Sunday, April 09, 2006

"More Details"

Very interesting site:

http://www.siliconchip.com.au/cms/A_30328/article.html
Motherboard Capacitor Problem Blows Up
By Peter Smith.


Quoting from that page:

     "Initially, there were only two clues to the mystery. First, the failing capacitors were more often that not to be found in the power supply section of motherboards. The capacitors used in this area are characterised by their need to have very low ESR (Equivalent Series Resistance- see panel).
     Second, most of the failing capacitors were identified as Taiwanese in origin. That's not too surprising at first glance, as Taiwan manufactures about 30% of the world's aluminium electrolytics (22.5 billion a year).
     In September, "Passive Industry Components Magazine" published a story that exposed the reasons behind the unusually high failure rates. They reported that the failures were directly related to the use of faulty electrolytes in the manufacturing process.
Industrial espionage?
     The story describing how the electrolytes came to be faulty reads like a lot of fiction. It begins in Japan, at a major capacitor manufacturer. A materials scientist for the Japanese company resigned and went to work for a Chinese capacitor manufacturer. While there, he reproduced one of the electrolytes used in his former employer's premium (low-ESR) aluminium electrolytic products.
     Staff working with the scientist then defected, taking the secret electrolyte formula with them. They used the formula to manufacture their own electrolyte, which they subsequently flogged to major Taiwanese capacitor manufacturers at bargain prices. Unfortunately, their reproduction of the formula was flawed and the rest is history."

Espionage!  Japanese scientist defect to China!  Trade secrets leaked to Taiwan!  Why isn't this in the mainstream news and on TV, with reporters crowded around the Japanese manufacturer (Nichicon, of Kyoto?) and with other reporters being sent to China and Taiwan to track down the scientist who defected and the company (companies?) in Taiwan which bought the technology from China?  If it's true, it's better than fiction.  If it's not true, let's here the companies involved refute it and give the real reasons!


     "As mentioned earlier, these capacitors are of a specific type; they have very low ESR. To understand the need for this requirement, let's take a brief look at the circuitry involved.
     Low-ESR capacitors.  Probably due to the fact that CPU core voltages change so often, designers have been forced to implement sections of the power supply circuitry on the motherboard. The standard power supply box still provides the usual 12V, 5V and 3.3V rails, but the lower voltages for the CPU core are provided by further step-down circuitry on the motherboard.
     This on-board switchmode step-down circuitry runs at high frequencies (over 100kHz) to minimise the required inductance and filter capacitance. A key ingredient in this recipe is physically small electrolytic capacitors that can handle high ripple currents at high frequencies. In short, they must have very low impedance at the switching frequency."

Here we go.  Just as I suspected!  They are not just any old capacitors, but rather of a special type!


     "The "Rubycon" (Japanese) brand ZL and ZA series ultra-low impedance capacitors will be suitable in most cases. They're available locally from Farnell Electronic Components - see Farnell's web site at www.farnell.com for more information.
     Caution: the standard ZL and ZA series may be marginally larger in diameter than the original parts (10mm versus 8mm).
     All electrolytic capacitors have a finite life, measured in thousands of hours. Unlike the exceptional cases discussed in this article, there are usually no external signs that a capacitor is nearing its end of life. However, it is possible to determine whether a capacitor is serviceable or not by measuring it's ESR (Equivalent Series Resistance)."

So... where to get the replacement capacitors for my new custom-built box?  Any recommendations?

Lyle

No comments: