Elcap / Alcap capacitors in a 80's crossover

[Antifon]

I've been taking a close look at a pair of B&W DM23s that I now own, which date from 1981. The drivers are all near-enough pristine, and they sound Ok, but I'm slightly worried about the 40 year old NP electrolytics in the crossovers.

Here's what they look like:



Any experience of these capacitors, and whether they will have aged well in 40 years? I can lift one end of each and do a simple test of capacitance, if necessary.

I'd be grateful for any advice on whether I should be looking to change them.

TIA!

 

[Tony L]

They’ll be toast. Just comedically off spec to the point you wouldn’t be able to guess what they were meant to be without looking at the writing on the side. I had the exact same type in my JR149s and they can go in either direction too, i.e. you’ll fine one high, one low. Normally by at least 80%!

PS If you want the speakers to sound as intended replace with as similar caps as you can find. I tried fancy film caps in the 149s and killed them (I have two sets of crossovers so very easy to A B). Film caps made them hopelessly over-analytical and dry. Very clear, but just wrong. I just stuck with the electrolytics from Falcon. They sound right.

 

[Antifon]

Thanks - I'll replace them. I'm only after something competent, not exotic, so I'll take a look at Farnell and see what they have first (because I have an account there!). Or are they no good for audio grade NP caps?

Edit: Ah, ok. Problem is that Farnell sell preferred values, and most are not preferred values. Thanks - I'll take a look at Falcon later.

 

[Antifon]

They aren't as bad as I expected, but then these particular speakers have spent a significant amount of their life in storage in a cupboard! All of the values were higher than the markings. Worst were 40MFD actually 50, and 20MFD actually 27. The smaller ones are closer. But it probably makes sense to change them now I've lifted one end.

Work on that 80's PCB reminds me how much easier it was to work with lead-based solder than anything used today!

One thing to confirm though - all of the black capacitors in the photo are non-polar electrolytic? I hardly ever use NP, so I don't know what each of the types look like!

 

[Gervais Cote]

They are all non polar and have to be replaced.
My favorites are these ones :

https://www.partsconnexion.com/jantzen-elko-cap.html

Very high quality, affordable and they won’t change the voicing of the speakers.

 

[Antifon]

Thanks - not yet found a UK-based supplier for those. Hificollective seem to sell the basic Jantzen product, which would probably be Ok. Falcon seem to sell Solen - they look to be a fair bit more expensive, but I'm not buying in big numbers, so cost isn't a major issue.

 

[DuncanF]

I took some Elcap's from some MS Pageant II crossovers and they were way out of spec. One nominally 5uF capacitor measured c. 40uF. I replaced them with Mundorfs ECAP70s from Wilmslow Audio. For 5uF I paralleled a 4.7uF Mundorf with a 330nF Epcos I picked up on eBay.

 

[Vinny]

they can go in either direction too, i.e. you’ll fine one high, one low. Normally by at least 80%!

Going high cap' is totally impossible. Think about it logically - to go high cap' you'd have to increase electrode surface areas in proportion. It is likely a consequence of high(er) leakage current as that screws the measurement method.

 

[a.palfreyman]

Vinny,
Interesting. As you say as they dry out, you'd expect them to get lower. Measured some caps in an old AR2ax. These all measured astonishingly high. Think the 6uF measured something like 25uF, but not certain now whether these were NP elcos or paper-in-oil. They were a square, encapsulated block with mid (6uF) and tweet (1uF) caps in one IIRC.
Photo from worthpoint:

 

[Tony L]

Going high cap' is totally impossible. Think about it logically - to go high cap' you'd have to increase electrode surface areas in proportion. It is likely a consequence of high(er) leakage current as that screws the measurement method.

