I cant understand this.

[gastorgrab]

I'm sorry, but how can one machine with leaking caps work so well while another machine stops working if you breath too hard on it?

I reassembled my fathers machine with the new Enhance power supply. It now seems incredibly stable. This is the system where the Antec SmartPower originated. After pushing the current machine into place i decided to shine a flashlight into the power supply of the EPoX 8RGA that i recently recapped. This was dads temporary replacement.

This machine showed no behavioral abnormalities but over half of the caps on the MB were leaking.

The first thing inside that i can identify as a capacitor through the fan grill is a blue 10mm radial.....................wearing a fur hat.


How the hell can this thing run so stable?
.

[Harvey]

Re: I cant understand this.

Are you sure its not just glue on the cap?

[willawake]

Re: I cant understand this.

can you do a favour, dont repair it just let it run and record the number of days until it fucks everything. i need this data.

[stevo1210]

Re: I cant understand this.

I'm sorry, but how can one machine with leaking caps work so well while another machine stops working if you breath too hard on it?

I reassembled my fathers machine with the new Enhance power supply. It now seems incredibly stable. This is the system where the Antec SmartPower originated. After pushing the current machine into place i decided to shine a flashlight into the power supply of the EPoX 8RGA that i recently recapped. This was dads temporary replacement.

This machine showed no behavioral abnormalities but over half of the caps on the MB were leaking.

The first thing inside that i can identify as a capacitor through the fan grill is a blue 10mm radial.....................wearing a fur hat.


How the hell can this thing run so stable?
.

I have the same idea as you. Why is it still stable with a bunch of bad caps??
I had an IBM netvista with a bunch of failed GSC caps and it was rock stable and I didn't know it had bad caps until I opened the case up....

So yeah, I have the same question. Why is a motherboard with a bunch of bad caps still stable??

[Spacedye69]

Re: I cant understand this.

Why do some General Motors cars still run? Sometimes its just pure blind luck.

[linuxguru]

Re: I cant understand this.

Depends on the CPU Vcore load and various other things, like the number of MLCCs below the CPU socket, the design margin on the VRM O/P, etc.

There are desktop P3/Celerons with a Pd of 12 - 20W or so - they'll probably run OK even if there's a single functional bypass cap on the output side of the VRM, as long as there are a number of large MLCCs below the CPU socket to take the core-clock and FSB ripple.

OTOH, all bets are off even with a single busted Vcore cap on a P4 board with a Prescott 100W+ CPU - the design margins are far tighter.

{\begin rant} Sometimes, it all depends on how much leeway the bean-counters have to fuck things up by omitting extra components that have been spec'ed by real engineers. That's also why bridges sometimes collapse. {\end rant}

[Tom41]

Re: I cant understand this.

I had several PCs at work that were all of the same basic design, same motherboard and components. Every PC had GSC caps, and in each case every single cap near the CPU was vented! Oddly, two of the PCs were still stable - but one of them was doing random reboots and crashes.
Surely the vented GSCs didn't help, but a contributing factor could have been that the fan in the PSU had been dead a long time. The other vented-GSC PCs still had PSUs with working fans - perhaps this prolonged their life?

[gastorgrab]

Re: I cant understand this.

can you do a favour, dont repair it just let it run and record the number of days until it fucks everything. i need this data.

With my luck with new parts these days i'm almost afraid to do anything about it.
.

[severach]

Re: I cant understand this.

Capacitance of a blown cap typically goes down but not always. If a few 3300uf caps turn into 330uf caps you can expect trouble every time. I pulled 4 vented Rubycon 3300uf from a Dell Dimension and the capacitances measured as high as 7500uf. Other than a mucked up Windows XP install, the Dell actually ran quite well. Surely the voltage drops as the capacitance rises but many caps are double to triple the voltage they need to be.

A particular Dell P4 power supply model with Arcon caps has a very high failure rate. I pull and test all these caps because many fail without showing signs. Some Arcon caps from these test more than double the printed value. Those are probably not the ones making the supply fail.

I also had an Asrock P4S61 motherboard that would not post because of two 1000uf filter caps. Each tested about 575uf. Looks like the bean counters left that one on the edge of the abyss.

[kc8adu]

Re: I cant understand this.

it may take awhile before the vented caps dry out enough to cause trouble.

[davmax]

Re: I cant understand this.

Severach. If capacitance does rise two things will happen. Ripple will reduce ( if ESR does not rise significantly) this may slightly raise the measured output voltage and VRM response time will increase.

[RJARRRPCGP]

Re: I cant understand this.

I have the same idea as you. Why is it still stable with a bunch of bad caps??
I had an IBM netvista with a bunch of failed GSC caps and it was rock stable and I didn't know it had bad caps until I opened the case up....

So yeah, I have the same question. Why is a motherboard with a bunch of bad caps still stable??

It probably wasn't stable, even when it appeared to be, bet you Prime95 would fail pronto!

[yanz]

Re: I cant understand this.

why, my ep 8k3a with more than 6 bloated gsc caps was still running f@h perfectly for months, untill it doing more bluescreen in the end..
it's just a matter of time.

[PeteS in CA]

Re: I cant understand this.

As lg pointed out, different loads and regulator circuit contexts make for a great deal of difference. Differences in cooling contribute. Low Z caps are also highly tuned electro-chemical systems. "Minor" differences can have major performance impacts.