wierd_w

No, it needs 2 simms for the full 32bits wide data bus. (30 pin simms are 8bit wide, 72pin are 16bit wide, DIMM is 32bit wide) If this was a 286, you could run it with just one, but this is a 486, and it needs 2. However, mismatched simms can cause all manner of problems. Do you have a matched set to try in it?

Those look like ceramic caps now that I have a better look. It is very rare for a ceramic cap to fail, and not ... explode.. so I doubt that is the issue. Wouldnt hurt to check to make sure none of them have full continuity across them, but they are probably fine. Does the problem persist with the backplane removed? also, those 72pin simms are miss-matched.

Re-reading the opening post, the OP has already disconnected all drives. The image is pretty grainy, but it looks like there are yellow tantalum caps near the AT power connector. Those should be inspected for any signs of scorching or cracks, and should be touched gently while power is on to see if any of them feel hot or warm. If the OP has (and knows how to use) an LCR meter, they should check them.

Indeed. It is TRYING to post, and just getting stuck. Possible sources of issue: Bad PSU providing insufficient power Bus conflict of some kind Bad RAM Corrosion on motherboard from dead caps (This system predates capacitor plague, but never hurts to check for spooge) "Out of tolerance", "Partially vented" caps Unstable / improper wall outlet power (wired backward, not grounded, noisy, irregular sinewave, out of tolerance voltage properties, etc...) Removing all other cards is part of diagnosing bus conflict as possible source, but the IDE cable on these older systems is another source of potential bus conflict.

Just being more clear about the rationale of the dumb question, that's all. (as concerns the alkaline battery replacement, an inexensive "3 AA battery holder" with the appropriate DuPont connector on the end can be had through any number of suppliers. 3 AA batteries in series adds up to 4.5v, and this was a very common solution to "dead battery" back in the day. Some sticker backed velcro to help stick it to the inside of the case, and bob's your uncle.)

I am rather serious about the upside-down IDE cable. In older systems like this, the IDE controller is likely not PCI. This means the data lines have very little between them and the CPU's data bus. (unlike PCI, which has a bridge controller chip in the way). If the cable is upside down, it can cause a bus conflict, which can have symptoms similar to this. As a diagnostic step, the OP should disconnect all IDE devices. (Leave power connected, but disconnect the data cable. Often the PSUs need a load.) See if the keyboard magically springs to life, if they can boot from a diskette, etc.

That is why I provide the STP files! Glad you made use of them! (I wish more people released CAD models along with the stereolitho models like I do.) In my case, I baked in supports, because Cura kept trying to generate support material in.... bad places. (like inside screw holes, inside the hinge mechanism, etc.) It was easier to just bake a support in, and disable supports.

Genuine article MT32 is very $$. I agree that it is also pretty, if you have the space.

As others have pointed out, you can convince a raspberry pi to emulate an mt32. The emulator in question is MUNT. (I have it compiled and running on my laptop as a softsynth, with dosbox's mpu401 aimed at it. It does more or less faithful mt32 emulation. Soundcanvas Is pretty good, but the MT32/LAPC1 is better still, and widely supported.) With a USB midi cable, MUNT can respond to real midi messages just like a real MT32.

Dumb question; Is the IDE ribbon upside down?

I can add more depth I guess. The issue with support removal has no real solution though. More air gap, and the supports will fail in printing. Try putting the printed model in the freezer, then hitting it down on a table afterward. That should leverage the unequal modulus of thermal contraction, to help separate the support. 'Shopper is correct about different printers giving different results, but it gets more tedious than even that. Extrusion temperature and brand of PLA also factor in. As does if you have cooling on or not, how hot the room the printer is in is, etc. Too many things can cause 'support stuck' syndrome. I did the best I could.

M3-6 3mm metric, 6mm long. works with the M3 brass stand offs.

