Kef 104 Crossover help.

[Mattias l]

Hi guys I'm doing a restoration of my girlfriends old Kef 104, the driver units are workin fine so I will change the internal wire and Capacitors and inductors and move it to a bigger wooden board because the Auricaps where much larger then the original.

I'm not good with reading shematics so I hope you guys can help me with that, when I look on my crossover at home I can see 3 capacitors and 3 inductors, but when I look on the schematic I got from internet one of the inductors have 3 frequencies? what does that mean? I've bought the capacitors now and will change the inductors in the future but I need to know what to buy

My first post here and I'm from Sweden so I hope you will have som understanding if I spell things wrong.

Happy New Year.
/Mattias

[Salas]

Welcome. Don't change the coils, in that category of commercial builds their (high enough) resistance was part of the correct crossover target analysis. The reasons are industrial economy. You can measure their DCR and order same DCR air core the most.
Maybe that woofer coil was multi tapped to be used with other models in the family. Keep it connected as originally. The Auricaps are worthwhile almost certainly.

[alan-1-b]

Hi Mattias,

keep that tapped inductor, at least till you hear the 104s with the new caps installed.
There should be a 3 position switch on the input terminals board on the back of the cabinet.
If it is there, it is connected to the tapped inductor.

Listen to all 3 positions of the switch after you have the new caps installed.
You may decide after extended listening to a lot of different types of recordings that you prefer one of the switch positions to the other two.
If so, then that is the one to change the inductor to.

For the inductor in the treble circuit, any new one must have the same DCR.
You can use lower DCR inductors in the bass circuit if you want a bit more bass output and slightly faster bass response,
but then you should add a resistor in Series with the capacitor in the bass circuit, or you may hear an unpleasant resonance in the lower midrange.

First listen with the new caps, then Post your findings here if you would like some follow-up advice.

[Mattias l]

Ok thanks both of you guys.

I will change the caps first, do some listening and then report here how it sounds.

/Mattias

[planet10]

Kef's improved HP filter for 104.

http://homepage.mac.com/planet10/TLS...Kef104abXO.pdf

dave

[alan-1-b]

Hi planet10,

I'm not convinced that Kef's ab version is the crossover is the best that can be done.
Audibly better may be able to be done with different, and perhaps with less, components.

I do not doubt the on-paper technically more correct response shape of the filter plus driver, but on-paper does not always equate directly with audible effects.

Ref. to Kef's Fig.2:
I think the middle diagram: "Acoustic Butterworth Circuit" may sound better than the "Practical Realization of ... Circuit".

It was clever to apply the transformation for the alternate circuit of the three capacitors, and that does result in smaller capacitances,
thus likely why Kef used that circuit,
and with the capacitors available at that time I have no doubt it sounded OK, but it may not now with modern polypropylene capacitors substituted.

For the middle circuit, C1 is 4.1uF; C2 is 12.4uF; C3 is 68.3uF.
At that time 68uF was not available in a reliable capacitor for critical audio-signal application, and the 68uF equivalents that were available did cost a lot more than the 0.6uF for the "Practical Realization ... Circuit", thus I think likely why Kef chose the latter circuit.
As regards accurate sound reproduction, I am suspect of the audible effect of the 0.6uF cap, because it will cause change to the leading edge of high frequency waveform, and particually so with a modern low ESR capacitor.
The old 0.6uF cap likely had sufficient ESR to slow down the rise time of the signal to similar rise time as through the 3.3uF and 10uF caps to not result in leading edge tizz or slight sibilence to the sound.
OK, some listeners like that change to the onset of a transient, and add 0.1uF bypass caps to their treble filters to achieve it,
however other listeners do not like that effect, and it is a sound-effect not present in the original waveform.
I don't like the sound it causes.

Now we can make a filter with 4.3uF; 12uF; 68uF from modern MKP type caps, and when I have time -{hopefully during this year}- I will try that and listen, as I have original Reference 104 with the original crossover.

Also, I have another idea to try for a slightly simpler circuit - I will post about it after I have tried it.

***************

Mattias,

did you buy 4.3uF and 5.1uF capacitors ? - or several smaller capacitances to connect in Parallel to sum to 4.2uF and 5uF ?
- and if so, what capacitance sizes did you buy ?

[planet10]

Originally Posted by alan-1-b

Hi planet10,
I'm not convinced that Kef's ab version is the crossover is the best that can be done. Audibly better may be able to be done with different, and perhaps with less, components.

I wouldn't argue with that, but the aB xo was definitely better than what KEF used before.

dave

[speaker dave]

Originally Posted by alan-1-b

As regards accurate sound reproduction, I am suspect of the audible effect of the 0.6uF cap, because it will cause change to the leading edge of high frequency waveform, and particually so with a modern low ESR capacitor.
The old 0.6uF cap likely had sufficient ESR to slow down the rise time of the signal to similar rise time as through the 3.3uF and 10uF caps to not result in leading edge tizz or slight sibilence to the sound. OK, some listeners like that change to the onset of a transient, and add 0.1uF bypass caps to their treble filters to achieve it, however other listeners do not like that effect, and it is a sound-effect not present in the original waveform.

I don't like the sound it causes.

Sounds like Pseudo-science to me.

Not buying the slow waveform arguement or effects causing "tiz or sibilence". Since both the star and the delta configuration are equivalents you can substitute either with the same sound qualities. Tiz or sibilence are effects well inside the audio band and the network gave just the performance desired in the low treble regions. The fact that the circuit gave the desired response is proof that there were no significant parasitic effects in place.

You should know that I was a KEF engineer shortly after this unit was designed and can vouch for the accuracy of computer modeling at KEF, and that every 0.5 dB effect not expected (such as might come from parasitic component effects like ESR) was painstakingly tracked down. Note that even when this unit was made, a capacitor value this small would have been a film capacitor. Even if not, lower ESR capacitors have been around for the 30 years I've been in the industry.

The configuration was used because the component values made more sense than adding 68uf caps needlessly. Good engineering and a competitive market would preclude using more expensive crossover components than necessary.

This particular T network isn't widely used today, but KEF should get credit for being the first to address the issue that it isn't the electrical filter shape that we should be pursuing, but the total acoustical shape of filter and driver together.

Pulling response down in the vicinty of the tweeter's resonance will have a much greater positive sound quality effect than any supposed crossover component signature.

David Smith
Former KEF Engineer

[Mattias l]

Originally Posted by alan-1-b

Hi planet10,

I'm not convinced that Kef's ab version is the crossover is the best that can be done.
Audibly better may be able to be done with different, and perhaps with less, components.

I do not doubt the on-paper technically more correct response shape of the filter plus driver, but on-paper does not always equate directly with audible effects.

Ref. to Kef's Fig.2:
I think the middle diagram: "Acoustic Butterworth Circuit" may sound better than the "Practical Realization of ... Circuit".

It was clever to apply the transformation for the alternate circuit of the three capacitors, and that does result in smaller capacitances,
thus likely why Kef used that circuit,
and with the capacitors available at that time I have no doubt it sounded OK, but it may not now with modern polypropylene capacitors substituted.

For the middle circuit, C1 is 4.1uF; C2 is 12.4uF; C3 is 68.3uF.
At that time 68uF was not available in a reliable capacitor for critical audio-signal application, and the 68uF equivalents that were available did cost a lot more than the 0.6uF for the "Practical Realization ... Circuit", thus I think likely why Kef chose the latter circuit.
As regards accurate sound reproduction, I am suspect of the audible effect of the 0.6uF cap, because it will cause change to the leading edge of high frequency waveform, and particually so with a modern low ESR capacitor.
The old 0.6uF cap likely had sufficient ESR to slow down the rise time of the signal to similar rise time as through the 3.3uF and 10uF caps to not result in leading edge tizz or slight sibilence to the sound.
OK, some listeners like that change to the onset of a transient, and add 0.1uF bypass caps to their treble filters to achieve it,
however other listeners do not like that effect, and it is a sound-effect not present in the original waveform.
I don't like the sound it causes.

Now we can make a filter with 4.3uF; 12uF; 68uF from modern MKP type caps, and when I have time -{hopefully during this year}- I will try that and listen, as I have original Reference 104 with the original crossover.

Also, I have another idea to try for a slightly simpler circuit - I will post about it after I have tried it.

***************

Mattias,

did you buy 4.3uF and 5.1uF capacitors ? - or several smaller capacitances to connect in Parallel to sum to 4.2uF and 5uF ?
- and if so, what capacitance sizes did you buy ?

I bought this 5.0uF AuriCap for the Mid/bas Capacitor change: AuriCap 5 0uF 10 400V AUR505K400VA

And some cheaper capacitors from ClarityCap for the tweeter:
https://www.madisound.com/store/prod...oducts_id=8148

https://www.madisound.com/store/prod...oducts_id=8147

What is really wierd is that the blue capacitor on my crossover is a 4.7uF capacitor and according to the schematic it should be a 4.2 so have not ordered the last capacitor yet, is it better to pick a 3.9uF?

/Mattias

[speaker dave]

Originally Posted by Mattias l

What is really wierd is that the blue capacitor on my crossover is a 4.7uF capacitor and according to the schematic it should be a 4.2 so have not ordered the last capacitor yet, is it better to pick a 3.9uF?

/Mattias

KEF tended to sort capacitors and pick non-standard values for drivers that needed a different particular vlaue. I would replace with 4.7 if that is what your unit had.

Its not worth getting too fussy over the values as the drivers can drift a little over time. It is also possible that KEF sorted through wide tolerance parts for a particular value: for example, 4.5 actual value, although marked as 4.7. You would have no way of knowing.

Get 4.7s and don't worry about it.

David S.