underspecified electrolytic capacitors,
 a new paradigm in electronic planned obsolescence

a rant by ralph klimek

keywords:  bulging electrolytic capacitors, ESR, equivalent series resistance,rant

This rant's subject concerns the usage of electrolytic capacitors in power supply smoothing applications to enforce planned obsolescence strategies in both professional and consumer grade electronic equipment.

The subject matter of today's rant  was inspired by the slowly degrading performance of equipment that I maintain at work. The equipment is supplied by a leading manufacturer of networking equipment. I have no quarrel with the vendor, they have done the utmost to help us  and will accept and replace damaged equipment without arguement. I think they are the innocent party in this problem, so I have taken pains to ensure that they cannot be identified by my rant. I have also replaced a number of damaged capacitors on expensive computer PC motherboards , restoring lost reliability.

However, I decided to mount my soapbox when I found this very nice PC ATX power supply in the dumpster. It appeared to be  almost new, it was the advanced atx model with sata connectors and the new wide atx motherboard connector, a desirable item. What was it doing in the rubbish ?   Knowing from past experience that good things are discarded . not for lack of function, but because they are no longer bright and shiny.  I took it home and put it on my dummy load.  It powered up for a few seconds and then crowbarred, ie, the overvoltage alarm had tripped and shut down the primary inverter. This power supply was effectively dead, but not irreversibly so. At least not to me.  

There were no obvious signs of burnt out components and a quick multimeter pass revealed that the inverter transistors, fuses, diode et al  were good.  I then noted one peculiar thing.  There seemed to be very few electrolytic capacitors on the low  voltage  output side.  Good design practice dictates that the 5V rail have not merely one but 5 or 6 capacitors in parralel. The reason for using so many is because of the high current spikes that are coming out of the step down transformer  and shcotky rectifiers are of many tens of amperes of peak current.  Your logic circuits dont want this kind of rubbish, they prefer nice clean DC please.  Real electrolytic capacitors have a parasitic resistance that is frequency dependant, electrode geometry dependant and electrolyte dependant. This parasitic resistance is real, and, being real, really turns amps into watts!  These high current pulses heat these devices.  You put many capacitors in parallel to spread the quite onerous burden of sinking these multi ampere pulses.  

Electrolytic capacitors are sold with two primary specifications, maximum sustained working voltage, and capacitance.
There are other specifications, sometimes published by the vendor, usually not, and that is equivalent series resistance, (ESR).   For simple smoothing applications , for example,  linear PSUs (plugpacks et al), audio frequency coupling and decoupling,  electrolytic caps  can have moderate to high ESR  and work perfectly fine for these applications and have an arbitrarily long life if their voltage rating is respected.  This life can be thirty years or more. I have seen fifty years life in some pioneering equipment.

Modern switch mode power supplies with their extreme pulse currents demand low ESR devices. These first become available about 30 years ago with the first generation of switchmode regulators.  The equivalent series resistance is never indicated on the device itself, unlike the voltage and capacitance rating. It is only ever specified in the manufacturers data sheets...if available.  

Like all electronic components, the power dissipated, temperature and lifetime relationship  is well characterized and governed by some simple rules of thumb, allowed for by statistical analysis.  The general rule of thumb is that for every 10 degree C increase in operating temperature, the average lifetime of an commercial electronic component is halved.  Devices supplied for defense or advanced commercial needs are very tightly specified with temperature derating curves and accelerated aging testing that can reliably predict device lifetimes given a service temperature and operating condition.  Are you smelling a rat here ? Put on your tin-foil hat.  I recenly acquired a capacitance meter that could measure electrolytics precisely, by timing how long a charge took at constant current.  It revealled the obvious. A ruptured electrolytic has lost over 99 percent of its rated capacitance.  One that had merely bulged, but not actually ruptured, had lost anywhere between 50 to 90 percent of its rated capacitance.

I have recently come to the conclusion that the cause of the "bulging" is due to the absence of one key stabilizer in proprietary electrolytes. I believe ( without proof ) that the missing ingredient is a gas recombination catalyst. The gas is hydrogen and oxygen from simple leakage current induced electrolysis.  That catalyst is most likely a trace of platinum or palladium. That alone is sufficient economic reason to exclude it !

The conditions under which "low ESR" devices must labor are onerous.  The thermal conditions inside the average consumer PC PSU are particularly onerous.  The pathetic excuse of a cooling fan serves only to exhaust pre-heated air from the  PC enclosure to the outside.  You cant cool something with hot air, unless your thing that you are trying to cool is hotter already! Power electronics must be kept cool, this is not news.  An electrolytic capacitor is particularly vulnerable to  over temperature conditions especially in switchmode smoothing applications.

some theory

The capacitance is a measure of how much charge an object can store. It is a strong function of geometry and material properties. The capacitance is inversely proportional to the seperation of the electrodes. In the real world, electrodes are seperated by air, oil, paper, assorted polymers. Such capacitors do not have the capacitance required to perform the SMPS
smoothing function.  However, what if the seperator was an molecular thickness of an impermiable substance...say aluminium oxide ?   It has a very high band gap, is mechnically strong, it adheres to a metallic aluminium substrate with great tenacity. It is easy to form and better still it is cheap!

An aluminium electrolytic capacitor was invented in the 1920s, its nothing new.  There are two metal foil electrodes, a mecanical sepparrator , ususally paper or now advnaced polymer mesh and a water based electrolyte, usually based on ammonium borate and  proprietary stabilizers. The electrodes and seperator is rolled up into a cylinder, canned with electrolyte solution and then hermetically sealed.  The actual insulator is a molecular film of aluminum oxide which is grown be electrochemical action after the first "forming charge" is applied.  One electrode of the capacitor is the metal with oxide film, the other electrode is the metal, it is the conductive electrolyte.  The "plates" of the capacitor being only a couple of molecules thick, and being so close to each other, the geometry of the device now acquires a very high capacitance and large amounts of charge  can be stored at moderate voltage.

The ESR of such a device is governed by the specific molar conductivity of the electrolyte and the thickness of the metal foil in the electrodes.  A low ESR device is necessarily larger.  A high temperature and high voltage will lead to electrolysis of the electrolyte and the inevitable generation of gas when the water is broken down into its hydrogen and oxygen. A good capacitor may even have a catalyst to recombine the gases. (that is speculation on my part)

Many of you reading this may now be going "ah-ha!" .  You will have come across the recent phenomena of the "bulging capacitor".  The modern cap has inbiult fracture lines stamped into the little metal can. When internal gas pressure is extreme, the can will gently rupture and expell their contents. They used to detonate! When a capacitor ruptures it irreversibly looses the electrolyte that is essential to its operation. Its capacitance goes down and its ESR goes up, thus dissipating even more power and getting hotter and hotter.  The DC that it is trying to smooth becomes noisier and riddled with transients. Your equipment becomes intermittant, behaves like its got gremlins, your pooter keeps blue screening untill it just stops working. Is this familiar....anyone ?  Open up your power supply and have a look.  Examine your computer motherboard. If you are getting aberant behaviour it could be due to bulging caps. They can be replaced, any competant electronic hobbyist is capable of doing so.

for the  tin-foil hatted or the bright-young-things-with-harvard-mba
conspiracy theory one
Equipment lifetimes can be controlled by carefull selection of switch mode regulator smoothing capacitors.
conspiracy theory two
This has already happened

images of damaged equipment

thumb.imgp4305-bulging-caps.jpgthis otherwise well made network gear had become erratic and intermittent as a result of noisy power rails caused by these damaged capacitorsthumb.imgp4377-psu-bulging-caps.jpg
damaged caps from this PC PSU these have ruptured and expelled their contents
damaged caps from PC PSU. The measured capacitance was only 150uF instead of 3300uFafter my repairsafter my repairs close-up
thumb.imgp4376-psu.jpgThe real question is why were there so few output smoothing caps in this PSU. No doubt the ESR and capacitance rating was in spec when this was assembled but it clearly did not stay working for very long. My replacements came from the junkbox. For completeness and piece of mind, I will be  making an outboard PCB containing extra paralled capacitors to ensure that the PC will have clean 5V and 12V railsthumb.imgp4379-psu.jpg
otherwise a really nice, expensive PSUotherwise a really nice, expensive PSU

Ha ! Thats nothing. I have recently recovered some nice lcd flatscreen monitors that were discarded because they were none functional.
The business part, the lcd flat screen had in all cases been carefully and well constructed by the Japanese and Taiwanese flat screen businesses that do this sort of thing very well. However, the power supplies had been made by the cheapest contractor that could be found on the day.  The problem and solution should come as no surprise.  These otherwise perfectly good lcd screens were dead because of bulging electrolytics  in the rubbishy power supplies.  Replacing these has now brought an embarrassing over-abundance of good flat screen monitors.  Gosh this world is weird.  Here is another tip. If you find a discarded monitor in the dump that uses an external power brick, the usual reason for discarding the monitor is that the previous owner has lost the power brick, experience has shown that that is allmost never anything wrong with the flat screen.

rear of a Mitsubishi 17 inch lcd monitor the psu board is made by Benq replacing these 3 bulging caps got it going

copyright declaration: none, this article is copyleft, quote it, copy it, use it as you see fit

mod record page created Tue Jun  2 19:17:04 EST 2009
added flat-screen monitor image

it was a dark and stormy night...