Recap question - KEF 105

[writethis]

Hey gang. Longtime lurker here, former semi-active participant, jumping back in and hope to be joining in more conversations in the days to come now that I have a bit more mental bandwidth available. Soooo after falling HARD for KEF 103.2's a few years ago, I stumbled across a pair of original KEF 105's here in Nashville. They've been great, and I see these as lifetime keepers. They do what I love about the 103.2, but with more authority, scale and bottom end. They've been sounding more dull and less punchy as of late, so I finally decided to pull them out of my main system and take a look at the crossovers. When I opened up the woofer cabs, a couple of big electrolytic caps were literally hanging off their circuit boards by one lead. I'm not a recap fanatic, but this was all it took for me to commit to replacing all the 'lytics in one go. They are 40 years old, after all.

This thread is about determining what values to replace with.

The crossovers on KEF's reference series speakers in the late 70s were hand-matched part by part to hit KEF's reference spec. So I know these caps aren't even supposed to match the listed schematic values. And as it turns out, all of the caps on the speaker A mid/tweeter crossover board still match their counterparts on the B crossover within 1%. But all of them but one pair happen to be higher than listed spec – as much as 20% higher. Is it possible they've all drifted somewhat, and drifted higher? Or am I safe to go with the measured values of the old caps when ordering replacements?

Thanks in advance.

 

[Pat McGinty]

Interesting question, complex.

Rising cap values? In a film cap, no. But an electrolytic has a wet dielectric that has the potential to dry out. I've never observed that happening, but haven't actually looked for it, either. Capacitance is an inverse square function of the distance between the plates, so a tiny decrease would cause a large increase in value. You'd normally expect to find values both above and below spec, a probability thing, but as the sample quantity increases so would my suspicion that the values have risen.

Just me: a guy who says he's trimming filter part values to tighten his unit-to-unit variation is making an odd statement. If you're intent on tightening that, you'd prefer to work on the transducer side for two reasons. 1) It is much less expensive to tighten the requirement on the driver supplier than to spend resources on it at the QC phase of production. 2) Adjusting filter values is a not a good substitute for correct driver behavior.

Also, there is a lack of value in reducing unit-to-unit variation below a practical limit imposed by the huge variations in end users' environments. It's just silly to concern yourself with variations so small that they can't be handled easily on the driver side of the build. So a grain of salt.

At any rate, you're not going to see the adjusted values deviate by 20% unless their drivers were coming in waaaay off spec, which almost certainly wasn't the case.

In a low pass filter rising shunt capacitance will shift the corner down and lower the Q which is benign compared with the opposite. You'd experience a 'sin of omission' at the top of the band that would be nice to correct.

Hopefully all of these 'lytics are shunts, not series caps?

One off-subject thought: because capacitance varies with plate distance, caps are naturally microphonic so they don't want to be inside the pressure vessel because their values will be instantaneously modulated by the acoustic signal - a distortion engine. Maybe move 'em out into the room where the pressure is lower. It's kinda silly to pay big bucks for nice parts then pound on them while they're trying to do their jobs.

 

[Kpatch]

One off-subject thought: because capacitance varies with plate distance, caps are naturally microphonic so they don't want to be inside the pressure vessel because their values will be instantaneously modulated by the acoustic signal - a distortion engine. “

Uh, Pat, could you say that in a different way so that someone like me who don’t no wurds reel gud could understand it?

 

[Pat McGinty]

Sorry. Am the son of an EE tech writer.

I'm suggesting that the engineering reasons for individually tweaking each crossover could perhaps be subordinate to the marketing reasons. Just me, I'm suggesting to @writethis that he disregard that and go with the spec'd values. At least start there.

A capacitor comprises two conductive plates with a space between them. In most cases the space is filled, and the distance set, by a thing called the dielectric, because it does not conduct. In most cases the two plates are sheets of metal foil and the dielectric is a sheet of plastic film, the whole thing rolled up into a cylinder. You can make a cap like that from the stuff in your kitchen.

The value of the cap depends upon the surface area of the plates and the distance between them, as set by the thickness of the dielectric.

For a visual, here's an educational tool for teaching students about the math involved in changing the thickness of the dielectric, it's a square function, so the cap's value goes up really fast as the plates approach each other. See the two plates and the stackable sheets of dielectric:




In any ordinary fixed value cap the plates' areas can't change, but the space separating them can be slightly squeezed or stretched by the changing air pressure around it. That's why caps are a microphonic; they change their value with air pressure. Many devices that are made and sold as microphones are actually capacitors that are made to have an easily effected plate distance.

