KEF R300 playground - measurements of raw drivers

And the winner is KEF! Coax unit is the larger one 17cm, R300 has 14cm. It looks like the Elac might benefit from some xo tweaks...
The new Q series speakers are 2 or 2.5 way speakers with the coaxial acting as a midwoofer reproducing the lowest frequencies. This is generally not a good idea because the large deflections of the midwoofer cone at low frequencies changes the shape of the waveguide for the tweeter. This will be lessened by using a larger diameter cone like the 200mm one in the Q950 but this large aluminium cone then needs to cross successfully to the tweeter. Interestingly the tweeter in Q950 has a large 38mm diameter and is crossed at 2.2kHz. With the phase plug and waveguide action of the midrange cone to handle directivity this looks a good design for a large 2 way.

The Elac is a 3 way and therefore can use a smaller midrange cone with a motor designed for midrange duty. The tweeter is a 1" crossed at 2.7 kHz and the midrange is a 4" minus the area for the tweeter making it equivalent to something like a 3". This small midrange is crossed to the 3 x 5" tiny woofers at 270 Hz! A conventional crossover for a 3" midrange would be twice that frequency. This is going to hurt the waveguide action on the tweeter and the midrange performance at higher SPLs. A rather unusual design decision.

After only a brief peruse, the KEF looks well designed given the decision to go for a 2 or 2.5 way midwoofer+tweeter coaxial. The Elac does not look so well designed given the decision to go for a midrange+tweeter coaxial in a 3 way with the potential for higher performance. However, it seems to be about 15-20% cheaper in the UK which is perhaps relevant.
 
Q550 is so well built that i really don't know how they pull it of for that price. Woofer is similar to R300 only smaller cone. Large voice coils and large magnets, Passive radiators with double suspension to achieve better symmetry - geez :)

In white - they are stunning

YouTube
 
Last edited:
I beg to differ. I think that Kef did much better job regarding diffraction than Goran with his test baffle, which is kinda expected given that he positioned R300 mid-hi on the center of the baffle. Problem is that low resolution on X axis and low resolution on Y axis (the way Goran showed his measurement) leads to wrong conclusions.

When i see 5dB or (worse) 10dB resolution on Y axis and i see, what i believe is, a problem - it always magnifies when i bring measurement resolution to what Geddes proposed: 2dB resolution with 40dB high Y axis grid.

Look how Goran's measurements look now...

Hi,

Here's three different graph resolutions (40/50/80dB, no smoothing) of the same measurement of the KEF SP1632 mid-range:

Measurement conditions:
Baffle: 90x20cm (se pictures).
Mic position: 1m on mid center-axis.

Regards

/Göran
 

Attachments

  • KEF SP1632 Midrange - 40dB Scale.jpg
    KEF SP1632 Midrange - 40dB Scale.jpg
    198.3 KB · Views: 607
  • KEF SP1632 Midrange - 50dB Scale.jpg
    KEF SP1632 Midrange - 50dB Scale.jpg
    204.6 KB · Views: 583
  • KEF SP1632 Midrange - 80dB Scale.jpg
    KEF SP1632 Midrange - 80dB Scale.jpg
    213.9 KB · Views: 630
  • CT-F3_3.JPG
    CT-F3_3.JPG
    62.8 KB · Views: 609
  • CT-F3_5.JPG
    CT-F3_5.JPG
    60.3 KB · Views: 614
Here's my measurement that's a little more comparable to yours but lower frequency limit is set to 200Hz, unlike yours that is 300Hz. Gate is good to 250Hz (yours is 275 if i'm not mistaken), 1m distance, mic at tweeter axis - on axis and 45 degrees off axis. Anyhow, midranges measure superb on both baffles my opinion. Tweeters differ somewhat and Kef's looks flatter on axis. Not that that is of great importance because off axis is by far more dominant at farfield listening but it was worth noting when Juha said that yours look much better regarding diffraction than measurements done in original Kef cabinet - which was product of displayed measurement resolution, in my opinion. When i posted it my way and side by side, it looked different enough and could be compared in a rightful manner.
 

Attachments

  • image.png
    image.png
    16.6 KB · Views: 213
Last edited:
^^ OK, but I was looking at/commenting tweeter responses!

Hi,

Here's some frequency responses for the KEF SP1632 tweeter.

The response is slightly better on the 56x23cm baffle with baffle edge round-overs compared to the 90x20cm baffle with sharp edges and other driver cones in the proximity of the co-axial.

Regards

/Göran
 

Attachments

  • KEF SP1632 Tweeter - 80dB Scale 15deg off-axis Baffle 56x23 vs 90x20.jpg
    KEF SP1632 Tweeter - 80dB Scale 15deg off-axis Baffle 56x23 vs 90x20.jpg
    204.9 KB · Views: 137
  • KEF SP1632 Tweeter - 80dB Scale on-axis Baffle 56x23 vs 90x20.jpg
    KEF SP1632 Tweeter - 80dB Scale on-axis Baffle 56x23 vs 90x20.jpg
    210.8 KB · Views: 148
  • KEF SP1632 Tweeter - 50dB Scale 15deg off-axis Baffle 56x23 vs 90x20.jpg
    KEF SP1632 Tweeter - 50dB Scale 15deg off-axis Baffle 56x23 vs 90x20.jpg
    193 KB · Views: 130
  • KEF SP1632 Tweeter - 50dB Scale on-axis Baffle 56x23 vs 90x20.jpg
    KEF SP1632 Tweeter - 50dB Scale on-axis Baffle 56x23 vs 90x20.jpg
    201.9 KB · Views: 136
  • KEF SP1632 Tweeter - 80dB Scale Baffle 56x23cm.jpg
    KEF SP1632 Tweeter - 80dB Scale Baffle 56x23cm.jpg
    250.5 KB · Views: 158
  • KEF SP1632 Tweeter - 50dB Scale Baffle 56x23cm.jpg
    KEF SP1632 Tweeter - 50dB Scale Baffle 56x23cm.jpg
    251.3 KB · Views: 171
  • KEF SP1632 Tweeter - 40dB Scale Baffle 56x23cm.jpg
    KEF SP1632 Tweeter - 40dB Scale Baffle 56x23cm.jpg
    249.5 KB · Views: 214
OWhen ever i made sim for R300 crossover that was flat on axis, i got a bump at 2-3KHz off axis. So, little irregularities on axis is quite acceptable if the dips fills up off axis. Since it is a diffraction issue there's no need to try to solve it on axis.

