Some measurements of Kef R300 drivers

[Zvu]

Before and after removing impedance flattening notch filter ?

I did give it a go last night but i can't say if there was any difference since i did not do A/B test. Music sounded pretty much the same - i did touch the heat sink after a while and it was barely less warm than without, or i didn't push it that much last night.

I will try to switch it on and off during playback... We'll see.

[Dave Bullet]

Do you think impedance comp helps audibly? What part of the frequency spectrum is your comp targeting?

[Zvu]

I'm not sure if it does. I'd say that it depends. I think that my speakers are linear enough and my system transparent enough to be able to reveal changes like that if there were any. Thing is, i listen most of my music from 1.5-2 meter distance (that's the main reason i gave up on floor standing loudspeakers while i live in this apartment). It is possible that at higher volumes it would be more obvious as amplifier starts to struggle but mine just doesn't because i can't push it that loud and keep my hearing undamaged.

Notch is targeting that broad impedance peak between 400Hz and 9KHz. It is very low Q notch and there is so much energy that i wouldn't recommend resistor in that notch filter to be less than 20W. 11W resistor was a warmish when i pushed the speaker to listen how would it behave and measure at higher loudness.

[Zvu]

experimenting a bit...

The reason i bought R300 is that i thought it's concept is quite similar to Reference 1 - which is something i've been looking for. It is a three-way loudspeaker that has enough punch in my relatively small room, it's a standmount so later i can add subs if i need them, it has waveguide loaded tweeter and all metal cones/dome. Ref 1 uses somewhat better bass driver (larger magnet, better power handling and different coupling voice coil to the cone) and quite similar mid-hi driver (differences being Nd magnet and more copper in the motor to reduce distortion at higher listening levels - which is kinda intuitive since more copper in the motor means that magnetic gap has to be bigger and then you need larger magnet to retain the sensitivity and i don't think that they could make bigger Fe magnet on that midrange than what it already is hence Nd - additional benefits being less sound reflexion etc.). Difference is also in the cabinet - Ref 1 has aluminium front plate that is not rigidly coupled to the cabinet and presents a continuation of the waveguide for tweeter dome and midrange cone (which is the main reason for seeing some diffraction products in my tweeter measurements between 2-4KHz and there aren't any in Ref 1).

Few days ago i ran into an interview with Jack Oclee-Brown where he spoke about Reference series. Being asked what is his favorite loudspeaker of all Kef's UniQ's he answered Ref 1. Reason is that he is still surprised when he listens to it what comes out of that loudspeaker regarding it's size. I also ran into few very good reviews on the web and on one German website with quite good measurements of Ref 1.

https://www.fidelity-magazin.de/2016/08/23/kef-reference-1-messungen/

As can be seen, they measured it's frequency response (on axis only, but they provided polar response - both horizontal and vertical) and they provided bass frequency response and mid-hi frequency response separated. What surprised me was that at 200-250Hz you can easily spot a "knee" that is a product of resonance of closed midrange chamber. That "knee" is only ~10dB under the fundament. Since i'm trying to run as fast and as far away of any resonance i can see, i was a bit surprised that they left it there since it can't be pleasant for ears. Then again, when i reread reviews i didn't notice any bad comment that could imply that resonance there is noticeable. So i decided to check it for myself and see if it is really that benign.

Crossover demanded a bit of rework to lower crossover frequency between bass and mid so i came up with something like this:




Then i played some music through it. All in all it wasn't so bad but something bothered me that i couldn't pinpoint what. Then i played some music with piano in it (Esbjorn Svensson trio to be precise) and it behaved as awful as i expected it to be. Phase tracking is quite good, fr is linear and without aberrations on and off axis but the music was unlistenable when it hits a note or two. I've let it a day and then put the "old" crossover back.

I tried to find Kef Ref 1 here to compare it to my modified speakers, but there aren't any. I found Blade and Blade II, every other loudspeaker from Q and R series but not one from Reference series (which kinda implies that middle-class is dead but that's already another story.)

My guess still is that driver/chamber resonance combo was too close to the crossover frequency and too high in level to be ignored. I've posted this to check if anyone can give another perspective to this which could lead to a different conclusion.

[Juhazi]

Was there too much distrotion on the mid driver, with low xo?

Another thing that you could try is LR2 MT xo. The mid might need a notch or two, but i like LR2 much more, specially for piano music!

[Zvu]

Midrange can go as low as 200Hz without being stressed.

