I'm stuck - recapped whole board gx620

[edspegeddd]

I've been recapping a lot of the small Dell GX620s and about a third of them do not POST even after swapping out ALL caps with a mix of brand new rubycon MCZs and Panasonic (not sure on the series). Is there anything else I could be missing? I swap out RAM, CPUs, and nothin.. I guess I was under the impression that replacing the caps were a guaranteed fix heh. Should I just be giving up on these..?

Some background on the PCs: they are on 24/7 eventually they shut off and will not power back on. This is all I really know. When I get them they all have visibly busted caps. Maybe a chip or something get busted too?

[Toasty]

Re: I'm stuck - recapped whole board gx620

Did you inspect the PSU's? If you swap board to a working unit, same problem?

[mockingbird]

Re: I'm stuck - recapped whole board gx620

1) Did you get the polarity right?
2) Did you push out trace pads by forcing the new caps in?

[severach]

Re: I'm stuck - recapped whole board gx620

Two easy problems that come to mind are board damage as already mentioned and incomplete soldering. I've fixed some boards repaired by others by adding more heat. They didn't add enough heat to get the lead free solder to melt and combine with the lead filled solder. Over many joints too little heat or bad technique occasionally produces one that doesn't connect.

The GX620 around here must be built to a less low standard because I haven't had to rebuild any yet. They should be similar to the GX280 boards many of which are dead after a recap. Topcat does plenty of boards so will have the best failure rate estimates.

On a GX280 I see 3 SMPS: 1.8v for memory, 1.5v for Southbridge, and variable for the CPU. Two sets of capacitors with an inductor and some MOSFETs in between is a clear indication that it's a SMPS circuit. Measuring the voltage and looking what else is nearby gives a good indication what the purpose of that supply is.

For the GX280 the 1.8v SMPS caps near the memory slots rarely fail. The 1.5v SMPS caps and the CPU caps fail often. A well designed SMPS will shut down when the circuit goes bad. A poorly designed SMPS can go over voltage (like Bestec) and burn nearby circuits. For the few I've taken apart to test I haven't been able to find faults in the CPU MOSFETs. I blame the Southbridge because I can't test it for less than the cost of a replacement board. Anything bad caps were powering is suspect.

For example, the 1.5v Southbridge supply consists of 3 12v input caps and 2 1.5v output caps. One possible scenario is that if the 12v caps go bad faster then the supply will just shut down with no damage but if the 1.5v caps go bad faster then the voltage could swing wildly destroying the South Bridge. If this scenario is correct then 2/5 of all GX280 motherboards with failures in the 1.5v supply should be dead. ESR on all 5 caps is likely to be rising and each side may have different tolerance of bad ESR so determining which side is going bad faster is not straightforward.

The last easy problem is that any Nichicon HM is suspect including the small and non bulged ones. Most of my caps come from used boards and I test all 1000uf and above caps before they go into the cap box. I've had to discard some good looking Nichicon HM caps because the ESR was too high. They were not bulged so didn't draw my attention at the time. Had I been more hostile to HM I might have saved the board. Now I test any HM before it goes into the box.

[kc8adu]

Re: I'm stuck - recapped whole board gx620

reusing caps is not worth it unless its an emergency.

[MXM]

Re: I'm stuck - recapped whole board gx620

nice replies, i appluase sharing such info

@op: ive been in your situation and i'd advise you to take a good read of the tips and directions given here, i personally do not have a better opinion than them, but this what i'd start doing: what you read above, testing components from a workin board, reheating the boards, also reheat cpu area only if still nothing....

and remember the name of the game is patience

goodluck!

peace

[PCBONEZ]

Re: I'm stuck - recapped whole board gx620

GX620 blow the small caps too. - Especially if they are 85C.

https://www.badcaps.net/forum/showthread.php?t=6055

Specifically:
https://www.badcaps.net/forum/showpo...6&postcount=14

[edspegeddd]

Re: I'm stuck - recapped whole board gx620

Thanks for all the input guys im not giving up. Well I am guilty of knocking out contact pad when recapping. What can I do to repair this?

I am sure that the polarities are correct. I've read the gx620 thread and had become aware of the 85c caps, so I make sure my new caps are all 105c. I replace everything, from 2200uf all the way down to 22uf. With ESR tests I find that the 2200uf 820uf and 220uf almost always test high. As for the psu , I keep a working that I use on all the recapped machines just like my RAM and CPU.

As for my soldering, it is pretty complete the caps don't budge (except when I knock out a contact point - again how to avoid this?).

I want to try to fix there contact points and see what happens?

[PCBONEZ]

Re: I'm stuck - recapped whole board gx620

Burned off contact pads are from too much heat.
Not doing it is just an experience thing.

If there is a trace leading from the pad then scrape the coating off a section of trace and solder a small bare copper wire to it. [Small wire like pulled out of some old phone or network cable.] Then solder the other end to the cap lead.
You may want to get a 30 watt iron for that repair or at least switch to a really small tip.
.