Time Alignmnent in Cornwalls

[TBrennan]

Here is a photo of a Western Electric speaker sysyem such as that used at MGM during the "double tap" incident. Note the path differences between the drivers of the wide-range "snail" horns, the Bostwick horn tweeters and the direct-radiating subwoofers.

Those of us who've been to the museum in Hope have seen the WE "snail" horns there.

[fini]

Tom,

I believe Mace does understand. In your example, you've reproduced an original sound through two devices. Think of the tweeter being the violin in the front row of the orchestra, and the basshorn driver being the snare in the rear. Maybe it would be clearer to think of two drummers in these positions, playing the exact note, at exactly the same time. A listener halfway between the two will hear the beats at the same time. Not so for the conductor, the audience, or the two players. Same reason it's so hard to stay in unison, singing the National Anthem at a ball game (pitchers and catchers are at Spring Training right now, BTW!!!). We could be talking about light from stars, for that matter.

fini

[Colin]

Very good points, Gil, thanks for sorting out the snakes and lining them all up too.

Many recordings over use the echo effect. My favorite pet peeve is the echo many female singers use. Adding echo may improve the sound of a singer, but if it is not applied to the accompanying instruments, it creates a cognitive dissonance. Sure, it sounds nice, but where is the singer standing? In a concert hall? If so, then where are the other instruments, why dont they ring too, arent they there also, or are they off in a studio somewhere while the singer is alone in the large empty hall?

ALK may right, the loudspeakers which I heard had time aligned drivers, in backwards leaning cabinets, and they did exhibit a cohesion which made them less colored or obvious. But that is an illogical syllogism. I do not KNOW that their cohesiveness was due to their time alignment or some other variable. In fact, the sound from all but the deepest of subwoofers arrives at such a high cycle per second speed, that nobody sitting near a loudspeaker should determine a difference in arrival and alignment times.

Is time alignment an obvious problem? Consider this: if two drivers produce the entire waveform of a note and they were 12 feet apart do you think that we would hear a delay so large that it sounds like double taps? I think yes. Now divide the lower half of the signal to one driver and the other, upper half of the signal to the second driver. Think you would hear a disruption? I think yes again.

The important thing to remember here is that every note created has harmonics which descend and ascend over the frequency range. A 40-Hz bass note doesnt just have a first harmonic at the first node, it has multiple harmonics descending lower and higher. The waveform of a low note ascends well up to the mid-range, and beyond. Even when divided over two drivers, the note is still created by both of them. The mid-range woofer alone doesnt reproduce a bass note, the mid-range tweeter does too.

How noticeable or measurable this is in the real world, I do not know. I see all of the snakes Gil lined up, but I am not familiar with many of them. My point is that the time delay in the critical mid-range region may have to be extremely slow to be noticeable (3ms is very slow), but if it is, it should affect the sound because both lower and upper drivers reproduce the notes.

[Clipped and Shorn]

Time delay can be adjusted with some electronic crossovers (like the Rane AC23) a component often used in concert (and club) situations where multiamplification is employed in the PA system. The literature which comes with this unit describes the problem of drivers being at different distances and how to compensate for this.

-C&S

[djk]

Around 1978 there was a company called UlraPhase, Don McDonald was the CEO.

They made a line of time aligned speakers. Good drivers, good cabinets, good crossovers, good design, etc.

They had a showroom display that was quite interesting. It was an actual crossover plus the midrange and tweeter out of one of their speakers, all wired up. The point was to play this at a reasonable volume level and move the midrange and tweeter backwards and forwards in your hands. There was a point where the sound just came into 'focus'. The range of motion to hear the sound pop in and out of focus was on the order of ± 1/2".

The crossover frequency here was about 3.5Khz. As the frequency goes down more delay can lived with (the wavelengths are longer so the number of degrees of phase rotation remain relativly constant).

The human ear is most sensitive around 3.5Khz too. As the frequency goes up from here the ear seems to be less sensitive to time off-set.

