[Kevin Warne]

Nice graphs and I can see your reasoning based on this limited information but let’s look at the bigger picture with some real FS measurements and a quick calculation shall we:

Distance between tweeter and FR driver: 10 cm

Reasonable listening distance for fantastic stereo image: 300 cm

Therefore using Pythagoras distance differential between tweeter/FR and listener is:  0.166 cm

Take a frequency of 10KHz when the tweeter is fully implemented

This gives a wavelength of 3.43cm

Now frequency cancellation would occur at ½ wavelength differential, that’s 1.72 cm

So we are 1/10th of half a wavelength differential at 3M, and that’s for the very worst case scenario wavelength shift (moving your head up/down until you receive the largest dip).

Please correct me if I have made an error in these calculations (I only had 5 mins) but it is no wonder you can’t hear any hint of combing at 3M, given the real perspective of in a non-perfect room conditions of 2 line sources using non perfect drivers. If you can hear any such effect you qualify for the Golden Ear award! 

Kevin

[JohnR]

I have a question for Jim Griffin about the power response. Toole says that flat on-axis and smoothly-varying power response are preferred. So, by that, any "comb filtering" at any particular off-axis point is irrelevant, is it not? What counts is the average over some vertical/horizontal window.

So, considering just the question of the two Jordans running full-range for the moment, why is there a problem exactly? It seems that the on-axis response (with the listener assumed to be within a certain window) and the power response would meet Toole's criteria?

I do realize that every loudspeaker designer has an ego the size of a planet, but from a technical and fairly impartial perspective, I have to question what the issue really is with the two Jordans run in parallel, assuming the constraint that a listener is in the "sweet spot"...

Thanks...

[Danny Richie]

Nice graphs and I can see your reasoning based on this limited information but let’s look at the bigger picture with some real FS measurements and a quick calculation shall we:

Distance between tweeter and FR driver: 10 cm

Reasonable listening distance for fantastic stereo image: 300 cm

Therefore using Pythagoras distance differential between tweeter/FR and listener is:  0.166 cm

Take a frequency of 10KHz when the tweeter is fully implemented

This gives a wavelength of 3.43cm

Now frequency cancellation would occur at ½ wavelength differential, that’s 1.72 cm

So we are 1/10th of half a wavelength differential at 3M, and that’s for the very worst case scenario wavelength shift (moving your head up/down until you receive the largest dip).

Please correct me if I have made an error in these calculations (I only had 5 mins) but it is no wonder you can’t hear any hint of combing at 3M, given the real perspective of in a non-perfect room conditions of 2 line sources using non perfect drivers. If you can hear any such effect you qualify for the Golden Ear award! 

Kevin

Okay, I like your response. JohnR basically asked the same question so I can understand the thought behind this. Let me answer this and JohnR's at the same time.

What you said above about the deviation to the on axis being very slight is true. Moving around a few inches up or down at 3 meters away may only cause a few db of loss up top. However, we don't just hear the on axis response. If we did then we wouldn't need any room treatment. What we really hear is a combination of the on and off axis response and how it is reflected in the room.

As JohnR pointed out "Toole says that flat on-axis and smoothly-varying power response are preferred."  A smoothly-varying power response requires a pretty smooth off axis response.

Let me use a few more examples.

Here is an MTM design with a little too much driver spacing and a high crossover point. Looks good on axis right.



Horizontal off axis looks good.



Now look at the vertical off axis.



Now here is a full range driver that is about the size of the Jordan. It is a Fostex driver if I remember correctly. Yeah, it is a little rough and has some baffle step loss...



Now the off axis.



Now a well designed two way with tight driver spacing and low crossover point. On axis...



Now horizontal....



And the vertical....



Now which speaker will have a smooth power response? Which one will have the best "average over some vertical/horizontal window."

No ego's, just trying to help out.

[Danny Richie]

 

[JohnR]

Now which speaker will have a smooth power response?

At the risk of being dragged into the bottomless hole...

Any which measure that way. You didn't provide that though.

[Danny Richie]

At the risk of being dragged into the bottomless hole...

Any which measure that way. You didn't provide that though.

I don't have in room response measurements on those same speakers. I don't usually save in room response measurements because they are a product of the speaker AND the room. So the in room response is only relative in "my" room. But I have evaluated speakers looking at in room responses and compared them to off axis responses made with gated time windows. The speakers with the smooth off axis responses have the smoothest in room response. This is especially true in the vertical as the ceiling is a large untreated reflection point. And my ceiling is 9 feet high. Any major deviation from the speakers off axis response shows up in the in room response.

