Toasted PSU

[grumpysmurf58]

It took a while to find this site and I hope I am posting this in the proper forum and that perhaps someone can help me.

I have an Antec SL400 psu (31/2 years old, so it is out of warranty) which did a slow burn on me a couple of months ago. Essentially it took a full ten hours for it to fry some wires. It managed to do this over a ten hour period. Anyway, all the +5vdc wires are burned (wires 4, 6, 19, 20 on the psu connector). Obviously, pins 4, 6, 19 & 20 on the motherboard side are also toast. The psu and motherboard connectors were fused together just enough, that I had to remove the the two as a single unit in order to separate them. When I pulled the motherboard connector off the board, the male portion of pin 4 came with it, stuck in its male pin.

After doing some lengthy research ( and coming upon this site) and reading the forums, I opened the psu and found at least 2 bad caps.

Seems most of the caps in this particular psu are of the Fuhjyyu brand. At least two of them have popped tops and the eloctrlytic is leaking out. The two huge Fuhjyyu caps have some goopy looking stuff at their bottoms, but can't really tell what it is.

I replaced the motherboard and psu (had a spare lying around). I purchased a new motherboard, swapped the cpu on to it and connected everything else up. I then powered the system up and everything seems to be ok.

So, the questions are:
1) were the shorted wires caused by the bad caps? (I am not an electrician or super-savvy pc expert),
2) would it be worthwhile to have this psu recapped?,
3) who would I contact about having it recapped?
and 4) any ideas where I might be able to get my motherboard repaired? It is a Gigabyte GA-7VAXP v1.0 socket A. Yeah, I know, socket As are out, but I like this board and I don't really want to upgrade to a newer board. Everything was working fine until the psu went bad. And still is.

[dood]

Re: Toasted PSU

Get a different supply. I wouldn't fix one that's had melted pins.

[linuxguru]

Re: Toasted PSU

> were the shorted wires caused by the bad caps?

Yup, fairly likely. When the caps in the PSU failed, the ripple current on the +5VDC rail that would normally have been bypassed to ground by those caps, was forced through the ATX connector and through the motherboard caps on +5V instead. This current can be sufficiently large (tens of Amps or more) to cause ohmic heating at the ATX connector, leading over time to overheating, oxidation, melting and failure.

These kinds of failures can be prevented by timely recapping of the PSU, but can be expensive to fix after they happen. At the least, a full recapping of both the PSU and mobo is indicated.

[999999999]

Re: Toasted PSU

It took a while to find this site and I hope I am posting this in the proper forum and that perhaps someone can help me.

I have an Antec SL400 psu (31/2 years old, so it is out of warranty) which did a slow burn on me a couple of months ago. Essentially it took a full ten hours for it to fry some wires.

Fried wires = one of two things
1) Bad contacts (less likely when more than one)
2) Problem component drawing far more current than it should.

It managed to do this over a ten hour period. Anyway, all the +5vdc wires are burned (wires 4, 6, 19, 20 on the psu connector). Obviously, pins 4, 6, 19 & 20 on the motherboard side are also toast. The psu and motherboard connectors were fused together just enough, that I had to remove the the two as a single unit in order to separate them. When I pulled the motherboard connector off the board, the male portion of pin 4 came with it, stuck in its male pin.

After doing some lengthy research ( and coming upon this site) and reading the forums, I opened the psu and found at least 2 bad caps.

These units are known to have cap problems, but that will not fry the wiring. It is very likely you have a damaged part powered by the PSU and this higher current exaccerbated, accelerated the demise of the PSU which already had a weakness.

[[QUOTE]Seems most of the caps in this particular psu are of the Fuhjyyu brand. At least two of them have popped tops and the eloctrlytic is leaking out. The two huge Fuhjyyu caps have some goopy looking stuff at their bottoms, but can't really tell what it is.[/QOUTE]

I feel you have two problems, the part consuming excess current, and a PSU weakness that caused it to fail under this load.

I replaced the motherboard and psu (had a spare lying around). I purchased a new motherboard, swapped the cpu on to it and connected everything else up. I then powered the system up and everything seems to be ok.

So, the questions are:
1) were the shorted wires caused by the bad caps? (I am not an electrician or super-savvy pc expert).

No, higher ripple would shut off the PSU long before that resultant current fried wiring. There were two separate problems, though the problem causing melting wires might have accelerated the PSU demise.

