It looks like I bricked my Kayak XU800 board. BIOS crisis recovery doesn't work either. There is no floppy activity. The BIOS chip is not socketed unfortunately, and I don't want to place an empty socket on the board with the risk of ruining the whole thing. So I figured I could hook up an empty socket in mirror-image. Test drive using a spare EEPROM:

Only a few pins left who are making bad or no contact... I slightly bent the contact points in the socket using a modified staple.

Sometimes you need to cut the bottom out for sockets to seat deep enough to make contact when on top of a chip.

BitWrangler wrote on 2026-10-26, 19:35:

Sometimes you need to cut the bottom out for sockets to seat deep enough to make contact when on top of a chip.

I see what you mean, but here's our brick:

Not many tall components, just a few low profile SMD caps or resistors but these are not obstructing.

I had mixed luck flashing chips while still in system, stuff like whole mobo trying to power off injected voltage etc, sometimes powering on and keeping it in reset helps

Well, I did something pretty dumb and impulse-bought a GPU for my retro PC, even though someone already offered to give me the same model I was looking for without paying £50 for it.
(I may be stupid)

Installing the 7600 GS has given my retro PC a good boost and saved the 5700LE for if I want to make a Pentium 3 PC with it one day, though I think I'm also running into some driver issues. (I just chose the most up-to-date drivers, which I know is not a good choice so I'll be looking into some better ones soonish)
Sonic Heroes (running at 640x480) does run better, but also noticeably runs way worse than I'd expect for a game that needs 16mb of VRAM and a Pentium 1, with more the rising platform gimmick elements in the first level causing a whole lot of lag.
Spore runs better but running in fullscreen constantly forces me out of it, and Peggle's hardware acceleration actually works better than software mode but the game looks like it was stretched slightly with nearest-neighbour filtering.
I've not yet tried any other games, although Discord Messanger does still work and Mypal does too and slightly better (but not by much)

There's definitely some kind of bottleneck in this computer, though I'm not actually 100% sure if it is drivers or if it's the CPU (3.06ghz Pentium 4 HT), or the motherboard (SiS chipset + no actual microcode for hyperthreading), or the USB 3.0 PCI card is saturating the PCI bus, or maybe Windows Vista is just too much for this thing after all and I should cut my losses & give up on it.
(or maybe I accidentally didn't put the molex connector in properly and it's throttling performance. I did try to put it in as far as I could but I didn't want to risk damaging the GPU board...)

(At the very least, I think hardware-wise this system is basically maxed out now for what the motherboard can feasibly support. there's definitely room to improve if I get a better motherboard, but I think that'd ruin the charm given how that's the only thing left in the build from the original computer.)

I had to print some government forms, which helpfully always have some fine print that isn't legible on the trusty dotmatrix printer. My Canon ip2700 is not working. It might just be a dead ink cartridge but it's hard to know with these things. Same deal with my retro inkjet, a Canon BJC-1000 that I got in 2000. But I had a 'photo' ink cartridge sitting around that was still sealed in the container, so I gave it a try. I was a bit surprised, it actually works...

VanillaFairy wrote on 2026-10-26, 20:50:

There's definitely some kind of bottleneck in this computer

SiS chipset

Mystery solved

CD-ROM repairs (sigh...)
Among all the retro repairs, individual points of failure and maintenance, CD-ROM drives are the ones that frustrate me the most. Even more than trace repair! Battery leakage causes tangible damage - you either succeed or you don’t with the repair. CD-ROM drives on the other hand, often don’t even properly fail - they rather get tired, don’t open or refuse to cooperate in countless other ways. And by the way, I never had a single CD-Drive that was fixed by cleaning the laser with IPA. Today I had a drive that would only reluctantly eject - but keeping the CD inside! (I have no idea how that is even possible - maybe the drive decided it was from now on a multi disk player???) I also had the glorious idea (after many years) to test my precious PLEXTOR burner only to find disks with heavy data corruption. At the end of such a day I always ponder the ZULU option but that’s not the same thing…(angry rant over)

The last time I had to repair a CD-ROM drive it were the rubber grommets of the mechanism that had developed too much sag, causing the mechanism to hit the static part of the drive. Added grommets and now it works again. And on the CD drive of my PS2, I had to do the well documented trick with the tape on the clamp to have it read PS1 discs perfectly.

