Some measurements of Kef R300 drivers

[Zvu]

Measurements are done in factory cabinets, gated, at 1m distance mic at tweeter axis - gate is good to 250Hz. Holm-impulse, calibrated ECM8000, M-Audio Fast Track II USB interface Dell Inspiron notebook. Vertical grid is 40dB, with 1dB resolution.

Link to gallery:

https://postimg.cc/gallery/3dpsmrwd4/

Tweeter on axis and 30 deg off axis:



Midrange on axis and 30 deg off axis:



Woofer an axis and 30 deg off axis:



Woofer and midrange on axis:



Midrange and tweeter 30 deg off axis:



Woofer impedance:



Midrange impedance:



Tweeter impedance:



TS parameters for woofer:

[Zvu]

Few quick and dirty sims are in order.

In and out of phase on axis - simulated distance 2.2m




In and out of phase off axis 30 deg - simulated distance 2.2m



It'll be a challenge to keep the impedance over 3 ohm at all frequencies.

[Juhazi]

Nice to see raw responses, thanks Zvu! I have noted and also heard, that most KEFs have a bit dull treble. Looks like the tweeter's sensitivity is the limiting factor. A straight on-axis response for a moderately directive speaker leads to descending room response, in KEF's case too much descending IMO.

They might have responded to this issue, at least if we believe that the new Q350 represents their new ideal.

https://www.stereophile.com/content/kef-q350-loudspeaker-measurements

[Zvu]

I listened the Q350 and was not impressed. Much more ordinary design - Focal Chorus 706 - had better treble and overall balance.

I think that frequency response flatness on and off axis is the problem because treble of older Q100 i would not describe as dull - more on the bight side. But then again, its on axis response it tilted upwards to higher frequencies at least 2dB too much.

If anyone said to me that q100 is dull, i'd attribute it to electronic used and/or listening conditions.

[PANDINUS]

Real Measurement R300

https://www.soundandvision.com/content/kef-r300-speaker-system-ht-labs-measures

[mjmacka]

Zvu, Are you perhaps intending to rework the crossover in this speaker ?
Thanks MJ

[Zvu]

That is my intention. Drivers have stellar performance per my standards. Tweeter in a waveguide, dedicated midrange, low difraction, good vertical off axis and relatively small visual footprint. When will that be done though is anyone's guess - due to lack of free time. Crossover will be a bit more complex than what people are used to see but i'll do what ever is necessary to achieve good performance.

[Zvu]

I'm here at this moment. I do need to make full set of measurements 0-180 degrees but i'm waiting to find the measurement room with 5+ meters ceiling height so i can move the gate a bit lower.

[mjmacka]

Thanks for starting this thread. I was considering buying one of the KEF R series and having a try at this myself . As I have promised myself that I wouldnt mess any longer with DIY in particular passive crossovers I will resist , for the present, and keep a watch on how you go.
MJ

[Zvu]

I'm the opposite of that - i find there are too few good factory loudspeakers and the ones that are good are priced ridiculously.

Problem with diy is that there are too many unrefined projects. I'm 33 and diy-hifi enthusiast from 15th year of age. When i got in to the diy world i saw an ocean of good projects anyone would be happy with. As i learned bit by bit i came to realize that there is enormous amount of projects but the good ones are scarce. Now i'm finding two designers and their decisions interesting. One would be Alexandar Heissmann and other would be moderator on this forum, JonMarsh. There are few others that i suspect are good but their projects are poorly documented and can't really make anything certain out of it.

Let me give you an example. Zaph SR71 looked great to me when it appeared and it is well regarded even today (which in itself tells a story about brand name loudspeakers of today). Now that i look at it i see a midwoofer breakup at -18dB bellow fundament left untreated and then the talk about HD that is -45dB or -50dB as noticeable. Baffle step could and should be solved better and the cabinet should be bigger to let the midwoofer play lower because it can. That is not because he didn't know better but, in his own words, he wanted to keep it simple - but it reflected on performance in my opinion.

