Why is it that secondary filter caps fail that frequently?

[technix]

I have to repair my 2011 Apple Time Capsule again today, replacing four of the seven secondary output filter caps on its internal Delta mains power supply. And this is the second time that power supply suffered from failing secondary filter caps, the first time of which I replaced the other three. Similarly, my cable TV box also had two secondary filter cap replacements already, first time with a crappy cap, then I went with a Nippon Chemicon one. All of those devices are long out of warranty already, so it is cheaper to fix it myself.

Why the hell does those secondary filter caps fail so frequently?

[bob91343]

They are subjected to large currents and run warm, thus lowering reliability.  Get parts with higher temperature ratings and, if possible, don't use one big single part but several in parallel.  That gives lower temperature.  Heat is the main culprit.

Capacitors with lower ESR will run cooler at the expense of greater currents through the rest of the circuit.

Further, electrolytic capacitors have a high failure rate to begin with.  They deteriorate with age due to internal chemical action.

[Haenk]

Apple (and almost every other manufacturer at that time) did use cheap chinese caps, which were manufactured using a stolen (but incomplete) formula for the electrolytics fluid.
They worked fine for a couple of years, but their failure rate over time is absurd. Unfortunately, the lifespan of a couple of years means, the "flaw" in quality was only detected after a couple of years, so lots of mass markt electronic from that era *will* fail, if they are using those caps.
Here is another notorious-to-fail Apple product:

http://www.maccetera.com/macstuff/imac_g5.html

Again, this is not limited to Apple, I had plenty of major brand servers and mainboards (not even talking about cosumer electronics) failing due to those caps...

I wouldn't say that failing caps are a problem itself, it's rather the manufacturers using the cheapest stuff they could find. There are "good" caps with enourmous lifespans on high current and temp. but they are not cheap (enough).

[magic]

May also be poor airflow/cooling in the area.

I had an ATX PSU whose capacitors cooked themselves and I decided it's not worth repairing because:
- they were some super compact "miniature" capacitors with higher ESR and less heat dissipation area than normal stuff, and no larger replacement would physically fit in their place
- they were hidden under a jungle of output cables, with little airflow
- the whole PSU didn't even have that great efficiency to begin with

[TERRA Operative]

Be sure to use 105 degreeC rated, Low ESR, Low ESL capacitors with the highest ripple current rating and highest voltage rating you can fit (you can often go up to the next standard voltage rating) from a reputable company.
I like United Chemicon/Nippon Chemicon and Rubycon, but any good brand will be ok.

Also, don't buy from ebay, banggod, aliexpress, amazon, etc. Buy from a real electronics distributor to ensure you get genuine parts.

[fcb]

Apple (and almost every other manufacturer at that time) did use cheap chinese caps, which were manufactured using a stolen (but incomplete) formula for the electrolytics fluid.
They worked fine for a couple of years, but their failure rate over time is absurd. Unfortunately, the lifespan of a couple of years means, the "flaw" in quality was only detected after a couple of years, so lots of mass markt electronic from that era *will* fail, if they are using those caps.

Can you point to an article that explains this? I'm intrigued.

[I wanted a rude username]

Here's a 2003 news report about it.

There's even a Wikipedia article.

Two questions remain: how many of the faulty capacitors are still in use, waiting to fail? And which company supplied the flawed formula? Mr Zogbi said most of the faulty parts had been tracked down. "But some of the large computer manufacturers just won't admit the problem," he said.

[fcb]

Here's a 2003 news report about it.

There's even a Wikipedia article.

Two questions remain: how many of the faulty capacitors are still in use, waiting to fail? And which company supplied the flawed formula? Mr Zogbi said most of the faulty parts had been tracked down. "But some of the large computer manufacturers just won't admit the problem," he said.

Fascinating - thank you.

[technix]

Also, don't buy from ebay, banggod, aliexpress, amazon, etc. Buy from a real electronics distributor to ensure you get genuine parts.

Since I live in China, so the best source I have is LCSC (the English version is actually the same store front as the Chinese version and shares the same internal stock, just the English version has a Digi-Key inspired design while the Chinese version is based on Taobao's design.) Sources like Digi-Key and Mouser has crazy shipping and handling fees here.

Although when buying caps I usually just hit the Shanghai SEG market and buy them over the counter, face to face.

[David Hess]

The reason that the output capacitors tend to fail before the input capacitors is that they are selected differently.

The input capacitors are selected for minimum holdup time between AC power line cycles.  This requires a minimum of capacitance which is determined by the input current and allowable ripple voltage.  The result is that these capacitors have a much higher ripple current rating than necessary.

The output capacitors are selected for ESR or ripple current rating depending on which is a limiting factor so they operate either close to or at their ripple current rating are wear out much more quickly.  Their capacitance is largely irrelevant.