Ahh. Well, I have the generic shoe style for the beige speech also. I can make a "narrow" version of that quite easily. (nevermind me, I tend to prattle. I just really like this project, and want to be helpful any way I can. I lack the skills to contribute other ways. Enclosures I can totally do though)

One thing to note--- The major difference between a modern ATX case, and an older AT style case, is basically the form card in the back, and the kind of power button. Most AT boards will mount just fine on an ATX mounting surface. (may need a few plastic stand offs with the tips cut off.) Either a metal sheet with a simple hole cut in it stuffed into the form card hole, (or a 3D printed version there of), and it would be off to the races. Well, aside from the nostalgia factor of having yellowed plastic. Replacing the button with a 2-throw switch is just a few bucks on amazon away. A suitable set of switches (since the power switch on AT supplies is really just a low voltage ON signal at 5v) can be obtained off Amazon for just a few bucks. (I am actually embarking on an ambitious project to 3D print an entire AT style computer case, since I have a printer large enough for the task... but I find that if I am going to have the printer tied up for a solid month to produce such a thing, I want it to be extra special, and so have been working on trying to produce a neat textured set of doors and such-- which is taking a lot of time)

If you dont mind sharing a STP file of the PEB card's PCB (with some general keepout clearances), I can probably modify the existing data I have for my generic PEB shell, to produce a printable housing for your interface card. Or, I suppose you could just lift the STP files for that housing, and modify it yourself. It might save you some time. Given that you are using energy efficient power regulators, PLA should be an acceptable material. The big shells print quite nicely when PLA is used. It's the "tolerates heat" materials that warp so damn bad.

The issue with data bottlenecking was the reason why AGP has the GART. https://en.wikipedia.org/wiki/Graphics_address_remapping_table Regardless of what standards convention you use (DirectX, VESA, et al) to get "hardware agnostic" graphics, despite the bespoke implementations present-- the lack of a pathway to get data into the card, or for the card to signal it is ready for data, etc-- meant that such devices had no choice but to have the CPU involved, and to make use of the main system bus. This tied up the CPU, which made the simple act of drawing the screen very intensive. Some tricks could be done with hardware blitters and buffers, but any time the data inside the card needed to be updated, meant the CPU had to be involved, and OOF-- there's that operation penalty. As the wikipedia article points out, AGP introduced a new kind of MMU, which could move and map memory, allowing the card and chipset to work together for the card and its drivers to offload memory operations from the CPU. As video cards have become more and more complex (approaching general purpose computing engines in their own right-- which is why bitcoin mining and super computing applications can leverage them today), this capability has really demonstrated the power of that kind of offloading. Back in the day, the rate that you could get pixels on the screen, was tied to how fast your CPU could move bits of memory around. Something like the word size of the CPU, or the speed of the front side bus, was very important--- it was why CPU speed was such a major factor, even with higher end cards.

Yes and no. Duke3D specifically uses VESA mode calls, which call either the card's rom handler, or univbe. (Warcraft II also uses vesa modes, but calls univbe to be sure a vesa handler is ALWAYS present. Duke just makes blind vesa calls. Quake I think, also makes vesa calls for higher res. Doom is too old, and does mode X, as does Heretic, IIRC.) Assuming the driver servicing the call is written right, 2D acceleration should be in play. The bigger issue is/was getting data into the card. This was before bus arbitration, so the CPU had to do that lifting. The cards could not bus master, nor DMA that data.

By the PIII era, I had been victimized by the downsizing of the dot-com bubble bursting, and was relegated to the role of "free family babysistter" for several years. (GAK). I kinda stopped caring about things for quite awhile until I got gainfully employed again, but by then, I lacked the convenient "employee discount" I used to enjoy for parts. I was even more frugal on my purchases, and have been on the "Somebody discarded it, but it is still fully servicable" path basically ever since. I occasionally splurge, but do so with the intent of the modern "computers can legitimately last 10 years now, who ever would have figured?" reality that now exists. A friend of mine is/was a janitor, and was working for a city hall of a medium-ish sized midwestern city (in another state). He would dumpster pick when they did upgrades. Lots of P3 and P4 boards and chips that way. While he no longer does the city hall gig (because it was toxic AF,) I still did the "generally second-hand" route all the way up into the i3-i7 era, before I started doing the "splurge then wait" cycle I am on now, when I splurged on an i7, and sat on it for a very long time, before splurging again about a year ago. I fully expect my AMD Ryzen 5950x splurge to last me about that long, before it becomes "low range" again. (now I just need the crypto morons to stop driving up GPU prices, so I can invest in a vastly needed replacement for my GTX 960... Without having to sell several vital internal organs to pay the current retail price.)