Anyhow, i realized that the best loudspeaker for me is the one i make crossover for - irrelevant if it is dsp or passive.

[mjmacka]

My situation is a bit different. I have been DIYing speakers for over 50 years. A while ago I decided my current effort would my last one . However like you it seems to me the KEF drivers have enormous potential. Whilst I no longer have the energy or design tools to pursue an implementation of my own I could persuade myself to implement someone elses well thought out solution.
I look forward to reading about your progress .
MJ

[Zvu]

Every project of mine is the last one - or so i've been telling myself that every time. I gave up on giving up - now i'd like to make some drivers of my own.

Can you give me some details about your latest endeavours ?

[Zvu]

Few quick and dirty measurements of test crossover (not the one simulated above). There is a chandelier in the room which made some diffraction issues (that can be seen on measurements) but the overall trend is noticeable. Much of the parts from original crossover are reused - but some coils are modified and series caps are of better quality. Measurements are done in factory cabinets, gated, at 1m distance mic at tweeter axis - gate is good to about 250Hz. Holm-impulse, calibrated ECM8000, M-Audio Fast Track II USB interface, Dell Inspiron notebook. Vertical grid is 40dB, with 2dB resolution.

Factory crossover:




Modified crossover crossover:




Modified on axis, 15 and 30 deg:



Crossover points are 500 and 2600Hz. Tweeter could be atenuated a 0.5-1dB and i have to bypass tweeter attenuation resistor with a capacitor to tilt up the response above 15KHz. Critiques are welcome.

[5th element]

I'd really like to start making some drivers of my own too. It's not easy getting all the soft parts that you'd want, ironically it's somewhat easier to find places to CNC mill all the metal bits!

[Zvu]

I found quite a bit here:

http://www.audioxpress.com/news/feel-the-beat-loudspeaker-industry-sourcebook-2018-now-available

I've contacted the closest ones with questions - for now.

[Juhazi]

Pretty smooth, that's it! Have you measured distortion at high spl? I guess that high treble is ok without 15k tilt, at least for adults.

[Zvu]

I intend to measure it but i don't expect for anything to surprise me. I still have to order the components for the crossover because these are from my stash of coils, resistors and capacitors i use for testing - so not the highest quality, but good for proof of concept.

[Zvu]

Few more measurements done at 1m/2.83V. Gated with 6ms with Hamming.

New crossover on axis in and out of phase, 2dB grid resolution:



Factory vs new crossover:



Factory crossover distortion:



New crossover distortion:



New crossover waterfall plot:



New crossover step response:

[PANDINUS]

Good work.
Can you please send frd and zma files?

[Zvu]

Can't do that. Sorry.

[PANDINUS]

What is the reason?

[Zvu]

My files aren't precise enough. I've made sims that did not match the measured response. If i can't provide good data, i won't provide any data. I've made the last test crossover with sim but then i had to do quite a bit of tweaking to get there.

I'm in the process of finding high enough room for measuring or i'll have to do it outside. What i can promise is that i'll post all files for download when i'm satisfied with consistency between simulation and measurement.

[PANDINUS]

I understand and thank you for the answer.

[Zvu]

No problem.

Do you have Kef R300 or planning to buy them ?

[PANDINUS]

Dual post.

[PANDINUS]

I do not have Kef. I will not even buy a small R300. I have a large room I want a bass area of 800cm2 and a resulting speakers sensitivity of at least 90dB / 2.83v
I look for suitable drivers from 500-700Hz up. I consider the horn with a 1.4 "driver or coax.

[Zvu]

Coax from Q700 or Q900 looks good.

Seas C18EN002/A could work - it is pricey though.

[PANDINUS]

I love coax just midrange series R. Will it achieve 90dB sensitivity?

The SEAS price is not an obstacle. But it has a low sensitivity.

[Zvu]

As far as i can see it will. Kef R900 has the same coax and is declared 90dB/2.83V

[Zvu]

I did some more measurements. Rain prevented making full set of measurements but i found quiet enough neighbourhood so yesterday at about 01:00am some sweeps were heard from the parking lot. Results are quite satisfactory and merging will be much easier now and lower in frequency too. Pics added.