[tooki]

Apple (and almost every other manufacturer at that time) did use cheap chinese caps, which were manufactured using a stolen (but incomplete) formula for the electrolytics fluid.
They worked fine for a couple of years, but their failure rate over time is absurd. Unfortunately, the lifespan of a couple of years means, the "flaw" in quality was only detected after a couple of years, so lots of mass markt electronic from that era *will* fail, if they are using those caps.
Here is another notorious-to-fail Apple product:

http://www.maccetera.com/macstuff/imac_g5.html

Again, this is not limited to Apple, I had plenty of major brand servers and mainboards (not even talking about cosumer electronics) failing due to those caps...

I wouldn't say that failing caps are a problem itself, it's rather the manufacturers using the cheapest stuff they could find. There are "good" caps with enourmous lifespans on high current and temp. but they are not cheap (enough).

The stolen-recipe electrolytics were in the very early 2000s, and it’s not even clear that it affected only “cheap Chinese” caps, since Apple didn’t, and doesn’t, use cheap caps. I had to replace the PSU in my sister’s 2011 time capsule (she drowned it when watering a plant), and it had all Nippon Chemicon caps. (It was a Delta Electronics PSU. Delta is a premium PSU company, and one Apple has used extensively for decades.)

So between Chemicon not being “cheap Chinese” crap, and the electrolyte scandal having been a decade before this product, you’re telling a story that, while true, is not even slightly applicable here.

[technix]

I wonder that for such a power supply, for example one based on Mornsun LS03-15B05R2S which requires me to supply the caps and additional secondary filters, can I use ceramic secondary caps? For example this 100µF 6.3V X5R 1206 ceramic, in place of aluminums? Will my ceramic secondaries have similar problems as the aluminums?

[trobbins]

Magic points out the issue of local heat transfer away from the region of the e-cap(s).  The application typically requires output e-caps to be as close to the power transformer secondary and rectification parts as possible.  Often the pcb copper in that region is used to transfer as much heat as possible away from power dissipating parts, and may influence e-cap temperature by either minimising any e-cap conductive cooling (from ESR heating) via its leads or even conduct heat in to the e-cap.

Technix, I would be cautious about replacing an e-cap with a ceramic directly on the converter output unless you know what the impact on converter regulation and transient stability will be, as some feedback systems require a certain minimum ESR.

[technix]

Technix, I would be cautious about replacing an e-cap with a ceramic directly on the converter output unless you know what the impact on converter regulation and transient stability will be, as some feedback systems require a certain minimum ESR.

Can I put a small series resistor on my ceramic caps?

[james_s]

I repaired piles of gear affected by the cap fiasco, motherboards, TVs, DVD players, external hard drives, IT gear, I lost count. 2011 sounds too late for that though, the motherboards I was repairing were PIII stuff back when that was still reasonably current in the early 2000's.

Even today I replace a lot of capacitors, switchmode power supplies are notoriously hard on capacitors, the high frequency square wave results in a high ripple current and compact gear uses the smallest suitable parts and runs them hard and hot.

[technix]

I repaired piles of gear affected by the cap fiasco, motherboards, TVs, DVD players, external hard drives, IT gear, I lost count. 2011 sounds too late for that though, the motherboards I was repairing were PIII stuff back when that was still reasonably current in the early 2000's.

Even today I replace a lot of capacitors, switchmode power supplies are notoriously hard on capacitors, the high frequency square wave results in a high ripple current and compact gear uses the smallest suitable parts and runs them hard and hot.

For point of load DC-DC converters I have entirely switched to ceramics for a few years already. Newer chips like MP2122 they just specifies ceramic output filters. It is just the mains rated ones I am wishing I can switch over to ceramics.

[David Hess]

I wonder that for such a power supply, for example one based on Mornsun LS03-15B05R2S which requires me to supply the caps and additional secondary filters, can I use ceramic secondary caps? For example this 100µF 6.3V X5R 1206 ceramic, in place of aluminums? Will my ceramic secondaries have similar problems as the aluminums?

Ceramic output capacitors can be made to work and some high reliability power supplies are built that way but read below.  Electrolytic capacitors with a higher ripple current rating would be easier.  This can be achieved with better capacitors, more smaller capacitors in parallel, and higher voltage capacitors which are physically larger.

Technix, I would be cautious about replacing an e-cap with a ceramic directly on the converter output unless you know what the impact on converter regulation and transient stability will be, as some feedback systems require a certain minimum ESR.

Can I put a small series resistor on my ceramic caps?

That will work if stability of the regulator requires a minimum ESR and sometimes that is done.  For new designs, either a series resistor is added or the regulator is designed to operate into a very low ESR.  Ceramic capacitors are available which have a resistor built in for a controlled higher ESR.