High Efficiency Speaker Asylum Need speakers that can rock with just one watt? You found da place.

---

[ Follow Ups ] Thread: [ Display  All  Email ] [ High Efficiency Speaker Asylum ]

---

[ Alert Moderator ]

---

Re: Horn stuff and a post back for Tom

Posted by Wayne Parham (M) on July 8, 2002 at 02:21:20

In Reply to: Re: Horn stuff and a post back for Tom posted by tomservo on July 7, 2002 at 21:07:05:

Hi Tom!

You wrote:

>> The upper and lower drivers DO NOT operate simultaneously except
>> in the crossover region.

From this discussion, I can see that you put a lot of effort into covering as many things as possible to bring the separate sound sources into alignment. But what you have said here is key. The crossover region is a critical region, because it is here that the system is most vulnerable to anomalies. There are more diaphragms in play.

>> By first measuring the inherent fixed delay of the upper and lower
>> drivers, a physical off set can be found which compensates the FIXED
>> delay time of the driver.

The delay of a motor is fixed at a fixed frequency, but not over its entire frequency range. It is also different depending on how it is loaded. Delay is a function of a filter, and the entire system creates a filter. I assume you find the delay curve in your particular application, yes? And motor QC is tightly controlled, yes?

Also, please define what you are calling "upper" and "lower" drivers in this system.

>> That fixed delay is EXACTLY the same as if the driver were moved
>> back in time the same amount.

At what frequency? At resonance of the driver, or midband? Because the delay shifts quite a bit through the mechanical resonance region.

>> Since the delays of all the drivers can be measured, the delays
>> can be compensated by an inverse physical distance.

Assuming the midband delay is fairly constant, and assuming you only use the drivers through regions where this is true.

>> At the same time one must have that physical distance correspond
>> to the frequency range within the horn that the drivers are
>> intended to drive so getting all of it tied in took some doing.

That's a problem. As frequency changes, the critical distance does too.

>> Also, that physical distance can be used to compensate for the
>> fact that high pass signals emerge from the crossover ahead of
>> the low pass signal.

High pass signals are shifted forward, that's certain. And the phase is pretty easy to find. But the problem is that the shift is a moving target that grows smaller as frequency goes up. So you have a range of distrances, not a fixed distance. That makes any "compensation" a generalization or approximation. That's the "best fit" a person can do here.

>> Since one knows the compression driver has the lowest delay and is
>> ahead of the lf by so many degrees, it is logical to put it to the
>> rear of the lower section.

Agreed. But any fixed position where it is placed will make it perfectly phased at exactly one frequecy. As frequecy goes up, its apparent position will shift one direction and as frequency goes down, it will shift the other direction.

>> At crossover, when the spacing and phase are correct, the
>> compression driver sound arrives in synchrony with the lower
>> system and as one goes to either side of crossover, there is
>> little or no change in phase or amplitude.

You are defining the crossover point as the median, and where you align phase between two subsystems. That's the way most time alignment is done, and is something that defines a set frequency - the crossover point, usually - and a set position - several feet in front, perfectly on-axis, usually. The idea then is that at this one frequency, the parallax is zero and the system is perfectly phased on this axis. This then too is the best compromise, and time mis-alignment is moved away from the crossover point, on either side.

>> Remember it is the response slopes of the horn sections that are
>> what add, who's phase and time matters, not what ever electrical
>> filter slope it took to make it work.

Yes. But there are two things that make slope significant:

1. Slope indicates overlap region, and
2. Slope also indicates number of nodes which is in turn derives speed of phase change (i.e. degrees per octave)

>> Put it this way, if you had 3 different lf point sources and you
>> could put each one at exactly the same distance and then measured
>> them and found that they each had a separate delay, wouldn't this
>> mean that the sound from an impulse would arrive at different times
>> from each?

There would be three separate signals that would sum. The signal measured (or heard) would be different at every position in 3D space. That's the key. With the scenario you just described, there is no possible time alignment in all positions at all frequencies.

I know that you are limiting the number of positions by using a horn. And you are limiting the number of frequencies by using a crossover. This very topic has been the largest driving force in loudspeaker development, particularly horns. I know that you're aware how many of your competition (and mine) have done work to reduce frequency anomalies caused by interactions of multiple sound sources. And I know that's something near and dear to your heart as well. I suspect your design deals with this situation adequately. But to actually call it a point source, that's like announcing to Heisenburg that you had found a way to measure an electron's position and momentum simultaneously.

>> Knowing how much each delayed the signal, now can one account for
>> that fixed delay by an adjustment of the physical position.