The UltraPhase demo was so powerful that they were hard to sell against unless you had something time aligned like the big B&W or the Dahlquist DQ10.

[Randy Bey]

Dennis,

could that be phase as much as time alignment?

I noticed a substantial improvement going from a modified AA network to a simple A, and I (in my ignorance) attribute that to the phase relationship being similar for the squawker and tweeter. Every capacitor moves the signal 90 degrees out of phase, and every inductor the converse, so by my pea brain estimates my squawker and tweeter are in phase.

I have heard time aligned conventional speakers (Vandersteins) and was not terribly impressed.

[sunnysal]

forgive my total ignorance of this issue but I still have a question. I would imagine that phase would have as much of an audible effect as delay. the only problem begins in that phase relationship between drivers as dictated by the cross-over varies by frequency does it not? or does it remain the same throughout the audio band? tony

[mace]

tony...lesssee... This can be much more easily explainded with a graphic. But, for example, lets say that driver A is playing one tone, say 40 Hz. Driver B is varying in frequency (pitch) from 1000-2000 Hz. As Driver B changes tones you'll hear "beats" in the resulting combined pitch which will increase and decrease in frequency. This is because the "peaks" of driver's B tones will match up with and counteract "peaks" in driver's A tones. The interference is changing because the difference (Hz of A minus Hz of is constantly changing as driver's B tone is changing.

To answer your question, over the freuency spectrum, the different frequencies of the different drivers will go in and out of phase, depending on the differences in frequencies. Crossovers can't help this. However, some crossovers are designed to "reverse" the phase of one driver relative to another. This difference would be most noticable if both drivers were in the same plane playing the same tone, then the net tone would have about zero dB. However, a tweeter and a squaker wired out of phase will sound "different" but the exact amount of differentness will depend on the frequencies going through each.

man, this turned into a long post...

Fini: thanks for understanding. Although, I hadn't really thought about intra-instrument time delays going through a 3 way speaker, like the tap dancer example. However, my point (or question) still stands...can anyone *really* hear a delay of 0.87 ms??? The 12 ms is more than an order of magnitude (almost 2) longer than 0.87, i.e. a lot more delay.

I need to find a recording with rim shots to see if I can hear this delay.

Al klappenberger: What is the difference between time delay and group delay?

McDermott: informative post.

Interesting thread,

mace

[Chris A]

There was a point where the sound just came into 'focus'. The range of motion to hear the sound pop in and out of focus was on the order of ± 1/2".

Sorry to have to open up this thread from so many years ago, but this is something that has my attention. I thought that ± 6" was "good enough", but this was based on very little.

Thanks to djk, once again, for this info. This means to me that time alignment to within 0.04 milliseconds is important at 3.5 KHz. Wow...Now I understand why crossover filters in this region are so difficult to do well, and 2-way speakers sound so good, IMHO.

Chris

[JL Sargent]

"and 2-way speakers sound so good,"

That's especially true if you "live in the midrange".

[Chris A]

Yes, and if you think about this a little bit, it's now clearer to me why some of those European folks like those big-ol' Tannoy "Dual Concentric" loudspeakers in their Prestige line (e.g., Westminster Royal SE). These models have a very large delay on the horn-loaded backwave of this design, but the woofer is directly coupled to the front of the cabinet, so you do get much lower time delay from this vs. a Khorn bass bin design.

An implication of the discussion above is that probably a lot of Khorn owner's dissatisfaction with their speakers could be remedied if they went to an active crossover and tri-amp their speaker, thus taking out the delays. I believe that Rudy81 is already triamping his using a EV DC-One crossover (a 2-in, 6 out design). Since the delays on the Khorn design were published by Richard Heyser in the mid-80s (~8.4 ms on the tweeter, ~6.7 ms on the midrange, which are really huge compared to a direct radiator system), then the corrections for these path length differences could be made easily- just dial them in. Active crossovers do not drift over time like passive crossovers, and more importantly they do not drift when pushed hard due to heating--they remain linear in their output.