[JohnR]

AFAIK the power response is not dependent on the room, it's a characteristic of the speaker itself. I'll try to find more references to support this.

[Danny Richie]

If you want to call the "power response" a summation of on and off axis response (independent of the room) that's fine. But the reason we look at the off axis response is to gauge what its interaction will be in the room. Otherwise why even look at off axis or why even use room treatment?

The off axis response is a strong contributor to the "in room response".

[AJinFLA]

I have a question for Jim Griffin about the power response. Toole says that flat on-axis and smoothly-varying power response are preferred. So, by that, any "comb filtering" at any particular off-axis point is irrelevant, is it not? What counts is the average over some vertical/horizontal window.

So, considering just the question of the two Jordans running full-range for the moment, why is there a problem exactly?

Hi John,

Yes, Toole does say that. Yes, there will be averaging, the total sound power at the LP will consist of the direct field and reflected fields (plus modal interaction, etc.). As arranged, the Jordans will be ok on "design" axis and directly horizontal off axis. But anywhere else will be as below, or worse (I'm sure you've used Edge before, fairly easy to sim):





Now, Kevin also stated the tweeter is run with a single cap, the Jordans only with a (4db) shelving filter (making the stated avg 1 watt sensitivity impossible). This will yield a large phase mismatch on (and off) axis, leading to some large swings in the FR around XO. Higher in frequency (9k IIRC?), yes, but not complying with Tooles on axis requirement either.
Whether any of this is going to be audible and detract from the listening experience of the average audiophile is entirely debatable (I'm going to side with Kevin on this, based on my experiences with audiophiles). But the effects on the physical soundfield, are not.
My apologies if this seems like "rival" designer egotism or just sheer "piling on" by another manufacturer, but technically, that's my 2c.

cheers,

AJ

[JohnR]

If you want to call the "power response" a summation of on and off axis response (independent of the room)...

I'm pretty sure I didn't invent that. Although the definition is not quite that. I'll look it up.

[JohnR]

Hi John,

Yes, Toole does say that. Yes, there will be averaging, the total sound power at the LP will consist of the direct field and reflected fields (plus modal interaction, etc.). As arranged, the Jordans will be ok on "design" axis and directly horizontal off axis. But anywhere else will be as below, or worse (I'm sure you've used Edge before, fairly easy to sim):

Hi AJ. Thanks but aren't you avoiding the question?

[AJinFLA]

Hi AJ. Thanks but aren't you avoiding the question?

Which one?

[JohnR]

OK, the question I had is, given that the on-axis response of two full-range point sources is flat, what does the power response look like.

[AJinFLA]

OK, the question I had is, given that the on-axis response of two full-range point sources is flat, what does the power response look like.

When Toole refers to "sound power", it is the average of the speakers frequency responses taken in 15 degree intervals, both horizontally and vertically, over a 360 degree sweep (anechoic).
As such, the 2x5 fullrange will have a sound power dip, as shown in my sims, >5k or so, mainly due to the vertical cancellations, which will be particularly pronounced by the configuration.
This of course, does not consider the effect of the added tweeter, both on and off axis, surrounding the 9k (?) region.
As with nearly all speakers (especially boxes), it will taper from LF (where it is omni) to HF (directional), with dips/bumps created by cancellations/summations.

cheers,

AJ

[JohnR]

I thought it was less than +/- 180. Regardless, any plots to share? Just to  couneract the thousand meamingless ones so far.

[Danny Richie]

Here is a relative thread that involved helping a guy design a set of desk top speakers using two full range 3" drivers per speaker. One driver was facing upward to minimize cancellation patterns. The thread includes measurements and explanations that might help one understand some of the things that I, Jim, and AJ have pointed out. It is an interesting thread.

http://www.audiocircle.com/index.php?topic=103789.0

And I don't sell these or offer anything like them and this not a commercial advertisement. It was just helping a guy with a fun and interesting project.

[AJinFLA]

I thought it was less than +/- 180.

Well, it was 360 way back when he was doing it at the NRC, which is when I was following his work. Not 100% certain about now, or if that's how Harman does it.

Regardless, any plots to share? Just to  couneract the thousand meamingless ones so far.

