I can haz plan nao!

[unixronin]

As recently reported, vorlon (my Athlon64/Win2KPro gamebox and writing/accounting machine) is currently down due to a BIOS upgrade that failed in a non-recoverable manner.  I filed a tech support ticket with GigaByte, but according to their support site they haven't even read the ticket yet, much less responded to it.

So, after doing some research, I have a plan.

The BIOS on my motherboard is in 4-megabit 32-pin PLCC flash memory chips that look ... well, precisely like these, actually.  (This photo is even from a K8N board just like mine — not actually mine, because my camera won't focus close enough for macrophotography without a macro lens, and I don't own one).  GigaByte evidently decided to solder them directly in place to save the cost of sockets.

Now, it should be immediately apparent that neither desoldering nor soldering these is going to be much fun.  Except that, well, they're hosed already ... and I can buy new ones online, already programmed with an updated BIOS for my board.  That means I don't care about preserving the existing chips.

So.  I order a set of replacement flash BIOS chips (ten dollars plus shipping), and a pair of these 32-pin surface-mount PLCC sockets (fifty cents each plus shipping).  I cut the leads on the existing PLCCs, throw them away, and individually remove the left-behind leads.  Then I solder the sockets in place of the PLCCs (which I can do MUCH more easily than soldering on new PLCCs because I have direct access through the socket to the solder pads), snap the new BIOS chips into the sockets, and ... everything should be fine.  As long as I don't screw up the soldering or damage the board.

If all goes well, I'll end up with a better board, with socketed BIOS as it should have been in the first place, for a total cost to me of around $25.  Sure, it'll void the warranty on the board ... but the three-year warranty expired almost a year ago anyway, so I don't really see that as a problem.

Comments

[zaitcev.livejournal.com]

I've done this sort of thing before. Here's a picture:
http://people.redhat.com/zaitcev/pics/sunray_3.jpg

Unfortunately, the angle is not very good, but look at the bunch of red wires: they connect the socket to the pads on the mobo. I went this way because it otherwise requires wave soldering, as the socket overlays the pads and I cannot reach with the soldering iron.

I actually saved the chip and was able to read its contents (which was the objective of the excercise in the first place). Since I cut the wires very close to mobo's pads, the chip worked fine in the PLCC socket later and the system worked as if nothing happened.

[unixronin.livejournal.com]

That's an interesting approach. I don't think it's necessary in this case, but it's something to keep in mind. I don't know how well it would work on a surface-mount device though.

[zaitcev.livejournal.com]

Oh definitely you won't need wires, since your socket allows access to the pads. Main points are:
0. It was exactly THE SAME chip and packaging as yours.
1. At least on PCBs made in 2001 it was possible to remove the pin stubs and re-attach the wires without melting maylar, which was a concern expressed by a commenter previously. But be careful.
2. You can re-use the chips with shortened pins in such socket, if you have a programmer and can download the image from somewhere.

[unixronin.livejournal.com]

That's useful information, thanks. I don't have any plans to try to re-use the primary BIOS chip, though, because since about half a dozen successive flash attempts all failed at the same point, I strongly suspect the flash chip itself has failed.