KEF LS50

[Melo theory]

Ok, 37 posts and this discussion hasn't started yet.....
I am thinking of using one of these drivers as an omni.

[jerryo]

Hi Melo theory,
I think that the discussion has started; I'm not sure what you are looking for, discussion wise, in this thread. What would you like to discuss about these drivers/speakers?
I do not think that these drivers are really aimed at an omni-directional set up. I could be wrong though. This is the point actually, none of us know exactly what these drivers will be capable of until we can get our hands on them. This is really what lies at the core of the discussion.

[jackocleebrown]

Hi Pallas,

Originally Posted by Pallas

First, the convex baffle shape.....

Imaging is one of the more difficult aspects of performance to quantify, but it seems to be related to dispersion and in particular getting consistent smooth responses on- and off-axis. Diffraction is an important consideration in achieving this, however, it is not the only factor.

Sorry to recap things you probably already know, the irregularity you get because of the 4pi-2pi transition occurs because of secondary radiation from the edges of the cabinet. In fact the whole diffraction step can be pretty accurately approximated in this way (see Urban et al. dipole edge diffraction model on AES). The amount of HF scattered from the edge is really important, if you can make the edge scattered radiation only have low frequencies then you'll get a smoother diffraction characteristic. This can be done by making the edge smooth (rounding the corners or giving the baffle a convex shape) but it is also affected by the dispersion characteristic of the driver. If the driver is directional then the diffraction step will also be smoother. With the 3005 the baffle is quite small and the 4pi-2pi transition occurs at quite a high frequency where the directivity control of the Uni-Q driver is already in effect. So even though the concave baffle is not the theoretical ideal, it does not cause a big problem in this case.

Originally Posted by Pallas

...I wasn't at all surprised to see the Blade's Uni-Q in a similarly-shaped baffle....

The baffle on the Blade has a slightly different intention to the 3005. With the Uni-Q layout the MF cone forms a smooth waveguide for the HF. With the Blade this idea is continued into the baffle shape - the baffle continues tangentially and smoothly from the outside diameter of the cone. In effect the whole cabinet is a smooth continuous waveguide for the tweeter. We had to specifically design the driver to work in this situation, previous UniQs were designed assuming a flat 2pi baffle. The LS50 is similar to the blade approach, the baffle is a smooth continuation of the MF cone and then the edges are rounded to continue this as far as was practically possible for this particular product.

Originally Posted by Pallas

My second question is about voicing...

It is hard to correlate frequency response measurements to the perceptual balance of the loudspeaker. The only way that I have found to consistently do this is to take measurements in many different directions and then to look at power responses and listening window averages. There are many acoustical phenomena in a conventional box loudspeaker which tend to be most visible on-axis. For example, diffraction ripples are worst on axis as this observation position is typically equal distance from the left and right edges. The on-axis response can be quite misleading as a result of this. Particularly when the system has a high degree of symmetry, as in this case, the 15 degree or 30 degree can be much more representative of the average or power response. It would be great if all published measurements also showed off-axis data but it is time consuming to capture this data so I understand why most choose just one measurement.

We try and balance our loudspeakers as neutrally as possible whilst still being enjoyable to listen to for long periods of time. In our experience, the better the drivers and cabinet the less you need to choose trade off these two aspects against one another. I think that this is one area where the LS50 is particularly successful.

All the best, Jack.

[Face]

Originally Posted by Melo theory

Why is this in the full range forum?

It would be great if it could be moved to the appropriate area.

[Pallas]

Jack, thanks for the response. It took me a while to find it, because I forgot that this thread is in the wrong forum. (Mods?)

Before I ask follow-ups, I'd just like to say that I really admire the stuff KEF has put out recently. I think there was a bit of a lull (products designed between the time Andrew Jones left and your current regime came in?) but you guys seem to be churning out winners of late. Even the Q100 - I'm currently listening to a pair - is a darned good speaker within its size/output limits. (Cheap looking, but great sounding.)