BitWrangler wrote on 2026-10-26, 14:03:

A couple of my older style ATX with the fan flush to a flat stamped grille in the back, the stamping had fat ribs so seemed to block a lot of airflow, and these PSU had a definite woooooooo noise when running. Just cut out the stamping altogether and mounted wire grille external and they went very quiet and case temps seemed to drop. (Pre 2000, only airflow is the PSU type systems) ...

^ That's one of the shortcomings in 1990s vintage hardware that I can't withstand: Small size fans, often choked by ridiculously restrictive grilles = Atrocious ventilation = High temps inside. Just what old electrolytic capacitors need [/SARCASM].

^ Pick your poison.

I don't make prisoners amongst exhaust grilles, and only use the raised wire type on some selected intakes:

The spice must flow!

Aui wrote on 2026-10-27, 08:27:

CD-ROM repairs (sigh...)
Among all the retro repairs, individual points of failure and maintenance, CD-ROM drives are the ones that frustrate me the most. Even more than trace repair! Battery leakage causes tangible damage - you either succeed or you don’t with the repair. CD-ROM drives on the other hand, often don’t even properly fail - they rather get tired, don’t open or refuse to cooperate in countless other ways. And by the way, I never had a single CD-Drive that was fixed by cleaning the laser with IPA. Today I had a drive that would only reluctantly eject - but keeping the CD inside! (I have no idea how that is even possible - maybe the drive decided it was from now on a multi disk player???) I also had the glorious idea (after many years) to test my precious PLEXTOR burner only to find disks with heavy data corruption. At the end of such a day I always ponder the ZULU option but that’s not the same thing…(angry rant over)

Sounds to me like the hub that clamps down got somehow "sticky" either literally, from a CD that had a bath in coke or something, or rubber on it that has gone off, or the plastic got grazed and is catching on inside ring of CDs....

But yeah, CD mechanisms are a royal pain in the ass. I picked up an inch thick book a few years back. "Compact Disc Player Maintenance and Repair" hoping that that would tell me the magic tricks to successful troubleshooting in the pickup and laser head. But no, it has about one paragraph total about the actual business end stuff that goes wrong and tells you to leave it alone. The rest seems to be all about troubleshooting the power supply and audio amp.... pretty much the generic parts that would be in a tape deck etc also. A good chunk of the beginning of the book is a complete waste of paper, like describing a CD player to someone from 1920 who has never seen one.

Edit: hah, yeah, knew I'd said something about that book before, found the image I posted, this is all the help you get in there...
download/file.php?id=195028&mode=view

Some old rubber literally melts into a goo not unlike fresh silicone glue.

myne wrote on 2026-10-27, 12:57:

Some old rubber literally melts into a goo not unlike fresh silicone glue.

Yeah it can exude sticky oils as well, if there's a rubber ring/washer under the spindle it might ooze down onto it.

Recently got a prototype FPGA MIDI synth board made and have been experimenting with porting classic MIDI synthesizer emulators to it.
It’s based on a Lattice ECP5-45F with 32 MiB parallel NOR flash and 1–2 MiB SRAM. No MCU, just FPGA gateware, which so far has been developed using LLMs based on the work of Nuked, gearmulator, and MAME.
The board fits a Wave Blaster header but can also run standalone.
Working so far: Yamaha FB-01, Roland MT-32/CM-32L, CM-64 (including U-110 cards), SC-55/SC-55mkII, SC-88/SC-88 Pro, and Yamaha DB50XG.
Still experimenting with adding more synths.

Line straddling again...

I ain't not no Englishologist, preferring in the tradition of Billy Shakespeare to adopt and adapt contemporary slang and abuse the hell out of regular verbiage for entertainment and amusement. Therefive*, I am unsure of the correct construction of the terms to describe close to retro and close to modern, I am thinking something like para-retro/para-modern, ad-retro/ad-modern, or ante-modern/post-retro or pre-modern/semi-retro. (* Therefore, but one bigger.)

But yes, stranded in no-mans land I are be. Trying to sort out case priority for machines on the one hand, para-retro, they could run modern stuff slowly, but I'm tilting them to Vista era, but not necessarily Vista, and para-modern, I wanna use them more as dailies on Win10 up stuff, harder, better, faster, stronger, or whatever Daft Punk said. El problemo grande is that they all want more or less the same case/PSU facilities in the way of power, ventilation, room to install longer GPU etc. The second massive problem, is that yaws terribly (A conflation of yours truly and a boating concept that I might be prone to swerving off course.) has a tendency to start "building up" the things he's supposed to be tearing down, possibly due to having extra hardware options since built. So a lame AM2 with lowest X3 in it, and I'm thinking "Oh, I have an Athlon II x4 that would be better, and I haven't thought of where to put that R5 240 yet..." Aaaargh, I was gonna evict you!