A lot of this is because the VGA font comes from the video ROM. There are utilities to upload a different one into the CGA/EGA video memory, and tell dos to use it--- (there were some CGA cards that lacked a proper 80col mode, and this was the software fix, IIRC) Refresh rate though I can totally get behind. My Trident 9440 had very crisp looking graphics, but the refresh rate control left much to be desired. The next step in the upgrade path I took was an ATI RagePro II, paired with a TV tuner and a 3DFX voodoo II. (though these were PCI). The refresh control was much better with the ATI card.

The layout resembles snes pads of the era. The two buttons there are placed just like the START and SELECT buttons. As for gravis pads... apparently still being made? (Amazon: discontinued:No) https://www.amazon.com/Gravis-40011-NFR-PC-Gamepad/dp/B00000JKMB The QA section at the bottom clarifies this is indeed for PC gameport, not USB.

technically, there was a standard called DVD-RAM. https://en.wikipedia.org/wiki/DVD-RAM What you are describing sounds more like multisession. https://www.gburner.com/online-help/what-is-multisession-disc.htm With a multisession disc, the computer has to read all the sessions, with their TOCs, then interpret the filesystem virtually. This takes lots of read time, is slow and laborious, and as you noted, prone to going FUBAR.

The pizzabox was a perfectly good solution to the "Desk realestate" problem, when CRTs were king. The CRT had to sit on the desk, and if the pizzabox was not much bigger (footprint wise), it served two roles simultaneously. 1) Elevate the monitor to eye level. 2) Does not consume any additional surface of the desk, or in the foot area. tower PCs took over because they could ventilate better, and monitors became thin wispy things when LCDs took over. For an authentic 90s experience, the pizzabox is really where it was at for most computing devices. Towers were mostly for DIY enthusiasts in the 90s. Most consumer grade devices were pizzaboxes well up to the 2000s.

For drivespace afflicted computers, I find a good old fashioned laplink cable, and a suitably old laptop (with a real parallel port) is still very workable as a solution for file transfers. FastLynx (FX.EXE) is very robust in that regard. The serial cable route can even self-bootstrap getting the client to the destination without need of a working diskette drive. It just needs CTTY to be present in the dos deployment.

As concerns old apple devices, there are a number of ways to get data in and out of them relatively painlessly. One is to attach an avante' (or other brand) localtalk bridge directly to a localtalk adapter, then plug it into your local LAN. (then either run the appletalk daemon for linux, or set up a win2k box with Services for Macintosh). That way you can use a bog standard fileshare with them. There is some great utility in using an SDCard or CF card as the HDD, especially if it is easily removed (such as poking out the front, or out the back), since it is easily attached to a modern computer and meddled with, however, that has issue when you have installed Doublespace/Drivespace on the volume. (Modern machines have no idea how to handle a compressed volume file.)

I would see about using some of the various boot catalog utilities on the offending drive. After XP, Microsoft switched away from NTLDR based booting, to a more complicated and byzantine "boot catalog" based boot loader. This method requires much more data to be available, and that data is stored in that special OS partition that rides alongside the primary system volume. For a rundown, this is how win7 boots. https://social.technet.microsoft.com/wiki/contents/articles/11341.windows-7-the-boot-process-explained.aspx The name Microsoft uses for this boot catalog is "BCD". Or, Boot Configuration Data. The tool that is provided by MS is called BCDEdit https://southjerseytechies.net/blog/modifying-the-windows-7-boot-loader-with-the-boot-configuration-data-editor-tool/ There are various GUI tools that offer automation. Since this process is very.... Fragile... and very easy to screw up with fiddling... I would recommend using an automated tool, rather than trying to prod at it manually with BCDEdit. Hilariously, this is what "Startup repair wizard" does (or tries to do). However, because MS are such paranoid cranks, they wont/dont allow you to just fix a horked up BCD entry for an OS install from the recovery media. If the recovery media thinks it is not a valid windows installation for some reason, it wont touch it. That's where the 3rd party utilities come into play, as they can identify installed versions of windows that lack suitable BCD entries in the hidden system partition, and then create them for you. (allowing them to boot.)