View Single Post
  #16 (permalink)  
Old 7th February, 2008, 11:52 PM
Gizmo's Avatar
Gizmo Gizmo is offline
Chief BBS Administrator
Join Date: May 2003
Location: Webb City, Mo
Posts: 16,178
Send a message via ICQ to Gizmo Send a message via AIM to Gizmo Send a message via MSN to Gizmo Send a message via Yahoo to Gizmo Send a message via Skype™ to Gizmo

Originally Posted by trodas View Post
Gizmo - whoa, whoa, whoa Carefull out there. Technology make a great leap forward in ceramics.
Hmm.......I admit it has been a while since I was in the game. The last I knew, surface-mounts in that size stopped about about .1 uF, but that was 15 years ago.....

Now bear in mind, that year has only 8 544h Hence well-cooled cap has to stay alive at least 14,9 years with every parameter in limits. Caps failing in good environment in like year or two are most common. Some excellent designs with Teapos (mine ABit ST6R) make 5 years. But that is still far from what they should last.
Hence they are bad caps.
I hope we are getting tripped up by language here.

A 'bad cap' is one that fails prematurely, when operated within it's normal parameters. Now, again, it's been a while since I was in the game, but I can't imagine the likes of Rubycon, Nichicon, or United Chemicon putting out parts that fail prematurely, even today. I submit that the vast, VAST majority of 'bad caps' are in fact not being operated within their design parameters, and THAT is why they are failing. I base that opinion on several years spent doing failure analysis and design of switch-mode power supplies. Granted, that experience was 15 years ago, but the basic principles of switching technology remain unchanged; you use a switching transistor to drive current through an inductor, charging a field. When you turn off the transistor, the field of the inductor collapses and the resulting current flow is used to charge a capacitor. By varying the frequency and/or duration of the switching pulse from the transistor, you control the voltage impressed upon the capacitor.

In addition, my own observations of various mobos and PSUs that I have disassembled has shown that the components that tend to fail are in fact being operated at or very near to their design limits. As you yourself have already noted, even quality parts are going to fail after about 8,000 or less (i.e. 1 year) when operated near their limits, and parts can be operated near those limits without the user even realizing it. Case in point: my Abit NF7-S mobo, running water cooling on the CPU, was running the VREG at about 100C because there was no air flow across that part of the mobo. Had I continued operating it that way, I would have had a dead board in less than a year (actually, I'd probably have written it off as a crap mobo when it started getting unstable in about 4 months). Instead, I put a fan on that section, got the operating temp down to about 45C, and I ran it for 4 years, until it was finally destroyed by a fire I had in my house.

In short, parts generally don't fail prematurely (with the exception of a small percentage of 'infant mortality', which a good manufacturing process will weed out), unless they are being operated outside their design limits. When you start seeing a rash of such failures, you have to look at why the design is pushing the parts past their limit, not just assume the part is 'bad'.
Reply With Quote