What's interesting to me is that virtually all HT preamp/processors correct for path length differences and most people take advantage of this feature when they set up their systems. But somehow the same problem when applied to their most important speakers (the fronts) is ignored between the low/mid and hf drivers. Go figure.

Chris

[djk]

All food tastes the same(if you have a zinc deficency), all women kiss the
same(if you're a eunuch), VHS and Beta look the same(to Ray Charles and
Stevie Wonder), and all amplifers sound the same(to the deaf old men at
Stereo Review).

WHO?

Tell the deaf/dumb/blind kid to go
play pinball.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Time alignment

Some people can hear it, and some can't.

On a speaker like the Cornwall it can be very obvious on certain program material, repetitive percussive click sounds being the easiest (drum-stick clicks, snare rim taps, triangle strikes, etc). On time-coherent speakers these click sounds are coherent, you only hear one click sound. On the Cornwall you can hear three distinct arrival times, tweeter first, then woofer, and lastly the midrange.

With passive correction only the woofer may be delayed so its arrival time is the same as the midrange, the tweeter is too far ahead to correct.

The circuit used is called an all-pass delay network, and is installed on the woofer.

The balanced type is easiest to use, the un-balanced requires a bi-filar inductor.

On a double 12 design (with an EV 8HD + Atlas mid, and a tweeter), adding the all-pass to the woofer made a very audible difference in how the midrange sounded. Unfortunatey, I see no way to fix the tweeter problem without tri-amping and a DSP crossover.

Should you attempt to modify your Klipsch speaker this way?

Probably not, it really requires a lot more knowledge that what I have outlined here, and test equipment.

[Chris A]

I see no way to fix the tweeter problem without tri-amping and a DSP crossover.

Okay, I'm game--do you think this is a bad approach? Are you "not into digital anything", or have you heard an audible difference? If so, could you relate the story/equipment?

Chris

[djk]

Most people are not willing to make the investment in the extra amplifiers and the DSP that would be required to correct the tweeter off-set, not to mention the extra cabling and input terminals required.

Another consideration is the audio quality of the DSP devices available. A Behringer DCX 2496 has about 15 cheap opamps and 17 polarized electrolytics in the signal path, all surface mount and hard to find quality parts. Not the last word in transparency.

Going passive require a complete re-design of the network, and for the specific group of components in the specific Cornwall, one size does not fit all.

[Chris A]

Thanks. BTW, I concur.

Chris

[Rudy81]

Chris, in fact I am now tri-amping, active filtering and time aligning my Khorns on a full time basis. I recently put my ALK universals up for sale which finalized my commitment to the active technology. As you mentioned, I originally used the EV DC-ONE, but had some difficulty with a high noise floor. Since that was my first shot at working with an active filter, there is a chance I did not properly set the gain structure and did not give the DC-ONE a fair shake. I since went to the much maligned DCX-2496 and could not be happier. I would like to try the EV once again, but have yet to find one at a reasonable price.

Right now my system sounds as good as I have ever heard. I believe much of the improvement is the benefit derived from having the amps directly control the drivers. The sound is crisp, precise and just a pleasure to listen to. Mind you, it wasn't bad before, but each little improvement helps. Using either crossover I was able to quickly A/B the driver alignment and there is an audible difference when you compare the modes. After quite a bit of testing, I preferred the driver aligned setting. It is not a huge difference, but it is noticeable.

I will say that getting to this point was a long and sometimes frustrating journey. As a novice to this technology, I had a steep learning curve. In all honesty I fully expected to find little difference between passive and active filtering with driver alignment. That was not the case. I now have an expensive Parasound amp sitting idle which previously was the lone stereo amp for the Khorns. I now use two Hafler amps for the tweeters and mid range. A crown K1 powers the bass bin.

I have looked into the 'upgrades' to the DCX crossover, but those are generally way more expensive than the crossover itself. It also got me wondering about the supposed poor quality of the op amps etc. How do they compare with what folks have in their source units, pre-amps, processors and/or amplifiers? I suspect the components in the DCX are no worse than many of the components found in most consumer level gear affordable by the average listener. This doesn't even address the quality of the room acoustics in a given listening room. So, I look at the complaints with a little skepticism.