Originally Posted by jackocleebrown

The amount of HF scattered from the edge is really important, if you can make the edge scattered radiation only have low frequencies then you'll get a smoother diffraction characteristic. This can be done by making the edge smooth (rounding the corners or giving the baffle a convex shape) but it is also affected by the dispersion characteristic of the driver. If the driver is directional then the diffraction step will also be smoother. With the 3005 the baffle is quite small and the 4pi-2pi transition occurs at quite a high frequency where the directivity control of the Uni-Q driver is already in effect. So even though the concave baffle is not the theoretical ideal, it does not cause a big problem in this case.

Interesting. So basically the 3005's image like they do despite the baffle shape, not because of it? If so, then what was the reason for the concave shape there? It really doesn't look much different from a flat baffle at that size, so I wouldn't think cosmetics. Does the concave shape make the cabinet more rigidi?

Originally Posted by jackocleebrown

The baffle on the Blade has a slightly different intention to the 3005. With the Uni-Q layout the MF cone forms a smooth waveguide for the HF. With the Blade this idea is continued into the baffle shape - the baffle continues tangentially and smoothly from the outside diameter of the cone. In effect the whole cabinet is a smooth continuous waveguide for the tweeter.

The Blade is one of the speakers I would really like to hear.

Originally Posted by jackocleebrown

We had to specifically design the driver to work in this situation, previous UniQs were designed assuming a flat 2pi baffle. The LS50 is similar to the blade approach, the baffle is a smooth continuation of the MF cone and then the edges are rounded to continue this as far as was practically possible for this particular product.

Hmm. Not quite sure I see that, though I haven't seen the LS50's in person. It looks to me like the roundover starts basically at the trim ring. You know, if the LS50's had silver (or better yet, black) diaphragms or at least grills, and were sold individually, I would love to buy a trio.

(Something you may wish to pass on: KEF IMO does a disservice to its customers by only selling bookshelf speakers in pairs. "Matching" centers, even KEF's Uni-Q based units, just don't match. Now, they may be fine for home cinema. Perhaps even concert videos. But when there's no screen to "patch together" the soundstage, as in the case of music either recorded in discrete multichannel or expanded from stereo via Dolby Pro Logic II or other means, that doesn't work. I know someone who went in a direction other than - and markedly inferior, to my ears - KEF Ref 201/2's simply because he couldn't have three 201/2's unless he bought two pairs and sold the surplus speaker. He just couldn't afford to do that. Even the LS50, which seems ideal for for a small multichannel music system, seems to only be available in pairs.)

Originally Posted by jackocleebrown

It is hard to correlate frequency response measurements to the perceptual balance of the loudspeaker. The only way that I have found to consistently do this is to take measurements in many different directions and then to look at power responses and listening window averages. *** It would be great if all published measurements also showed off-axis data but it is time consuming to capture this data so I understand why most choose just one measurement.

I agree.

I'm curious, how exactly does KEF currently do measurements? (At least, that you're willing/able to disclose.) Is it similar to Harman, with a series of measurements (I think they take 70) in a sphere?

For a DIYer without access to anechoic chambers or things like that, is there anything would you recommend taking besides the standard gated horizontal and vertical polars to help us characterize our designs. (I tend to take nearfield, gated measurements 10, 15, 20, 30, 40, 50, 60, and 90 deg horizontal, and 10, 15, 20, 30 vertical. Unless it's an axisymmetric speaker using a Uni-Q or Dual concentric, in which case I just do the "horizontal" measurements)

[Bare]

Odd all this. Over the years I've had passing interest in KEF.
They were making Coax Drivers. Which I like :-)
Sadly every one of their Uniq drivers I managed to audition was to me at least, Far short of the mark.. quite disapointing actually..especially so given their aggressive pricings.
Which I imagine, would have allowed genuine quality manufacture and Innovative R&D.
Shame but just another also ran, a footnoted maker IMO.

Want a Cheap Coax?? Buy a Car Audio one. It won't sound lesser :-)

[talaerts]

Originally Posted by Bare

Odd all this. Over the years I've had passing interest in KEF.
They were making Coax Drivers. Which I like :-)
Sadly every one of their Uniq drivers I managed to audition was to me at least, Far short of the mark.. quite disapointing actually..especially so given their aggressive pricings.
Which I imagine, would have allowed genuine quality manufacture and Innovative R&D.
Shame but just another also ran, a footnoted maker IMO.

