Gregory's Serverfor teardown and inspection i had to remove the entire plastic case. the monitor inside is shielded entirely with metal held shut with security "flat" screws. no bad caps inside, no RIFA caps, nothing else looked damaged.
|
51 comments
 the hard drive is this weird IBM 30MB model. the interface is a weird primitive form of IDE, but totally proprietary (apparently) the computer boots with RAM errors. i decoded these the hard way by swapping SIMMS and chips around. memory is strange in this system. there are two 256KB SIMMS. yes, kilobyte. that still doesn't add up to 640KB, so IBM added 128K using discrete memory chips! looks like one of these chips has gone bad. this RAM chip has a little attitude! actually it was working just fine. the bad chip was a different one. i replaced it and the machine booted! even though it has a hard drive, it's not the 286 version of the Model 25 (the only versions to come with a stock hard drive). in fact it has been upgraded with the famous NEC V30! an 8087 has also been installed. someone tricked out this machine. it was easy to remove, but i couldn't find out what kind it is. nothing at the local hardware store matched. spoiler: it is something special. after a ton of research, it turns out to be a Medeco Biaxial lock (original) but with a rare 3-pin arrangement. since the front face does not have a logo, this is likely a special version made just for IBM. https://www.lockwiki.com/index.php/Medeco_Biaxial  @tubetime Even with tumbler disassembly, I'm so surprised you got it to work. I also wonder if it's cheaper to have a locksmith make a key from that or try to use it for a mold. @tubetime I guess you could get a metal key cut from it at the local hardware. @tubetime your Medeco Biaxial key reverse engineering feels like something @deviantollam would be interested in. He might also be able to suggest where you could get a suitable metal blank to suit your pinning (or to suit a repinned version). (3D printed sintered metal probably also isn’t strong enough for frequent use, but a metal key cast from the negative of your reverse engineered key model might be sufficient for light use.) @ewenmcneill @deviantollam i found his video on the pinning tray, so that was helpful. @deviantollam @ewenmcneill thanks for the kind words. right now it is just some random notes and a crappy solidworks file. i am not convinced i have the depths right. i also have another lock (with no key) of very similar make, so i'll probably take that one apart and compare the pinning @tubetime Wow. This is genius. Amazing work. Thanks for sharing.  @tubetime Pretty impressive reverse engineering there. Impressive! First idea would be James from Clough42 on YouTube. He seems really skilled with 3D CAE and delicate CNC machining.  @tubetime you’ve already got a working model for an FDM printer, so I’m wondering if this would also work in a SLS metal printer? No idea what the cost of the print would be though. Or you could have a go at making a high temperature silicone mould of the key and casting it in pewter.  @tubetime neat! Could you get an outfit to print you a copy in metal with DMLS? I have no idea how much such a thing would cost
[DATA EXPUNGED]
@HoustonDog if you don't have a compatible keyboard but the port itself works, maybe build an Arduino-based converter from PS2. if the port is dead, well, fix it already 😉
[DATA EXPUNGED]
@tubetime Oooh interesting. My 25 doesn't have this lock. Mine was a single floppy model, and the right bay is just 100% blank. @tubetime Seems like a majority of the PS/2 Model 25 I find for sale are the 8086 monochrome variant. Kinda want to get one of the 286 color variant sometime—just like the ones my middle school had—although it’s a low priority. (Apparently there was a 386 variant too but I expect it to be hard to find.) @tubetime A lot of PS/2 used ESDI drives, but that looks like a standard MFM or RLL to me - the ESDI had a wider connector. @ThomDenholm it's kinda like ESDI but with a subset of the ISA bus. but not quite standard IDE. |
the floppy drive needed some work, however. the motor boards in these 720KB 3.5" drives have leaky SMD electrolytic caps. i had to clean the board and replace them.