Knowing how much each delayed and being able to correct for it is entirely a different matter. Each sound source is a radiating diaphragm in 3D space. Each is driven by a signal that moves in time with respect to frequency. And three subsystems are each doing this in different directions. They are also at different initial positions. I would say you can determine the phase, and you can limit the anomalies in the frequency domain caused by the movement in the time domain. But to limit the anomalies and to cause the system to act as a point source are two very different things.

>> Put the fastest speaker at the end of a closed pipe and then had
>> the second fastest in the middle and then the slowest one at the
>> open end. Now a signal fed to the system is radiated first by the
>> fastest speaker and then as the wave travels down the pipe, it is
>> reinforced by the second speaker and then finally by the slowest
>> speaker.

I knew you had that analogy running around in your mind!

During one of our dialogs, I was thinking about your approach as I was getting ready to brush my teeth. (Bear with me now, this isn't really as stupid as it sounds - you probably already know where I'm going with it.) As I squeezed the tube of toothpaste, I recalled the Hunt for Red October and its "hydrodynamic worm drive." I realized that what you were attempting to do with this system, was to "squeeze" the pressure first at the bottom, then the middle and finally at the exit of the device.

It is really a cool mental image; It's also why you often say "think in terms of pressures."

But the problem is that these aren't static pressures, they're zones of pressures. We're not so much pushing a pressure as we are jiggling a pressure; It's more like the synchronized movements of a centipede's legs, when moving at a constantly changing pace. No one distance defines the movement of each leg per second - the rate is changing. Likewise, since the phase changes as a function of frequency, the "squeeze" only works at a specific frequency. Beyond that in either direction, and we fall further and further out of sync. Now I know that your goal in this design was to minimize the amount it moves out of sync. But it isn't possible to keep them in sync over a large band of frequencies and for a large field of positions.

>> In the Unity, there is never a situation where more than two
>> ranges produce any single frequency and even then it is only
>> in a narrow crossover region where the time and phase right,
>> hence the crossover does not show up.

Of course. If what you're saying is what I think you're saying, most multi-way loudspeakers limit output to a single subsystem except at the overlap region, which is as narrow as possible. That is usually a primary design goal.

>> Again, many of our users have test equipment and measure speakers,
>> if anything people say our stuff "does what it says", ask around.

I'm sure it is good equipment. I'm also sure that no one is unhappy with their purchase.

>> What your not getting is that except for acoustic phase which is
>> only + - 90 degrees if minimum phase, that the positions of the
>> drivers do not change with frequency, hence the static part can be
>> removed by physical displacement. Also that the electrical filter
>> IS NOT the response and phase which matters, it is the acoustic
>> response and phase, where your focussing is a step removed from
>> where the combining takes place.

That's where the problem is, and it's where it has always been. One item in this matter is a fixed position or offset, and the other is a moving phase or offset. If this were not so, then a fixed position would work. But as frequency rises, wavelength size is reduced. It's a moving target in the overlap region. Parallax between the sound sources and the listener forms a distance offset. And changing phase sent to the motors forms another moving offset.

>> You couldn't be more right about difficult but I have partly by
>> trial and error, partly by model found a set of "rules" which work.
>> You are right about dimension issues and so on but again that is
>> what the design involves.

Honestly, I'm sure that it is a good design. After having this discussion with you, I am even more convinced that you have minimized the anomalous effects to a point where the system is working as best it can.

>> Remember it could have any number of divisions and the arrangement
>> allows that when any two ranges are interacting, that they are the
>> correct time and phase apart and that the region where they
>> interact be small enough to still be a point source so far as
>> loading the horn.

A point source is a point source. It is a physical imposibility, like a magnetic monopole. You can say your device approaches the behaviour of a point source, but then single driver speaker owners often say that too. It is really not - neither case - because the sound source is not a point.

Run a swept square wave through this thing. None of us passes that test, but a point source would have no trouble generating it.

[ about my statement, "For a 4 inch radiator to have a "2 inch ]
[ maximum path" to holes makes the assumption that the hole is ]
[ infinitelty small. There will actually be a window of distances ]
[ from the shortest path to the longest path. This is significant ]
[ to a person attempting to make a time aligned speaker. ]

>> Yes that would be true and is lower in frequency (where it isn't
>> an issue) but at the high end the holes and the trapped compression
>> volume are in play which makes the air mass in the hole driven by
>> net pressure, higher in frequency the air mass in the hole chokes
>> off the mobility and the hole gradually becomes more acoustically "solid"

What is the distance from the compression driver diaphragm to the midrange cluster? Please specify whether the dimension is given along the horn axis or along its wall.