So yeah, going in circles.

New plan tho, spur of the moment, so absolutely no unforeseen problems, naturally, Operation Garbage Bag Box.. just noticed I have a sturdy box I can put a populated AM2 board in and run it on a whim.

Edit: aaargh, the stripe not striping. I was making prep to remove, very next job was to take any useful data off the Raid 0 pair I had in there, now non booty. hoping it's just cable problems...

EditII: I think the ends of the SATA are a bit worn, glitchfest. Getting it booted... buuuut, plug another cable into the spare SATA hole and it's gone again.. not gonna get a drive for Xfer in there like that. Next time I see a thousand gallon aquarium tank for free, Ima fill it with deoxit and flourinert and just run everything submerged in it 🤣

EditIII: I need a shower after dealing with Windows shit that long. For some reason, it doesn't like the WHQL Radeon drivers fresh from AMD, (well the 2022 ones which supposed to be latest for R5 240) says they are unsigned, then locks up on boot when I force it... so yeah not running the R5 240 in it, turned into a whole thing. Possibly the board is a little old?? Dell weirdness?? IDK, but I gave up while I still had the 2 milligrams of sanity I started with. Checked what the nVidia driver that was previously installed supported, and picked a maxi-min (highest gen, lowest power) of a 440GT which is a bit yawn, but working is good. Damn and I still haven't rearranged the drive and backed anything up, never mind moved it (as a going concern as it were).

EditIV: So apparently there are many folks who have had troubles getting right R5 240 drivers going with similar hangs and blackscreens. Anyway, for next attempt with that, I found some on Driver Scape (do we trust them?) that are 2015 released, so in the theoretical 2 years after hardware release debugged sweet spot.

Then also I have been having 2nd..9th thoughts about the direction for the evicting to a box machine. If I am bending it to more retro, I probably want higher single core, so my 2.6Ghz of Athlon II in the 4 core is a bit lame compared to the X2 I can put in it.

Discovered a 90's TV website called SPELUNKR TV. It's pretty cool, can cast to shield box and with TV remote can flip or 'SPELUNK' through channels with 90's content

The AM2+ machine continues to be more of a pain in the butt than I was bargaining on. I am running out of patience for the "boxing it up as fully sorted" idea, and might just leave it kinda working. I updated the BIOS to a June 2009 version, which supports Regor cores.... but only up to the A2 X2 250. Then a little while after that dropped, so did the 260 I have, but no major updates just a clock bump. Anyway, so I swap that in figuring it should be good. Was working in an AM3 board fine last year. POST screen says unknown processor, which I figured, gets into windows. Now first time I reran the performance index it said it couldn't do it, errored out, but there was background thrash because update decided to do random shit at that point in time. So got that closed down, ran again and it completed. Surprisingly, it got a whole 10% over the x3 8450, score of 6.6 vs 6.0. Cpu world comparison said it would be about 4% better. Also only about 4% worse overall vs an A2 X4 620, though that one caned it on multithread. Level or ahead on older 3Dmarks though.

So one glitch, seemingly overcome, but then I can't run 3dbenchmarkok which was a simple graphics benchmark. Ran it on the 8450 when I put the 440GT in and it completed then. What I am not sure of, is if I ran it before or after the BIOS update. Anyway so could finger the BIOS update, OR the cpu change.

I am realising though I might have forgot to reset CMOS, so will have to try doing that and reconfiguring.

Anyway, that's probably last try before condemning it to the box "as is" anyway and since the whole idea is it can run from in there, I can sort it out in there some other time.

Edit: Dammit, just twigged another flaw in my plan, the 440GT only has DVI-D, when I want it somewhat "portable" run anywhere, not having a home, then I want more flexibility. So I gotta swap the graphics out again, up/down/sidegrading to a 520GT being most convenient ATM... not sure if it might get something a little better when the game of musical chairs slots ends among the GPU.

EditII: So resetting CMOS to defaults and putting settings back in again seemed to fix the no-load of that benchmark and everything else looks hunky dory. So I feel like a bit of a maroon, maybe even a nimrod for forgetting that. Now it's running blazingly mid with 96 cores of Fermi power, I need to cut that in half with the 520.... yeah don't worry this ain't a 1080P gamer.