Yes, I’m sure you are right and to be honest I may have misremembered as it was many years ago. I tested very quickly with the capacitance setting on my Fluke multimeter, I don’t have a proper capacitance meter. I can’t remember the values off the top of my head (though I may actually have retained the caps), I just remember none being even remotely close to the 10% spec. They struck me as particularly poor caps as the ones I pull out of old 8 bit computers that are getting on for the same age often seem to measure fine. I’ve no idea if their being bipolar caps impacts measurement.

 

[Vinny]

The only way to actually, really measure actual cap' is by comparison to a standard. Meters calculate cap' from something like rate of inflow current change, or some such. It is very broadly similar to measuting current - in very many circumstance you cannot, must not, measure current directly - you insert a 1R resistor into the circuit and measure voltage drop across it.

When working in cap' R&D, I reviewed many, many, many thousands of test results and modest leakage caused high cap' figures. Once you became familiar with what was going on, you could skim read sheets of results and determine ACTUAL failure modes very swiftly.

Eventually, the sorting was computerised but as we needed to know what the actual cause of failure was, sorting had to be done in a heirarchy - LI (leakage) was the first to sort by as it caused nonsense numbers on other characteristics that would then also be out of spec'.

Changes to cap' over time depend immensely unpon design. Anything using a wet (gel) electrolyte will gradually dry out although if they are metal cased and sealed with glass plugs, that will be VERY slow. Things like dry tantalum electrolytics suffer from electrode migration - the MnO2, if unused for an age, will move, which will cause leakage, as will migration due to moisture absorption, of infinitesimally small amounts of ionic internal contaminants. The secret with these is to ramp up the volts very slowly and they will "heal" (lots of folklore about what happens but when I worked with them over 20 year ago, no-one ACTUALLY knew what happens, except that leakage drops - the step was part of the production process as the manganese was slowly built-up inside the cap' and was called reforming, but that was just the name as it was the effect seen, quite possibly not what was actually happening).
Things like film caps will suffer from degredation of the actual materials used and that depends on lots of things - damp will destroy aluminised surfaces for instance, and the resulting oxide could cause leakage, depending on design.

 

[a.palfreyman]

Vinny,
Interesting to get the inside story and rings true with something I was reading on rechargeable batteries some time ago. If I've understood you correctly there is a similar leakage process in such as NiCad batteries where the compounds that do the energy storage on the plates / cells can build up into clusters on a microscopic level (there were some SEM images) thus bridging between the cells, causing leakage. Apparently if you discharge NiCads to about 0.8V, you can break these leakage paths as it 'refines' the compounds on the surfaces of the cells.

 

[Vinny]

Interesting to get the inside story and rings true with something I was reading on rechargeable batteries some time ago

I don't know how many battery technologies are affected but certainly at least one uses a sintered electrode - so like a sponge. Over time, breaks appear in the sintered matrix so that bits of the electrode become disconnected, so the mAhr rating falls - a full charge last a shorter time. In many uses, it is mechanical damage that causes the disconnections.

That much I picked up from someone working on one (or more) battery technologies, probably at one of the UK universities, who was interviewed on something like "Inside Science" on R4, maybe 2-3-4 years back. If I remember correctly, he was, amongst other things, no doubt, looking at ways to try and reconnect the isolated bits of electrode, but memory..............................

Leakage in electrolytic Ta cap's is down to imperfections in the dielectric, which is tantalum pentoxide. Even though very, very pure tantalum powder is used, it will never be 100.000% and it will pick up contaminants during processing - no Ta cap' factory is clean beyond normal house-keeping standards. These imperfections wil allow current to pass. MnO2 is a semi-conductor, which helps in the design, in part because this means it is not especially mobile, so does not migrate to the flaws in the dielectric. Any conducting (ionic) contamination, will migrate and cause leakage.

I was once given two new encapsulation resin samples to test. Both worked fine and batches of cap's went to test. There were two standard tests - HAST and DHSS. HAST I cannot remember the meaning of but DHSS is damp heat, steady state. The results for HAST came back absolutely brilliant for both resins. But one had dreadful leakage on DHSS.
I asked the supplier to go back to see if they could idenify a mobile conducting species in the dud, which they thought very unlikely to begin with, but their chemists did eventually find one, and eliminated it. It was probably there in ppm.