Want a Cheap Coax?? Buy a Car Audio one. It won't sound lesser :-)

I have listened to many, many speakers over the years and cannot agree with your (aggressively voiced) opinion...

[Melo theory]

I had a 5.1 system once, using these.......



They where awful. I really wanted to like them because they are so nice looking and I like the coax design.....
But, they where the most boxy sounding speakers I've ever owned. I'm sure it was the porting and x-over design.
I really like the coax, and I'm sure if handled actively, they would have really shined, but I didn't have the equipment at the time.

[jackocleebrown]

Hi Pallas,

Originally Posted by Pallas

Interesting. So basically the 3005's image like they do despite the baffle shape, not because of it? If so, then what was the reason for the concave shape there? It really doesn't look much different from a flat baffle at that size, so I wouldn't think cosmetics. Does the concave shape make the cabinet more rigid?

The concave was part of the design, as you say it is not really very far from flat but it is noticeable when the product is viewed from the side - acoustically it is not much different from a flat baffle. The diffraction from the edge of the MF is very good on the 3005 because of the way that the surround is dropped back a little so that it does not protrude much in-front of the cone. This was more or less the fore-runner to the z-flex surround on the latest Uni-Qs.

Originally Posted by Pallas

Hmm. Not quite sure I see that, though I haven't seen the LS50's in person. It looks to me like the roundover starts basically at the trim ring.

The round off does start from the trim edge as you say. What I was trying to articulate (badly) is that everything forms a waveguide for the tweeter - starting with the small plastic molding around the tweeter OD, then the MF cone, then the MF surround, then the MF trim then the baffle of the cabinet etc... On the LS50, just like on the Blade, each of these surfaces smoothly flow into one another with as little irregularity as possible.

Originally Posted by Pallas

You know, if the LS50's had silver (or better yet, black) diaphragms or at least grills, and were sold individually, I would love to buy a trio.

I am a convert to the "rose gold", after initial being sceptical I'm now quite fond of it! Point taken about the individual sales.

Originally Posted by Pallas

I'm curious, how exactly does KEF currently do measurements? (At least, that you're willing/able to disclose.) Is it similar to Harman, with a series of measurements (I think they take 70) in a sphere?

For a DIYer without access to anechoic chambers or things like that, is there anything would you recommend taking besides the standard gated horizontal and vertical polars to help us characterize our designs. (I tend to take nearfield, gated measurements 10, 15, 20, 30, 40, 50, 60, and 90 deg horizontal, and 10, 15, 20, 30 vertical. Unless it's an axisymmetric speaker using a Uni-Q or Dual concentric, in which case I just do the "horizontal" measurements)

I believe that Harman have standard positions for measurements and then a set way to deduce a response curve from these. By contrast we don't have a fixed standard for measurements. Typically for xover design we would measure horizontal and vertical polars, from 0-180 deg, and then look at the individual directions and some weighted averages of these while doing the design work.

It is also helpful sometimes to also measure equatorial polars (starting at 90deg left then to 90deg above then 90deg right then 90deg below and back to the start) it is a pretty easy measurement to take with a turntable and gives a lot of additional data.

Personally I don't think that you can really distil it into one curve. Listening window averages seem to be really good for getting an idea of the balance but you still need to look at the individual responses so that you can check you don't have a particular direction with a horrible irregularity at some freq. The other thing to say is that by the time you have got to the xover design you've already made most of the decisions affecting the dispersion of the loudspeaker - such as driver selection/design/mounting, positions, baffle shapes etc. We do a lot of simulation work all the way through the design of a loudspeaker and try and come up with system configurations which have nice dispersion characteristics.

All the best, Jack.

[randybes]

I have AJ speakers and the LS50's. I like both and of course the bass on AJ's is better, but the LS50's are the best minimonitor I have ever heard and I think as time goes by, people are going to buy these up in droves. Just my opinion.