>> And by horn loading, how do you say that would be in evidence?

The output should rise dramatically, i.e. more than simply directed energies caused by focusing the source. Horn loading is impedance matching, not just focused directionality. Taking all the energy and directing it in one direction will raise output in that one direction, but acoustic loading of the diaphragm will cause it to actually be more efficient. So the evidence of horn loading is a rise in efficiency, over what is generated by direct radiation of the diaphragm on a waveguide having the same dispersion as the horn under test.

>> The response of the compression driver (a B&C DE-25) on this horn
>> raw shows a broad peak at about 2.5 KHz, about a 6 dB/oct roll off
>> above that (the drivers REAL power response) to about 10 kHz where
>> the series inductance adds a second bend. Below about 2 KHz there
>> is a rolloff which reaches about 4DB/oct to about 1 kHz where it
>> reaches about 10 dB/oct roll off.

I'd say that this is the peak caused by the conical flare. It's roughly a 1Khz horn.

>> The 1 kHz slope reflects the low cutoff of the horn driven at the apex.

Agreed.

>> the 2.5KHz slope is the mass rolloff dictated power response.

Agreed again.

>> A TAD2001 on the same horn has this rolloff at about 4 - 4.2KHz fwiw.

OK.

>> The crossovers job is to flatten the response and trade off the
>> upper/lower frequencies.

A bypassed attenuator. I use similar compensation. I gues you've probably noticed a side benefit is that strange (relatively) linear phase shift it generates.

>> Since one can only attenuate with passive circuits most of the
>> time, it is the sensitivity of the driver at the high cutoff that
>> matters.

Yes. The unimproved performance of the compression driver. The horn isn't participating past a couple octaves above its flare rate - 1Khz. After that, your compresson driver is operating as a direct radiator. I run 'em this way too over a part of their bandwidth. I find it is a design solution that has merit, especially in the top octave.

>> The design starts with 0 attenuation at say 18 kHz and then "rolls
>> off" with decreasing frequency.

Again, our implementation of the compression driver is exactly the same.

>> What one is left with when the final acoustic response is measured
>> is different than the electrical filter because the drivers
>> acoustic response is added to the filter's.

Of course. The two response curves sum together. I describe this as subsystem "participation" in the overall response curve.

>> I don't claim acoustic impedance matching over two decades.

It's late, but I could have sworn that was one of your claims. No matter; Good enough. I'll tell you that I don't either - I run my horns with compensation exactly as you do. Some of my LF horns have motors with rising response, the horn's participation in the overall response curve sums flat, or relatively flat. On all of my systems that use a compression driver, I use a compensation circuit for tis purpose. It does two things - It has the bypassed attenuator that matches levels and removes attenuation in the top octave. It also acts to set the amount of peaking in the HF crossover, so that the first coupole of octaves are flat and then the amplitude response of the crossover starts to climb as a diagonal line.

>> Yeah more or less, the horn has to be mostly solid everywhere or
>> it isn't a horn (or is a lossy horn).

Hmm. Maybe we started off on the wrong foot. Your description of a lossy horn is very similar to my calling it a "mal-formed horn." I've seen a few other similarities in our approaches in this dialog.

Ahh, but I've still got to get you back for the steam in the campfire comment. [grin]

[your explaination of the anomaly measured in the Lambda Unity]
>> Nick made his horns more slick looking than ours, in doing so
>> filled in the corners more, nice and smooth, very nice looking.
>> Ours had very little smoothing in the corners, only where it was
>> needed near the compression driver. This difference changed the
>> distribution of energy by making the directivity increase a little
>> high up, so when he tried the crossover we came up with, it had a
>> plateau in the response. Nick went ahead and designed a crossover
>> and also sent me one of his deluxe horns to play with. When I had
>> time (I do have a job too) I used his horn and came up with another
>> crossover which I sent him. Nicks horn shape also made the directivity
>> perturbation somewhat more of an issue as well, some time later I
>> found that the reflection could be largely or entirely eliminated
>> by placing a pad of acoustic foam at the right point of the outer
>> mouth. Nicks measurements did not include that foam and I don't
>> recall which crossover (his or my first one) it was. I do remember
>> telling him that it looked like periodic reflections and to try it
>> outside.

I see. Lots of trial and error to get the phase angles that would cause anomalous behaviour outside the vectors defined by the boundary, i.e. the flare walls. That is what you have endeavored to do, and I always understood that. If nothing else, our discussion has illuminated the issue of the angles where anomalous behaviour is found.

But the problem is that the angles aren't really straight angles. They are caternary planes, and they are a shape that has a wider angle near the loudspeaker and a more narrow cabinet as you move away. It's easy to see why this is, since they are formed by a parallax distance.

