Atari 520STe death by soldering iron

This is somekind of hardware horror story. Maybe most of other upgrades I’ve made before turned out nice and without bigger problems, but I’ve got some serious screw ups earlier along the way and this is the one. I’ve got this 520 STE since a long time, I’m very attached to it, because it was firstly used by my father in early 90’s for DTP in his company and it was very first machine I began coding on and it accompanied me on several demo parties too. So, exchanging whole motherboard was out of question.

Atari originally had few adjustments like 1,44 MiB floppy drive modification made by TOMS from Warsaw / Poland (,which is better known for it’s 8-bit Atari world from it’s very sought after floppy drives) and was extended to 4MiB. The problem was that I’ve started to tinker with it when I was younger, I didn’t have proper skills, training and equipment. By improper tools, soldering / desoldering of some elements I’ve broke some onboard via’s and components, which ended up in partially functioning computer system and it gave me alot of troubles later. So, work on this board took me several years, on and off, depending on my patience and time I had. But I’ve told myself that if I could break it, I could fix it. I’m not an electronic engineer and if you’re one, then before looking on following photos, please take some sedation pills.

Basically, everything started when I’ve got some issues with floppy drive, I couldn’t find new working replacement. Atari services or individuals, who knew something about this hardware vanished. I’ve started with removing TOMS HD floppy extension to exchange it for something different that could work with newer floppy drives and desoldered floppy ribbon cable, which was soldered directly to motherboard and couldn’t be easily replaced. Of course without proper background, with very poor soldering iron and basic desoldering pump, I just used what I had at hand at time. Also original simm sockets began to crumble (literally) and for long time simms were held only by few pieces of paper folded several times and inserted between them – when computer was moved from place to place and it ceased to work completely or less ram was detected.

So, I’ve replaced RAM sockets with new ones, I think that I’ve done it at least twice. First time something went wrong and system couldn’t detect more memory than 1,5MiB, when it should detect full 4MiB. Of course I’ve damaged motherboard slightly, when doing it.

Removing floppy ribbon wasn’t easy without equipment, I’ve also damaged the surface of the motherboard (what is seen on photo). I don’t remember correctly, but new header pins were also resoldered several times. In the meantime I had another HD floppy mod, but much later before mounting internal Gotek I’ve decided to remove it, because hd detect signal had to be redirected somehow outside.

I’ve replaced bugged TOS 1.6 GER with dual 1.62 / 2.06 UK one, later on I’ve replaced keycaps with UK layout and dual TOS was replaced with newer and faster one manufactured by Exxos.

Removed old PSU and inserted brand new one.

I’ve exchanged all electrolitic capacitors on whole motherboard and exchanged resistor arrays from 10kOhm to 2,2kOhm, which resulted in write protect errors on disc access, so I had to add 10kOhm resistor array on WD1772, which fixed the issue.

As I decided to install internal Gotek floppy emulator I’ve had to add internal / external floppy boot switch to have a possibility to boot from SF314 and use physical floppy discs.

I’ve done additional mandatory fixes related to STe video ghosting and mixing volume of YM2149 with DMA. The only one is missing is fix for DAC noise, which can be removed by adding small pcb, unfortunately it cannot be bought ready made, but it can be assembled, so probably I will have to make several in the future. I really recommend doing all those fixes to STe / MSTe. Those are very noticable improvements and really make a difference.

Some ICs received new precision sockets (mostly from ACSI / floppy circuits without maybe DMA), but I’ve removed them badly. After desoldering ICs I lifted them with a screwdriver and applied too much force (don’t do this!), which resulted in ripped out vias and damaged througholes. Additionally some pcb traces vere broken in not very obvious way. Afterwards I’ve soldered precision sockets, which was another dumb thing to do. Effect was that internal / external booting has worked, but external ACSI port started to malfunction – UltraSatan SD drive didn’t want to boot at all. So, to revert it I had to remove precision sockets and I’ve made more additional damages . This was really bad. So, project went into hibernation for some time.

When I’ve retrieved it from the drawer after loger hiatus with a mission to ultimately fix everything I’ve had to change approach. The only way to bring back full functionaluty was to sit down with original schematics from original service manuals and check every connection in DMA port and floppy circuits. This turned out to be more difficult than I’ve anticipated, because not all published schematics were the same as my motherboard revision, there were some bigger and smaller differences. But checking all the traces between ICs using schematics with multimeter, seeking out damaged vias and other physical damages on board turned out to be a good strategy. After identifying problems I’ve fixed or replaced broken traces with insulated, enameled copper wire, which is used commonly in transformers – transparent lacquer can be scratched with exactoknife, aligned with, old, broken via and soldered to endpoints or not damaged fragment. After that I’ve replaced sockets on WD1772 and YM2149 and once again tested trace continuity with multimeter.

When I’ve put everything in place DMA port / ACSI started to work, but some letters in names of devices were wrongly displayed and data on drives couldn’t be accessed. This turned out to be easy to trace, I’ve taken all the letters that were changed and compared their bit patterns, this in turn exposed problem with one dataline (#3), which was broken and I’ve missed it somehow, when checking continuity.

Floppy worked, but I’ve got another weird issue. Couldn’t boot from external floppy drive, but internal drive worked somehow. After rechecking traces, connections and switch everything looked fine. I’ve swapped signals A and B on chip itself and external drive worked! In the end source of problems was YM2149 chip. When inserting and deinserting existing, fully working chip I’ve damaged one leg and couldn’t use it further, so I’ve exchanged it with another one from my spare parts box. Unfortunately it looked good / unused, but was not used and tested before. After exchanging it for different one internal / external boot switch allowed me to boot from internal and external floppy.

Rear panel with TOS version and internal / external boot selection

Further tests uncovered another problem with mouse movement – sometimes it ceased to move or moved by itself or in wrong direction, so I had to reopen keyboard, reflow / resolder solder points on all connectors and replaced motherboard connection cable. Additionally I’ve fixed one pin in mouse port, because somehow it was shorter than other pins, probably it was pushed in somehow.

With correctly working mouse and keyboard I could assemble everything after several years :-). So, everything ended fine, but unfortunately motherboard isn’t that pretty like before, repairing mistakes and backtracing took much more time than it should. So, if you are about to repair some retro computer ask someone skilled or get some proper tools before and excercise soldering / desoldering on something less valuable, than your precious vintage hardware ;-).

The only missing things is DAC fix and case could look better. It’s no longer grey and got few scratches, but after my rather failed experiments with deyellowing keys on ST I think I will leave it as is.


About this entry