So apparently chip revision somewhat matters depending on the board. Going through my parts box I've found a nforce2 socket A board that probably died of natural causes. The south bridge simply fried itself over time. Though this motherboard turns on the south bridge gets to about 50C in a second and it's in constant reset. Never mind since it uses revision 'A' and I need revision 'S' south bridge. Tomorrow I'll have time to take off the bad chip. Might make some pictures along the process.

Is your POST diagnostics card new or did you have it for a while before it went faulty?

Tetrium wrote on 2023-05-31, 10:29:

Is your POST diagnostics card new or did you have it for a while before it went faulty?

It's from Ebay. Single chip (CH311L) 2015. PCI/ISA type. I have a bunch of different ones. After the accidents I tried it in a crappy P3 board and it couldn't post with the diag card. Shut it off immediately. Never going to use this one ever again.

Decided to remove the bad chips while I try to source replacements. I used an aging setup consisting of a ceramic preheather and a hot air rework station. I used this ghetto setup before on larger things and there were never any issues.

First I practiced on a beaten up Gigabyte GA-7n400. Figured out the temperatures and stuff and the chip came off nicely. PCB is fine.

The next was one of the boards that got killed by that diag card. MSI K7N2. Also no issues. Clenaed the pads on the board. Looks good.

Nforce2 MCP rev A and S removed

And last the other victim, Gigabyte GA-7n400 Pro2. This has the most features.

This time it went not so well. Like before the chip came off the board... and took some pads with it. Must be something with the solder expanding, some stress or I don't know. I was dissapointed. Though it seem that these pads don't connect any circuit traces it is still worrying.

Also the resist mask got damaged on the board a bit so I have to fix it up before the rework.

This is how the damage looks on the board and the chip:

DAVE86 wrote on 2023-05-31, 11:42:

Tetrium wrote on 2023-05-31, 10:29:

Is your POST diagnostics card new or did you have it for a while before it went faulty?

It's from Ebay. Single chip (CH311L) 2015. PCI/ISA type. I have a bunch of different ones. After the accidents I tried it in a crappy P3 board and it couldn't post with the diag card. Shut it off immediately. Never going to use this one ever again.

Was the card like this? I hope it is not common that this diag cards cause such damage.

janih wrote on 2023-05-31, 17:53:

DAVE86 wrote on 2023-05-31, 11:42:

Tetrium wrote on 2023-05-31, 10:29:

Is your POST diagnostics card new or did you have it for a while before it went faulty?

It's from Ebay. Single chip (CH311L) 2015. PCI/ISA type. I have a bunch of different ones. After the accidents I tried it in a crappy P3 board and it couldn't post with the diag card. Shut it off immediately. Never going to use this one ever again.

Was the card like this? I hope it is not common that this diag cards cause such damage.

If you have that card, and if that is OP's card, then have I got bad news for you.
The ISA part isn't connected underneath the chip, the traces run there and then just end. Also the ISA connector isn't quite right and could potentially short adjacent pins.
The pci part worked for me, although it does get detected as a PCI device as well.
I know I have some photos of the card somewhere with the chip removed, I'll add /link them when I find them .

On more than one occasion I've put the POST card into ISA slot the wrong way around. PSU would trip right away, but there doesn't seem to be any mobo damage. Wasn't such a "modern" PCI mobo though. Also, different POST card.

As for the lifted pads, could be weak glue on the PCB. I don't deal with BGA often (in fact I try to stay away from such packages, not enough experience with those and no access to Xray machine to inspect the soldering job afterwards) but I've seen pads that come off the PCB with temperatures that were not excessive. Could be too long heating or not enough thermal soak from the bottom preheater but frankly there is always some risk with PCBs that have seen a lot of thermal cycles.

janih wrote on 2023-05-31, 17:53:

DAVE86 wrote on 2023-05-31, 11:42:

Tetrium wrote on 2023-05-31, 10:29:

Is your POST diagnostics card new or did you have it for a while before it went faulty?

It's from Ebay. Single chip (CH311L) 2015. PCI/ISA type. I have a bunch of different ones. After the accidents I tried it in a crappy P3 board and it couldn't post with the diag card. Shut it off immediately. Never going to use this one ever again.

