My sub-woofer hums...

[Andy Dingley]

You must be incredibly young then!

[Clive Mitchell]

In message , "Dave Plowman (News)" writes

The tantalum bead capacitors have a notorious reputation for unreliability.

[Joe]

I don't believe it's actually a law, but it's a general feeling among the electronics people I've worked with, and it's a matter of reliability. I wouldn't trust valuable data to a hard drive more than seven years old (less than that, in fact) even though it may well go on for many more years. Similarly, electronics kit doing a fairly important job would not be considered reliable beyond about seven years, with electrolytics being the parts most affected by ageing.

Last year I finally replaced a 33 year old tuner amplifier, mostly because I couldn't maintain the switches any longer, and replacement was not an option. I didn't have electrolytic trouble. Had the device been been doing a vital job rather than providing music in my house, there's no way it would have been allowed to reach that age.

I do actually replace a lot of electrolytics, the wretched surface mount 1000 or 2000 hour types, in equipment typically between five and twelve years old. I've seen them die in under two years. And those buggers don't dry up, they leak highly conductive (naturally) copper-dissolving gel.

But even the good old radial types are the weak link in many things. The power supply of a particular range of oscilloscopes would need a couple replaced at around seven to ten years. Despite being switch- mode types, and working within their published specs, the ripple current dried them out, and a fuse would blow despite carrying significantly less DC than its rating.

I'm well aware that electrolytics *can* last a long time. I just wouldn't bet significant amounts of money on, say, thirty particular specimens all lasting more than seven years.

[meow2222]

I think you'll find the high early failure rate is a modern phenomenon, caused sometimes by bad caps, but more often borderline penny squeezing circuit designs.

We know that isnt the case, most run faultlessly until the equipment is scrapped.

big problems with those, but due to getting damp rather than to drying out

Not really. The temp rise this causes is minimal, and far inside lytic temp ratings. We can see this isnt the issue by considering the reliable service of the many lytics in valve kit, which run at far higher temps.

The issue with high current smoothing caps is different: consumer grade equipment is liable to exceed capacitor ripple specs, with the result that as ESR slowly climbs over time, which it will do whether you use (ie warm) the equipment or not, ESR falls out of spec for circuit operation. IOW the problem is one of circuit design, of cutting corners.

I am surprised, the milspec stuff I worked with was aluminium lyitcs. Not surprising since tants are known for reliability issues.

The fact that reservoir tants dont exist would also seem to make the existence of all-tant equipment difficult to achieve.

One among many, yes. But a fairly minor player in equipment failures.

exactly, ie dont exceed their ripple specs. Naturally this can be done for high ripple circuits too, when one is willing to pay the price for reliability.

and decoupling pins on amp ICs. And in the various control tronics, where there is no reason to use anything more expensive. And in some cases on speaker outputs.

My oldest mains radio (1930s) has not a singly lytic in it, and its mains powered. But certainly its not very practical today.

NT

[meow2222]

I agree there

Well, this has never cropped up for me, and if it existed we would have learnt about it very early on considering what we were designing. We never saw a 7 year failure problem, the equipment was designed with a service life of decades.

That mechanical hdds fail doesnt mean lytics do

Why then do we still use historic electronics in old nuclear reactors? Why do we still use electronics over 7 years old in life support medical equipment? Why do electroni engine control systems not become unreliable after 7 years? I'm afraid your rule is a myth.

Any lytic rated at 1000-2000 hr life is in a garbage quality or disposable product. Thats barely even 2 months. I expect one can manufacture most things in garbage grades, but that doesnt make the properly made items flawed.

I'm open to being educated here, as I've never found any electronic device where cap failures exceeded poor connection failures. Before you say mobos, most cap related problems with mobos (but certainly not all) were due to borderline designs that failed to account for the normal rising ESR over time that all lytics have, and which any genuinely competent design accounts for.

Guesses then.

Spec sheets are available for lytic caps, and they show what the tested lifetimes are, and under what conditions they hold true. These spec sheets are normally correct, as misinformation would put off a considerable amount of potential custom, namely any reputable equipment manufacturer, and lay the supplier open to expensive litigation.

At the bottom end things are different, with sometimes a no questions asked policy, out of spec circuit designs, and sometimes use of known substandard goods. But even this does not make electrolytic cap technology bad any more than the Trabant made cars bad, all it tells us is some folk are very determined to cut costs regardless of result. These kind of components show up on PC mobos, novelty goods, clock radios, and sometimes off brand consumer goods, ie in equipment not worth repairing.

NT

[Ian White]

That can be a problem, true, but there can be many others.

Electrolytics in valve equipment have relatively little ripple-current heating, and are mostly heated by the plentiful hot air circulating outside. In contrast, electrolytics in high-current semiconductor amplifiers are mostly heated internally by ripple-current losses. Heat transfer inside electrolytics is very poor, so for the same perceived surface temperature, the internal temperature will be significantly higher.

[Joe]

And there we have it. I haven't worked in military electronics for over thirty years.

I haven't worked in consumer electronics, either. Ever.

[The Natural Philosopher]

Probably cos they used the cheapo bead types.

The metal can ones - in the 60's a quid or more each..were pretty good ;)

[The Natural Philosopher]

Old..

that was in the 60's and very early 70's

[The Natural Philosopher]

Valve radios.

Each cinnectein hand made by skilled popel and not that many of them

The things that go in order are

scratchy switches and pots

corroded valve bases and valve pins - this is all cured with switch cleaner

electrolytics.

[The Natural Philosopher]

Run off 48v DC.

IC's? In power amps. Sacrilege! No way.

Not on quality kit. Not on a sub woofer..

What control electronics? a volume control is all there will be.

So how did they smooth the mains?

[Serge Auckland]

Paper in oil is one possibility, 8ufd 350 volt were common, ISTR.

S.

[Nick Gorham]

And chokes of course (often the field coil of the loudspeaker).

Also, output stages and loudspeakers that didn't go down to 100hz helped as well.

[Dave Plowman (News)]

Bead type. Although BEEB designed and made gear was fairly 'regardless of cost' even they had to draw the line somewhere.

[meow2222]

Right

seems a bit arbitrary.

volume, balance, signal path switching, bass, treble, often electronically conrtolled these days.

A C-R-C filter using 2uF paper caps, followed by another RC for each stage further toward the front end. And of course the iron diaphragm horn speaker doesnt reproduce anything below 400Hz. (Or as high as 4 kHz.)

The radio is built in 2 layers, with all the RC filtering taking up as much space as the rest of the radio. Its designed to run on 200v dc, which of course was not clean dc but rectified 3 phase ac, hence several 2uFs are passable.

As Serge & Nick said, late 30s radios used the loudspeaker's field coil for smoothing, when luxurious moving coil speakers were used - but they had lytics by then.

NT

[Arny Krueger]

That was true of the miltary gear I worked on in the late 1960s. Any cap that had too much capacitance to be a film type was a hermetically-sealed tanalum.

Even the cheapo bead type could last wonderfully - if they were made right.

Tantalums are fine for audio as long as they are kept biased. For example, an amp or preamp with a single-ended power supply.