66
\$\begingroup\$

We can sometimes see decades-old capacitors (such as ones made in the USSR) still working. They are bigger and heavier, but durable and not desiccating. Modern aluminium capacitors serve for about 11 years, if you are lucky, then become dry and quietly fail. I remember early 2000s devices where capacitors failed after 3–4 years of service, and not necessarily low-end devices (one example is E-TECH ICE-200 cable modem worth ∼ 240 USD in 2000). A repair due to failed electrolytic capacitors became a commonplace, something uncharacteristic for 1980s.

Was this 1990s degradation caused by cheap mass production? Or by poorly-tested technologies related to miniaturization? Or many manufacturers just didn’t care?

It appears that the trend is by now reversed, and recent capacitors are a bit better than the ones from 1994–2002. Can experts confirm it?

\$\endgroup\$
  • 1
    \$\begingroup\$ Only a wild-arsed guess, so not an answer. I wonder if as manufacturing became better controlled, the margins between service and overload were reduced, and as it turns out, by too much. That and cheap mass production. \$\endgroup\$ – Neil_UK Aug 24 '16 at 16:56
  • 4
    \$\begingroup\$ (1) What happened, you ask? Cost optimization, I answer. (2) The surviving old stuff from USSR is often mil-spec. By the same token, XXI century western mil-spec will last for decades too. \$\endgroup\$ – Nick Alexeev Aug 24 '16 at 16:58
  • 6
    \$\begingroup\$ I remember this, it was due to poorly manufactured parts from some foreign manufacturer. They called it the capacitor plague. "Bulging capacitors" was added to the list of common things to check for when a computer was super slow after a reformat, alongside "hard-drive failure" and "heatsink dust buildup" \$\endgroup\$ – BlueRaja - Danny Pflughoeft Aug 24 '16 at 20:34
71
\$\begingroup\$

There was a period of time where lots of capacitors were made with a dodgy electrolyte, especially by some large Taiwanese manufacturers. The capacitors looked OK in a wide variety of tests when new, but they didn't age well. Because it took a few years for the capacitors to fail, and the high failure rate to become known, an awful lot of them had been produced and built into things before people realised there was a problem. It then took a few more years to for the things to leave circulation.

Exactly why these manufacturers had electrolyte problems is not completely clear. They were using new, water based electrolytes which had been developed in Japan and worked very well. Presumably the cheaper manufacturers had missed something or cut some corners while reproducing (or ripping off) the Japanese research.

The type of capacitor affected was cheap, large capacitance, low ESR capacitors. These are the kind of thing that appears in huge numbers of consumer devices, so the problem became known in the wider community. Plus, the failure mode of these capacitors was rupture and venting, so it was easy for even people unfamiliar with electronics to see which component was at fault when their motherboard stopped working.

Wikipedia has an article about it: Capacitor Plague

\$\endgroup\$
  • 1
    \$\begingroup\$ Additional Source: theguardian.com/technology/blog/2010/jun/29/… \$\endgroup\$ – Byron Jones Aug 24 '16 at 21:15
  • 1
    \$\begingroup\$ Another source: badcaps.net/index.php?pageid=causes \$\endgroup\$ – rdtsc Aug 25 '16 at 13:19
  • 3
    \$\begingroup\$ @rdtsc: the host is down by now – possibly, slashdot effect? \$\endgroup\$ – Incnis Mrsi Aug 25 '16 at 14:08
  • 1
    \$\begingroup\$ @IncnisMrsi: this question has only been viewed 3728 times as I write, so if that site has gone down under the load of however many of those people followed the link, it may be the feeblest successful DDoS of all time ;-) \$\endgroup\$ – Steve Jessop Aug 26 '16 at 11:52
  • 3
    \$\begingroup\$ @SteveJessop yeah, but seeing "index.php" I can believe it. \$\endgroup\$ – Segfault Aug 26 '16 at 15:52
32
\$\begingroup\$

Industrial espionage gone wrong. Verified many years after the fact.

Though it was suspected from nearly the beginning. (Article courstesy of The Wayback Machine, since the original is gone from the web.)

Basic story: Guy leaves Japanese capacitor manufacturer Rubycon and goes to work for a company in China, taking a copy of the electrolyte formula for high performance aluminum electrolytic capacitors with him.

Later, part of his chinese staff leaves and goes to work for a capacitor manufacturer in Taiwan. They also took a copy of the Rubycon formula, but mangled it somewhere along the way.

So, the manufacturer in Taiwan builds what he thinks are valuable, high quality caps manufactured using Rubycon's formula. He sells them at a good price, but cheaper than Rubycon and promises the same quality.

Loads of companies buy and install these caps, then the things start failing in droves.

