Amiga 4000 CH04: Damage Inspection 

Yeah, I suppose, with smaller electronics, it could make it worse, allowing the electrolyte to go upwards. But for the earlier generations of machines, where components are larger and more spaced apart, yeah, I think its a bit of a life saver. Its a bit like what talcum powder was once used for, soaking up sweat. Eeew...

I think with care, some self training and a LOT of patience, even the badly trashed ones can be made to work. Of course, there must have been examples that really were beyond repair. I suppose we tend to only be shown the ones which made it. For the A3640's, sounds like the, I think its GALs, they are programmable chips which hold connections for the CPU, to the main board. One of those is probably borked. Some of them get VERY hot when its running, so difficult to tell by touch if a chip is shorted. Unless you know how to diagnose, its best to get them to a repairer. Not easy to repair, as to replace a GAL you need to programme it with the same code as on the dead chip, and for that, you need a desoldered working chip. But of course, it could be many other things... If you have two broken A3640's consider lending them to a channel, and see if they will repair them as a project. Might be an idea to offer some £$£$ as well, as it will be a lot of work to fix two of those.

I have blank replacements for all the GALs, new sockets and a programmer for it with stock code and Haynie's SpeedGeek code. I tried installing a socket on clean U203 pads with both hot air and a fine tip iron and so far melted one socket, cracked a second one and the third is soldered onto the board, but caused a partial boot to become a no boot with three different U203 GALs. The small 20 pin PLCC sockets are super hard to solder on the boards because of access room for the iron. I have so far resurrected a 1200 and my "BBS" 4000, but I still have an A3000 and a 4000T to work on (both motherboards look absolutely pristine) as well as the roached 4000D (I have an original from 30 years ago with a bad mouse port that C= replaced under warranty) and two A500's, one of which is my much hacked on first Amiga from 1989. so the 3640's are well down on the priority list right now. I also have a whole slew of ZORRO cards of all types that need to be tried. I doubt I'll live long enough to repair everything!

Carumba it does sound like you have your plate full there. Certainly. Me, I'm still a total noob at soldering and all this repair stuff. I have a PC, with very fine pitch pins, from 98, and I thought I might have to de-solder it. Its really scary.... Have you tried finding any videos were someone solders on those kinds of sockets? I'm assuming that to get those to work, you might need a heated mat, to get the card up to close to 200 degrees, clean pads and then hot air the socket, with capton tape to shield everything around it, including the plastic upper on the socket. And then it will just magically float into place, honest... Now I spell that out, it does sound like a mucho pain, considering how many chips there are. Did you manage to ascertain WHICH GAL might be borked? Also, Im assuming you checked all the traces on the before before going for socket replacements. Anyway, I have no idea what I am talking about. But certainly all of that stuff will keep you very busy. I want to keep the channel more general, with presentations and documentaries, and there is the day job and other projects. So I really don't have that much time to repair machines. Especially when it gets very gnarly But I do have the 2nd part of the Fujitsu series coming up, where I did learn quite a lot in the process of attempting a fix. That was quite rewarding...

I used the hot air to remove the U203 GAL and then used an iron and braid to clean the pads, it came out very nice. I first tried kapton tape and hot air with solder paste to mount the socket and managed to melt it even though it didn't get the solder paste hot enough to flow much. Once again, the access to the area needing solder on the 20 pin PLCC socket is just too restricted. I had looked at a number of vids of people installing PLCC sockets, but they were all larger sockets with more access to the pins/pads. That 3640 had been booting most tries, but only chip ram would be available, no fast mem at all no matter what mobo sockets were populated. I plan to replace all the GALs and decided to take a bit of a stab in the dark with U203 as that was the first one I tried burning with SpeedGeek code. I thought I had managed to get all 20 pins to flow and when it got a black screen no-boot I reflowed them again with no joy. I tried the original U203 GAL and still nothing. I had previously recapped the card and examined all the traces with my optivisor. I found one iffy trace that I bodge-wired. I have a lot of old motherboards kicking around. I gave away my 286 30 years ago, but I still have a 486 DX4/100 and a 120 too I think. I might have a pentium 90 and may a 133. I have most of the parts for the boxes I owned in the '90s. A bunch were in use as FreeBSD servers though the years. I probably won't rebuild any of those. Regardless, Amigas have always been my favorites.