A time-aligned top end - Part II

[Rudy81]

Rudy,

" Is my understanding correct that a round horn would be optimum?"

Actually no! A round horn is only optimum for reflections off the mouth and probably diffraction too. BUT: It has no dispersion control. It has the same directivity pattern horizontally and vertically. You would like the horizontal dispersion to be wider than the vertical. I think the Elliptic is the best compromise. Lee Clinton said he would be doing both horizontal and vertical polar plots on the first Eliptrac 400 prototype and HF200 driver next week. That should illustrate the point.

Al K.

Well, I'm really glad I asked then. Sounds like every option is some sort of compromise. In that case, no use in pursuing a round one when an eliptrac will do.

[Al Klappenberger]

"Sounds like every option is some sort of compromise."

This is the very root of all good engineering. It's picking the best balance of compromises.

AL K.

[DrWho]

Does the fact that, except for the tweeter, the rest of the drivers being within 1/4 wavelength of each other and close to the throat of the horn--have any bearing on the throat delay/bounce comment?

Well the drivers being close is what allows the steady state response to look good. A steady square wave just consists of several steady sine waves...being able to make a square wave doesn't guarantee phase coherancy. In a linear time-invariant system it does, but this scenario isn't time invariant. You'll have to look at the first and last cycle of the square wave tone to see the effect of the reflections...

I'm still not sure about the impedance mismatch issue (I think that I understand, but I'm not sure -- it would be nice to understand the concept of the dynamics that you're envisioning).

The offset horn model assumes that pressure is magically created equally across the cross-section at the offset distance. When you load the horn on the side, the cross-sectional area doesn't get loaded equally: more pressure near the throat, and then the particle velocity isn't all parallel to the intended wave direction...so you end up relying on all of the reflections to be similar in phase as the pressure is built up, but there's a small delay to it. Basically though, you don't have as much control over the air.

Clearly, FM distortion is the trade-off or downside of co-axially aligned or full-range driver systems that must be offset by other performance improvements gained by using those design approaches. Using horn-loaded drivers minimize the needed displacements of the drivers' diaphragms, thus minimizing FM distortion (i.e., this is PWK's argument--in fact he said it would decrease FM distortion by 25 dB over a direct radiator's FM distortion, but you already know this argument).

So at the end of the day we need to decide which set of tradeoffs we want to go with. I think designs should be measured against the target tradeoffs, but I suppose discussions about the relative weighting of tradeoffs is valuable too. Danley is pretty solid at busting out his goals.

[djk]

Horn loaded above 100hz:

http://www.theaudioeagle.com/columns/column03.html

[Rudy81]

Rudy,

" Is my understanding correct that a round horn would be optimum?"

Actually no! A round horn is only optimum for reflections off the mouth and probably diffraction too. BUT: It has no dispersion control. It has the same directivity pattern horizontally and vertically. You would like the horizontal dispersion to be wider than the vertical. I think the Elliptic is the best compromise. Lee Clinton said he would be doing both horizontal and vertical polar plots on the first Eliptrac 400 prototype and HF200 driver next week. That should illustrate the point.

Al K.

Al, why are we concerned on the directivity pattern? Looking at it from the perspective of an orchestra, the only 'control' on the sound is the room itself. Is it done to attempt and control room reflections in a small environment, ceiling and floor?

I can go with either a round horn or the elliptical horn, it really doesn't matter in terms of WAF or speaker positioning. I would like to strive for the best sound. My room has quite a bit of acoustic treatment and I use Audyssey Pro to control room modes.

In the small space of an average room, I would think the lateral dispersion of the horn dimensions you mentioned above would not present problems in imaging.

[Al Klappenberger]

Rudy.

Now you are getting into an area I know very little about. The "common wisdom" seems to imply that wide horizontal dispersion is good. I really don't know why and "common wisdom" is often wrong!

Al k.

[Al Klappenberger]

Well folks .. This is embarrassing, but I just had to try out my hair-brained acoustic tweeter delay line pipe idea! I made one out of layers of particle board squares with a 1 inch hole in each. The total length is 9 3/4 inches. I stuck it between a B&C DE10 driver and ME10 horn. I measured the frequency response with and without the delay pipe keeping the driver 1 Meter from the mike. Now I know why nobody has mentioned this idea before. It louses up the high frequency response. Oh well, it was a fun project!

Al K.