Was the card like this? I hope it is not common that this diag cards cause such damage.

Yes that's the one. I didn't expect it to go bad. Something got shorted internally. I remember from the very begining it never worked in ISA slots, showing post codes that is. Just as a voltages present indicator.
Other chinese POST diag cards that has gals/pals/logic, etc never failed me thankfully...

Deunan wrote on 2023-05-31, 18:45:

As for the lifted pads, could be weak glue on the PCB. I don't deal with BGA often (in fact I try to stay away from such packages, not enough experience with those and no access to Xray machine to inspect the soldering job afterwards) but I've seen pads that come off the PCB with temperatures that were not excessive. Could be too long heating or not enough thermal soak from the bottom preheater but frankly there is always some risk with PCBs that have seen a lot of thermal cycles.

Of course such things show up on the one board that has the nicer features... I've seen some Nforce2 based motherboards that had both the north and south bridge heatsinked. Gigabyte and Asus didn't seem to bother.

Managed to get two badly beaten up abit boards as donors. It was a bargain and hopefully the southbridge chips are fully functional.

So I have two reballed chips.
Haven't got the time to place them on the patient boards though...

Fixed up the solder mask.

Chip on the board

Aaaand both boards are still as dead as doornails. Either the chips were dead to begin with or I screwed something up. Anyway it was an interesting experiment.

might be worth investigating that post card to see what got shorted to what?
is 3.3V supply working on those boards right now? maaaybe there are fuses popped somewhere? or PCI clock buffer chip?

Thanks for the tips! I have injected 5V into that post card and the voltage appeared on some of the data bus pins (where it shouldn't I guess). I haven't examined it further.
I'm bummed out on these boards right now but will probably get back to them in the future...

The pads usually rip when all the solder has not been fully liquified under the chip, when you pull off the chip do you first gently touch it to see if it floats then take it off?

How do you prepare the chip that goes back on?
The pads on the replacement chip need to be wicked off from the old solder then add some solder paste with a stencil, wipe it down then use the heat gun to create solder balls on the pads

DAVE86 wrote on 2023-06-01, 07:40:

janih wrote on 2023-05-31, 17:53:

DAVE86 wrote on 2023-05-31, 11:42:

It's from Ebay. Single chip (CH311L) 2015. PCI/ISA type. I have a bunch of different ones. After the accidents I tried it in a crappy P3 board and it couldn't post with the diag card. Shut it off immediately. Never going to use this one ever again.

Was the card like this? I hope it is not common that this diag cards cause such damage.

Yes that's the one. I didn't expect it to go bad. Something got shorted internally. I remember from the very begining it never worked in ISA slots, showing post codes that is. Just as a voltages present indicator.
Other chinese POST diag cards that has gals/pals/logic, etc never failed me thankfully...

Those single chip Diag cards are true garbage ! There are other topics here all about how crappy they are and should be avoided at all cost (unless you like to fry yer boards). Just saying the 5 chip versions are the best like you say

treeman wrote on 2023-07-07, 00:25:

The pads usually rip when all the solder has not been fully liquified under the chip, when you pull off the chip do you first gently touch it to see if it floats then take it off?

How do you prepare the chip that goes back on?
The pads on the replacement chip need to be wicked off from the old solder then add some solder paste with a stencil, wipe it down then use the heat gun to create solder balls on the pads

The original and replecemnt chips were floating nicely before I took them off. On the replacement chips I used a stencil and leaded balls. I do bga work often altough not as "big" as a nf2 southbridge. I was suprised it didn't work out on either boards. As rasz_pl suggested there must be something else broken on the boards.

sounds like you know what to do, I have never done bga on anything bigger then mobile phone pcbs but suspect this is where the problem is. The heatgun you are using may not be delivering enough heat, altho when the chips are floating it usually means its all ok to move.

Maybe the age of the hw also has something to do with the thermal resistance of the epoxy under the pads being weak, if those are all non connected pads? Taking a guess and also veering off the topic.
Thats all from me

I didn't do anything differently with the boards during the reworks. The Gigabyte GA-7n400 Pro2 board have probably gone through many stressful thermal cycles during operation. The other boards let go of the SB chips fine.