My cousin was mostly into games and didn't bother a bit about programming or even the most basic things beyond how to load the next game but I got kinda attached to the idea to have a machine that I could shape to my needs.
The machine in question was a "mobile" Commodore 64, the SX-64 model. It was never (to my knowledge) used in a mobile fashion but it took quit some beating by my cousin... he usually smashed the joystick into it when he lost a game. Quite a tamper.
Back to present times, I wanted to own one of these for quite some time but only ebay made me realize just how rare they are... and for that just how expensive even used or broken ones are. Meh. But eventually, one popped up whih was below my own limit of around 400 USD including shipping. Yay!
The only problem with it was: US model. I need the European model (240V, 50Hz, PAL) and not the US one (110V, 60Hz, NTSC)... so I did some thinking: The SX has a built in screen, so the TV standard wouldn't be an issue. Also for the external monitor (TV) most modern TVs support both standards so also not an issue. The power supply however was a bit more of a challang. I hoped to convert the one built into the unit to the required voltage/frequency but it turns out that the required transformers aren't available at all. Note even the specs could be found. Oh well... let's look at the schematics then: There is a 9V AC transformer that is just routed to the user port with no internal connections whatsoever. Everything else is powered from either 12V DC (including the TV screen) or 5V DC. There should be an alternative, as these are very common voltages. And behold: I managed to find a power brick that had exactly that: 5V and 12V output with enough amperage to spare a bit. Add a tiny 240/9V transformer and everything is set! I just needed a way of mounting those things into the casing...
When it arrived, I was already a bit worried... the seller put it into a box, but only included a miniscule amount of bubble wrap and lots of newspaper for packaging. But with a 10+kg machine, the newspaper got compressed and it was basically banging around in the box. This also lead to a damaged keyboard case, with the missing part nowhere to be found. It must have fallen out during shipping. Highly annoying, as the SX was in very good condition originally.
I decided to go for a complete cleaning job first: disassemble, wash and rebuild. Holy cow, there are a lot of screws in that thing. And while it is pretty servicable, using different screws for everything isn't really helping a complete rebuild.
While I had all the electronics out, I changed all the electrolytic capacitors, just to make sure, as I have no good means of checking them. None of them had any visual issues, but I think averaga caps of today still outperform high end caps of the mid 80s.
I also removed all the chips from their sockets and used contact cleaner to make sure every connection is nice and crisp. The SX has the entire TTL bus routed over two boards and two ribbon cables, so any additional contact issue would probably be the end of any signal, even if it is "just" 1MHz.
During the PSU mod, I made the mistake of cutting the original wires instead of desoldering them... The wires are a pain in the behind to solder and turned out slightly too short for comfort. So if you ever put in a modern PSU: desoder, don't cut!
Since I know (from first hand experience) how mains voltage feels like, I got myself a 12VDC->240VAC converter and a cheap car battery some time ago, so I can test drive mains equipment with some insulation to the actual mains. Also, I can attach my scope without any special probe. This is how I fired up the SX for its first few tests.
The fabrication of the brackets needed to tie down the new PSU was lost in digital limbo, I'm afraid. I did record every step of the process, but somehow managed to not copy the files before erasing the SD card...
The screen did turn on and the typical frame arund the actual content was recognizable, but the content was just flickering garbage. To me, it was clear: the VIC chip is not reading the correct RAM data but rather random data with every refresh. Hmmm... this could have a number of reasons, but investigation with my scope quickly put the PLA in the spotlight of shame: it outputs a constant DC level of 2.4V instead of any changing signal. On all output pins. And here's where I made a slight mistake that caused some delay: I looked at the chip's designation and it said (something like) UE3. I didn't realize at that time that Commodore used some sort of matrix for labelling these and just looked up the "#3 chip on the schematic". Which I ordered through ebay again. And when it arrived, I had two Basic ROM chips...
I put in the new PLA connected everything up again and turned it on: Start screen! Whoopie! Success! But even before I could get my cell phone ready to snap a quick picture, the top line disappeared... And after a restart, it only showed a white screen... Bummer #2. SOMETHING else was wrong. I found out that a removed SID chip gave me the start screen again. This is also when I ordered the "Dead Test" cartridge to check for any memory issues. Again probing the bus with my scope, the A13 line showed some weird response... not a rectangle but rather like the charge curve of a capacitor... I put that up to the CPU being buggy and ordered a replacement. Same for the SID since adding it in would immediately lock down the system.
By now I had almost everything replaced and it seemed to work. I decided to run the dead test and it looked good: no memory issues found, the SID plays. Even the original copy of Maniac Mansion loads and the intro plays... So it seemes perfect now.
About that time, the car battery ran out of juice and I decided that it was time to plug it in for real. All the mains stuff was already neatly tucked away anyway. So: plug in, Turn on and ZAP! Listening to the frantic noises of UPSs and standing in a pitch black room in the middle of the night, trying to find my way to the fuse box. It turns out that one of the two heat sinks of the new PSU is not (as I thought) connected to ground/earth but rather to one of the rectified inputs. Ahem. Move alon, nothing to see, just some stupid mistake in tracing a PCB track... Thus I put in a piece of inulation sheet and now that also worked fine.
One weird thing about the process was that the SX worked fine when I had it open, but crashed or didn't even go to the start screen when I put the case back on. I found out later, that the screws for the case actually squeezed againt the CPU board, but I'm still not sure if that was the issue.
The only thing left was that the dead test cartridge reported "all OK" but the count in the test iterations went like this: 00-01-02-03-04-XX-11-12-13-14-XX-etc... where "XX" are some of the graphic characters. It seemed that something was wrong anyway and since the dead test doesn't do a full RAM check and my test harness was still in the mail, I decided to use one of the C64s I got for parts and swap all the RAM chips. After 2*8*16 desoldering and one 8*16 resoldering... the problem was still there. ARGH!
I replaced the CPU next: and it worked! Somehow the replacement from earlier (which said "tested" on the ebay listing) didn't quite work. Or my PSU mishap caused the damage.. Anyhow, to make it even stranger, NOW the original CPU also works fine...
That's when I wrapped things up and put the thing together for the final time. I lost count on how often I opend/closed that case...
Anyhow, here's a little video I made during the process: