KEF driver phase and the H-B generated phase

[dlr]

I thought I'd try to get some discussion going on something I've never encountered before. I've got two of the more recent KEF coaxials, a true, pure midrange. The output drops like a rock below about 400Hz. The slope exceeds 48db at 400Hz.

The issue is that I cannot, no matter what I set up in the model, get the H-B generated phase to match the measured response. Below about 600Hz it diverges drastically. To get close I used a 36db slope. This gets the low end closer, but of coarse the middle of the passband then won't match. Likewise, the upper end above about 6K will not match, no matter what I do.

I've never encountered such a case. I'm guessing that it has to do with the newer profile and the drastically limited Xmax. There is no roll in the surround whatsoever. Being a true midrange, it's not meant to have any significant displacement. BTW, it is a 6.5" driver even so.

Anyone else ever run into this?

dlr

[AndrewJ]

Hi dlr,
can you give more info, such as measuring equipment and conditions, and show your graphs. It is difficult to diagnose with the info provided.

Andrew

[AJinFLA]

Dave,

Are you saying the acoustic roll off of the driver on your test baffle approaches 48db/oct??
As previously asked, do you have some SE graphs of the response and impedance curve? Did you measure the T/S parameters?
Is the driver old enough where it would have been one of Andrews designs? Sorry I missed it earlier. Enquiring minds want to know .

cheers,

AJ

[Feyz]

Mostly speculation guessing on my part:

A few months back I remember reading a nice article on AES journal which was describing and modelling cone modes. I don't remember which article it was right now, or the author's name. The author had modeled and divided the sound generated by the speaker cone into two parts, one is generated by the movement of the cone back and forth as piston and the other the movement of the cone surface with waves going on the surface. He had calculated and graphed the frequency response of each of these sound sources seperately. I don't recall that he had mentioned about minimum phaseness, but as soon as I saw those graphs, I started thinking there it is what could make a driver non-minimum phase. The driver even though is a single vibrating source, with this way of analyzing it, consists of two different sources and each having different frequency responses of its own. What the driver generates as sound is the sum of the two. Depending on how much each source's frequency response differs and how they overlap, the sum can become non-minimum phase. In your example, the driver having no roll surround limiting its travel, the sound generated by the waves travelling on the surface of the cone might become a significant sound source to to cause the driver towards non-minimum phase.

As I said at the beginning this is just speculation on my part, take it with a big grain of salt

[dlr]

Originally posted by AJinFLA
[B]Dave,

Are you saying the acoustic roll off of the driver on your test baffle approaches 48db/oct??

Yes, but it's the tweeter. I am actually working with some in-box measurements, but the tweeter is not much different in-box due to the horn loading. Excuse me, make the "waveguide".

Is the driver old enough where it would have been one of Andrews designs? Sorry I missed it earlier. Enquiring minds want to know .

That I don't know.

The way I wrote it made it look like the mid was the issue. Here's what I had previously posted at PE:

BTW, I tried to work some with the KEF midrange units I have. The woofer models very easily in SoundEasy, but the tweeter has me stumped! I can easily import the SPL file to set it up, but the H-B generated phase just will NOT match the measured phase. There must be something going on with the horn loading and the woofer profile.

The problem is the tweeter. I had forgotten that I had some graphs already at my site (using LAUD). Here's the page:

KEF measurements

Here's the tweeter response on the large baffle:




The in-box is not all that different due to the horn loading. BTW, my curves look to be very close to those of KEF. And note that there there's an overlayed curve. These represent the two drivers I have, so unit-to-unit consistency (even though I had to fix the domes on both) is excellent.

It's not that I need assistant in modeling them, it's that the H-B generated phase using the model that matches the frequency response does not match the measured phase. There's an excess-phase component that is not constant across the passband and that makes the response not minimum-phase, at least as far as I've been able to determine.

What this is leading my to believe is that the horn loading is affecting the low end response in an unexpected way. But I need to do some more work, maybe re-measure and try to determine what's going on. I think that it may have to do with the time-delays of the radiation as the wave expands and the lower frequency energy is directed more forward for constant dispersion.

But I could be all wet as I've always found breakup of standard drivers (woofers, mids, doesn't matter) to always be minimum-phase, hence my puzzlement.

dlr

[AJinFLA]

Originally posted by dlr

The way I wrote it made it look like the mid was the issue.

