Friday will be here soon enough!
With all the attention I've given the Mitac Paragon 88 Turbo XT lately, I guess its time the Ugly 286 was finally given some. While it may be a little rusty, yellowed, and have "Tasman" engraved in the plastic, its not a bad machine overall.
Current specs are: 12MHz 286, 10MHz IIT 2C87 FPU, ST-238R RLL HDD (on an MFM controller for some reason, so only formatted as 20MB rather than the full 33MB), 2MB RAM, and some Trident VGA card. The yellow FDD is 1.2MB, while the black one is 360K.
I went digging and these are the parts I found. I'll probably regret the NIC choice; it looks suitably old for the machine, but old usenet posts suggest it may not be quite as NE2000-compatible as it could be.
I've no idea if the RAM works, but I don't think I have any other machines that take this kind, so it may as well go in if it does.
And a black 1.2MB FDD to match the existing 360K drive (not that there is anything wrong with the existing 1.2M drive aside from being yellow not black).
Well, I regretted that NIC for an entirely different reason than anticipated. After the machine had been running fine for 10 minutes or so, I briefly stepped out of the room only to come back to smoke pouring out the computer.
Thankfully the computer is totally fine - no damage to the cards on either side of the NIC and everything else tests OK.
I expect the cause was probably one of those blue tantalum capacitors - there *was* one right in the middle of the black mess.
Due to the NIC failure, I had to go digging for an alternate and ended up with an 3Com EtherLink III. A little newer than I'd like, but I don't have all my ISA NICs here with me.
I had to switch to a different unfortunately somewhat newer (33.6k) modem too due to space constraints. The 14.4k modem I had before interfered with the RAM slots, and a few other smaller 14.4k modems had issues (couldn't hear the dialtone, or couldn't establish a connection for some reason)
So this is what the inside looks like now with the new cards installed. The left-most slot, and the two right-most slots are only 8-bit ISA so only the SoundBlaster 2.0 or modem would fit there, but the SoundBlaster is too long for any of those slots. The SIP memory and HDD means only short cards in the three 8-bit ISA slots.
What started with a "lets make the floppy drives the same colour" turned out to be a bit more of an ordeal. The black 360K Chinon FZ-502 drive already in there turned out to be no good, so I set out to replace it with a slightly older Chinon F-502L II.
But this drive would only step the heads once during the seek test and again while booting. Couldn't read disks at all. Almost gave up on it, but decided to try cleaning the PCB edge connector. Now its works nicely.
Then the 1.2MB Pansonic JU-475-3A14 I wanted to replace the old yellow 1.2MB drive with was misbehaving. I was sure I'd used this drive recently with a greaseweazle but now it couldn't seek properly past the middle of the disk. It would just start jumping grooves on the worm gear.
After cleaning the worm gear a bunch and helping the heads travel the full length of the disk, the drive is now working fine too. The ROM-based diagnostics were helpful in verifying that all was working as it should.
So after all that, the ugly 286 now at least has matching floppy drives, and both of them even work too! 1.2MB on the top, 360K on the bottom.
Still a few things left to do on it (like figuring out how to switch turbo on and off, and making the power LED work), but this is enough for one night I think.
@neozeed Yeah, after this tantalum on a Socket 8 board shot glowing shrapnel into the air one piece of which left a burn mark on a piece of paper nearby I started thinking perhaps some eye protection is a good idea when powering up old stuff for the first time. That and doing it outside so I don't smoke out my office!
Old CRTs I'm the most nervous about powering up for the first time, though I don't know why as I don't think I've ever had one do anything exciting.
I'm just happy to still have the board.
a crt is just a write only dram framebuffer
@mothcompute I swear there was a very, VERY early computer that used a CRT as an actual *readable* memory element but I forget where I read about this and I can't remember what computer that was
@techokami @mothcompute I think you're thinking of Williams tube memory, as famously used by the Manchester Baby. (I dragged myself through the entirety of ENIAC in Action and I am ecstatic to be able to share this trivia.)
@MrDOS @mothcompute THAT'S what I was thinking of!
@techokami @MrDOS oh the way this works makes a *lot* of sense. thats cool
@mothcompute @techokami So many early memory and storage schemes are just... not only technically impressive, but wildly creative. Yeah, no big, just gonna store these bits as sound waves travelling through mercury??? Pass me the program, please, yeah, it's all of those rings embroidered together?????? Modern storage is no less arcane (“we taught sand to remember sixteen different voltage levels!”) but doesn't feel anywhere near as off-the-wall.
@MrDOS @techokami i think the scariest part of what you just said is my computer might be storing nibbles of data as *analog*
@mothcompute @techokami I used to work with a mixed-signals EE whose e-mail signature firmly attested that “Contrary to popular belief, the world is ANALOGUE!” And I mostly try to retain my innocence by forgetting that.
@MrDOS @mothcompute @techokami meanwhile, I'm diving in head-first and snorting all the EMI/SI texts I can get my hands on :D
(it's so true though. not only is it analogue, but everything is high frequency now. that 1kHz clock signal? 300MHz bursts in the frequency domain due to the edge rate!)
