DIY with Kef and Celestion

Status
Not open for further replies.
Whew, all this tweeter filter discussion is very tedious and missing the point. This is DIY and you try simple stuff and refine stuff later if not happy. Perfectly easy to make little tweaks later, for, say, a rising top end.

The main thing is to have a circuit that basically WORKS. Flat response, good phase alignment. Good rolloff before the nasties creep in. 🙂

The BEST filter topology I have found for 200mm paper bass uses a 4th order tweeter filter and a steep crossover around 2.8kHz. Phase alignment is a dream, so imaging will be good, and it does about the best you can with the bass cone breakup, which is the bugbear of all 2 ways.

But this KEF/Celestion 3rd order idea is pretty good. I know because I have built it. The smaller coil does good things IMO. Very direct and clear sound on my 3.5kHz crossover. Just a tidge of bass breakup, but not offensive, and better than a simple coil and tank on the bass. Impedance is good too, never below 7 ohms.

BTW, DrBoar, this is the paper Elac sourced B200 SP1070, not the bextrene SP1014, but the pictures seem to confirm our crossover ideas. I see two coils and 3 capacitors, and maybe a resistor.
 

Attachments

  • KEF B200 SP1014.JPG
    KEF B200 SP1014.JPG
    48.6 KB · Views: 539
  • KEF B200 SP1070.JPG
    KEF B200 SP1070.JPG
    19.1 KB · Views: 441
  • KEF Celeste III crossover.JPG
    KEF Celeste III crossover.JPG
    42.2 KB · Views: 423
  • KEF Celeste III.JPG
    KEF Celeste III.JPG
    23.2 KB · Views: 1,372
vans60, I don't know how your own work is going this weekend, but I have been simming some other things. 🙂

I wondered if we could get a 3kHz crossover on a 1.5mH coil, since I have another 200mm paper bass speaker awaiting some work. The drive unit just happens to be an Elac sourced one like yours.

Below is how it models in Visaton Boxsim. I like the look of that. To allow for a smaller Le (1mH?) bass unit, I think the 8R shunt (actually a sort of damping resistor) becomes about 4.7R and that is where I would start.

The 4uF tweeter capacitor controls the crossover frequency. 3.3uF is probably too small, 4.7uF is a bit peaky.

The 4.7uF bass shunt capacitor looks small, but the 33R/0.33uF tank (actually a 7kHz notch) helps rolloff and phase alignment too. 2dB @3kHz, 8dB @7kHz. It works better overall than a single 6.8uF capacitor.

The 1R input resistor is select on test. I'd guess I won't need it at all with an 88dB tweeter.

I would expect this to be a slightly better speaker than my current 3.5kHz build. Less cone breakup, but still quite efficient.

Salut.
 

Attachments

  • KEF_Celeste_Type_3kHz_Phase.PNG
    KEF_Celeste_Type_3kHz_Phase.PNG
    23.3 KB · Views: 814
  • KEF_Celeste_Type_3kHz_FR.PNG
    KEF_Celeste_Type_3kHz_FR.PNG
    20 KB · Views: 889
  • KEF_Celeste_Type_3kHz_Filter.PNG
    KEF_Celeste_Type_3kHz_Filter.PNG
    7.3 KB · Views: 1,064
Whew, all this tweeter filter discussion is very tedious and missing the point.
This is DIY and you try simple stuff and refine stuff later if not happy.

Hi,

No you seem to be missing the point. You use all the information
available as a starting point and take it from there. Alternatively
you can start from scratch using a shedload of assumptions
and hope it turns out alright, and then run around in circles
if its pretty OK but you want to try to make it any better.

Your simple stuff is too simple, and nothing follows from it.
You can't talk about crossovers as being generically good.
You can't talk about phase alignment of a x/o without
detailed frequency response (hence phase) of the drivers.

I'd advise anyone who thinks by whatever method they can
build speakers with random drivers and no measurements
to build a fully documented and measured design and then
compare their musings to a decent reference design.

rgds, sreten.
 
