Article on the K-Horn and Interview

[Islander]

Although the 9kHz glitch certainly seems measurable, just how audible is it, and is it audible at most typical listening levels?

[BEC]

Although the 9kHz glitch certainly seems measurable, just how audible is it, and is it audible at most typical listening levels?

Ah, now that is the right question.  I have done a little test on trying to hear that 9 khz glitch.  I measured a K-55V with push pin connectors and found that the output at 9 khz was about 10 db below the "in range" output of the K-55V.  With this driver installed, I wired a switch into the squawker circuit that allowed the squawker to be switched off and on.  I then input a 9 khz tone into the speaker.  With me at a normal listening position, and an assistant alternately turning the squawker off and on, I could not hear a difference between just the tweeter and both the tweeter and squawker playing that tone.  Then, my son with much younger ears tried to see if he could identify a difference and he was also unable to do that.  

So, I figure that if it is that way with a steady tone, actual music would be even more difficult since the occurance of the 9 khz frequency would be even more masked.

I think if you can find a K-55V that has the 9 khz output closer to the output "in range" you may have a better chance of hearing it at least with a pure tone.  Still might not be able to hear it with music.  Just my opinion.

Bob Crites

[DrWho]

Did you have the squawker and tweeter time-aligned? If not, then there's all sorts of phase issues that could come into play that mask the effect of a constant tone that wouldn't be there with dynamically changing music.

[BEC]

Did you have the squawker and tweeter time-aligned? If not, then there's all sorts of phase issues that could come into play that mask the effect of a constant tone that wouldn't be there with dynamically changing music.

That is an interesting thought.  Seems to me there would be more masking with a dynamically changing signal.  At least to the ear.

Bob

[ZAKO]

On my system i can shift the tweeter back to align with the mid driver plus change xover slopes 6db, 12db, 18db, 24db and i cant hear the phase issues..constant tones or dynamic changes in music....I believe Paul Klipsch made this test also...He also did a blind test relating shift of bass bin in relation to HF units...It didnt become apparent till the bass bin was at your knees.

[DrWho]

Wow.

About twice a week I set up portable sound systems at work where I'll time-align by ear. Lately I've been measuring my results and it seems I'm usually within ~1ms. Don't matter if its aligning bass bins or squawker or tweeters. Being unsure as to whether or not I've "over-trained" my ears, I'll have my coworkers turn the knob, not knowing what the knob is doing or really knowing anything about audio and ask them to stop when it sounds better. They usually get within ~2ms. And this is just playing whatever random CD happens to be in the CD player (stressing that it's not something anyone is familiar with).

There's no way in heck you can do it by ear playing constant test tones though.

Btw, I think many would find PWK's latest thoughts about the time domain a lot more interesting than what he was trying to push back in the 70's. Perhaps I've misinterpretted what Roy has told me, but PWK was a supporter of time-alignment in his later years.

[ZAKO]

It seems that with the Smith horn Distributed source horn,, sounds appear at the mouth rather than eminating from the throat,,, Bob Smith suggested alighning at the mouth of horn rather than back at driver....Ill need to borrow your ears. Every thing sounds cohesive though..

[DrWho]

My measurements are only showing relative time-arrivals so I have no

way of knowing where the physical acoustical centers are. I can only

tell when everything is lined up. Does Smith provided any reasons for

why his acoustical centers are pushed forward so far? That sounds like

something that horn designers would want to take advantage of.

Anyways,

that's getting a bit off topic. I was just trying to point out that the

difference in acoustical centers between the K400 and K77 has the

ability to introduce significant amounts of phase shift at 9kHz, where

the wavelength is on the order of 1.5". All it would take is 180

degrees of phase, or any multiple of .75" for maximum destructive

interference to occur. 90 and 270 degrees of phase rotation are both

going to result in no change, so anywhere between 90 and 270 (.3" to

1.1"), or any integer multiple is going to be have less output than if

perfect summation was occuring. Or more simple to think about, any

integer multiple of 1.5" is the only frequency where full summation

will occur.

When dealing with a steady state signal (like a test

tone), it's just a basic superposition problem. But for a transient

signal, you have to deal with the absence of superposition at the

beginning and end of the group response (smearing the sound), and the

comb-filtering during the group response (changing the timbre).

With

a steady tone you don't get the smear, and the comb-filtering is just

the superposition problem. With a transient signal (like cymbals or

snares), you get to hear the smearing in addition to the

comb-filtering. So if the comb-filtering isn't present at a given

listening position, then the smearing is the only disparity left to

hear...which I think is a fairly evident distortion, but it's more

readily heard when you can adjust the time offsets in real time. When

swinging the knob, you'll hear the comb-filtering shift (sounds like a

flange) but you'll also hear the duration of the high-hat getting

shorter until it doesn't sound hashy anymore.