 

[Tony L]

What is the failure mode in a tantalum cap that causes them to short and/or explode? They are notorious on some old computer mainboards though I can’t find any suggestion as to expected lifespan, they just seem to have two states of perfectly fine and exploded. There’s about seven in my BBC B Micro, but those don’t seem to have a reputation of failing so I’ll likely leave them be.

 

[Gervais Cote]

Thanks - not yet found a UK-based supplier for those. Hificollective seem to sell the basic Jantzen product, which would probably be Ok. Falcon seem to sell Solen - they look to be a fair bit more expensive, but I'm not buying in big numbers, so cost isn't a major issue.

Solen are much higher quality and they last a lifetime so go for it if you can afford and if they fit on the crossover’s boards. They may change the sound slightly but most of the time, it’s for the best anyway.

 

[Vinny]

What is the failure mode in a tantalum cap that causes them to short and/or explode?

The short answer (pardon the pun) is leakage, but exactly how they come to leak can be any one of numerous reasons. Any tantalum cap needs to be brought to volts very slowly if it has been unused for a long period - otherwise damp may well have got through the encapsulant and that is what will allow leakage, which runs away and you get the explosion.

I am using the term leakage curent unexplained, apologies to the non-electronically-minded - in theory, a capacitor is two conducting plates seperated by an insulator (dielectric). In an ideal world the insulator is perfect, no current passes from electrode to electrode. Any current that, in real life, does flow, is leakage current - LI.

Tantalum is extremely, violently combustible - it was or is the material used to tip armour-piercing artillery rounds - on impact it burns very fast and hot and just melts straight through armour plate.

Hit a cap' hard with a hammer on something solid, like concrete, and you will get a small explosion - a reasonable pop or bang, depending on the size of the cap'. It became the game of choice amongst local kids when the plant that I worked at was moved to the Czech Republic, until management got wind - tantalum is VERY expensive for a start.

 

[Tony L]

That makes sense as from what I can tell when they blow it is on kit that has stood unused for years, so very much a thing in retro computing loft finds etc. I’m likely ok as long as I keep firing the computer up every few months. I tend to do this with my stored electronics to keep electrolytics formed etc. I try to remember to fire audio kit in the ‘pile’ up for several hours at least once a year.

 

[Vinny]

^^^Virtually essential - do not leave elctronics idle for long perods under general household conditions.

I do not know/understand the implications on the machines in general, but from the point of view of things like wound components and cap's ONLY, loft finds should be brought to volts over days, not hours.

 

[Antifon]

Thanks - a lot of interesting info. At a more mundane level...

I'm now facing the dilemma that I suspect a lot of people rebuilding old crossovers do: the fact that the originals are non-preferred values. It means a load of faff to replace with ~exactly~ the same values.

Despite the apparent fact that capacitor values cannot go up, they apparently have *** - or they were way out of spec to start with. The smaller two are within 10% spec, but the bigger ones are not. Here are marked values versus measured values (one end lifted from the PCB, of course):

3MFD marked, 3.3MFD measured
4MFD marked, 4.2MFD measured
5MFD marked, 6MFD measured
20MFD marked, 27MFD measured
40MFD marked, 50MFD measured

I'm thinking that I will just leave the 2 smallest ones as they are, and replace the three bigger ones.

(*** but I understand the point that what I may be seeing is just leakage.)

 

[a.palfreyman]

Are the 3MFD and 4MFD (bottom left of picture in original post) in parallel? If so, one 6.8µF is fine to replace the pair. Pretty certain these would be 10% at best. As Tony has said Falcon Acoustics are good and do the 5MFD [never understood MFD, is this American?]. They also do 20µF so use two in parallel for 40µF