Last edited:
Sighs...let's talk about the KEF Celeste III filter in a way that a simple DIY enthusiast can understand. 😱

The reason it is GOOD is that it has very good phase alignment and bass cone breakup suppression above 3kHz. Phase alignment is extremely difficult with a 200mm bass on a flat baffle. This is because the bass acoustic centre is set back about 5cms relative to the tweeter, which just happens to be half a wavelength at 3kHz. 😎

Flipping the polarity of the tweeter works here wonders. This filter, IMO, is a step up from the more common 1st order bass with first or second order treble arrangement. Modifying it for different drivers is really not that hard. I'd say my Visaton equivalent with a modern louder ferrofluid tweeter does much the same things. As vans60 has noted, it is however rather inefficient and bassy. That, along with matching the different tweeter, is what this thread is addressing. 🙂
 

Attachments

  • Orig_Celeste_Filter.PNG
    Orig_Celeste_Filter.PNG
    6.6 KB · Views: 165
  • Orig_Celeste_FreqResp.PNG
    Orig_Celeste_FreqResp.PNG
    19.3 KB · Views: 155
  • Orig_Celeste_Phase.PNG
    Orig_Celeste_Phase.PNG
    23.4 KB · Views: 149
Hi,
I tried a lot of combinations this week end and i agree with Sreten : it's very difficult to obtain a good result. But, like say System7, when it's not good you retry😉It's long and sometimes the differents are inaudible😡
I have a good sound (surely not perfect) with this diagram.
I try, for the B200, with and without the R+capa and it's better with. The mid is less "projected" but still a good presence. The sound is more "flat". With the 3 order for the tweeter, the speakers are singing more "together".

I'm aware it's a light process for build a good speaker but now it's sound relatively good for me.
I try to post some photos of them and of the crossover when i have the courage😀

Thanks a lot for all your contributions to System7 for your patience and comprehension with a begginer like me.
Thanks a lot too to Sreten for your technical cautions and advices even if you don't approves this method without accurate informations😛
On the futur, a friend will instal all the equipement for measuring and i will not miss to ask you a lot of questions😱😀

Cordially.
 

Attachments

  • 2013-11-15_111457.jpg
    2013-11-15_111457.jpg
    16.6 KB · Views: 171
That is similar to the circuit I am running on my Monitor Audio speakers:
http://www.diyaudio.com/forums/multi-way/203461-restoring-monitor-audio-r300-bookshelf-speakers-4.html#post3673721

An externally hosted image should be here but it was not working when we last tested it.


The 10uF capacitor doesn't really do much except give the (1 kHz) tweeter Fs resonance a bit more protection at low frequencies, but is a nice touch. I used a mylar tweeter here because it has the time alignment I wanted. A soft dome tweeter is a bit more difficult. But actually if you set it back just 1cm or listen in the right place, it all lines up very nicely.

I also found that the Visaton TW70 cone tweeter was a slot in replacement for the DT94 with a very nice sound. Monacor do one similar.

What is interesting, is that if you get the phase lined up as well as you can, the drivers sound very INTEGRATED behind the grille. You can then pick out the sound of bass cone breakup very easily. It sounds a bit harsh on male vocals.

But that's the nature of this speaker. That's about as good as you can get it. If you want better, you'd build a three way or might try a polypropylene coned bass like Morel or Vifa which has more rolloff. I have enjoyed this thread, Bon Chance. 🙂
 

Attachments

  • vans60_sim_TimeAligned..PNG
    vans60_sim_TimeAligned..PNG
    13.7 KB · Views: 89
  • vans60_sim_phase.PNG
    vans60_sim_phase.PNG
    19.5 KB · Views: 108
  • vans60_sim.PNG
    vans60_sim.PNG
    19.3 KB · Views: 105
Hi everybody, i'm back.
Last friday, a friend have measure my speakers.
It was the first experience for me and it's really interesting.
I put one speaker down to him and he'll can work about the crossover for me😉
On the photo you can see two curves : a bright and a dark.
The dark represent the tweeter out phase and the bright on phase.
Incredible isn't it ? This guy is really proficient. He has heard this problem immediatly and propose the good solution before measuring.
See you.
 

Attachments

  • courbe enceinte david.jpg
    courbe enceinte david.jpg
    66.4 KB · Views: 129
Glad you're still interested in this project. I'm not sure what you're saying there. What filter are you using?

