Life of Caps.

[chinoy]

So for the past few months I have been playing with an restoring vintage audio equipment.
Like a Hitach HT-350 turn table, An akai amp all from the late 70s.

And it goes without saying the first thing you do is replace all the caps.
never mind that they used the best quality brands in 10-20 years most caps need to be replaced.

This got me thinking about my own product. Which uses a lot of caps. And is encased in epoxy.
Does this mean that in 20 years my products are also going to stop working. ?

How do I address this issue. Some of my customers want to know what if they buy my product and never use it. Will the caps stay ?.
Is there a better cap I can turn to ?. Im already using the best and highest grade caps money can buy.

Should I shift to using Tantalum Caps because they will last longer ?.
Should I use higher voltage caps ?. Im already using 65v when all I need is 5V.

One thing I did notice between the caps used in Audio in the 70s vs what you get today is the size. Modern caps seem to have shrunk by a huge margin for the same values and voltages.

Any guidance or help in this regard much appreciated.

[El Rubio]

I like to restore older audio gear also. I am a believer in the positive results from recapping older gear. 10-20 years doesn’t require a wholesale recap in my opinion. Most electrolytic’s have a suggested lifespan of 30 years. I don’t know about you, but most of the caps I replace from 70’s gear are still within spec. For me it is a little insurance because not many other components have a specified shelf life, at least not as short as 30 years. Most of the out of spec caps I have found are usually those low voltage (5-16v) caps. As you are aware going up in voltage is fine, but after a point, the advantages diminish and putting 65v caps in may qualify as overkill. Why are your products encased in epoxy? Is it for water-proofing, preventing arcing, or tamper-proofing. By encapsulating in epoxy, you have decided the device is not repairable if something were to fail. I wouldn’t be too concerned about longevity of electrolytics in that case.

[Jeff eelcr]

Time, tempature, type, manufacturer, size, voltage and other factors can affect longevity.
Parts like this lasting 25-30 years is possiable but not expected all you can do is use the
best parts (at price) you can and hope they last as long as needed.
I have one product that is potted in epoxy with one capacitor (tant) 35V in a 12V circuit
so far in about 10 years we have 0 failures but I have the same results with electrolitics in
the last almost 40 years (as far as I know).
Jeff

[golden_labels]

Others will answer the question directly, which I can’t do, but I want to remind about some things.

A failure will happen. Accept it and be prepared to deal with the situation, instead of desperately hoping to evade the issue altogether. Diminishing returns of the latter hit badly. Make sure your customers can easily diagnose and address potential problems. I would see this as much more valuable than avoiding breakage in 30yr equipment. I am sure that in 2040 a customer of yours will be much more happy having schematics to replace a dried out electrolytic capacitor, than the capacitor surviving until 2050, but the unit being impossible to diagnose.

Many states have laws against offering excessively long warranties. For a good reason. Even if the manufacturer is not deceptive,⁽¹⁾ chances of building devices lasting that long, the company being able to survive or wanting to deal with the long-term liability are very tiny. The first of those is relevant here. It doesn’t matter how much you try: the reality shows that even despite the efforts, not cutting corners, using best quality components, stuff lasting decades is a rare occurrence. Probabilistics work against you here.

I’m not saying you should build crap. It’s about being realistic about the goals and methods to deal with potential issues.

⁽¹⁾ For example the infamous “lifetime warranty” cases, where “lifetime” basically means “expected lifetime of the device”.

[chinoy]

Ive been doing this product since the 90s.
So far we have had one case with a failed cap.
And that was because the customers bat voltage crossed 30 volts. So the cap swelled up.
Its encased in a special dielectric Epoxy to prevent arcing.

I am new in Audio. And was suprised to see the number of quality caps in audio equipment that were not in spec.
We have hundred if not thousands of people running the product for over 20 years.

Just got freaked by the audio scene.
Already using higher than need voltage, and high temp quality caps.  By top brands. But that did not seem to save many of the caps on audio gear. I guess your right its mostly the under 25 volt caps. i.e. 10v and 16 volt.

What about switching to tantalum. I dont like that bit that they explode when they fail. Over the years we have had sales from time to time where we replace a failed unit with a working unit at a deep discount. That seems to keep the few unhappy customers happy.

