the same thing though. Each instrument is producing its own complete sound. You don't have half the sound of the strings coming at you from one place on the stage and half the sound coming from further away on stage.

DJK,

"The steeper the slope of the crossover, the closer the levels must be matched. 6dB is more forgiving in that regard, it has to be because you can't adjust the levels in fine enough steps with any autoformer."

Good point, a sudden jump in level with a sharp crossover would tend to highlight that the crossover happened. Yet another benefit of active... I can adjust driver levels in 0.1dB increments.

Good to hear the DCX2496s are back in production.

Shawn

[DrWho]

The "orchestra illusion" is a rather misleading analogy.

The spacing of the orchestra has a huge effect on the sound and is

completely intentional. Composers have been known for moving sections

around to achieve a different type of sound and different styles of

music has different seating arrangements. You don't want every

instrument in an orchestra to be time-aligned because then it will

sound like one instrument - no longer achieving that symphonic sound.

The sound of overdubbed guitars does not sound the same as a huge

ensemble of them or even mic'ing dozens of cabinets.

The person in the studio is choosing mic positions and making timbre

adjustments that are very sensitive to the time-domain. He probably

isn't sitting there thinking "hmmm, gotta time-align everything" but

the time-alignment he ends up with is going to have a particular sound.

And our playback systems aren't supposed to alter that choice.

Another problem with the orchestra analogy is that it's talking about a

delay on the entire passband of every instrument. Sure, the upright

bass is a bass instrument - but it has plenty of information above

5kHz. In other words, it's just not going to sound right without a

tweeter in the playback system (even an electric bass is that way). The

argument about time-alignment, especially in regards to horns, is that parts of the passband are getting delayed differently relative to each other. Delaying information only below 400Hz is going to sound different than equally delaying 0 Hz to infinity Hz.

In regards to the 6dB slope crossover frequencies...I would argue that

the biggest advantage there is the increase in IMD. By increasing the

passband of a driver, you naturally increase the cone-movement for that

driver at any given SPL which is going to result in an increase in

distortion. But I would argue this to be a good thing because the

studio and mastering engineers were hearing similar distortions on all

the different pairs of speakers used for referencing the mix. The distortion then becomes a part of the intended sound.

Another reason I think slower slopes sound better is that the polar

response and distortion characteristics don't change as abrubtly

resulting in a speaker sounding like one unit and not two seperate

entities.

There are a lot of variables involved - all of which changing one

changes another. Even horizontal time-alignment doesn't account for the

vertical path length differences (which will always show up in the

off-axis response). If people are looking for how important the

difference is, then I think the best thing to do would be to compare it

to the other variables involved. For example, efficiency is rather high

up on the list along with frequency response. I think time-alignment

makes a much bigger difference than the difference between any two

quality amplifiers. Will spending half as much on an amplifier, so you

can afford two and then time-aligning the system, sound better than

spending twice as much on the amp? I think so.