This is where I ended up with that KEF filter using a Monacor HT22/8 tweeter. Wired in negative polarity. I like it. 🙂
 

Attachments

  • Modified_MA-R300MD_HT22-8.jpg
    Modified_MA-R300MD_HT22-8.jpg
    79 KB · Views: 188
  • Modified_MA300-MD.PNG
    Modified_MA300-MD.PNG
    7.1 KB · Views: 208
Glad you're still interested in this project. I'm not sure what you're saying there. What filter are you using?

This is where I ended up with that KEF filter using a Monacor HT22/8 tweeter. Wired in negative polarity. I like it. 🙂


Regarding my other post that you replied to: What is the difference in sensitivity of the two drivers that you're using? I'd like to try this circuit but my drivers are very close in sensitivity (~.5dB).
 
The Monacor HT22/8 tweeter is about 90dB sensitivity AFAIK. Which makes it fairly middling as these things go. Probably on a par with a Morel CAT 308.

If your tweeter isn't keeping up with the sensitivity of the woofer, you increase the bass coil usually. But it's obviously hard to fix a total mismatch. My bass is quite efficient.

As I said, I'd need to see some data on your Focal bass to be more exact. But I'm already attenuating the tweeter quite a bit, so I think you could get this working.
 
Last edited:
Regarding my other post that you replied to: What is the difference in sensitivity of the two drivers that you're using? I'd like to try this circuit but my drivers are very close in sensitivity (~.5dB).

Having given it some thought, I'd guess my woofer is 88dB efficiency. I do use a higher efficiency Eminence style woofer in another speaker. This gets the following crossover to help the tweeter keep up:

An externally hosted image should be here but it was not working when we last tested it.


I subsequently added 22R shunt to adjust tweeter level. The second tweeter capacitor gets a bit bigger without the input resistor, which is just how it all works. All quite close to the 5uF and 7uF caps the KEF filter started with. The bass shunt doesn't need such a big resistor because the woofer inductance is low. Impedance doesn't fall below 5 ohms and it's above 7 ohms most everywhere. Not a fussy circuit at all. I adjusted mine by ear.

Ah, correction to the circuit below, I'm currently using 3.9uF and 6.8uF on the tweeter filter after some experimentation. Sorry about that.
 

Attachments

  • KEF_HighEfficiencyFilter.PNG
    KEF_HighEfficiencyFilter.PNG
    7.4 KB · Views: 255
Last edited:
Hi
system7 say :
"Glad you're still interested in this project. I'm not sure what you're saying there. What filter are you using?"

Of course I'm still interested😉 The measure are with the filter on the first photo
On the second photo you can see in black the kef without x/o, in pink the kef with my x/o, in red the entire speaker, in mauve the tweeter with x/o
But there's some news with the help of my friend.
On the third photo you can see some inductance trying
On the fourth the new x/o
On the fifth the measure with the new x/o
I'm not listen to them yet but i'm impatient
And i'm impatient too to have your opinion😀
Sreten will be happy with this work😱
 

Attachments

  • 2013-11-15_111457.jpg
    2013-11-15_111457.jpg
    16.6 KB · Views: 233
  • filtré et non filtré.jpg
    filtré et non filtré.jpg
    72.5 KB · Views: 205
  • mêmes courbes fenêtrées.jpg
    mêmes courbes fenêtrées.jpg
    55.1 KB · Views: 219
  • Filtre Kef-Celestion.jpg
    Filtre Kef-Celestion.jpg
    19.1 KB · Views: 147
  • enceinte david 30cm axe rouge courbe non fenêtrée-noir fenêtrée.jpg
    enceinte david 30cm axe rouge courbe non fenêtrée-noir fenêtrée.jpg
    59.1 KB · Views: 115
vans60, I think 1.5mH is right. But that notch is better at 7kHz (22R and 0.47uF tank with a 1.2mH coil), not 5kHz I know now. But actually I have abandoned it altogether these days in favour of an RC shunt. 🙂

Like this: http://www.diyaudio.com/forums/multi-way/245917-diy-kef-celestion-5.html#post3865139

You can use 3.9uF /4.7uf like me, or 3.9uF/6.8uF in the tweeter filter for an easier impedance for your amplifier to drive. But wire the tweeter in negative polarity. This is the real KEF trick. 😀
 
Status
Not open for further replies.