[Chris A]

As you mentioned, I originally used the EV DC-ONE, but had some difficulty with a high noise floor.

Do you think that it could have been some defect in the DC-One that you had?

In all honesty I fully expected to find little difference between passive and active filtering with driver alignment. That was not the case.

Interesting comment.

I now have an expensive Parasound amp sitting idle which previously was the lone stereo amp for the Khorns.

Have you thought about Audiogon?

I have looked into the 'upgrades' to the DCX crossover, but those are generally way more expensive than the crossover itself.

I have not actually heard the DCX2496 in action. I have another Behringer product, an Ultracurve Pro DEQ2496, in storage that I use as an Audyssey-like device to understand room acoustics better (it has an auto-EQ mode that continuously measures with pink noise input until you get tired of it trying to find an optimal EQ). I used it to help me understand the settings that I was using when I had the Crown XTi's biamping the Jubs. I learned a lot from those experiments. As you know, I presently use the EV Dx38 with Roy's anechoic measurements for the TAD-4002s/K402 horns and Jub bass bins, and I'm really happy with those settings.

[Rudy81]

Chris, I don't know if anything was wrong with the DC-ONE. At the time, I was working with consumer amps which caused all sorts of gain issues as well as interconnect issues. The DC-ONE outputs are XLR and one of my consumer amps had both XLR and RCA while a third Parasound amp had only RCA inputs. That led to unbalanced connections on four channels and balanced on two. This ended up causing noise problems as well and I could not attenuate amp sensitivity on the multi-channel amp.

Shortly after getting rid of the DC-ONE I decided to do this the right way and get three pro amps that were properly matched to the DCX output voltage. I truly doubt there was any problem with the DC-ONE. It was sold to another forum member and he has not mentioned any problems. So, the issues were likely due to my inexperience and the lack of proper pro amplification. I still look around for one since I would like to try it again. The DC-ONE has a much nicer GUI than the DCX. Otherwise, they are rather similar in what they do. The DCX does have a few added features, like the ability to use a mic to auto align the drivers as well as a few added crossover orders.

I did use the auto align feature together with some excel spreadsheet calculators to come up with a particular crossover setup that is really amazing. I discuss that process in the Active Crossovers thread I started in the 2 channel forum. In fact, this setup is what I am still using.

I started the active setup as an experiment to see if I could use the system to test a Jubilee 'like' bass bin, or a Jamboree with my top section. I have yet to build either of these, but this was the first step before I could get into that area. Hence my comment that I anticipated going back to the passive ALK setup when I finished the experiment. However, several members jumped in to help me with the setup and after several months of learning and tweaking, I ended up loving the sound so much I put my ALK's up for sale and have gone way over to the 'dark side'.

[Don Richard]

Is time alignment an obvious problem? Consider this: if two drivers produce the entire waveform of a note and they were 12’ feet apart – do you think that we would hear a delay so large that it sounds like double taps? I think yes. Now divide the lower half of the signal to one driver and the other, upper half of the signal to the second driver. Think you would hear a disruption? I think yes again.

Time alignment in loudspeakers is not so much trying to eliminate any "double tap" problem. When two (or more) transducers in a single cabinet or in multiple cabinet arrays reproduce the same frequency at the same amplitude but are separate in time arrival several things happen:

1) When the sound waves are not in phase, accurate addition of those waves is not possible. For example, a phase shift of 90 degrees between two waves equal in frequency and amplitude will not result in double the amplitude but rather 1.4 times the amplitude.

2) Separate arrival times of the same sound will cause the lobe null angles to shift, which shifts the position of the dominant lobe and causes side lobing.

3) Separate arrival times of the same sound will cause comb filtering.

Time alignment is exact for one position in space, but is made closer at any point within the coverage area of the system. Vertically stacking the elements of a loudspeaker system makes the time alignment very close or exact horizontally.

[fini]

Don, you forgot

4) Your head might blow up.