[stereohermit]

When Ifirst bought my La Scalas back in 1979, I time aligned them, kept them that way ever since. Whenever I go back to stock, I find them forward, dry and two demensional. I also can tell you the tweet /mid interaction problem, mentioned in this thread, is mostly due to the time stagger.

[Islander]

Bob, thanks for clearing up the issue of the glitch, to my satisfaction at least.

On the time alignment issue, I seem to recall seeing a reference to an inexpensive ($250 range?) active crossover with adjustable delay on the forum in the last month or so. Does anyone remember it, and would it improve the sound (and not be a piece of junk at that price)?

[stereohermit]

The xover you?re probably referring to is a Rane AC-22, or maybe one of the Behringers. Not very transparent, two demnsional...not so good. I know this cause ive used them in pro situations. Also, the "time delay" is not the same as physically ajigning the driversxxxxxxxxx

[mas]

One of the active crossovers for $250 is the Behringer DCX2496 or a used Rane AC23.

The advantage of the newer digital and computer interfaced units is simply that it can be much easier to precisely dial in the amount of delay rather than trying to turn a rotary knob while trying to address the precision alignment desired.

And signal alignment is the same as physically offsetting and aligning the acoustical centers of the drivers - albeit in one plane (usually the vertical XY plane).

We have been doing this for years, and an impulse response or an ETC makes aligning the centers a quick and simple adjustment.

Note: The problem with the physical alignment above is the radical amount of reflections and diffraction that will occur from the cabinet within the coverage of the mid horn... Electronic signal alignment via delay avoids this.

[stereohermit]

Point well taken, however I find with the Klipsch, there is a problem with tambral shift with distance, since, stock, you are differing distances from the three sources. The tweeter falls more rapidly with increasing distance, relitive to inverse square law. Only physical alignment will correct this...

[mas]

Point well taken, however I find with the Klipsch, there is a problem with tambral shift with distance, since, stock, you are differing distances from the three sources. The tweeter falls more rapidly with increasing distance, relitive to inverse square law. Only physical alignment will correct this...

Huh?

You measure the IR or ETC at the listening position. The acoustic centers are aligned in time relative to the distances of each acoustic center to that spot. That is the ONLY way signal alignment is done. And electronic delay works just fine. In fact, it often offers additional benefits that physical alignment may not, as evidenced above.

When aligning signals in the time domain, time and distance are related linearly. The amount if delay in time is related directly to the distance. Period.

The idea that you can align a signal physically within the time domain that you cannot align via electronic signal delay within time is totally without merit. But you can avoid obstructions and secondary reflections and diffraction with the signal delay performed electronically that you may not be able to do physically. And such errors can effectively negate the benefits of the signal alignment.

And ALL acoustic energy (within our simple example here and without introducing other variables! ...and assuming that our listening position is within the 3Space polar coverage pattern of the drivers.) decreases in intensity by virtue of the inverse square rule relative to the distance....lows, mids and highs. And given our initial conditions, certain frequencies are not behaving differently than others in this regard.

[stereohermit]

e analog crossover Islander was asking about dosent have time delay, only edge band phase adjustment...which would not be the same as signal delay or physical delay.

[mas]

OK - this discussion has taken a very strange tack...[:S]

"tambral shift" ???

and..."edge band delay"???

What does edge band delay, a facet of digital filter design have to do with an analog crossover?

I don't have any idea what it is you are talking about regarding the above that is relevant to the acoustical signal alignment of acoustic centers of origin - or even apparent acoustic centers of origin!.

Are we back at the beginning, addressing what can and what cannot provide for the alignment of a multiple signals in the time domain?

If you mean that a particular crossover lacks the components to provide signal delay other than what a few passive components can typically provide, then you are most likely dealing with a passive crossover and it is not sufficient to effectively serve as a delay line for use in signal alignment. A few high quality active crossovers (ie from Marchant) also lack signal delay capability features as well. As such, these crossovers lack the capability to provide signal delay of the type that has been heretofore referenced and discussed. They are insufficient to align the acoustic centers or origin of the various low and mid frequency drivers in a speaker such as a LaScala or a KHorn.

There is a fundamental difference between circuits that provide for only the cursory adjustment of phase using passive components (capacitors and inductors) and a circuit that is designed to provide for signal delay. Passive crossovers cannot provide for the type of signal delay to which several have referred. And analog or digital have no fundamental or necessary bearing on this capability.

But I think we are begging the question of whether a crossover can provide signal delay if it does not have the circuitry necessary to provide signal delay. And if it does not, then I will stick my neck out and state that you cannot align the signal in the time domain using electronic means.

[*-)]

But if an active crossover, be it analog or digital, has the componentry necessary to provide for the electronic delay of signals corresponding to a separation of many feet, then it can provide (within reason) for the alignment of the various passband signals in the time domain. (and we are talking within reason here!)

[]

[DrWho]

e analog crossover Islander was asking about

dosent have time delay, only edge band phase adjustment...which would

not be the same as signal delay or physical delay.

I don't recall Islander referring to a specific model number? Mas is talking about this unit:

http://www.behringer.com/DCX2496/index.cfm?lang=ENG

which does have time-alignment.

The tweeter falls more rapidly with increasing distance, relitive to

inverse square law. Only physical alignment will correct this...

Do you have any measurements that quantify this? Or...on what basis do you make this claim?

If your claim is true, then the relative tonal balance between the

speakers will change as a function of the relative listener position -

further back and you get less tweeter, and further forward and you get

more tweeter. That disparity can easily be adjusted for with the volume

knob.

Putting the mouth of a horn next to a surface like in your pictures

changes the effective horn flare-rate. So in addition to all of the

reflections off the top surface of the cabinet, you are also going to

change the frequency response of the direct sound of the horn. And the

frequency response change isn't going to be condusive to improving the

response - rather it will introduce a ripple very similar to that of

comb-filtering (only not as steep slopes which will be more audible).

In other words, the baffle is very much a part of the flare rate of the

speaker and is an intrinsic part of the design process. Change the

baffle, and you change the horn for the worse...

Seems like a rather large tradeoff in light of the small effects of

signal alignment. But if you say it sounds better, I almost have to

wonder what kind of room acoustics you've got going on. Perhaps the

nasty frequency response is perfectly aligning with the frequency

aberrations in your room? That wouldn't explain your sensitivity to the

time-domain though...

I think you'd find it interesting to take your solution outdoors and

compare directly to that of a proper active crossover alignment.

Speaking of which - I've often wondered if the process of

time-alignment in the home is futile in light of all the other early

reflections inside a typical room. Time-aligning a system in such an

environment isn't going to fix all the problems. However, in a

sound-reinforcement situation there aren't as many early reflections.

So perhaps there is more capability to fix the problem which would in

turn make the changes more audible. I don't think anyone would deny

that acoustical treatment vastly improves the sound. Misaligned drivers are functionally equivalent in how they alter the sound.

[stereohermit]

So sorry to confuse the issue. I thought Islander was inquiring about an active crossover that was non digital. Analog based active crossovers of course dont have signal delay capability, generally... Ashly XR series, Rane AC series, TDM, etc., but as BSS calls it Band Edge Phase Adjustment see: http://www.jesther.com/BSS%20FDS%20360%20manual.pdf, page 6. We are very much in agreement on the digital side of the crossover issue, including your methodology. The timbral shift I was mentioning, is just an unfortunate side effect of over 40" of time/distance between woofer and tweet in stock La Scala. Remember, when you are 3' in front of the speaker, you are almost 6 and a half feet from the actual woofer. Time-distance measurments I have made show lo freq level drop to distance to track driver as source, not horn mouth as source. If youve ever wondered why they sound so thin way up close, this is the reason,(among other factors).The radial mid was designed for no baffle just plenty of moretite.Scalas were never known to be good near field monitors...much of the reason is they were likely voiced further back.

[pauln]

We are discussing the effect of two sources at different distances
from the listener. When the many players in an orchestra spread out on
a stage or studio do they play in time to what they induvidually hear
or do they compensate (those furthest from the conductor leading their
notes, those closest leading less) so the conductor, michrophone, or
audience receives the sound of all the musicians at the same time? I
think the answer is the former and the latter couldn't be accomplished
even if desired unless all the musicians wore headphones each
differentially adjusted from back to front with decreasing amounts of
compensatory signal delay. This is not a difference of inches, it's
more like 50 feet from the back row of the orchestra to the front.

Is time alignment important to live music? Does reproducing live music need to pay attention to time alignment?

[bodcaw boy]

Point well taken, however I find with the Klipsch, there is a problem with tambral shift with distance, since, stock, you are differing distances from the three sources. The tweeter falls more rapidly with increasing distance, relitive to inverse square law. Only physical alignment will correct this...

so are you saying that not only did time align but you also decreased the level of tweeter and mid by moving them back?

have a blessed day,

roy