In my crossover document, I find an offset similar to what you have in the Unity. But I make no pretension that it will create a point source, or even a system that acts like one. Instead, I say that it will limit anomalies in the frequency domain for points inside the caternary plane. It basically defines listening positions that are more uniform, similar to the positions you've tried to limit your dispersion to in your horn, to avoid the possibility of anomalous output. But as the listener moves further and further away from the device, the more parallax will grow and anomalies will appear.

So if you take measurements at greater distances, you'll find anomalies there. This being so doesn't make the speaker any less of an accomplishment, but it does prove that it doesn't act as a point source. No hard feelings; No speaker can be.

>> No Fluke, HP, TEF machine mostly :-)

(drum sounds) Two snares, splash!

[about the "LAB" horn]

>> With that design, as it is so small acoustically even a large pile
>> has little directivity. Not caring about that, I used a hyperbolic
>> flare.

That's Bruce Edgar's favorite for bass horns as well.

>> This kind of woofer is used for concerts, special effects Raves
>> etc, with our BT-7's only 16 were needed for the largest sound
>> systems used for bands like Garth Brooks, U-2, Michael Jackson
>> and so on.

Do you mean the woofer system or the woofer motor? I assume you mean the BT-7, which I also assume is a Servodrive bass subsystem cabinet, yes?

>> That is to say that those 16 horns (two groups of 8) produced
>> enough bass to keep up with 140-180 speaker box per side sound
>> system. The lab sub should be very similar.

But with an Eminence woofer, right? I like Eminence products, and they are more popular by far than my more expensive product offerings. But have you considered using any other motor for the LAB device? I am very aware that there is an intimate relationship between the motor, diaphragm and system. But there are some pretty good existing systems using significantly more expensive components. Do you really think it will be a "fair fight?" Are comparisons really warranted?

>> Should one not want the 200+ acoustic watts at 30 HZ per box,
>> I think then one could indeed build a full rage horn this way,
>> figure about 2-3 octaves range per.

All of my systems used to be three and four way systems, and I still make 'em that way when maximum output is required. For home use, I began to promote simpler two-way systems with compensation networks, for many of the same reasons that probably made you want to build the Unity.

But for all-out performance, I always find myself putting together four-way systems too. Turns out to be about 2.5 octaves per subsystem. A three-way must push each subsystem to roughly a decade, and that is taxing on at least one subsystem.

>> If you were one of the companies who sold big bass horns and saw
>> what this one looks like it was going to be, wouldn't you be scared?

Not if the performance of my device were up to it. Again, you said that you were planning to use a motor that is likely very good but that is also probably somewhat average in terms of distortion performance. There are several more expensive motors that will outperform it, and that are designed for horn loading. Since the biggest cost is the cabinet (labor), I think it is important not to skimp on the motor.

>> why don't you think I could push the envelope with an Eminence driver?

I like Eminence, and particularly for their price. I think performance is quite good. But I don't think their best motors rise to the level of a comparible JBL Professional motor.

Now price/performance - That's an entirely different matter. Give me just a couple hundred bucks and I can make a better sounding system using all-Eminence than I can with JBL. In fact, the two-hundred dollars worth of Eminence parts will have better extension than three times the money in JBL. But when we get to "nosebleed expensive" and "price is no object" - there are some solutions available from other sources that Eminence just can't touch. They are not in that market, and maybe with good reason.

>> What my ears tell me I want is to have everything above about 80 - 100 Hz
>> come from one spot, I have a large set of flares here but as it is
>> not an "official project"

That's the reasoning of the satelite and subwoofer crowd too.

In a sound production environment, it is not a matter of choice - You've got a lot of hardware to set up so subsystems are physically distant. Installation - done well - is a more difficult job than building the speakers, in my opinion.

But in the home, I would prefer to keep the components together as much as possible, and it is usually much easier to do since the sizes aren't as big.

The market position of consumer audio makes it very apparent that my preference is in the mionority. [grin]

>> I worked on a project doing something like that for Gary Kendall
>> at Northwestern U. and then helped Gene Pitts write an article
>> about for Audio a few years ago. I used a focussing electrostatic
>> transducer, was able to get ~155dB at the focal point at 75 KHz.

That's an idea that may come together someday. I'm impressed with it.

>> The thing worked but gave people headaches and used a lot of power.

At 155dB, I'd guess that it probably did.

Thanks for your time, and for continuing this discussion with me. It's been interesting.

Wayne Parham

This post is made possible by the generous support of people like you and our sponsors:

Parts Connexion

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2025, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---