2) would it be worthwhile to have this psu recapped?,

If it has the approprate specs for current to power your present parts, then yes.

3) who would I contact about having it recapped?

It's really only cost effective if you can DIY.

4) any ideas where I might be able to get my motherboard repaired? It is a Gigabyte GA-7VAXP v1.0 socket A. Yeah, I know, socket As are out, but I like this board and I don't really want to upgrade to a newer board. Everything was working fine until the psu went bad. And still is.

Whether motherboard is repairable depends on what's wrong, excessive damaging ripple from PSU may make it unworthwhile as a repair candidate. If it just had bad caps it would be more straightforward.

[linuxguru]

Re: Toasted PSU

>No, higher ripple would shut off the PSU long before that resultant current fried wiring.

The total ripple current in the PSU doesn't increase when the caps start failing in the PSU, though the ripple voltage will increase. The ripple current flowing through the PSU cables to the mobo will definitely increase significantly - the mobo caps just appear to be another section of the secondary pi filter of the PSU, albeit with a small inductance (from the ATX cables to the mobo) in series.

Depending on the quality of the mobo caps on +5v and +12v, this arrangement can work for quite a long time (months to years), but will eventually cause cascading failures, usually due to one of two (or more) reasons:

1) The ATX connector is unable to handle the extra current, and the contacts heat up, leading to oxidation, higher contact resistance, and further heating leading to thermal runaway and contact meltdown.

2) The capacitors on the mobo on the affected rails are unable to handle the extra ripple current from the PSU, causing premature failures of caps on those rails. This causes progressive cascading failures through all the mobo caps on those rails, leading to increased ripple *voltage* on those rails. Components on those rails (Hard Drives, Northbridge, Southbridge, memory, PCI cards, etc.) will then start failing. Usually, hard drives fail first, followed by optical drives, memory, etc.

Depending on where the PSU protection circuits sense their voltages (usually at the ATX connector), the PSU won't shut down protectively in this failure mode until the amplitude of the ripple voltage gets high enough to trigger the over-voltage protection, if it has been provided/implemented correctly in the PSU. That's usually set at 15-18v for the +12v rail, so there can be a ripple voltage of +/- 6v on the +12v rail before the OVP triggers. Hard drives will fail long before that, when the ripple gets to +/- 2v. VRMs running on +12v can also fail, sending full rail voltages to Vcore and destroying the CPU.

[999999999]

Re: Toasted PSU

>No, higher ripple would shut off the PSU long before that resultant current fried wiring.

The total ripple current in the PSU doesn't increase when the caps start failing in the PSU, though the ripple voltage will increase.

Whenever the ripple voltage increases, so does the ripple current because these aren't all constant current parts, current ~ voltage^2 or at least a rise in quiescent. However there is a breaking point in that PSU still has an inherant threshold for shutdown.

The ripple current flowing through the PSU cables to the mobo will definitely increase significantly - the mobo caps just appear to be another section of the secondary pi filter of the PSU, albeit with a small inductance (from the ATX cables to the mobo) in series.

Certainly the motherboard caps get more of a workout, but the ripple current factor somewhat contradicts what you had written in the preceeding paragraph, that if it doesn't increase in the PSU there's no way it's going to increase on the wiring harness out of the PSU. I feel it does increase, but not so much that wires will fry just from failing caps, and plenty of us have seen PSU with failed caps and no wire or contact problems. PSU are not so different that this would be a cause:effect on one unit unless very unusual, but then not quite a lot of them too.

Depending on the quality of the mobo caps on +5v and +12v, this arrangement can work for quite a long time (months to years), but will eventually cause cascading failures, usually due to one of two (or more) reasons:

1) The ATX connector is unable to handle the extra current, and the contacts heat up, leading to oxidation, higher contact resistance, and further heating leading to thermal runaway and contact meltdown.

2) The capacitors on the mobo on the affected rails are unable to handle the extra ripple current from the PSU, causing premature failures of caps on those rails. This causes progressive cascading failures through all the mobo caps on those rails, leading to increased ripple *voltage* on those rails. Components on those rails (Hard Drives, Northbridge, Southbridge, memory, PCI cards, etc.) will then start failing. Usually, hard drives fail first, followed by optical drives, memory, etc.

Yes IF the connector can't handle a large current increase it will fail but there is no indication a failed (5V in this case) cap in the PSU will cause this. Perhaps all the connector contacts in the PSU plastic block connector were marginal and 5V was the only rail with a significant problem because it was the only one with much of a load. Perhaps a downstream part consuming 5V was faulty.