Though it may be that the sound with LR2 MT crossover would be different, i don't believe that is the problem here.

[Dave Bullet]

Do you know which particular piano notes were causing the problem? Can you isolate which driver is contributing more?

I am having problems with a Seas 27TDFC/G tweeter on certain tracks. I can blame bad engineering to a point, but certain instrument types, or the track has a very forward upper midrange / lower treble. It's almost as if someone has removed the L-Pad on the tweeter. I can then play other music and the balance is restored.

Annoying.

[Zvu]

If you compare the crossover from post 40 to the one in post 49, bass crossover is marginally changed. The only thing that is really different is bass/mid crossover point and midrange slope. Crossover from post 40 works great by itself. There are recordings where it doesn't work (in which case i know the recording is to blame so i don't worry too much) but EST i have is good and it doesn't sound bad on any decent loudspeaker.

[Dave Bullet]

It's good to have a reference recording that you know should sound great. Do you have a particular track from the Esbjorn Svensson trio that is a good example of the issue (and time from start of track?) Selfishly, I'd be keen to try these on my 3 ways to see how they sound.

I suppose post #40 has a tapered slope and no peaks in the sensitive 1-3KHz area whereas, #49 crossover is flatter and has those peaks.

I don't know whether it is crossover slope / position or the peaks / flattness. How does #49 look off axis (you did say no issues there)? or is the EST recording still troublesome regardless of on or off axis listening?

[Juhazi]

ECM records are now on Spotify, including EST. I have many CDs by Iiro Rantala, and piano sound is different on everyone, because of differrent pianos, halls and obviously also mic locations. Jazz piano is voiced typically more nearfield than classical piano. In my taste, most ECM records have unnatural piano sound, too much like the head inside of the piano!

[Zvu]

@Dave: Waveguide loaded speakers i use toed in - i point the left speaker to my right ear and right speaker to my left ear. It works the best with my room.

For evaluation i start with something like Allan Taylor (The traveler) or even Alison Krauss (Down by the river to pray - soundtrack from "O brother, where art thou?") - great recordings, and few more to see how male and female voices sound. Then i move to EST (only three instruments yes, but piano, contrabass (playing both arco and pizzicato) and drums - so quite revealing), then i play more complex (Dave Bruebeck - Take five; Chuck Mangione - Children of Sanchez) etc. then to classic music and i finish it with Daft Punk - Random access memories, checking how it sounds with modern music.

With Seven days of falling i had problem with what left hand does on the piano, not the right. EST-Tuesday Wonderland had its moments. Play it and you'll realize that piano is heard almost through the entire song so you can't pinpoint exact moment or part of the song that is troublesome. Contrabass, when played arco (with a bow) can be a little forward sounding, but that's how they wanted it to sound. It doesn't present an issue but it is on the verge so it keeps you on your toes - i like that feeling very much

Bare in mind that what you are looking at is 1dB resolution so it looks worse than it actually is. Especially if you move 10 deg off axis when products of diffraction disappear. Bad experience wasn't dependent on listening angle and you hear that it's lower in frequency.

@Juha: As you said it: it is a matter of taste. I quite enjoy jazz piano like the one in EST recordings.

[sdl2112]

Nice work Zvu, thanks for sharing your design and measurements.

[Zvu]

Not as nice as your loudspeakers and i'm not saying that out of politeness. Thanks, that means a lot, coming from you.

You've linked your thread few months ago and, though it is a bit longer, i read it in one breath. It may just seemed longer to me because of problems you've ran into. That level of persistence when you've discovered that lead sheets aren't glued good enough left me in awe. I don't know if i could find will to continue as dedicated as you did.

[sdl2112]

Thanks for the complement....really...thank you. It has been a very rewarding journey.

[Dave Bullet]

My 3 ways are way to forward / bright on Brubeck's take 5. The sax is just too in your face and dominates . Is it meant to? It does however have a nice weight to it. The sax i not all treble / top end which is good. I am blaming the tweeter using the pillow covering test

[Zvu]

Sax is in your face but that song to me is about bass weight and resolution. If there is a little more or just a little less bass than it should, it is easily heard. Also, you have to be able to hear the pads opening and closing the holes and the air going through them when sax plays. The drums are quite revealing too. Hardly any time hitting it sounds the same.