Woofer on axis at 1m:



Midrange on axis at 1m:



Tweeter on axis at 1m:



This amount of quasi anechoic resolution makes it much easier to simulate the crossover between woofer and midrange.

[Zvu]

I'm quite satisfied with consistency now so here are files for simulators as i promised.

https://www.dropbox.com/s/dzg8gzfclgh8h92/Kef%20R300%20sim%20files.zip?dl=0

Few words about it.

Only woofers are merged between 100-200Hz. Mids and tweeters aren't because there's no need to. You have everything you need to derive Z axis offset. You'll find that frequency response of midrange+tweeter, that is needed to derive Z offset between them, drops bellow 300Hz. I am protecting the tweeter that way so, when deriving Z offset, look only how combined frequency response behaves over 300Hz in regard to reference. As far as X and Y offsets all you need to know is that woofer and midrange centers are 26cm apart. There are impedance files there. Just in case you don't like how i extracted my sim files i added HolmImpulse measurements .zip file that you can import into your HolmImpulse and set the gate to derive files as you see fit.

Play and enjoy yourself (:

[PANDINUS]

Thanks you

Представљам акцију на паркингу
Добар посао. Хвала вам пуно !

[Zvu]

You're welcome (:

Nemáš zač.

EDIT:

I've played with it a little so it could be a good starting point.

[PANDINUS]

I am interested in the scheme for the factory crossover. I did not find it on the web. Can you provide it?

[Zvu]

For woofer i can provide it but for mid-hi i didn't even bother to take the schematic given how bad it performed.

[PANDINUS]

The speakers were only lent?
For my understanding why did you measure them and make a new crossover?

[Zvu]

They aren't lent, they're mine.

I made new crossovers because the Kef's sounded really bad. Drivers look cool though and behave excellent.

[Zvu]

Here's Kef woofer crossover:

[Zvu]

Simulated impedance response:



Measured impedance:



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

[Zvu]

Measurements are done in room so 4,5ms gate is best i can do. 1m distance. Measurements done 0-90 degrees in 10deg increments. I don't believe i'll be dragging these to the parking lot any time soon, but who knows. Until then:




Averaging them:







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.

[Dave Bullet]

~ 2.5 ohms in the midbass. not an easy drive. your 220uF cap will be the culprit here. What are the target slopes for each driver?

[Zvu]

Here's the sim:




One can remove the impedance flattening circuit. It would look like this:




Kef's are 4 ohm rated so nothing strange about 3 ohm impedance in midbass there. Luckily i have an amplifier that doesn't care too much about impedance load presented to it.

[Dave Bullet]

All good if your amp doesn't care. But keep an eye on heat output. As you know, the hotter something runs, the shorter its life expectancy (that's my experience).

I'd remove he impedance flattening circuit as it is lowering your impedance and if your amp is a tank the upper midrange / lower treble swings shouldn't matter?

Listen to both I suppose and see if you can hear it.

Also if you are low volume listener with a tank of an amplifier, impedance may not matter anyway.

[Zvu]

My amplifier was designed for a guy that makes his own voice coils - he winds his drivers with copper ribbon to be 1-1.5 ohm max. I listened that amplifier in my system. Sounded the same or better than some First Watt products so i got rid of A class amplifiers. I liked the fact that i (almost) don't have to worry about the impedance ever again.

[Dave Bullet]

My amplifier was designed for a guy that makes his own voice coils - he winds his drivers with copper ribbon to be 1-1.5 ohm max. I listened that amplifier in my system. Sounded the same or better than some First Watt products so i got rid of A class amplifiers. I liked the fact that i (almost) don't have to worry about the impedance ever again.

We'd all like one of those!

I'd be interested in your subjective opinion on before and after listening. These are your speakers after all so it would be interesting to hear where the gains have been made (i.e. whether certain types of instruments, vocals, FR balance (i.e. bafflestep) off-axis behaviour etc...