There’s a dedicated group of users out there that aren’t ready to let their beloved IBM PC110 go to that Great Big Data Center in the Sky. Unfortunately, between the limited available technical information and rarity of replacement parts, repairing the diminutive palmtops can be tricky.
Which is why [Ahmad Byagowi] has started a project that aims to not only collect all the available schematics and datasheets that pertain to the machine, but to reverse engineer all of the computer’s original circuit boards. Working from optical and x-ray scans, the project has already recreated the motherboard, power supply, modem, keyboard, and RAM module PCBs in KiCad.
Just last week the project released production-ready Gerbers for all the boards, but considering there have been 45+ commits to the repository since then, we’re going to assume they weren’t quite finalized. Of course, with a project of this magnitude, you’d expect it to take a few revisions to get everything right. (Hell, we’ve managed to screw up board layouts that had fewer than a dozen components on them.)
If you’d like to lend a hand, [Ahmad] says he could use the help. Beyond checking the boards for problems and reporting issues, he’s also on the hunt for any datasheets or other documentation that can be found for the PC110 or its components. It looks like there’s still schematic work that needs to be done as well, so if your idea of zen is figuring out how ~30 year old computers were wired up internally, this might be the perfect summer project for you.
Interestingly, our very own [Arya Voronova] has been working on creating a drop-in replacement motherboard for the Sony Vaio P using KiCad and imported board images. That hobbyists are now able to do this kind of work using free and open source tools is a reminder of just how far things have come in the last few years.
Thanks to [adistuder] for the tip.
I´m impatient to see AI taking over PCB reverse engineering. This is a rewarding and interesting task for bubbling AI specialists.
Identifying chips, pulling datasheets, building a BOM, a netlist from visible tracks, rebuilding the circuit with KiCAD…
Next level would be a CNC probing robot trying to figure out values of SMD passives, figuring out the interfaces, glitching voltages or sending targeted EMPs…
Alas, most of the AI gimmicks I see here are gadget of dubious usefulness.
But for sure it will come. I hope it will come soon.
Snork
Ha.
Haha.
Hahahahahahahahaha.
Hahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahaha
Thanks. I needed a good laugh.
I guess we’re still long way. I asked AI to design a basic 3D-printed toilet seat. Instead I got something that looked like giant fish-hook, probably suitable to BDSM dungeon, not a toilet.
Realistic automated PCB reverse engineering (especially if being done en masse) would be to make a detailed optical “clone” of the board before depopulating the PCB. You then make a record as you slice the PCB along the edge. You’ll be able to see every layer, trace, and buried via within the board and be able to reconstruct it using the record of each slice. The passive components can also be tested destructively but having an AI to dig out component values from datasheets would result in the most accurate results.
I do find it a bit peculiar that people hold on to run-of-the-mill machines like this because it’s become fairly trivial to use smaller and more capable chips to simply replace the internals of machines like this. Don’t get me wrong, if it’s a completely custom device like a console, that would make sense to me but this is merely an x86 computer. This would be a good target for a Vortex86-based replacement board which would allow you to also upgrade everything else about it too.
One could say the same for old automobiles. But strangely enough, more than half the fun is making the old stuff work as it used to work. (And computers have the benefit that one doesn’t need to be unpoor, or own a farm, to be able to collect them.)
There’s two tribes in the classic car argument – the purist “rivet counters” who cherish total originality and the pragmatic ones who would fit modern components to an old vehicle to keep it running, perhaps improving reliability, performance, safety, etc. too – the “worst” example of course being EV swapping which honestly is an ever better looking way to keep a lot of classics usable and on the road.
I feel the same could be said for old hardware like this – but a lot depends where the novelty factor is – is it in the electronics, the external design, the software/games it ran? If dropping in a modern board with everything done in one FPGA isn’t throwing out what made a machine special or beloved then I see no big problem.
Swapping an uninteresting and unreliable engine out of a classic car for an EV motor is fine, but you would be ruining something special if you did the same to a V12 Ferrari, and a similar argument applies to old electronics.
my friend started buying late 1940s / early 1950s ford trucks and, with some significant limitations, he drives them around town as he goes about his business. it’s not the practical thing to do, and there is a lot to disrecommend it. but it is nonetheless useful.
that’s what makes me sad about old computers…i struggle to even use one for a few moments as a novelty, let alone to actually get benefit out of one for day-to-day chores. they used to be so magical but now they are not at all. just in my experience.
even when i was a kid, i was always acutely aware of the fact that i was using less of a machine than i wanted to be using. the best thing about using an old machine i dug out of my basement was buying the period-correct add-on that my period-correct budget wouldn’t’ve tolerated.
Great to have some more visibility for the project. There is still an ask that we are trying to locate the datasheet for the VLSI VL82C420 as it seems to be missing.
If anyone would happen to have information then please reach out to the above linked project on git thanks Mike (also involved from an advocate point of view)