Just thought I would expand with a 'lessons learnt' here, in the hope that my experience with this A500 might help someone else trying to repair one in the future.
When I first got the A500 it was described as working and in good condition. Upon booting the machine with a few games it seemed OK, but some games crashed at the same point each time. Now originally it had KS1.2 ROM so I assumed that all it needed was a KS1.3 and a good clean etc.
Stripped it down, fairly clean inside but the metal shielding had some corrosion - not a good sign as the flakes get onto the motherboard. Proceeded to remove the shielding and take a look at the motherboard. I could see several 'dark' coloured pins on Denise, the CIA nearest Paula, Paula, Agnus, and spill marks on the PCB. It smelt and looked like coffee from many years ago.
After removing those sockets chips and cleaning up the corroded pins with a wire brush I noticed the sockets were much worse - that green colour you get with corrosion in the same pins in the socket where the pins were 'dark' on the associated chips. EDIT: Cleaned the PCB using Isopropanol achohol, and after soldering a socket / chip, cleaned with the same stuff but used a fine wire brush to remove flux (PCB looks untouched as a result).
Replaced the Paula and Denise sockets with new ones (48 pin DIP), then replaced the CIA socket (40 pin DIP). Finally re-tested, it was a bit better but some games still crashed at the same point and it seemed to be sensitive to movement. Used PLCC extractor to remove Agnus... Agnus came out with the removal tool WITH the PLCC socket plastic component. So what was left on the board were 84 pins sat there on their lonesome, with the plastic part of the socket in my hands.
The socket then just litterally crubbled appart - I am not kidding the plastic was mega brittle with coffee and ageing. On inspection I could see what had gone on in the past. Firstly, the Agnus had at some point been levered out of the socket using the wrong kind of tool and the corner had split and been SUPER GLUED (OMG). The owner then proceeded to use a MEGA STICKY DOUBLE SIDED PAD to glue the Agnus INTO the socket (hence why Agnus was sat in my hand with a plastic socket glued to her ass). With coffee, corrosion and a partly split corner held with glue the little lass stood no chance of lasting.
I replaced the Agnus socket - this was the most difficult to do as you cannot just cut the plastic part on the top of the PCB. I mentioned the plastic came off already, it did, but that was the square part that holds all the pins to the chip, underneath that is a very very thin piece of plastic that holds all the pins flat against the board, that doesnt come off as its pretty much molded with the pins. For DIP 40 and 48 sockets a nice safe way of removing them I found was to desolder all the pins, once the solder hole looks empty, wiggle the pin with a very small screw driver, and if the socket doesnt just fall off, proceed to CUT the black plastic of the socket into blocks of 2 / 3 pins - this way they come out much easier and if you get a stuck pin you can just heat that pin on its own whilst using the other hand to pull slightly at the block of pins you are trying to remove. Anyway, one 84 pin PLCC through pin hole socket later and Agnus had a new home, free from any problems - phew...
Testing again revealed a significant improvement, a few games worked now that didnt previously, but I just had problems with many games and they all had similar problems whereby they would crash at a specific point every time.
Using the piggyback method (just sit a new 41256 RAM chip on top of the existing one so it snuggly hugs the bottom chips pins), and testing Turrican one chip per reboot I found a chip that when it was piggybacked caused Turrican to NOT hang on the part it had before. To make this story a bit shorter, I repeated that process around 4 times until Turrican was almost flawless, and extended the testing to Lemmings, Days of Thunder, Rolling Thunder, and some shoot em up I can't remember the name of (coded by DMA), each exhibited either a hang at a certain point loading, or graphical corruptions. Now the graphical corruptions were by far the hardest to address because after piggybacking all chips the shootemup game and Turrican still had graphical wierdness with some sprites. At this point I had replaced like 3 or 4 RAM chips in sockets, but now I had to replace all of the remaining chips with sockets. After carrying out this work I started replacing the testing the newly socketed 'old RAM' one chip at a time against those games with garbage sprites. This approach worked well and at the end I had 8 X NEW RAM and 8 X OLD RAM. The system was 99.8% now, everything loaded, next to no crashes at all, but Lotus III only worked for a few minutes before a random reboot. I thought about this and came to the conclussion that my approach - ie. to try and save old RAM chips and only replace what was required had long since become a flawed approach, since I had ended up replacing 8 chips already (50%), and out of those 8, 4 allowed the piggyback approach, meaning that an address line or data line had gone open circuit, and the other 4 chips didn't allow that approach, meaning that they were bringing back random results (eg. a high instead of a low), and its not like these were all 8 of the top or bottom bank, they were randomly situated. So my reasoning at this point was, yes its much much better now but these really obscure crashes in Lotus that are very hard to re-create must be related to RAM timing, and possibly one of those remaining chips is glitching on an infrequent basis.
So I finally just replaced the last 8 old chips with 8 new ones. Unfortunately I didnt have any more - 15 (150ns) chips, only -12's. I seperated them into the 2 logical banks, 1 row of -12's and 1 row of -15's. It shouldnt make any difference at all, just one bank will be perkier than the other lol, but both within the design limits. For the first time I was able to play Lotus 3 until I SWITCHED IT OFF, rather than the system dictating I would have only 30 seconds to race. It worked flawless, no glitches of any kind, very stable. I also left it on the demo screens of Lotus 3 for a good 20 minutes with no crashes.
I declare this A500, 'back from the dead', to live another 20 to 25 years hopefully.
Final steps for me now are to solve a loose connection on the floppy power cable (probably needs a bit of solder), and to 'retrobrite' the case, a process which i've had good success with so far on the A1200 case I am doing at the mo. I've already serviced the floppy drive, and also added a KS1.3 ROM in very early on btw. I also plan to add a DSDT switch to toggle the expansion RAM between 1Mb chip or 50/50, but I will need the 1Mb Agnus before I do that. I may also add one of those 8Mb fastram boards to the CPU but I am not sure as I probably would never use it lol
EDIT: Forgot to mention, that corrosion on the metal shielding - all dealt with early on, using metal polish (Brasso). Wiped down afterwards with paper towels to remove residue, then wiped again with a very fine oil in order stop future corrosion (don't use something like WD40, unless you want it to smell like a car engine forever lol).
