KEF 104/2 refurb

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have a pair of 104/2 purchased around 6 years ago that i enjoy a lot. Last year I replaced the doughnuts and the tweeters ferrofluid. Highs improved significantly. Since then I have been wanting to redo the caps and restain the enclosure. For months I have been waiting for falcon acoustics to have the caps in stocks but still have not been able place an order. So, i have decided to start the refinishing and removed all components. I was curious to see how bad were the capacitors. Lifted one leg on each and checked with my DE-5000. To my surprise the first board seem to be pretty good, surprised for the age unless the previous owner recapped it. There was no sign of work on the board..

I have not dissassembled the second speaker but if it measures equally good I guess it will save me the recap.


  • image.jpeg
    121.1 KB · Views: 244
  • image.jpg
    472.7 KB · Views: 240
Last edited:
Electrolytics have a limited half life. If you are going to pull out the crossovers consider a re-cap just to save you time in the future. :)

Mundorf MKP's are very high value and good sounding. I would at a minimum do everything in series with the tweeters. Assuming 2 caps each, that's probably $30 total.

Good luck,

Electrolytic caps only drop in capacitance once they are well and truely fubar. Before that happens the ESR will skyrocket.
What you need to measure is their ESR to know if they are bad or not. On your DE5000 you want to set it to manual series resistance ("Cs") then scroll the secondary display through to ESR. I use a measurement frequency of 10kHz - this may not give an accurate capacitance value for larger caps but it will give an accurate ESR value. Google "capacitor ESR chart" for a chart of typical ESR values for a given capacitance and voltage rating.
Last edited:
As 104/2 owner I believe it's a wize advice : better ESR at some aera will not be necessarilly an improvement, as MKP over electrolytic (which is the idea)!

In these ref series, the caps/résistors were matched to be close at less 5% than the original former and matched by pair : so be extremly carefull when swaping anything ! Most of the time, changing the original crossover of the old reference series will change the sound but not surely for an improvement !

Imho, the best is to measure all with precise Tools and follow the ESR path :)

To try to improve the genuine crossover which is certainly possible but ask an expensive and huge and long work : keep the genuine XO but outside of the cabinet then clone it for swapping it with the caps, resistor, and coils you want but at least with the same C/R/L values !

Check your Tools with 1% R, L and C to be sure of your initial measurement !

At least I would proceed like that if I want to work on my 104/2 ref (have the mono wirered version with genuine filter and tweeter :) )
Mundorf MKP's tend to be within 1%.

It's at worst an equivalent that will last a hundred years. At best many report a better sound.

If you are going to go through a thorough replacement, which I'm not recommending, at most I'm suggesting replacing 1 or 2 caps as may be in the tweeter section, then from experience, it's best to do a thorough speaker analysis. You'd be surprised how many well-regarded speakers have heinous choices in the crossovers. I have gone down the part for part replacement route. It was an expensive mess compared to doing a complete analysis. Still, that's all much more than I'm suggesting.

Avincenty, try my idea. Easy to back out if you don't like it, and you may learn more than the $30 would suggest.


It's not about the 1%, it's about the result, so I have to disagree, jumping each time for a Mundorf Supreme will not give you each time a better result !

At least one has to try to be sure ! But I doubt it's reliable to say : buy a SUpreme and forgett it, this 104/2 was made to sound good, it's not a crappy filter than the simpliest tweak will improve. Here the filter have caps with different ESR in relation to the aera... it was not made by luck.

And this speaker is more than an other sensible to the room..
Well, you will learn a great deal so that will make it worthwhile.. do a thorough job analyzing the complete speaker as is first so you have something to compare to. o use omnimic and dats both for analysis and minidsp configuration. i also use xsim for crossover modelling. very straightforward.


Not time for that, maybe in some years ! It's not my main speaker... but it's a good one, non fatiguing, very sensible to power but not only : for instance when I putt a Chord amp on it... I am disapointed despite it's a powerfull one ! It's rocks more odly with a Nad which has some headroom for peaks demand but which is only rated to 60 W ! :eek:

Many reviewers say a big Pass, a Quad 909, or even a Bryston 7 is more suited to this speaker gives it's best (I use mine without the Box and the relativ lack of deep end in an average living room (here between 20-30 m² in my country) is enough because the mid-bass rocks !:yummy:... with a total 3D soundstage.... not bad for a 80s speaker :)

So it can be fun with some inexpansive class D to putt three or more of them per box with an active setup !
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.