A few months ago, I bought a pallet full of GX620 USFF systems from an off-lease recycler, dirt cheap, condition unknown. Turns out they were in near mint condition, only thing missing from them were the hard drives, was a good score. Only a couple of them actually had bad caps, all the rest worked on arrival.
I'm sure most all of you are familiar with the infamous SX280 and its bad cap problem.....well the GX620 USFF is basically the next step up from an SX280, even the cases are interchangeable. The motherboards are a little different as far as cap layout, the GX620 actually has less 2200uF and more 1800uF caps.
Anyway, the problem areas on these systems are the same as the SX280, in that the larger caps on the south end of the system are always fried. I've seen these systems fry UCC's, Rubycon's, and even Panasonics. The common failures are the 3x 820uF 6.3v caps around the northbridge chip (the 2 in front by the power button usually don't cause any trouble, but I changed them with polies anyway), 7x 2200uF 6.3v, 1500uF 16v, and if the fans become clogged with dust bunnies, they'll also roast the 1800uF 6.3v & 1000uF 16v caps around the processor.
It came to me......I've poly modded a lot of SX280's, but unfortunately, theres not a whole lot of poly options for the south end of the board, which is where the majority of the failures are.... What could I use in their places... Well I fired one up, upside down on the bench and started measuring voltages...
All of the 820uF 6.3v caps had 3.3v on them. All the 1500uF 16v caps had 12v. The 2200uF caps were a mix ranging between 1v to 5v. Of course, all the 1800uF 6.3v caps were just VCORE. The 1000uF 16v caps behind the heatsink had 12v on them. The 1000uF 16v caps were the only lytics I didn't change to a poly, as I didn't have anything polymer in a 16v 8mm config.
Anyway, here's what went where:
the 820uF 6.3v lytics were changed to 560uF 4v polies
the 1800uF 6.3v lytics were changed to 820uF 2.5v polies
the 1500uF 16v lytics were changed to 330uF 16v polies
the 2200uF 6.3v lytics were changed to 1500uF 6v polies
Here's some pics of the test subject....
The original UCC's:
Some rubies too:
After the poly recapping:
Ok, now to see if this is going to work....
YAY, it POST's
It ran some burn-in cycles for a while without any hiccups...
Quick OS install, which went without a hitch....
All done, up and running!
I'll find some use in the shop for this system, something that will keep it running 24/7. These don't last more than a couple years when they're allowed to run 24/7 with lytics, curious to see how long this one will live.
Stability will also be something to keep an eye on. Most values I subbed with polies were close to the originals, except the 330uF 16v in place of the 1500uF 16v, that was the biggest deviation in the system. I had a hand full of the 1500uF 6v polies, I bought them at a local electronics shop a few years ago, and had about a dozen of them left. That store is now out of business, and I haven't found another source for these yet... I may run another test using 330uF polies in place of the 2200uF lytics....that could be a more interesting one...
Anyway, cheers to the worlds first full polymer GX620!
I'm sure most all of you are familiar with the infamous SX280 and its bad cap problem.....well the GX620 USFF is basically the next step up from an SX280, even the cases are interchangeable. The motherboards are a little different as far as cap layout, the GX620 actually has less 2200uF and more 1800uF caps.
Anyway, the problem areas on these systems are the same as the SX280, in that the larger caps on the south end of the system are always fried. I've seen these systems fry UCC's, Rubycon's, and even Panasonics. The common failures are the 3x 820uF 6.3v caps around the northbridge chip (the 2 in front by the power button usually don't cause any trouble, but I changed them with polies anyway), 7x 2200uF 6.3v, 1500uF 16v, and if the fans become clogged with dust bunnies, they'll also roast the 1800uF 6.3v & 1000uF 16v caps around the processor.
It came to me......I've poly modded a lot of SX280's, but unfortunately, theres not a whole lot of poly options for the south end of the board, which is where the majority of the failures are.... What could I use in their places... Well I fired one up, upside down on the bench and started measuring voltages...
All of the 820uF 6.3v caps had 3.3v on them. All the 1500uF 16v caps had 12v. The 2200uF caps were a mix ranging between 1v to 5v. Of course, all the 1800uF 6.3v caps were just VCORE. The 1000uF 16v caps behind the heatsink had 12v on them. The 1000uF 16v caps were the only lytics I didn't change to a poly, as I didn't have anything polymer in a 16v 8mm config.
Anyway, here's what went where:
the 820uF 6.3v lytics were changed to 560uF 4v polies
the 1800uF 6.3v lytics were changed to 820uF 2.5v polies
the 1500uF 16v lytics were changed to 330uF 16v polies
the 2200uF 6.3v lytics were changed to 1500uF 6v polies
Here's some pics of the test subject....
The original UCC's:
Some rubies too:
After the poly recapping:
Ok, now to see if this is going to work....
YAY, it POST's
It ran some burn-in cycles for a while without any hiccups...
Quick OS install, which went without a hitch....
All done, up and running!
I'll find some use in the shop for this system, something that will keep it running 24/7. These don't last more than a couple years when they're allowed to run 24/7 with lytics, curious to see how long this one will live.
Stability will also be something to keep an eye on. Most values I subbed with polies were close to the originals, except the 330uF 16v in place of the 1500uF 16v, that was the biggest deviation in the system. I had a hand full of the 1500uF 6v polies, I bought them at a local electronics shop a few years ago, and had about a dozen of them left. That store is now out of business, and I haven't found another source for these yet... I may run another test using 330uF polies in place of the 2200uF lytics....that could be a more interesting one...
Anyway, cheers to the worlds first full polymer GX620!
Attached Files
Comment