[Rudy81]

Well folks .. This is embarrassing, but I just had to try out my hair-brained acoustic tweeter delay line pipe idea! I made one out of layers of particle board squares with a 1 inch hole in each. The total length is 9 3/4 inches. I stuck it between a B&C DE10 driver and ME10 horn. I measured the frequency response with and without the delay pipe keeping the driver 1 Meter from the mike. Now I know why nobody has mentioned this idea before. It louses up the high frequency response. Oh well, it was a fun project!

Al K.

Al, nothing ventured, nothing gained.....at least now we all know.

On the round tractrix question, I am very intrigued to try some out. I will see if Dave has the time to build a DIY kit that I use to slap my 2" P. Audio driver on and see how it sounds. I'm reading up on round tractrix horns right now. It seems the roundover at the end of the horn is very improtant to the sound it produces.....learning something new every minute.

[Al Klappenberger]

Rudy,

On the round tractrix, GO FOR IT! They have a good reputation.

Al K.

[Rudy81]

Rudy,

On the round tractrix, GO FOR IT! They have a good reputation.

Al K.

We'll see if I can talk Dave into it. It would be interesting to compare a round Tractrix with my current CD horns and/or the Eliptrac horn. I am surprised at how few places sell round Tractrix horns.

[Chris A]

I measured the frequency response with and without the delay pipe keeping the driver 1 Meter from the mike. Now I know why nobody has mentioned this idea before. It louses up the high frequency response. Oh well, it was a fun project!

Actually, what you have generated is data, and that isn't hair-brained.

One reason why I introduced the idea of a multiple-entry horn design is that it avoids the problem that you've now got data on (i.e., the waveguide-in-series design). However, I believe that what you will find is that, if a multiple-entry design is used, it will need to have built-in delay in the driver crossovers to compensate for the issues mentioned above. I believe that passive delays can be designed for the delay magnitudes discussed (...you tell me if that is possible..). One reason why I believe that is can be done it that Danley has stated that he is using delays on both active and passive crossover designs right now.

Chris

[DrWho]

A round horn is only optimum for reflections off the mouth and probably diffraction too.

All the reading I've done indicates that the mouth reflection is worst for cylindrical horns....but if you build a full-size tractrix, then the mouth reflection is a non-issue.

[DrWho]

Al, why are we concerned on the directivity pattern? Looking at it from the perspective of an orchestra, the only 'control' on the sound is the room itself. Is it done to attempt and control room reflections in a small environment, ceiling and floor?

You can't really compare against an orchestra because the stage they're on is technically part of the instrument making the sound we're referring to when we say "orchestra". In other words, when the orchestra is recorded, the ambience of the environment is captured in the recording. We're not trying to create that ambient effect in our listening rooms.

Once the directivity of the speaker is tight enough to get rid of early reflections, then the next step is to maximize the length of the initial time delay (amount of time between the direct sound arrival and the next reflection that arrives). After that, you want to maxmize the density of the semi-reveberant decay. You will always have some amount of decay and the size of the room will dictate the longest it can be. To make it shorter, you'll need to add absorption in the room, but then absorption can rob reflections that might arrive later in the decay portion...so the general premise is that absorption reduces the density of the decay, and the wider your polars, the more energy you have available to make it as dense as possible.

Also, the indirect sound constitutes a much larger portion of the acoustic energy that arrives at our ears. I think PWK calculated something on the order of 75% or higher. The tonal balance of the indirect sound will be a function of the speaker's off-axis frequency response and the frequency dependant behavior of the reflection paths in the room. It's important that the indirect sound mates well tonally with the direct sound, and that the decay of the indirect sound doesn't create distracting artifacts to the imaging. Absorption tends to have a low-pass nature since the higher frequencies have shorter wavelengths, and therefor see more attenuation (unless the frequencies are high enough to make the absorption material reflective again).

You should have all the tools you need to measure the ETC, so you could compare against what is shown on page 2 in this article:
http://www.rpginc.com/news/library/HT_AcD.pdf
The 'before' they present is incredibly horrible, and then the 'after' they present is pretty much run of the mill for a good sounding room.

[DrWho]

I believe that passive delays can be designed for the delay magnitudes discussed

I believe you're referring to an "all-pass filter" or "lattice phase equalizer".

http://en.wikipedia.org/wiki/All-pass_filter
http://en.wikipedia.org/wiki/Lattice_phase_equaliser

[djk]

"I believe that passive delays can be designed for the delay magnitudes
discussed"

For the woofer (to line up with the mid horn), but not a tweeter on the baffle.

