Why does the +12v cap always seem to die first?

[Maxxarcade]

I've had a handfull of power supplies apart the last couple weeks, and almost all of them had a bulging cap on the +12V and no where else. Does this one go bad because of the type of load (hard drives and CDROMS etc)? One was a cheap generic supply, and the other a Sparkle 250W.

[japlytic]

Re: Why does the +12v cap always seem to die first?

On P4 mainboards, there is an ATX12V connector, and the board uses a couple of amps of the 12V rail for the DC/DC converters for the CPU.

[linuxguru]

Re: Why does the +12v cap always seem to die first?

Yup, the ripple current on +12v is fairly high on P4 boards because that's used as the input (through the 4-pin ATX12v connector) to the VRM for supplying the high-current Vcore rail.

The +12v caps on the same PSU, connected to (say) an Athlon board without the 4-pin +12v connector, would probably have lasted longer - but the tradeoff would have been higher load on the +5v rail, which was usually used as the input to the Vcore VRM on Athlon boards. The +5v caps would have probably failed quicker instead.

[gonzo0815]

Re: Why does the +12v cap always seem to die first?

Most older PSu`s are desinged to cope with high currents at low voltages. Thus the 12v line was usually not considered to be under any seriouse load. And as usuall, as Intel and then AMD board vendors altered this, the psu vendors won`t do a complete ne desing. Adding a few wirres more and a p4 conector ins`t that expensive. Thus i think those failures are from an not well suited ripple current rating of the caps they used.

[Maxxarcade]

Re: Why does the +12v cap always seem to die first?

The thing is, none of these were used on ATX12V boards, only older 20 pin boards (PC Chips M810 v.5.0 and v.7.0a).

[gonzo0815]

Re: Why does the +12v cap always seem to die first?

May be the caps where very undersized & mediocre too. The 12v line had been very long the highest allowable rippl & noise in ATX specs. This means, that it was not the line an vendor would have had put mutch attention on it. The most attention from a desing perspective was on the lower voltage line, thus they skimped out on the 12v line. I have seen many psu`s with as low as may be 470uF to 1000uF on single cap without Phi filter. It needs probalby not mutch imagination, that this is even for a low load, may be 2 to 4 A for HDD &dvd isn`t suffizient. MAy be the input capazitors on a VRM unit would lessend the strain a bit from the psu`s ones.