[Zvu]

I'm jealous now

http://www.audioxpress.com/news/kef-announces-new-completely-revamped-r-series-loudspeakers

[Zvu]

Crossover tweaking is nearly completed. I didn't measure both loudspeaker cabinets but i have just now checked for production consistency with indoor measurements. There is slight difference between the midrange drivers frequency response.




This lead to slightly different crossover for midrange - on one loudspeaker series midrange coil is 1mH and on the other it is 0.9mH. Both loudspeakers now measure the same but it's always a good thing to check.

[Dave Bullet]

Crossover tweaking is nearly completed. I didn't measure both loudspeaker cabinets but i have just now checked for production consistency with indoor measurements. There is slight difference between the midrange drivers frequency response.



This lead to slightly different crossover for midrange - on one loudspeaker series midrange coil is 1mH and on the other it is 0.9mH. Both loudspeakers now measure the same but it's always a good thing to check.

Interesting and a good point to note - matching pairs of drivers is one thing - but matching acoustic slopes is what we are trying to achieve.

What explains the 5KHz breakup node on one driver and not the other? Is the crossover sufficiently attenuating that without requiring a specific notch?

[Zvu]

In an interview, with John Atkinson, Jack Oclee-Brown explains - voice coil of their drivers isn't rigidly coupled with the cone. They are interconnected with a damping ring that is incompressible at lower frequencies but acts as a "shock absorber" for frequencies above 5KHz to minimize the breakup. You and i know how the breakup looks in L15RLY but here with Kef coax there isn't any.

I had my suspicions about damping system of that sort because that fine details are dependent on compressibility of that ring and on amount of glue applied. It does a great job - I used only one small capacitor to bypass series mid coil but even that isn't necessary. It's just something i usually do to get rid of any hint of breakup.

I'll post sim with and without that capacitor later.

[Zvu]

Here it is... With a 0,33uF cap and without.

[Zvu]

I've been listening these speakers past two weeks and tweaking a bit. Nothing dramatically changed but few details for better phase matching and a bit more linearity. I'm at the stage of putting the crossovers into the cabinets so here's few screenshots:

Midrange in phase:





Midrange out of phase:





Simplistic crossover:





"Exploded" view:

[ergo]

Moving GIF is very nice for visualizing the effect, good idea :T

[Zvu]

Thanks Ergo. When i take them outside to measure them 0-180 in and out of phase i'll make a gif and post it here. It's so much more looking at that then dead curves one on top of another. I think that one curve at a time gives better perception of what's actually going on.

Anyhow, here's how the crossover will look like - few resistors missing but soon to be acquired and all electrolytic caps are going to be bypassed with 0.1uF mkc.





Now, i got quite a deal with these speakers because of finish damage it has. Does anyone have any idea how to fix this ? I can always apply some epoxy resin, sand it and repaint it (probably white, red would look cool also i think).

Lousy phone camera but here it goes...

[tecnogadget]

Zvu congrats for such a great job! This thread is quite interesting.
I have always been curious about the R series internal bracing build quality, especially the "constrained layer damping pads" KEF advertises in their own website and brochures. I could not find anything else than just the computer render...do you have any pictures of the inside of the cabinet when you took apart the crossover?

[Zvu]

Zvu congrats for such a great job! This thread is quite interesting.
I have always been curious about the R series internal bracing build quality, especially the "constrained layer damping pads" KEF advertises in their own website and brochures. I could not find anything else than just the computer render...do you have any pictures of the inside of the cabinet when you took apart the crossover?

Pics of inside that i have i'll post here. I can make a few more before i finish the loudspeakers. Maybe there are some CLD pads in R500/R700/R900 but in R300 there were none that i could see. Maybe on the upper side above the coax because there is nothing on the back panel (BR and terminal plate uses most of the back panel), nothing on the sides and bottom panel. No CLD strip on braces.

[Juhazi]

Obviously only the revised R series has more of CLD etc. , total of 1043 modifications! http://us.kef.com/media/doc/r2018_whitepaper_en.pdf

https://www.audioxpress.com/news/kef-announces-new-completely-revamped-r-series-loudspeakers

[Zvu]

I bet that they added different paint job as a modification (:

For 800€ more per pair there are a few that are worth mentioning as - redesigned magnet for woofer and cone, shadow flare, flexible port tube design, decent cld and more quality series coils. Everything else is simply marketing department mumbo jumbo.

[ergo]

The box cut image has these funny "Computer designed 3way crossovers" texts... I would hope that they are still "Computer aided design by a person" type of things as I would not yet trust a computer to make a good sounding job of it ... yet.... until of course they start throwing "machine learning" on the task and it makes wonders and no-one knows how.

[Zvu]

I've made some more measurements last night to see the agreement of simulator to measured response. There is a slight upward tilt toward high frequencies but that could be due to cold (6 degrees Celsius last night). By the looks of it, if it is real, the tweeter is a 1.5-2dB hotter but final level is to be determined indoors by listening. So, without further ado...

On axis:





Off axis 10 and 20 degrees




Up to 45 degrees





The crossover work is pretty much done so for the winter i'll be doing nothing but listening music and maybe a little tweaking by ear. This was very enjoyable experience

[Dave Bullet]

What sort of music do you listen to? Whats the "worst" sounding track?

[Zvu]

I listen pretty much everything.

The worst soundtrack is Crazy Mery - Big head Todd, without a doubt. It's all great but then at 02:05 just mayhem.

https://m.youtube.com/watch?v=Z8wI10LscDU

But then again i never heard it sounding good.

[Dave Bullet]

Thanks. Sorry I remember you posted some tracks above. I'll give that a play on mine and see what happens ��

[Zvu]

It looks alive

0, 10, 20, 30, 60 and 90 degrees.

https://postimg.cc/FdwdNkkL

[Bear]

Measurement boogie!

[Zvu]

Yup, it's kinda boring when they don't dance

These measurements are done indoors so not so pretty but i had to check the level of tweeter after adjusting. It is a bit better but i still get that off-axis peak at 8KHz. I'm trying to avoid notch if i can. I'll listen to it like this and we'll see. The only thing different is 2ohm series resistor on tweeter instead of 0.5 ohm.

[technodanvan]

Zvu - just read through your whole thread again. Thanks for all the input and detail in this endeavor! I've been considering a DIY variant of the R9 or R11, possibly using the same coaxial Jon did for his Minervas, so it seems a number of us are kind of going down this coaxial path. It's nice to see we are beginning to have a few decent options on the DIY front, though honestly I am tempted to just try the KEF itself as it looks competitive for the price.

Out of curiosity, what do you use to make the moving graphs?

[Zvu]

Hi Dan. Thanks for kind words.

I save frequency response pics from HolmImpulse and upload them on gifmaker.me

I mark the pics from 1-6 (in this case) and upload them in next order:

1234565432

I set the time betwen changes 120-180ms depending on number of pics. Voila.

[Juhazi]

GIF animations can be made with PhotoScape too, that's what I use!

Problem is that animations don't work on every intenet forum, like diyaudio.

[Zvu]

Here are some pics of R300's insides.

The only trace of what i presume they called constrained layer damping is this patch on the upper side of the loudspeaker, above the midrange chamber. Cabinet is upside down and i took these pics through woofer hole.









Here are integrated connectors and port. Port looks good, but the -link- buttons are something I'm going to skip.










When you turn the link thingy, a metal part slides and touches small PCB and connects mid-high and woofer crossovers in parallel. I'll probably install L-pad's on that plate or I'll use some jumpers and link on - just to be sure.

Now here's the highlight of the loudspeaker, a little concentric driver. Maybe not the best pics, but one can easily spot some of the design features.













Oh, and grill looks quite nice too. It is made out of MDF with small Nd magnets at the corners. Cool stuff.






I really like gloss Kef sign on the back. It is usually just black so it's nice that somebody thought of that.

Well that's it. I'm gonna put some bituminous coating on the outside of the midrange chamber and send the cabinets for painting.

[mordikai]

looks like these are on closeout for $1299 a pair.

[Zvu]

I'd say that 1300$ is not too shabby knowing what you get for that amount of money.

Friend asked me if i could make a crossover that's a little easier on the amplifier so i made one with higher impedance in lower midrange - 3.3 ohms to be exact.




I'd try it myself but there's no need to since my amp is ok with what i did before, but for the ones that are interested in modifying their speakers and don't have 2 ohm stable amps i think it'll be quite satisfactory. Any 4 ohm amp will do the job.

For the one's that want to use impedance flattening circuit, lower midrange doesn't go below 3 ohm. L6 is 0.1mH whatever wire gauge you find cheapest. C10 is not critical at all - anything from 60-80uF will work without any important difference. Take two 10W 12 ohm resistors, connect them in parallel and use it in this notch filter.




I'm still waiting cabinets to get back from painter. Either he is a perfectionist or he's lazy. We'll see in a few days i guess. About the color i chose - i was inspired by one of the projects on this forum. I'll snap some pics when it's done.