This effect is why good quality audio caps are wound as tightly as possible, but still it's a problem. Squeeze it and the value rises and visa versa. Let's look at a cap in series with a tweeter; the operative element in the 'high pass' circuit because, depending upon its value it blocks lower frequencies. Squeeze and the crossover frequency goes down, stretch and it goes up. Do that hundreds or thousands of times a second - as in the woofer chamber - and you're adding a nasty and totally unnecessary distortion by instantaneously changing the cap's value with the very signal that it's working on. Ugly stuff.

That's why, just me, I've never made a speaker with a capacitor inside the woofer chamber. Pardon the snide comment, but guys are out there fussing around with wire and spikes and so-on while they're pounding the crap out of a bunch of (possibly cheap) caps inside their speakers. Going with nice quality caps and isolating them goes a long way and is usually a bargain by comparison.

I'm hypothesizing that perhaps if the wettened dielectric in an electrolytic cap dries out over time, it could leave the dielectric less thick. If that were happening you'd expect the value to rise and, perhaps, it could become 'looser', hence more microphonic because the plates could be more free to move. Writethis's observation doesn't prove that, but it coincides.

FWIW: when you hear from guys who swear that their electronics sound better when placed on well damped platforms, they could be experiencing a reduction in the microphonic effect on the numerous and very large value electrolytic caps in their AB amp's linear power supply. This could also explain why many of us prefer the sound of oil filled caps in metal envelopes.

Any of that clap-trap make any sense?

 

[Kpatch]

Sorry. Am the son of an EE tech writer.

I'm suggesting that the engineering reasons for individually tweaking each crossover could perhaps be subordinate to the marketing reasons. Just me, I'm suggesting to @writethis that he disregard that and go with the spec'd values. At least start there.

A capacitor comprises two conductive plates with a space between them. In most cases the space is filled, and the distance set, by a thing called the dielectric, because it does not conduct. In most cases the two plates are sheets of metal foil and the dielectric is a sheet of plastic film, the whole thing rolled up into a cylinder. You can make a cap like that from the stuff in your kitchen.

The value of the cap depends upon the surface area of the plates and the distance between them, as set by the thickness of the dielectric.

For a visual, here's an educational tool for teaching students about the math involved in changing the thickness of the dielectric, it's a square function, so the cap's value goes up really fast as the plates approach each other. See the two plates and the stackable sheets of dielectric:

View attachment 18646


In any ordinary fixed value cap the plates' areas can't change, but the space separating them can be slightly squeezed or stretched by the changing air pressure around it. That's why caps are a microphonic; they change their value with air pressure. Many devices that are made and sold as microphones are actually capacitors that are made to have an easily effected plate distance.

This effect is why good quality audio caps are wound as tightly as possible, but still it's a problem. Squeeze it and the value rises and visa versa. Let's look at a cap in series with a tweeter; the operative element in the 'high pass' circuit because, depending upon its value it blocks lower frequencies. Squeeze and the crossover frequency goes down, stretch and it goes up. Do that hundreds or thousands of times a second - as in the woofer chamber - and you're adding a nasty and totally unnecessary distortion by instantaneously changing the cap's value with the very signal that it's working on. Ugly stuff.

That's why, just me, I've never made a speaker with a capacitor inside the woofer chamber. Pardon the snide comment, but guys are out there fussing around with wire and spikes and so-on while they're pounding the crap out of a bunch of (possibly cheap) caps inside their speakers. Going with nice quality caps and isolating them goes a long way and is usually a bargain by comparison.

I'm hypothesizing that perhaps if the wettened dielectric in an electrolytic cap dries out over time, it could leave the dielectric less thick. If that were happening you'd expect the value to rise and, perhaps, it could become 'looser', hence more microphonic because the plates could be more free to move. Writethis's observation doesn't prove that, but it coincides.

FWIW: when you hear from guys who swear that their electronics sound better when placed on well damped platforms, they could be experiencing a reduction in the microphonic effect on the numerous and very large value electrolytic caps in their AB amp's linear power supply. This could also explain why many of us prefer the sound of oil filled caps in metal envelopes.

Any of that clap-trap make any sense?

Yes very much so and very much appreciated. And not clap-trap in any sense. It was the “distortion engine” that threw me.
I could only be considered an ‘audiophile’ because I’ve acquired a lot of audio ‘stuff’ - and continue to do so - but not because I understand the why and how of the ‘stuff’. But thanks to forums like this I feel that I’m learning more about the technical side of things, knowledge that doesn’t come to me naturally. But baby steps and all that.