[Al Klappenberger]

The "T" and Lattice circuits above are 1st order group delay equalizers. They are NOT time delay equalizers. They will not function to equalize time in a crossover. Group delay is a measure of phase linearity, not TIME DELAY! The two are equal ONLY of the group delay is perfectly flat over the frequency range of interest. All-pass networks are NOT!

Here is an example if a "T" section equalizer designed to produce 1 mSec delay at 20 Hz. Note that it generates that group delay (not time delay) at only a single point. The group delay increases rapidly as you go down in frequency. To generate a flat equal-ripple group delay requires many SECOND order sections in series. This generates major component losses. As you go up to tweeter or mid-range frequency the amount of delay possible becomes lower requiring even more sections.

Note that the group delay at higher frequency drops to zero quickly.

The inductor required must have extremely tight coupling and be center tapped. It should also be bi-filer wound.

Have done may equalizers like this at high frequency I have found the usual requirement is a single 1st order equalizer low in frequency followed by two or more 2nd order sections to equalize a lowpass filter. Narrower frequency ranges require only 2nd order networks that require 3 caps and two inductors for high delay. These display a sharp peak dropping off on either side.

FORGET THIS!

Al K.

[Al Klappenberger]

Here is what is required to generate 1/4 of a millisecond group delay from 6 to 15 Khz for a tweeter.

4 2nd order sections requiring 8 inductor and 12 capacitors.

Notice the increasing loss as frequency goes up. This is assuming an inductor "Q" of about 40. That means Litz wire,

Al k.

[DrWho]

The "T" and Lattice circuits above are 1st order group delay equalizers. They are NOT time delay equalizers. They will not function to equalize time in a crossover. Group delay is a measure of phase linearity, not TIME DELAY! The two are equal ONLY of the group delay is perfectly flat over the frequency range of interest. All-pass networks are NOT!

I thought you only cared if the delays were lined up at the single xover frequency?

[Chris A]

I wonder what Danley is doing to compensate for delay in his passive crossovers? See this reference:

http://www.audioasylum.com/forums/hug/messages/15/153707.html

Chris

[djk]

"They are NOT time delay equalizers."

Beg to differ.

They work fine for audio. B&W uses them (the unbalanced with a bi-filar inductor) in many of their speakers

You must not have read the part about:

"but not a tweeter on the baffle."

I have built many speakers with the balanced delay circuit on the woofer to delay the signal at the point where it crosses to the deeper mid horn, it works, and it improves the sound.

It is not practical to try this for the tweeter, that is why I said "but not a tweeter on the baffle."

"I thought you only cared if the delays were lined up at the single xover frequency?"

Correct, I'm glad you understand this.

"I wonder what Danley is doing to compensate for delay in his passive crossovers? "

The low-pass filters add delay to their drivers, but they are mechanically ahead of the HF, so it works out OK. His (Unity and Synergy) speakers can pass a reasonable squarewave from about 200hz~2Khz. The leading edge of the square wave is odd harmonics out to infinity (in practice about the 29th and it doesn't look and better), so the HF driver bandwidth is the limiting factor at the 2Khz end.

Danley:

"The mid drivers are forward of the compression driver, by an amount
which is off set by the phase shift difference between the mid and high
sections, they are automatically "time correct".

I can say that when everything is "right" they are the best speaker I
have ever heard, no trace of horn sound, effortless dynamics, a huge
reduction in room interaction due to the directivity AND lack of
interference with multiple ranges and drivers. Imagine a single, really
really clean and dynamic horn that covers the most of the audible range
from a single exit and doesn't sound like a horn, it doesn't sound
"like" anything.."

http://www.cowanaudio.com/images/unity28.jpg

"I'm currently working on a 2 way 60x90 degree unity horn for the workshop. It uses two B&C 8NDL51 mid drivers and a BMS 4550 compression driver. The horn will cover 100Hz on up. Woodwork is complete on a pair, I just need to spend the time doing the crossovers in CALSOD. A few other projects are keeping me away from this low priority though very interesting speaker system."

The above looks like it would fit on the top of the Klipschorn, but four cheaper 5s could be used similar to his Lambda Unity system.

Similar idea: