Do electrolytics values stay constant ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello,

I want to recap a pair of 25 year old KEF 105/3 with new electrolytic caps (the high values and little space on the board leave me little choice). However, I've read that for these speakers, KEF used to select cap values specific to each driver, with a tolerance of 1%.

So ideally, I should measure the original caps and try to match them as closely as possible. Hence this question: do electrolytics capacitance drift when they age, making this measurement moot, or do their values stay constant, and only the ESR increases ?
 
I don't know your location, but Falcon Acoustics can supply a replacement capacitor set for the KEF 105/3.

KEF 105.3 REPLACEMENT CAPACITOR SET

Yes, thank you, I've ordered the set. The guy at Falcon acoustics told me that the caps are following the values in the schematics because there is no way to know the exact values for my pair (KEF supposedly has them but they won't bother answering such a request, quite honestly disappointed with this company). Also, because the drivers have aged, it doesn't matter anyway, according to him.

Nevertheless, I don't like that the crossover has so many caps and that the board is so small (a 19 cm square) that nothing else than small electrolytics will fit. Right now I am not listening to music, I'm listening to electrolytics.:guilty:
 
In my opinion, there's nothing wrong with the 'sound' of properly chosen electrolytics.

In fact, replacing electrolytic capacitors with film capacitors can upset the sound balance of vintage speakers because of changes in equivalent series resistance (ESR).

Well, I'm not going to lie, I don't like the sound of these speakers at the moment. They sound dull, lifeless, they lack detail (no air around instruments) and the treble can sound harsh. This certainly isn't the case with even entry level 2 way speakers I have. So I assume that's not how these well regarded reference speakers are supposed to sound. My personal experience is too many components in the signal path, and in particular electrolytics, tend to do that.
 
I don't like the sound of these speakers at the moment. They sound dull, lifeless, they lack detail.
It is probable that the original electrolytics have aged substantially, changing in both capacitance and ESR.

Replacement electrolytics, as included in the Falcon set, will hopefully restore the factory fresh sound.

Good luck with the renovation and please report back on your findings. :)

P.S. I'm no fan of overly complicated crossover circuits myself!

P.P.S. Are the KEFs a recent acquisition, or have you owned them from new?
 
If you studied my earlier diyAudio link, you will have come across this useful analysis of the complex 105/3 crossover circuit.

I presume your French is as good as mine! :)
 

Attachments

  • 105.3 Analysis.jpg
    105.3 Analysis.jpg
    244.4 KB · Views: 130
My personal experience is too many components in the signal path, and in particular electrolytics, tend to do that.
If the straightforward replacement of all the capacitors fails to produce the open and detailed sound you seek, then it may be interesting to recreate the crossover MINUS the various add-ons such as the impedance linearisation, contour and shelving networks shown in the above analysis.

This would leave the basic high pass, band pass and low pass filters, and produce a crossover with the low component count that you favour!

I wonder how that would sound? :tilt:
 
If you studied my earlier diyAudio link, you will have come across this useful analysis of the complex 105/3 crossover circuit.

I presume your French is as good as mine! :)

He he, I am the guy who put the original photo in that forum and that's what one of the guys responded. :) I am not sure why the speakers need such a complicated circuit. Anyway I can translate for you, from top left to bottom right:

Tweeter
Linearisation to the tweeter resonance freq. (not sure what that means)
Amortized Band cut ( <-- no idea why this is necessary :confused:)
3rd order pass high

For the Mf UniQ
2nd order band pass
Networks to compensate the resonance freq. of impedance (don't know what that means)
Shelving

For the Mf Up and Low woofers
2nd order low pass
Linearisation of impedance

For the 2nd circuit (internal woofers in "double cavity")
2nd order low pass
RLC networks to linearize impedance


I wish KEF published the frequency response of their drivers, I could simulate the circuit in VituixCAD. Following your idea, I went ahead and I've done the exercise with the Thiele/small parameters of each driver (except for the tweeter, for which I found no information online, so I used the tweeter of the 104/2 just to have an idea), and found out that if you remove all the compensations, you can almost halve the number of components. But without real response data, it's pretty pointless and it also means it's a completely new crossover. I wonder if KEF used to design and simulate their crossovers on computer at that time (1989), or if they simply designed them on paper after measurement., which would explain all the compensation subcircuits.

(btw, these speakers are a new acquisition for me)

edit: attached the simulation
 

Attachments

  • circuit simplifié.GIF
    circuit simplifié.GIF
    29.5 KB · Views: 124
  • circuit simplifié2.GIF
    circuit simplifié2.GIF
    36.2 KB · Views: 117
  • circuit simplifié3.GIF
    circuit simplifié3.GIF
    239.6 KB · Views: 122
  • circuit simplifié4.GIF
    circuit simplifié4.GIF
    126.3 KB · Views: 118
Last edited:
I bought a cheap capacitance / inductance meter and ESR meter on ebay and had some capacitors precision measured as a reference. For the record the capacitance meter was out by about 7% and the inductance was about 1% but once known was easy to compensate for.
I have used them to rebuild crossovers for KEF Concertos and Tannoy monitor Golds.
I found that some of the old Elcap electrolytic capacitors in the KEF held their values but had quite a high ESR value, while others were massively dried out and not doing much at all. The replacement Alcap electrolytics have a lower ESR value but of course I have no idea what the original ESR values would have been when the crossover was new. Plastic caps have a very low ESR and this can cause problems . A plastic cap used in a low pass filter after an inductor can cause a resonant peak and need a small resistance in series with the capacitor to avoid the worse effects.
 
Amortized Band cut ( <-- no idea why this is necessary :confused:)
Ha! Our posts obviously crossed!

The add-on circuits are basically to smooth the impedance of the bass drivers and to eliminate the midrange and tweeter resonances where they occur at specific frequencies.

The meaning of 'amortie' is 'damped'. The parallel LCR circuit is often called a contour circuit. It's purpose is to shape the frequency response of the driver, depressing, or damping, its output over a specific band of frequencies.
 
I bought a cheap capacitance / inductance meter and ESR meter on ebay and had some capacitors precision measured as a reference. For the record the capacitance meter was out by about 7% and the inductance was about 1% but once known was easy to compensate for.
I have used them to rebuild crossovers for KEF Concertos and Tannoy monitor Golds.
I found that some of the old Elcap electrolytic capacitors in the KEF held their values but had quite a high ESR value, while others were massively dried out and not doing much at all. The replacement Alcap electrolytics have a lower ESR value but of course I have no idea what the original ESR values would have been when the crossover was new. Plastic caps have a very low ESR and this can cause problems . A plastic cap used in a low pass filter after an inductor can cause a resonant peak and need a small resistance in series with the capacitor to avoid the worse effects.

My crossovers have shiny Alcap and KEF branded caps. I'm wondering if they have been recapped already, in which case recapping a second time likely will not benefit at all.:rolleyes:
 

Attachments

  • IMG_20190920_212602.jpg
    IMG_20190920_212602.jpg
    439.7 KB · Views: 58
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.