https://s15.postimg.cc/649wnq4qz/Webp.net-gifmaker.gif

Thanks for uploading the measurements.
I have been experimenting with them for a while with the following result.
The 2-3 kHz bump (45 deg characteristic, green) is still present but it's less than 1dB over the slope line.
On-axis (blue) response variations are witin +-1 dB around average level up to 17 kHz (except for the dip at 8kHz)
 

Attachments

  • Kef300XMach03.jpg
    Kef300XMach03.jpg
    444 KB · Views: 285
  • k300_onaxis_XMach03_dxo.zip
    63.3 KB · Views: 56
Thanks for uploading the measurements.
I have been experimenting with them for a while with the following result.
The 2-3 kHz bump (45 deg characteristic, green) is still present but it's less than 1dB over the slope line.
On-axis (blue) response variations are witin +-1 dB around average level up to 17 kHz (except for the dip at 8kHz)

Great stuff X. How's your phase response and for what distance did you optimize your crossover for ?

I'm still waiting for capacitors to arrive. Then i'll be able to play outside of simulator a little bit :)
 

Attachments

  • Webp.net-gifmaker.gif
    Webp.net-gifmaker.gif
    71.6 KB · Views: 680
Last edited:
How's your phase response and for what distance did you optimize your crossover for ?

I did not actually assume anything about the distance in this excercise. I just took the data "as measured". The only preprocessing was determnination of mod delay (with relation to the tweeter: 0.42in mid, 1.13in woofer)
 

Attachments

  • sysSplPhase_and_reversePolMid_Kef300XMach03.jpg
    sysSplPhase_and_reversePolMid_Kef300XMach03.jpg
    113.4 KB · Views: 646
I did not actually assume anything about the distance in this excercise. I just took the data "as measured". The only preprocessing was determnination of mod delay (with relation to the tweeter: 0.42in mid, 1.13in woofer)

Only now i see that i didn't include merged woofer files in my DB folder. I've modeled your crossover with my merged woofer files so it looks like this:
 

Attachments

  • Webp.net-gifmaker.gif
    Webp.net-gifmaker.gif
    189 KB · Views: 237
Few pics of factory crossover and woofer. Few capacitors are wrapped in some sticky material. It looks like the stuff that Kef uses for CLD in LS50.
 

Attachments

  • DSC_0057.jpg
    DSC_0057.jpg
    744.5 KB · Views: 338
  • DSC_0056.jpg
    DSC_0056.jpg
    664 KB · Views: 350
  • DSC_0055.jpg
    DSC_0055.jpg
    642.8 KB · Views: 411
  • DSC_0058.jpg
    DSC_0058.jpg
    917.2 KB · Views: 371
  • DSC_0052.jpg
    DSC_0052.jpg
    403.3 KB · Views: 381
  • DSC_0051.jpg
    DSC_0051.jpg
    392 KB · Views: 419
From your pic it appears that the driver hole is not chamfered/beveled. When I pulled the mid tweet section from my R300 the hole was not beveled either and I was amazed it was not. I was too lazy to set up my router so I left it alone as the speaker was listenable if not a bit dynamically constrained. Do you plan to bevel the inside of the driver hole?

I'm thinking of gutting my pair and making an outboard crossover and slowly tweak it and the box (as needed) over a period of time. Or not, just something to do when I'm bored albeit it is last in the que of speaker projects. Overall, I think Kef did a pretty good job.
 
I'm there - these are my test crossovers.

DSC_0060.jpg


Wires go through the port tube now and i'm listening for changes with slight changes of values. Actually crossover sounded really good without any tweaking but i want to see if i can achieve something better by tweaking it a little. For now i can make it different but not better. I'm here at the moment:

attachment.php


Transfer function of crossover:
attachment.php


Number of components can be used from factory crossover. C1 is made of 170uF and 50uF in parallel, C7 and C18 are from factory crossover, L6 is factory coil of 22mH unwinded to get 10mH, L4-L8-L10 are also salvaged unchanged from factory crossover.

Simulated impedance response:

https://s15.postimg.cc/u0yjku5y3/Untitled1.png

Measured impedance:

https://s15.postimg.cc/j1dc987t7/Untitled.png

In the next days i expect to find some time to measure frequency response.

As for the cabinet, there are no chamfers but the real question is if it needs it. If you do the chamfering i'd appreciate if you posted some frequency response measurements here - before/after.
 

Attachments

  • Untitled.png
    Untitled.png
    21.2 KB · Views: 1,158
  • Untitled1.png
    Untitled1.png
    45.5 KB · Views: 1,050
Last edited: