For the past year and a half, I have turned my DIY attention towards vintage Macintosh restoration and repair. One of the first Macs I picked up, in March 2022 as part of a job lot of old computers and accessories, was a SE/30.
It’s a special machine – a direct continuation of the iconic 1984 compact Macintosh, of which there are numerous models. The SE/30 is the fastest of these and therefore very sought-after by collectors today.
This one was sold as-seen, with no guarantees if it would actually boot. On my first attempt, I was greeted with a ‘simasimac’ display, but was able to get it to reliably boot off the included System 6.0.3 floppies after I’d reseated the ROM SIMM:
I went back to inspect the logic board, which had unfortunately had a botched recap job.
I carefully removed the components, and found that a solder pad and some traces (circled in pink) had been damaged. Someone had clearly struggled here.
I soldered in replacement capacitors across the whole board and ran a repair wire to patch the broken trace.
With the botched repair work now corrected, I thought that it would be smooth sailing. All seemed good for a while; I was able to get System 7.5 set up on a hard drive and use the Mac normally. But as I ran the machine I was finding that it exhibited odd behaviour and instability, particulary with above 8MB RAM installed and Mode32 enabled. And sometimes I would still get that simasimac display.
I spent ages extensively troubleshooting the board, swapping RAM, ROM and trying different hard drives; even recapping the power supply and analog boards, but none of these attempts got to the root cause. I was stumped – but determined.
In around October 2022 I made contact with Ben, a fellow vintage Mac enthusiast on Twitter, who offered me an SE/30 Reloaded PCB along with parts to go with it. The deal was that I could have all this stuff for free in exchange for building a second Reloaded board for him. I agreed as I thought it would be a fun project and a way to get a fully working SE/30 for myself. What followed has been an extreme test of my electronics assembly and troubleshooting skills, ending in a conundrum more perplexing than the one left to me by the original faulty logic board.
The two empty SE/30 Reloaded PCBs and donor boards arrived. Here they are laid out on my humble kitchen table. I would be moving parts from the old boards to new – a big undertaking and certainly on a scale I’d never attempted before – but I was excited to get going.
I had intended to transfer all components across, but I quickly realised that it was exceedingly difficult to remove the through-hole components without a heated desoldering gun such as the Hakko FR-301. I melted the first RAM slot I tried to remove – I got impatient and applied too much hot air – and then broke a pin when removing the 68030 CPU.
I was able to locate new old-stock RAM and ROM sockets; and brand new PDS, floppy and SCSI connectors, as well as other bits and pieces for both boards. All of the PLCC and DIP chips were moved across from the donor board.
It took about three months of concentrated work to put together, but by January 2023 I had a working first board. I tested it out and when I was satisfied that it was all working, I sent it off to Ben.
I started assembling my second board and this time I used sockets for all the DIPS and PLCCs. I’d found it really hard to troubleshoot bad donor chips in the first build and wanted a way to make this process easier. I also bought brand new surface-mount logic chips from Mouser such as the RAM and VRAM multiplexers and resistor networks.
I got the second board up and running a lot quicker than the first, but quickly realised I had a different sort of problem:
As you can see, my board was experiencing a kind of graphical interference/artifacting on-screen whenever there was movement. The issue is persistent across boots, but it’s also varying in severity and intermittent. Generally, it tends to show up for 5-10 minutes shortly after powering on, and then disappear until the next cold boot.
I’ve done many tens of hours of troubleshooting and it doesn’t seem to be a bad chip, or – where I’ve selected new parts – an incompatible chip. I assembled a third board, using a completely new set of parts, to see if it was an issue with my soldering work. But it’s not: the third board does exactly the same thing.
It was around this time that Ben got in touch to report that he was having graphical issues with his board, too. But the issues were different and seemingly unrelated to my own. After some minutes, or tens of minutes, the following artifacts would start to build up on his screen:
He sent me the board back to look at, and it took me quite a while to work out that the root cause of his problematic symptoms were the same as mine. Take a look at these photos:
They are the PAL chips at UE6 and UE7, responsible for driving the video output. They come as a pair, and depending on how old or new the donor board is you’ll either get parts ending in 0637-A and 0688-A, or 0754-A and 0755A. My two boards – i.e., builds 2 and 3 – have the 06xx chips installed, while the board I made for Ben has the 075x chips.
I recently got another battery bombed donor board which has the 075x chips and as an experiment I decided to fit them to one of my two boards. And the symptoms changed over to the other set that I’d separately been battling to fix on Ben’s board.
The screen doesn’t flash up with horizontal lines of corruption each time there’s movement, as in my own set of symptoms. Instead, these short ‘dashes’ gradually build up and remain on the screen, filling it with columns of garbage.
The clue all along was that both sets of symptoms are intermittent in exactly the same way – they come and go with varying severity. I did consider that they may be linked, but now I’m totally certain.
The discovery really makes me think that I may have faulty reloaded boards. It’s hard to believe it’s anything to do with how I’ve assembled them, because I would have had to make the same error three times. It also seems unlikely it’s anything I’ve fitted. Board 1 only has original, donor logic chips installed. I’ve tried combinations of almost everything including both original and brand new components. I’ve gone over my work with a magnifying glass countless times and reflowed any suspect joints.
Concluding that there is a fault with the PCBs themselves is a little tricky. After all, many people use JLCPCB and nobody complains about their quality control as far as I’ve seen. More detective work, and probably an oscilloscope, is required.
On a positive note, I did recently trace the instability on my original board to a faulty F240 logic chip, and it’s now running rock solid. I would rather have my Reloaded board in there, but it is nice to at least have a fully working SE/30.
DIY DIY Guy 