[jonpaul]

Western Electric manufacturer of telecommunications equipment for Bell System used in house special lytics, rated for high ripple current and derated temp and voltage

Subscribers equipment like 1A2 PBX had 20 yrs and CO Central Office gear 40 years

Use old MIL-217 to calculate reliability of a systems.

We specify 105 1nd 135 C caps from Rubycon, ChemCon, or other non Chinese suppliers.

Just the ramblings of an old retired EE

jon

[Doctorandus_P]

Have a close look at datasheets for electrolytics. They are by far the parts with the lowest life expectancy, and it can be as low as 2000hours, which is less than 3 months. But that is at their highest rated temperature and ripple current, and at lower temperature and ripple current life expectancy is (a lot) longer. Decent manufacturers have special series with longer life expectancy. In general a few smaller caps have a higher total ripple current then a single big one, and is thus further derated an has a longer life expectancy.

There is quite a lot of literature available about derating capacitors and calculating their life expectancy.

Maybe it is an option to keep the electrolyticys outside of the potted area?

[james_s]

And it goes without saying the first thing you do is replace all the caps.
never mind that they used the best quality brands in 10-20 years most caps need to be replaced.

No it doesn't go without saying. I almost never replace all of the caps, I got tired of working on gear where someone had replaced all the good quality caps with the cheapest new junk they could find, and messed something else up in the process. I measure the ESR and do a physical inspection, if the capacitor is marginal or shows signs of leaking or damage I replace it. If it checks good I leave it alone. If I do replace a capacitor I make sure I use a good quality part that is likely to last another 30 years. I have seen lots of 40+ year old electrolytic capcitors that still work fine, SMPS applications are really hard on them but in many other circuits they can last a very long time.

[chinoy]

This is why I keep asking about Tatlum as an option. Yes I have moved to 105c and 150c where possible.
Yes ESR values are high on most of the caps I pull from old audio equipment. I use a cheap component tester to test each cap.

https://resources.altium.com/p/which-type-capacitor-should-you-use

[Vovk_Z]

What about switching to tantalum. I dont like that bit that they explode when they fail.

Tantalum caps still may fail as usual caps.
Tantalum caps are good if you like that 'Tantalum' word magic or if your customers like it. Otherwise, I can't say they are really much-much better in longevity vs price.

[Shock]

The capacitors datasheets tell most of the story (or in some cases are missing story), other than that check the vendors website to see if they have any other information on the series.

It's beyond any reasonable expectation for someone here to predict what capacitors are suitable for your application and how much extra life overrating will provide. You need to do all that yourself while taking into consideration all environmental factors and failure modes combined with your testing methodology.

Certain tantalum and electrolytic's have improved longevity so I think it's hard to compare against the past especially when very new products have little real world data to go on. It sometimes takes a few years for these failures to show up. On the flip side an established product if they have changed the manufacturing process isn't immune either, so sometimes it's wise to compare historic specifications.

I'd personally be hesitant to use fringe brands if you can find adequate specifications in a major vendor. Most expensive is not always the best or most reliable and if it's a critical application I'd seek the opinion of major vendors about their products suitability and thoughts on the competition.

Lastly look at comparable products and see what the competition is using or changed and what fails and comes back. Retailers are a good source of this information. Obviously information provided can be anecdotal when there is a bias but you can still get an idea and weed out claims made without any validating evidence.

[JRosario]

I feel like the only way to really know if a cap is going to last 20 years is to buy for a company who has been around that long.

Does anyone know manufactures with that kind of track record?

[tooki]

Well, yeah… all of the big, top brands have been around for a long time. (Some are literally a century old: Cornell Dubilier has been around for 114 years, Sprague, now part of Vishay, is 96 years old, Chemi-Con is 90, etc.)

Panasonic, Vishay, Kemet, Cornell Dubilier, Chemi-Con, Rubycon, TDK-Epcos … I’m sure I’m forgetting some, but I think those are the big ones. Any of these is a top-notch product.