I think you fooled Feyz as well. When you mentioned 400hz and no surround, I thought it was the mids rolling off @ 48db/oct, which I would certainly have no explanation for.
The measurements he mentions sound interesting.
Now that you clarified it being the tweeters horn loading - I still don't have an explanation .
Do you have an impedance curve for the tweeter? What is the Fs?

cheers,

AJ

[dlr]

Originally posted by Feyz
Mostly speculation guessing on my part:

A few months back I remember reading a nice article on AES journal which was describing and modeling cone modes. I don't remember which article it was right now, or the author's name. The author had modeled and divided the sound generated by the speaker cone into two parts, one is generated by the movement of the cone back and forth as piston and the other the movement of the cone surface with waves going on the surface. He had calculated and graphed the frequency response of each of these sound sources seperately. I don't recall that he had mentioned about minimum phaseness, but as soon as I saw those graphs, I started thinking there it is what could make a driver non-minimum phase. The driver even though is a single vibrating source, with this way of analyzing it, consists of two different sources and each having different frequency responses of its own. What the driver generates as sound is the sum of the two. Depending on how much each source's frequency response differs and how they overlap, the sum can become non-minimum phase. In your example, the driver having no roll surround limiting its travel, the sound generated by the waves travelling on the surface of the cone might become a significant sound source to to cause the driver towards non-minimum phase.

Sorry, Feyz, I've been writing to hastily at times and not being very precise. Gotta slow down some and choose my words more deliberately.

That does sound like an interesting bit of research. I'd like to learn more about the details of wave propogation from drivers. My point about the surround was confusing, as I pointed it out to show that the driver was a pure mid with a special profile and lack of roll in the surround to show that it has little movement.

That profile and surround have quite an effect on the tweeter output, doing a good job of loading the tweeter more optimally. The steep negative slope of the tweeter SPL shows just how much it does direct the energy forward. There's almost no diffraction (other than step) from these when mounted on a normal, nominally sized baffle. When the frequency wavelength approaches the diameter of the cone, it starts to drop precipitously. The waveguide support goes away.

In essence, I think that the delayed energy directed forward throughout the area of the cone makes the tweeter appear to be other than a point source to some degree. A tweeter on a flat baffle is close to point source below frequencies dictated by the dome diameter, though modified by the profile and doping. But there's no support for energy being directed forward, only prevention of radiation (and loss) into 4-pi space.

The waveguide directs some of the energy more forward and radiates into something less than 2-pi. That is true only to the point of the frequency/diameter specifics mentioned earlier. So this re-directing of energy sums with the direct waves. The slope of the SPL response indicates gradually decreasing support at higher frequencies, but I'm not quite sure of the full relationship. I'm not sure if there's a gradually decreasing acoustic impedance responsible for this or if it's more a case of "reflected" energy to some degree. The latter would, I suspect, have an influence on the minimum-phase aspect.

Maybe I'm speculating way beyond reality, I don't know. I think that I'm going to have to measure again, though I've never had any measurement issues in all of these years. I do also still have all impulse data, the mic and the probe feedback, so with LAUD I can always re-load any measurement, going back years, and re-create any post-processed data, SPL, CSD, etc., with different window types, lengths, etc. I may just review that data first.

I was wondering whether anyone else had ever seen this kind of phenomenon with the better waveguide drivers.

Of course, I may have just screwed up the measurement!

Dave

[AJinFLA]

Originally posted by dlr

Of course, I may have just screwed up the measurement!

Now that is something I have a lot more experience with
I have the same type roll of with my JBL, albeit at slightly lower frequency.

An externally hosted image should be here but it no longer works. Please upload images instead of linking to them to prevent this.

cheers,

AJ

[Feyz]

The issue is that I cannot, no matter what I set up in the model, get the H-B generated phase to match the measured response. Below about 600Hz it diverges drastically

Yes, it sounded like you were talking about the mid's phase.

One suggestion if you haven't tried already. Does it get better if you model the response as eventually converging to 12db/octave on the low end? From your graphs it looks like the initial sharp slope before 2Khz then turns into a gradual slope starting around 800Hz and going below.

[john k...]

Hi Dave. What does the measure phase look like?