For the past month I have been very active with my retro builds, especially with my P233MMX industrial PC. Looking back at my few posts on this forum from the past 12 years they all seem to revolve around this build or at least some of the parts in it: issues with AWE64 Gold, then with Trio64V+, then again with AWE64 Gold with me thinking it had broken.

As I bought for my other build an Audigy 2 I decided to finally see if I could fix my AWE64, with the support from a - yeah, trigger warning, LLM. It didn't take too long for me to realize my card wasn't broken but instead all IRQs and DMAs between it and my Audician were a mess. Having then solved that, with my AWE64 Gold back in action I then tried my Trio64V+ again. I was quite sceptical of this card due to having faced so many freezes with it, but as the SBC onboard Cirrus Logic produced so horrible image with my newly bought Kramer scaler, I decided to try again.

The image quality was entirely different, like night and day in comparison to the Cirrus. But I also got freezes, many of them, in Windows and in some games such as Duke 3D. I had the LLM write me a diagnostic software to help me find out if the issue was with the Trio or with my sound cards or with something else, just to make sure. It tested memory, tested playing sound, burned the CPU. By this time I also had removed the memory modules, tested the Trio without them and the modules swapped. Many long nights I spent with the diagnostic software. Everything was surprisingly stable. However, ultimately I found out that starting Windows 3.11 with higher colour depth than 256 colours in 800x600 resulted in a freeze after clicking some buttons.

By this time I had already ordered a Virge DX, and when it arrived I was anxious to test it. Alas: it was much worse, it couldn't handle even 800x600 256 colors. And the image was super washed out. But fortunately that graphics card wasn't my last option.

My friend was throwing away some old parts and I was in need of an audio cable to put between my CD-ROM and AWE64 so I asked if I could go through his stuff. While I wasn't able to find that cable (it's so strange, I have the feeling I had many of these in the past, where have all of them gone??) I noticed there was a PCI GPU in one of the cases. A Geforce 2 MX400! And as I tested it in my setup, I was super happy: finally I had found a GPU that's stable and its DOS compatibility seems to be great. Sky in Lotus 3 has no bars any more and with VBESVGA.DRV I was able to test Win 3.11 stability. It was excellent, although the performance was slow and even the mouse and midi playback sluggish (I hadn't realized the buffering of the driver should be turned off in system.ini - doing that solves the issue for an older PC like mine).

Because I didn't realize the cause for the slow VBESVGA.DRV I thought it would be for the best if I had a driver with actual hardware acceleration. So I asked my LLM if it could write one for me, based on VBESVGA.DRV and Linux Nouveau driver. Building also upon the work it did previously with the diagnostic software, it has been building me that driver & a test bench for performance in Win 3.11 - and I have seen it go up a lot.

I have no issue confessing that I have absolutely no competence whatsoever in building drivers for GPUs, no know-how and no time for acquiring it, but the amount of time spent just moving the files over to my retro build first with CD-RW, then over null-modem and finally over Ethernet and bringing back the necessary logs of the tests made it all feel like work, even if I'm just a "meat-proxy", ha . Ultimately my Geforce2 hardware accelerated Windows 3.11 is probably going to be just a nice party trick, since I guess there aren't that many programs that would need that kind of performance, but at least I can have a high resolution with high bit depth and feel that my GF2 is supported 😉

Back to testing the next version and hardware cursor...

Okay AM2+ now GPU lobotomized down to GT520, GF119 rather than GF108 though. Core speeds are highish but the memory is crap 64bit DDR3... which is how come I am getting outbenched by a GT220 with 3/4 the core speeds, same number of cores, it's got 128bit GDDR3... ah well. At least it beats an HD5450 like the scrolls predicted... and I now have 3 choices of output DVI HDMI and good ole 15 pin VGA. Though I thought I remembered seeing a GT220 around somewhere, completely ignored as I thought it wasn't worth looking at vs 4, 5, 6 series low enders. It is kinda weird though that they would release replacement cards for points in the range that were slower than previous gens even if core clock faster.

To the box it goes now, as it is doing okay in that config... looks like laying out in there okay, mATX board, I am kinda wondering about cheap plastic crates in that size now.... could put maybe 3 or 4 mATX systems in those, stack them or rack them.

Wanted to use a socket A platform for some tests.
The PCI video card I used would make it freeze just after entering bios or boot screen. Either way I had limited time.

Put an AGP and ok, ISA and ok.
Video card is a S3 Trio64V+, board is https://theretroweb.com/motherboards/s/qdi-k7 … -k7e-kinetiz-7e
Maybe a chipset related issue?
1st time I had issues with this card.