\$\endgroup\$
  • \$\begingroup\$ Does this cover the thing Ī̲ told you 30 minutes ago? relatively modern (post-plague) high-voltage capacitors. \$\endgroup\$ – Incnis Mrsi Aug 25 '16 at 14:12
  • \$\begingroup\$ You mention the monitor is from 2005. That fits in perfectly well with the date (2002) mentioned in the linked articles. \$\endgroup\$ – JRE Aug 25 '16 at 14:16
  • 1
    \$\begingroup\$ Even if it were post plague (which it doesn't look like it is,) there would still be stocks of the crap things around that could get into newer devices. \$\endgroup\$ – JRE Aug 25 '16 at 14:17
  • 16
    \$\begingroup\$ Well, I am SORRY. My crystal freaking ball is out for repair and I couldn't read your mind to ascertain the details. I suggested a possible reason, it was wrong in light of information I didn't have. Excuse me for trying to help. \$\endgroup\$ – JRE Aug 25 '16 at 14:38
  • 11
    \$\begingroup\$ Actually, the story is even WORSE. After it was discovered that the "new" electrolyte formula was bad, existing stocks of it weren't discarded... it just got dumped on the market, and was bought up by bottom-feeding manufacturers in China who continued using it ANYWAY because it was so cheap compared to "good" electrolyte. This is a major reason why respectable CHINESE companies won't buy Chinese-made capacitors anymore. It ends up costing honest manufacturers more to keep testing and qualifying every new batch of capacitors than to just buy Japanese capacitors & hope they aren't counterfeit. \$\endgroup\$ – Bitbang3r Aug 25 '16 at 18:58
20
\$\begingroup\$

During the 70's the Mil-Std-HDBK217 calculations for MTBF included an acceleration factor inverse to the circuit ESR. This implied surge currents and thermal rise which in turn follow Arhennius effects of localized degradation. Out gassing is a primary early warning with a bulged lid.

Also recall,SMPS development were on the rise as material cost pressures demanded lower cost and lower ESR parts. This implies ignoring the natural failure modes of circuit ESR in order to get high efficiency converters.

So the trend to see more SMPS cap failures is partly due to th designers ignoring the aging effects on ESR and the inherent thermal runaway when self heating followed.

True, new technology electrolytics have improved as well as the conductor surface finish to lower ESR in the foil. Rising costs in tantalum from places like Russia forced companies to switch to Aluminum electrolytics.

one must evaluate MTBF on a case by case if the root cause was:

  • bad design,
  • bad parts,
  • bad process (no clean or Aqua clean flux with acidic residue, or excess thermal spike on reflow profile, etc.).

A high end modem does not validate if they used high quality qualified parts with MTBF verification done inhouse and maybe just trusted the supplier.

Generally the best capacitor MTBF comes from companies in Japan, Taiwan and China a distant 3rd due to the QA reliability and process control verification diligence required for long life parts. Material contamination is a major cause in cap fabrication.

**** The biggest improvement in Aluminum Electrolytic's is the charge/discharge time constant T=ESR x C has reduced to same or better as tantalum in low ESR grade, in some cases but not all. You have to compute this next time you choose a cap that needs to be low ESR and compare 10 parts with 1 @ 10x the value for a large bridge Cap. If it is smaller, chances you get a lower ESR and higher SRF or if in the same size voltage and family just the same constant T.
Ultra low ESR Caps. are now <1~20us while general purpose are 100us to >1ms.****

\$\endgroup\$
  • 8
    \$\begingroup\$ Chinese electrolytics even scare the Chinese. My last motherboard was made in China, but the electrolytics were all made in Japan. \$\endgroup\$ – Robert Endl Aug 25 '16 at 2:14
5
\$\begingroup\$

Principal reasons were:

  • The capacitor plague of 1999–2002 – an attempt to reproduce a stolen Rubycon Inc.’s formula for electrolyte, which went bad.
  • Otherwise changing composition of electrolyte; more H2O (useful to obtain lower ESR) makes it more corrosive.
  • Cost optimization due to increasingly mass production of electronics.
  • Errors in design, process, or low-quality materials; bad quality control.
\$\endgroup\$
4
\$\begingroup\$

One reason may have to do with the circuit around the capacitors, not the capacitors themselves. Until (circa) 1980, most power supplies were operating at mains frequency (50 or 60 Hz), using a large filter capacitor after the diode bridge, and a linear post-regulator, using some more capacitors at mostly DC, with only a mall AC component. Not much trouble caused by RMS current within the capacitors, and (very) low ESR wasn't a big concern because even with a high-ish internal resistance, capacitors would not heat up much just by themselves.

Later, switch-mode power supplies (and post-regulators, including point-of-load step-down converters) became more and more popular, and they put a much larger RMS current on the capacitors they use. Therefore, the right choice of capacitors became more and more important, and unwise design decisions mattered more. Also, with miniaturization, more components end up in smaller enclosures, making heat dissipation more critical. The smaller you build your device, the harder it becomes to separate hot components from heat-sensitive capacitors. A small (5 mm diameter) 10 µF / 16 V capacitor rated at 2000 h / 105 °C next to a big heat sink? Bad idea. A large (25 mm diameter) 47 µF / 400 V capacitor rated at 5000 h / 105 °C placed in cool spot of your switching power supply? Might never even become a noticeable problem.

Also, for a while, circuits may have demanded more than what advances in capacitor technology could keep up with. Designers may not have been aware of the importance of IRMS ratings and internal heating. Add in the constant pressure to save pennies on any component, consider the fact that capacitors tend to be your more expensive components, conclude that this makes caps a focus area when it comes to penny-counting, and you have a good explanation.

So, to be fair, it's not only the caps, it's also the overall design and the caps' application in more and more demanding circuits.

That being said, I have happily used some devices with commercial switching power supplies for years without problems, and I have also replaced a fair amount of capacitors (dated from the late 70ies, for example, in stuff like high-quality reel-to-reel tape recorders or test and measurement equipment).

Then, ceramic capacitors are catching up. Before circa 2005, 22 µF at 25 V in a 1206 SMD package were uncommon. Today, you can use them instead of electrolytic caps or tantalum types, and not even spend more money. This makes it possible to make very good overall design decicions: Avoid tantalum caps (because they are very sensitive to current or voltage spikes, even very tiny ones. Use electorlytic caps only when you need much capacitance, and when you are able to choose large cans which typically have a much longer useful life. Take good ceramic caps for everything else.

\$\endgroup\$
  • \$\begingroup\$ Linear power supplies were prevalent in some parts of the world (say, USSR) long after 1980 (year of production). \$\endgroup\$ – Incnis Mrsi Aug 30 '16 at 10:21
  • \$\begingroup\$ By the way, why you deem a SMPS puts a much larger RMS current on the capacitors? There may be some difference from LPS, but IMHO even not two-fold. \$\endgroup\$ – Incnis Mrsi Aug 30 '16 at 10:26
2
\$\begingroup\$

Capacitors vary wildly in failure modes due to ageing, and it is actually untrue that older designs were always better.

Folks who repair vintage equipment will almost always replace certain capacitors without even testing them, and make sure they test others just to be sure.

For example, the old wax rectangles are pretty much toxic landfill when you crack open an old amplifier. They will have aged well out of spec. Not to mention the same equipment assumed certain mains voltage quality that has changed over the decades, which will drive your power or high voltage signal or decoupling caps well beyond their rated specifications even when new.

But, as others have pointed out, it is a complicated thing. Materials, manufacturing, markets have all changed a lot, which has had an effect industry wide. In general, though, modern capacitors at a certain price point per lots of thousands are much better than the same devices from the past.

My only refs for this are YouTube channels like Mr. Carlson's Lab, as seen on SE.EE sidebar!

\$\endgroup\$
  • 1
    \$\begingroup\$ Please, specify, to which places are your “mains voltage quality” consideration relevant. First of all, mains quality varies between buildings since it depends on the nature of load and mains design. Second, Ī̲ don’t see much change in Russia during last decades: the quality of mains supply was (and still is) reasonably good in Moscow and was (and still is) awful in many regions, especially rural. \$\endgroup\$ – Incnis Mrsi Aug 25 '16 at 13:32
  • 2
    \$\begingroup\$ The bulged capacitors in the "capacitor plague" were all lower voltage aluminum electrolytics. Low voltage as in "this don't go no where near line voltage." So, not power caps. \$\endgroup\$ – JRE Aug 25 '16 at 13:37
  • \$\begingroup\$ @JRE: What do you denote as “power cap” – the one after a mains rectifier? Ī̲ recently had to replace namely this one, in a 2005 LED monitor. \$\endgroup\$ – Incnis Mrsi Aug 25 '16 at 13:41
  • 2
    \$\begingroup\$ In the answer above, power caps are implied to be someplace where line voltage can get to them. The ones most hit by the problem were in PCs around the regulators for the CPU, and in smaller devices (like DSL modems) that ran off of external regulated power supplies. Not what I'd call "power caps." \$\endgroup\$ – JRE Aug 25 '16 at 13:44
  • 2
    \$\begingroup\$ The "mains quality" might be a partial recollection of the changes to European voltage standards. It's currently 230V +/- 10%, a 20% margin where it used to be only 16% in the past (220 V+10%/-6%) . \$\endgroup\$ – MSalters Aug 26 '16 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.