Merely having a bit higher ripple will increase current but not this much, and it is still a ripple as-in changing to a lower value as well. For example a connector rated to 6A continuous can operate at 8A with 50% duty, because the associated heat is averaged over the whole period.

Depending on where the PSU protection circuits sense their voltages (usually at the ATX connector), the PSU won't shut down protectively in this failure mode until the amplitude of the ripple voltage gets high enough to trigger the over-voltage protection, if it has been provided/implemented correctly in the PSU.

The PSU would have to be missing shutdown features if it did not do so in a situation where delivering enough current to do this damage, it will not stay on waiting for ripple to exceed a value only at the end of the ATX cable because of overcurrent low voltage, not high, shutdown threshold. A higher voltage it can't detect is not a problem, it's the current. Wires could carry 24V instead of 5V and all would be fine from the connector standpoint, failing to detect this won't do that damage.

That's usually set at 15-18v for the +12v rail, so there can be a ripple voltage of +/- 6v on the +12v rail before the OVP triggers. Hard drives will fail long before that, when the ripple gets to +/- 2v. VRMs running on +12v can also fail, sending full rail voltages to Vcore and destroying the CPU.

Certainly a very high voltage can damage parts, but there is not clear cause:effect here. There's only one specimen with bad caps amoung many which had caps even worse (bad enough the PSU couldn't work at all at that point and shut off). I think it fairer to say the PSU lacks or has malfunctional shutdown circuits, that the cap failure is certainly a problem but not directly related. If it were, melting wiring harnesses and connectors would be a regular event around here as the motherboards plus installed parts, loads (per era of system) aren't so different.

[grumpysmurf58]

Re: Toasted PSU

Perhaps the best thing to do is just replace the psu as suggested by dood. I can get a Thermaltake Pure Power TR2 430W (W0070) power supply for $39.99 from several different dealers. Is this a good psu for the price? It also comes with a 5 year warranty. Price, warranty, wattage and availability. Sounds good to me. What are your thoughts on this psu?

[999999999]

Re: Toasted PSU

Perhaps the best thing to do is just replace the psu as suggested by dood. I can get a Thermaltake Pure Power TR2 430W (W0070) power supply for $39.99 from several different dealers. Is this a good psu for the price? It also comes with a 5 year warranty. Price, warranty, wattage and availability. Sounds good to me. What are your thoughts on this psu?

The question is not is it good for the price, it is whether it will be acceptible for the system you're trying to use it in. It is a low-middling quality unit which would not have been suited for your GA-&VAXP because of insufficient 5V current capability, but for running a more modern system of modest specs using 12V rail for CPU VRM subcircuit power, you could do worse for the same price, or better for more $$.

Don't forget about cause and effect. It seemed at first like the PSU killed the board but it could be the other way around that the board killed the PSU. That is what the evidence suggests, that the board was drawing an excessively high amount of 5V current and this is what accelerated the demise of the PSU. If you were to find the board fault and repair it, ok, but at what cost and time?

Determine exactly what parts the new PSU will power, and then go from there selecting one suited to that load. Many newer Antecs are better than they used to be, I would not rule out the company's products for that reason. The old PSU might be worth repairing except it's more work since the wires are fried. Since you asked who you would have do it (suggesting you won't be doing it yourself) then it does seem unworthwhile to pursue repair instead of replacement.

[dood]

Re: Toasted PSU

I've had a total of 3 sl300s Antec supplies die of bad caps, and two of them ruined the motherboard by melting the yellow pin at the ATX connector. Twice is too often to be coincidence. This is why I avoid Antec altogether anymore.

[999999999]

Re: Toasted PSU

Considering dood's losses, I'm starting to suspect some of these SL(n)00 PSU do have bad connector contacts in addition to bad caps. Too many PSU caps fail without any connector problem for there not to be an additional variable contributing to the cause.

Dood, which connector, the 20 pin or 4 pin? If the 20 pin I wonder what was being powered by that 12V lead. Then again we lack some info like whether both were powering same make and/or model or even same board in that the board contact might've been fouled before 2nd PSU was used. Even then it wouldn't excuse Antec using bad caps.

[dood]

Re: Toasted PSU

it was the 20 pin on a Socket A mobo without an Aux connector. To be fair, I guess that means the 12v was under a good load. Still, I've only had it happen with Antec supplies.