Cheers.

 

[writethis]

Pat, thanks so much for these responses. Like Kpatch, some of this is definitely over my head. So I really appreciate the simplified version. As to KEF's late-70s build philosophy, I can't speak to that with authority. But their reference series were pretty expensive at the time, and while they were making all their own drivers, the consistent story I've heard is that they did in fact fine-tune cap values in order to get the frequency response of every pair matched to +/- .5 dB.

That said, your suggestions make sense to my less technically inclined brain. And if there was ever a time to move the crossovers outside the boxes, this would be it. I'll report on progress as it happens.

 

[Kpatch]

Aren’t the x-0vers for the 105s located in a truly pain-in-the-butt place, like right behind the tweeters?

I have a pair of damaged 104.2s in the basement awaiting a rebuild. This will be one of my 2020 projects. I’ve been inside them and everything looks straightforward. I don’t know about putting the x-overs outside the box. Seems somewhat sacrilegious.

 

[mccarty350]

Greetings, I'm joining the thread as I'll be assisting Writethis in the recap. I have opinions on all of this and recommendations I've made but I will stand by and not voice them because I want to hear the consensus and opinion of others without introducing any bias.

 

[JimPA]

I think Pat did a good job explaining caps.

 

[mccarty350]

I think the question Writethis is asking and likely looking for an answer for regarding the KEF's:

When he replaces his electrolytic caps for the recap does he:

1. Look at the existing measured values and replace them with the same measured values because:
a. The values he has measured don't match the schematic but are still almost exactly the same between both speakers for the same cap types.
b. Since the values are still almost the same between left and right speakers that might indicate that the values are still the originals and have not drifted over the last 30-40 years.

or

2. Replace the existing electrolytic caps with values that match the rated values on the capacitors and the original schematics i.e. if the original cap is 50uf replace it with a 5% 50uf new cap.

Thought processes to validate:

1. Since the original capacitors were in theory matched between left and right speakers in KEF reference series products according to the drivers should we be deriving values from what we measure?
2. The drivers are 30-40 years old, are they likely to match the same specs that the capacitors were matched to in the original factory or is it likely that time has changed the original spec of the drivers?
3. Is it possible to reproduce this 'matching of capacitors to drivers' now or are the caps and drivers too old to reproduce the original matching?


These have been our conversations and Writethis is trying to determine the credence level of both strategies.

 

[writethis]

Thanks @mccarty350 for explaining this more clearly than I was able to. A couple of points I’ll add:

- I’m not OCD about any of this stuff. But these 105s will be in my main system from here on out, I imagine. I’d rather think this through and do it right the first time. No need to rush. I’m not striving for some mythological point of perfection, just want to be smart.

- each speaker has 2 crossover boards. The board for highs and mids lives behind the sealed mid enclosure. The woofer boards live inside the bass boxes. Easy to get to once you pull the front woofer panel. So actually, moving those to the floor behind each speaker would be easy enough, and not unsightly. Mildly sacrilegious, perhaps. But I do get Pat’s point.

 

[Golden Gate ER]

The values have drifted out of spec over the years, it's normal I recommend replacing the caps with the same value or near the same. Example: if the original cap is 5.3uF and all you can find is 5.0uF that difference is not significant enough to alter the sound to a noticeable level & besides the sound will change anyway simply by replacing these old caps. No need to stress over quality of cap either. A low to mid grade cap from any reputable brand is all you need for these speakers. Even if you went out a purchased the cheapest caps you could find its still a major upgrade as the modern caps have 40+ years of advanced materials and technology.

 

[writethis]

By way of perspective, here's the current system (though the 105s are out and pulled apart, with a pair of 103.2 swapped in). VPI HW-19 MKIV, Hana SL MC cart, SDS power supply; Modwright SWP 9.0se phono stage; Rega Elex-R integrated amp; Eastern Electric MiniMax CD player; Harman Kardon HK 730 receiver used as a tuner, as my Citation 15 is sidelined needing tech attention. There's also an old M&K MX-70 sub woofer adding just a touch of extreme bottom. All in all, it's a really fun system, huge holographic/borderline-hallucinogenic soundstage. No music ever sounds bad, and great recordings sound fantastic. The room is actually the most important component – I'm fortunate to have a 400 sq ft addition off the back of our house that was expressly designed from the ground up to sound good for 2-channel audio. Well tuned space (the Indian fabric covers corner bass traps done w 703 rigid fiberglass). The head in the foreground is Birdie, our small auxiliary canine. She's a 75-pound Pyrenees/Airedale.