[chinoy]

Thanks for all the  comments. When I got started a guru advised me to buy good brands that have a proven track record.
So we developed a list for each component i.e. caps from company x, diodes from Company y etc.
This has served me well. He also told me to source critical transistors from known sellers like RS Components and Digikey.
As these guys have their own quality checks.
I do find the price is normally 10X what they are in the open market. But with that extra costs comes some security.
Sadly there are a number of fakes in the market. i.e. small time makers marking their stuff as STI or other brands.
What I normally do is a destructive test of every component every time I order a new batch. i.e. push it past its design parameters and see how it holds up.
Once again thanks 

[David Hess]

And it goes without saying the first thing you do is replace all the caps.
never mind that they used the best quality brands in 10-20 years most caps need to be replaced.

The capacitors may be worn out without actually causing failure.  It is is not like gradual evaporation of the electrolyte causes sudden failure.  A circuit may be completely happy with 5 or 10 times the specified ESR and 1/2 or 1/4 of the capacitance.

In low ripple current applications where self heating is low, small capacitors wear out first through gradual evaporation of their electrolyte, but these applications also tend to be tolerant of increased ESR and lowered capacitance.  Large capacitors are more likely to fail in another way, see below.

This got me thinking about my own product. Which uses a lot of caps. And is encased in epoxy.
Does this mean that in 20 years my products are also going to stop working. ?

It is possible to design for an operating and storage life greater than 20 years, but many factors other than aluminum electrolytic capacitors drying up come into play.

How do I address this issue. Some of my customers want to know what if they buy my product and never use it. Will the caps stay ?.
Is there a better cap I can turn to ?. Im already using the best and highest grade caps money can buy.

Use the best capacitors and derate them for greater operating life.  Design the circuit to be tolerant of gradual deterioration.  In critical circuits where no single capacitor will provide the desired operating life, use multiple capacitors in series and/or parallel.

Should I shift to using Tantalum Caps because they will last longer ?.
Should I use higher voltage caps ?. Im already using 65v when all I need is 5V.

Those are both good options.  A higher voltage aluminum electrolytic capacitor has more volume and therefor more electrolyte, which will take longer to evaporate.  Tantalum capacitors have no electrolyte to evaporate but are not suitable in all applications.  High reliability and long operating life applications may avoid using aluminum electrolytic capacitors at all despite the cost.

One thing I did notice between the caps used in Audio in the 70s vs what you get today is the size. Modern caps seem to have shrunk by a huge margin for the same values and voltages.

And those physically larger capacitors last longer because they start with a greater volume of electrolyte, so it takes longer to finish evaporating.  You might have noticed that in applications where ripple current rating is not an issue, huge input capacitors used to filter the 50/60 Hz line last so long that their primary failure mode is water or oxygen getting in through a degraded rubber seal and corroding the connections to the pins until one breaks.

Tantalum caps still may fail as usual caps.
Tantalum caps are good if you like that 'Tantalum' word magic or if you customers may like it. Otherwise, I can't say they are really much-much better in longevity vs price.

The difference is that tantalum capacitors do not have an inherent wear-out mechanism limiting their life, except maybe defect growth in high voltage parts if you believe NASA.  Aluminum electrolytic capacitors literally dry out over time, and whether in use or not.

[Capernicus]

vacuum capacitors i think just last forever, they are indestructable.
same for gimmik capacitors they have no life duration, they just go forever.

[tooki]

vacuum capacitors i think just last forever, they are indestructable.
same for gimmik capacitors they have no life duration, they just go forever.

Most vacuum caps are variable, which means they have a mechanical seal, which means that they will lose vacuum over time. One manufacturer I saw lists a lifetime of 10 years. They are also almost all designed for very high voltages, so irrelevant as a substitute for electrolytics. (Ceramic caps also last essentially forever. Low capacitances are easy to get in “last forever” technologies like ceramic and film. It’s the large capacitances that are troublesome.)

The gimmick cap is certainly reliable, but of extremely low capacitance, so again, totally irrelevant to this application.

[MathWizard]

I haven't worked on very much stuff overall, but I'm surprised how few bad cap's I've found in stuff. I'm just getting used to my LCR meter tho. I should re-check a few in some PSU's I took apart but never finished yet.

Just the other day, I took apart some old cheap 24V power bricks I have, and I thought they were junk. But they are ok, and the caps are actually 105C rated, and measured great. I'm still a bit worried they will blow up tho, due to coil whine, even after I added some glue.

Usually tho, any time I'm working on something power related, or voltage rail related, I'll go ahead and pull the caps to check them. I should see if my meter actually works in circuit. I don't think I ever tried that.