[Zvu]

It's been a while since i've addressed the sound of modified Kef R300.

When the initial impressions of sound settle, then comes that long lasting feeling you get when you get to know your loudspeakers very well. Kef did an amazing job with these drivers, well almost. Almost everything sounded the best ever but i will not write about qualities of this loudspeaker but more on the weak spots. R300 when modified has two of them. Low bass - it's there but not really. In my room it worked perfect but when put in larger rooms 40sqm or more, it needs help - and i'm talking about hard walls, not wood or drywall - i'm guessing it would behave worse in that situation. Other one being low mids. Concentric at the top does such a good job that i wonder what could sound better than that. When Kef says -obsessed with high resolution- they really mean it. It's one of the finest midranges and tweeters i have ever heard, and i've heard quite a few loudspeakers. Woofer is crossed at about 450Hz to midrange, and there problems begin. I'll post a picture оf its construction first:




It's made by attaching large aluminium plate to a paper cone that is connected to voice coil former. In the paper cone there are elliptical openings to let the pressure out. Something similar can be seen in typical car subwoofer units. When it comes to bass frequencies boy does it deliver. It is one of the most punchy bass drivers i've laid my eyes (ears???) on. It would be almost no compromise loudspeaker if midrange could be crossed to it at say 120-150Hz but pushing it to 400-500Hz has taken its toll. Since i made quasi-anechoic measurements with almost 11ms gate time, i get very good resolution from about 90-100Hz. I've made the flattest measuring loudspeaker (on and off axis) up to about 2KHz i've seen. But something was missing. i first thought that midrange was a bit quieter than it should be so i reduced the attenuation in 0.2dB increments to set its level by ear. That's wasn't it. Then, after some time i started to look at the crossover frequency between midrange and woofer. With few slight adjustments i moved it to 380Hz - it was better. Then i moved it to 310Hz, even better still. All the time i kept linearity and good phase tracking. The more i let the midrange work, the better the rendering of that thing missing was. Then i just played some music i know well and got really close to loudspeaker listening to midrange and woofer up and down. Every tone that wasn't pure bass frequency (say 80 and up) it played virtually the same. Every note sounded the same. They've made it work well and deep enough at low frequencies but they've damaged the transients of lower midrange on the way. I guess the price had to be payed somewhere and it was right there where it hurts the most. That couple hundreds hertz it should be playing it doesn't do well.

Knowing that there were couple of options. To buy used Q300 and use its 6.5" woofer. I'd remove the tweeter and make a phase plug. But that's an investment of more or less 300-400 euros and leaves me with having to measure new trim rings dimensions and have them made out of plastic or aluminium, new crossover etc. Color of Q300 woofer looks really ugly when combined to concentric midrange in R300 so i gave up. Next thing would be trying to find a 4 ohm woofer that will work as well as factory one in about 15 liters of volume. That was cheaper option but i haven't found it.

Anyhow, i'd strongly recommend concentric driver from R series as it really sings but to use it with another woofer that will not absorb transients on the way. If you're listening electronic music, you'll never want for another loudspeaker. Couple of my friends does so i sold my R300's to one of them.

This was a very good experience. I've learned a lot and it showed me that Kef is a force to be reckoned with when concerning their concentric drivers.

cheers

[mihaiil]

Hi guys and happy new year! I see you have knowledge about R series of the KEF. Short story, I bought r300 recently and shortly after, one of the woofers started to make strange noises. I opened it to see that the spider was starting to came off of the voice coil. I opened the other one too just to see if everything is ok, and sure enough it wasn't. The previous owner did a terrible job by gluing the woofer back to the voice coil with hot gun glue.



That wasn't the problem, I cleaned the glue(it came out easy whit isoprophyl alcohol), but I'm concern that when I look in internet for bass driver for the R300, I see different part number and they are 8ohm (but there is a R300 kit, where the woofer is like mine SP1637 (6,5" 4ohm). So which is correct? Does the old owner swapped the woofers or these are original? I see that r700 use same bass drivers, but probably different impedance, since they are 2.
Can someone tell me for sure if the bass driver of KEF R300 is 8 or 4ohm. Thanks guys!

[Zvu]

In Kef R300 it is a 4 ohm woofer.

There is measured impedance inbox raw for all three drivers in the first post of this thread. Also, TS parameters are measured with Re value of around 3.2 ohms.