 

[Salectric]

You haven’t said what order slope the crossovers are. This will affect the sonic importance of exact value capacitors. For example, in a first-order (6db slopes) crossover, the cap value isn’t critical at all. With a second order (12db slope), the cap value is not quite as tolerant but still not that critical. With a 3rd or 4th order, the values become more critical and the matching aspect could possibly be important for the best imaging.

All that said, my advice is to go with the rated value on the caps, not the measured value, especially if this is the same value shown on an official KEF schematic.

In my experience, the type and brand of capacitor is more important sonically than the exact value.

 

[writethis]

@Salectric thanks for raising this. While I do understand slope, Q, etc. from a recording engineering perspective, my knowledge of crossover science is shallow, so I can’t answer your question. As for cap brands: the originals were aluminum Elcaps. Falcon in England sells new replacements as Alcaps, but their customer service is a bit wobbly. Any suggestions among other current brands of electrolytics? I’d like to spend no more than $100 on parts for this endeavor.

 

[Salectric]

I would contact KEF and get their recommendation for a replacement cap that is currently available. Since you liked how the speakers sounded years ago, I believe your goal should be to restore the sound to original condition, as opposed to improving on the original.

It’s popular these days for folks to “improve” their crossovers with expensive caps and coils. Unfortunately these new and different parts often sound so different from each other as well as from the original parts that the new parts may change the overall tonal balance and coherent blending of drivers in unacceptable ways so the upgrade is actually a step backwards. Even worse the sonic problems may not be obvious at first, but rather become annoying over time.

Unless you are prepared to spend a lot of time and money trying different caps to find the best sounding combination, I would stick with something as close to the originals as possible.

 

[writethis]

I would contact KEF and get their recommendation for a replacement cap that is currently available. Since you liked how the speakers sounded years ago, I believe your goal should be to restore the sound to original condition, as opposed to improving on the original.

It’s popular these days for folks to “improve” their crossovers with expensive caps and coils. Unfortunately these new and different parts often sound so different from each other as well as from the original parts that the new parts may change the overall tonal balance and coherent blending of drivers in unacceptable ways so the upgrade is actually a step backwards. Even worse the sonic problems may not be obvious at first, but rather become annoying over time.

Unless you are prepared to spend a lot of time and money trying different caps to find the best sounding combination, I would stick with something as close to the originals as possible.

Completely agree. I'm not looking to "improve," but rather to "restore." It's like boutique guitar amps that take something (rightfully) iconic like an early-60s Vox AC-15, and build a new version but with modern parts. The result might sound cool, but it won't sound like an original. A friend of mine in LA build the amp that I think sounds *most* like a real AC-15 or AC-30; among other tricks, he gets there by using carbon comp resistors, not today's "better" alternatives. I have no desire or motivation to second-guess the engineers who worked at KEF in the mid to late 70s. I also do not intend to replace the thin internal wiring with 12-guage. I'll need to add a couple of non-original wires, especially if I end up keeping the bass crossovers outside the boxes, but my likely choice at this point will be a thinner Duelund wire.

 

[writethis]

The values have drifted out of spec over the years, it's normal I recommend replacing the caps with the same value or near the same. Example: if the original cap is 5.3uF and all you can find is 5.0uF that difference is not significant enough to alter the sound to a noticeable level & besides the sound will change anyway simply by replacing these old caps. No need to stress over quality of cap either. A low to mid grade cap from any reputable brand is all you need for these speakers. Even if you went out a purchased the cheapest caps you could find its still a major upgrade as the modern caps have 40+ years of advanced materials and technology.

Thanks for this. It feels like I'm gaining a lot of perspective through this thread.

 

[Pat McGinty]

I would contact KEF and get their recommendation for a replacement cap that is currently available. Since you liked how the speakers sounded years ago, I believe your goal should be to restore the sound to original condition, as opposed to improving on the original.

It’s popular these days for folks to “improve” their crossovers with expensive caps and coils. Unfortunately these new and different parts often sound so different from each other as well as from the original parts that the new parts may change the overall tonal balance and coherent blending of drivers in unacceptable ways so the upgrade is actually a step backwards. Even worse the sonic problems may not be obvious at first, but rather become annoying over time.

Unless you are prepared to spend a lot of time and money trying different caps to find the best sounding combination, I would stick with something as close to the originals as possible.

Completely agree.