No program/sims that I am aware of, for sound power. Why do you feel what has been presented so far is meaningless? The sharp drop above 5k anywhere not directly on design or horz off design axis, will most certainly have consequences to the sound power.
I'm not (nor will I) saying this means the speakers sound horrible, etc. It's just that it will have consequences to the soundfield and perception of it, as defined by Toole et al.
Kevin seems to be satisfied, like I'm sure his customers are, so I really see no big deal, other than as a technical point, it's quite real.
As you know all speakers have issues. But as you also know, that doesn't make all those issues wishable away. I reiterate, not trying to single out Kevin or his design. Just focusing on the technical issues vs claims.

cheers,

AJ

[JohnR]

? The sharp drop above 5k anywhere not directly on design or horz off design axis, will most certainly have consequences to the sound power.

I'm sorry but I think I missed that. Where is the 5K+ drop demonstrated?

[Kevin Warne]

This seems to be going on a bit so this is my last post. I really have better things to do.
Sensitivity measurements are not simply a matter of reading the spec of the driver it is about measurement in a box configuration and a large cabinet actually enhances the sensitivity. We have no incentive at all to exaggerate sensitivity measurements as our speakers are often reviewed, with the result that the measured value agrees within around a db of our spec . It would look rather suspect if our measurements are considerably at odds with what reviewers find. As for review results I can't post them here but a cross section are on the website if interested - good and bad!.

Edge software. Yes I have looked at it and found it a bit useless for 2 drivers - too many assumptions. For instance its baffle-step calculator assumes the driver has a flat response infinite baffle and this is true for one but certainly not true for 2 such drivers where you get enhanced bass output dependent on square of total driver area. I contacted the helpdesk to point the issue out but hey did not seem to know what I was talking about.

As for room respose. It's a muddy topic. My position is that some of what Danny says is sensible but its not something I worry about too much. It certainly makes sense to have an even room response but I don't think room response is necessarily given by a few off-axis readings at different heights. By definition the room contains a myriad such measurement points so the readings will be averaged over the whole room, that's my definition of room response. Why are large vertical arrays used in rock auditoriums - it's actually to minimise room response and give a high ratio of direct to reflected sound over large distances. You can measure a very flat room response from a completely omnidirectional speaker, indeed the flattest, but what are you actually listening to - it's garbage, very room dependent, little imaging and certainly not something that was in the recording. You are just hearing many virtual copies of the loudspeaker appearing from different points all around you. It's not really a good design consideration to achieve a flat response in-room if you lose all the acoustic detail in the process. This is where the industry has actually gone so terribly wrong. The beauty of Ted Jordan's designs is that they appreciate a good degree of directionality is important in detail resolution, minimising any room response and maximising true stereo imaging (not ghost images). Having said that is is of some importance that the room does sound right and this is a small property of the speaker but a large property of the room. Common sense really! 

[Danny Richie]

My position is that some of what Danny says is sensible but its not something I worry about too much.

Hey if you are not too worried about those issues that's fine. If you are happy and your customers are happy then that's all that matters. But I wouldn't dismiss those issues as unimportant.

It certainly makes sense to have an even room response but I don't think room response is necessarily given by a few off-axis readings at different heights.

The reason we take off axis measurements is so that we have some idea of what the room interactions will be. It is pretty common. Even a lot of the review magazines take the time to take these measurements. Even John Atkinson of Stereophile has added some vertical off axis measurements to his standard measurements.

You can measure a very flat room response from a completely omnidirectional speaker, indeed the flattest, but what are you actually listening to - it's garbage, very room dependent, little imaging and certainly not something that was in the recording. You are just hearing many virtual copies of the loudspeaker appearing from different points all around you. It's not really a good design consideration to achieve a flat response in-room if you lose all the acoustic detail in the process. This is where the industry has actually gone so terribly wrong. The beauty of Ted Jordan's designs is that they appreciate a good degree of directionality is important in detail resolution, minimising any room response and maximising true stereo imaging (not ghost images). Having said that is is of some importance that the room does sound right and this is a small property of the speaker but a large property of the room. Common sense really! 

Actually that is not correct at all. Omni directional speakers don't produce the flattest room response measurements but they can be just as good as any other speakers. It depends on the design and execution of the speaker as much or more so than whether or not it is an omni design or not.

And the sound field that a well designed omni speaker produces is certainly not garbage, and is not a virtual copy appearing from different points all around you. Omni speakers really don't work that way.

They can present a large layered sound field that is very accurate. Take for instance the playing of any acoustic instrument. The sound radiates in all directions not just in one direction. Your standard speakers do as well in the lower ranges. The omni radiating pattern of an acoustic instrument does not create ghost images. That is their natural radiation pattern. And some omni designed speakers can create that same natural radiation pattern.

And "good degree of directionality" has nothing to do with detail resolution.