@gsuberland @MrDOS @techokami i will have you know that fact is unpleasant and scary
@mothcompute @MrDOS @techokami heh, yup. die shrinks and new FET constructions did it. modern IOs have single-digit nanosecond rise and fall times now, sometimes sub-nanosecond on faster parts, so the dV/dt (and thus dI/dt) has gone through the roof, and bandwidth rocketed up with it. even if you're using "old" logic like 74-series / 4xxx-series, if the ICs themselves were manufactured in the last decade their slew rates go off like a rocket.
@gsuberland @mothcompute @techokami Gonna get me a 1GHz Cortex-M and bit-bang a 433MHz radio signal just to upset EEs everywhere.
@mothcompute @MrDOS @techokami so now we've got all these old 80s/90s board designs where folks just ran 2 layer boards, do whatever you like, traces all over the place, zero thought to transmission line behaviours or return currents, who cares, it's all slow, and nobody ever had any SI or EMI issues! but then you make the exact same board with the exact same part numbers but with die production runs in 2020 and they absolutely *spew* EMI due to modern process nodes being so fast. whoops >_<
@gsuberland @MrDOS @techokami hmm. well i have zero ee experience and one of my project ideas involves recreating a 90s board with (some) new parts so this seems concerning
@mothcompute @MrDOS @techokami go 4 layer, solid ground plane on both inner layers, that'll get you 90% of the way there.
@gsuberland @MrDOS @techokami im staring at an snes again and now quickly realizing like. literally all of these parts are surface mounted
@gsuberland @MrDOS @techokami actually i dont think i can even replace any of these chips probably. though i still have no idea how to design video or audio circuitry so i still might have to worry about noise
@mothcompute @MrDOS @techokami surface mount is not as scary as it looks. get yourself some flux (not the pens, get paste, proper soldering stuff not plumber's flux) and some of those SMD practice kits off AliExpress. tin the iron, flux the board, heat the joint, feed solder in, you'll get the technique pretty quickly.
for analogue video stuff, again the signal/ground/ground/signal stackup will save you a ton of headaches. if you swap layers on a trace, place a ground via near the signal via.
@gsuberland @mothcompute @MrDOS not the pens? I personally prefer using a pen, haven't had any issues with surface mount stuff using one. I use a drag soldering method with a chisel tip.
@techokami @mothcompute @MrDOS I don't like the pens at all, they apply too little flux and tend to dry out fast. I also find that decent flux paste acts to tack parts down for hand soldering.
@gsuberland @techokami @mothcompute Plus, how do you get the satisfaction of cleaning the stray flux off your anti-static mat if your flux doesn't leak out of the syringe and get all over everything in the first place?
But my state media compells me to sharecrop culture!: | 0 |
Always own!: | 0 |
CeX Rox!: | 0 |
I don't even own a functional optical drive: | 0 |
I've got a mountain of cassettes: | 0 |
Ha a tonne of vinyls : | 0 |
Reel to reel, baby!: | 0 |
I just download FLACs from archive.org: | 0 |
Closed
Either way it's nice to have a differencing disk image that won't get destroyed, and a couple seconds from execution to login prompt.
Now the real question is, should I even try to building something with it?
While the GMA900 never set the world on fire, in terms of OpenGL performance, it's enough for simple stuff like ssystem.
The Mitac Paragon 88, a rather nice compact 10MHz Turbo XT. Integrated Paradise CGA video, RTC, Floppy controller, one serial port, one parallel port and 768K RAM (the top 512K-640K is bank-switchable to support a 128K RAM disk). Display is an ADI DM-14+ green monochrome. In 1988 with the 20MB Seagate ST225 hard disk, this thing would have cost somewhere in the region of NZ$10k in todays money.
As its 2025, I guess its about time this thing finally got online. I had a dig through the big box of modems, and this one seems suitable enough! A little newer than the computer itself, but it doesn't really look it. Fits right in between the NIC (a far more sensible connectivity option), and the Transteque MFM disk controller for the ST-225 20MB HD
Checking RNZ News at 2400bps! Its not fast, and RNZ is not as compatible with text-mode browsers as it perhaps could be. But it made horrible screeching noises as it connected and that's the important thing! https://www.youtube.com/shorts/DCfbWiuM7b4
I used to get booted from all the local BBS's as I was too slow, but I was so thrilled to step up from 1200 baud.
@neozeed I guess by '92 14.4k modems were probably taking over! I was thinking as I heard this modem dial for the first time it probably hasn't been used since the early 90s; I can't imagine 2400 baud would have been at all useful for internet access when that started to become available to the public so it was probably exclusively used for accessing BBS in the past.
I guess its why building mach wasn't so scarry.
It took a while but eventually I did get a 9600 baud 'field upgradable/dsp' modem that sometimes could 14.4 but it mostly did 9600 which I was still amazed with. It's also back when people would meet in person to swap disks, and then eventually I'd help other people install Linux into those linux install parites. And OS/2 as well, well before NT 3.5
One last upgrade for the Mitac: a second serial port. In part because I can't think what else I would put a one-port serial card in and I have two of them lying around. This leaves it with one 8-bit ISA slot free, maybe for a sound card someday - one of those adlib clones perhaps, though they're a little on the expensive side.
I mirrored it and fixed the pathing issues so it works again: https://virtuallyfun.com/pdp11/
I thought it was pretty cool though, shame to leave it broken.