Tried to boot Windows 98 SE with iSCSI off my NAS by booting on real hardware with iPXE. I followed this tutorial and adapted it to try to start Windows.

I managed to install Windows but could not get past the first boot splash screen with a Windows Protection Error. It was also dreadfully slow-going. But, I got a lot farther than I thought I would.

Shponglefan wrote on 2024-10-29, 23:46:

Swapped a couple more damaged RAM sockets on a pair of Sound Blaster AWE32 cards (CT3900 and CT3980). […]
Show full quote

Swapped a couple more damaged RAM sockets on a pair of Sound Blaster AWE32 cards (CT3900 and CT3980).

Had a bit of a panic when the CT3980 didn't seem to be recognizing the installed RAM. Spent about 30 minutes checking continuity, reflowing solder joints, etc. Until I finally realized I had forgot to set the jumper to enable external RAM. 😅

Before:

After:

Those new ones look far better, I also had to swap out the SIMM sockets on my CT3670 a while back since they were both broken. I did think that a 3d print would work to hold the SIMMs in place but then found some good replacements, but not metal clips like you have there so I treat them very gently.

Today I put my Toshiba T4850CT back together for hopefully the *LAST* time. It's been apart for capacitor replacements, floppy drive repair multiple times because 3d printed belts aren't perfect. 3d printed replacement parts had to be made for missing parts. The soundcard has given me so much trouble - I recapped the sound board since it had one of those Elna 6.3v 1000uF stubby caps that goes bad on just about every Toshiba from this era but it wasn't that that was faulty. (I had good reason to suspect it though, just fixed a T4700CT with no audio which was bad caps on the audio board...)
In fact my recapping of the sound board broke it for a while. Now I'm using a 560uF polymer cap in place of the original 1000uF to fit the height requirements of the sound board. The little electrolytics in plastic boxes have been replaced with MLCC caps so they won't go bad over time.

The ridiculous part is, I started recapping the sound board because it would go BZZZZ from the speaker sometimes after playing a game for a bit. Which would persist after a restart and made me think there was an electronics fault. There was corrosion all throughout the inner casing with corrosion evident on the speaker area too and although I recapped everything, it turned out that the little 0.5w 8 ohm 36mm diameter 4.4mm height mylar speaker must've had a point where it could get stuck in low resistance.
Those are pretty cheap to replace even though I did need to sand down the plastic frame of the new speaker to get it to fit the very limited height requirement of the original speaker:

Sounds great now and hasn't gone into that BZZZy state since.

Messing about with the sound card caused problems with the BIOS not detecting the Windows Sound System card / sound board because of initially a soldering error on my part with a tin whisker bridging a cap I replaced to a nearby cap. Thankfully that did no damage, but then it was board-to-board connections being bad stopping it from detecting. After re-seating and cleaning connectors a few times it's behaving and I hope I don't have to go into this laptop again any time soon. And this was with putting the laptop back together and tearing it down again several times. This final re-assembly I've tested sound at each major step.

So at long last I can actually play a game on it without the sound going messed up. Sadly this T4850CT is no perfect example, there's a big scratch on the polariser layer of the LCD which I might fix in the future but for now it's not causing much of a problem in use. Actually using this laptop is quite nice, it feels quite premium for its time since it can charge the battery while running and shows the remaining battery life on the status LCD.
The sound though, well it's a Windows Sound System with FM capability (the T4700 / T4800 are WSS with no FM) so it's not very soundblaster compatible. Running wssxlat gets audio part-way in that it can do sound in Wolfenstein 3D and some other games but there's no digital audio in DooM. This is where having a frickin ton of Toshiba laptops helps, I remembered that my T4900CT has the same WSS sound setup and could give sound in DooM, albeit only in Win95. It seems that Microsoft made a more extensive WSS translation layer / software / driver in Windows 95 which works well with DooM so there's digital audio and FM music all working well and reliably, but it comes with a speed penalty, I had to run DooM at a couple of notches smaller screen to get playable frame-rates:

Tried out FastDoom for a bit but that really does not play nice with Windows 95 or the WSS translation in either DOS or Windows, so it's classic Doom only after all.

Oh and been messing about with the battery on this laptop, I re-celled one of these Toshiba packs a while ago and it charges up and works pretty nicely on this T4850CT. It is quite nice being able to run the laptop from its batteries again, even if not very important.

Glued together the printed parts for my “new” front bezel, looks OK so far. To be honest, I imagined the 3D printing stuff to be less cumbersome and more precise with less warping, twisting, etc…

Next up it’s time for some plastic putty and grinding, one of my favorite activities - not 😒.

CMB75 wrote on 2024-10-31, 09:35:

Glued together the printed parts for my “new” front bezel, looks OK so far. To be honest, I imagined the 3D printing stuff to be less cumbersome and more precise with less warping, twisting, etc…

Next up it’s time for some plastic putty and grinding, one of my favorite activities - not 😒.

Different materials behave differently.

PLA tends to print very easily, but is prone to sagging and warping if it gets even the slightest bit warm after printing.
PETG prints "okay", and has good thermal tolerance after printing, but tends to "NEVER LET GO" of glass build plates.
ABS is abysmal to print, and I would avoid it like plague on principle. In theory it makes stronger prints that you can vapor smooth. I my experience, it's just a waste of money.
Nylon produces strong and flexible parts, but wants to warp, creep, and "elephant foot" like crazy unless you use a heated, temperature controlled printing volume on the outside of your printer. Paint wont stick to it, nor will glue.
PVC (Yes, you REALLY CAN buy it) makes fabulously hard and durable prints, but is SUPER HARD to get dialed in, and produces hydrochloric acid gas during printing that will corrode your print nozzle. It has a margin of tolerance in its melt window of like 5 degrees Celsius. It DOES work with plumbing aisle PVC adhesives, such as for pipes.

My guess is that you printed with either a more modern formulation of PLA, or with PETG.

wierd_w wrote on 2024-10-31, 10:09:

Different materials behave differently. […]
Show full quote

CMB75 wrote on 2024-10-31, 09:35:

Glued together the printed parts for my “new” front bezel, looks OK so far. To be honest, I imagined the 3D printing stuff to be less cumbersome and more precise with less warping, twisting, etc…

Next up it’s time for some plastic putty and grinding, one of my favorite activities - not 😒.

Different materials behave differently.

PLA tends to print very easily, but is prone to sagging and warping if it gets even the slightest bit warm after printing.
PETG prints "okay", and has good thermal tolerance after printing, but tends to "NEVER LET GO" of glass build plates.
ABS is abysmal to print, and I would avoid it like plague on principle. In theory it makes stronger prints that you can vapor smooth. I my experience, it's just a waste of money.
Nylon produces strong and flexible parts, but wants to warp, creep, and "elephant foot" like crazy unless you use a heated, temperature controlled printing volume on the outside of your printer. Paint wont stick to it, nor will glue.
PVC (Yes, you REALLY CAN buy it) makes fabulously hard and durable prints, but is SUPER HARD to get dialed in, and produces hydrochloric acid gas during printing that will corrode your print nozzle. It has a margin of tolerance in its melt window of like 5 degrees Celsius. It DOES work with plumbing aisle PVC adhesives, such as for pipes.

My guess is that you printed with either a more modern formulation of PLA, or with PETG.

No real secret there, I used ABS-like resin because I've had even worse results using standard resins. It does print within very nice tolerances and I've mastered the shrinkage (more or less) but I haven't really mastered the bending and twisting which seems to occur while curing after a couple of weeks. That's why I have to do a lot of touching up with plastic putty and grinding.

Started some work on my fresh EPoX EP-51MVP3E-M.

- added a K6-II+ 500MHz, 2.1V Vcore as it seems the board doesn't go lower (@Sphere478, you happen to know a lower voltage for this board?)
- 256MB RAM spanned across 2x64MB and a 128MB stick
- ESS Audiodrive 1869 😀
- Intel Pro100 LAN (funny innit)
- 3D Blaster Banshee CT6750 16MB

Currently workin' on a 98 reinstall. HDD came with a dead K6BV3+/66 that, despite fitting the EPoX like a glove, is in dire need of a refresh. I have all me aces up me sleeves.

Also dug out my AOC 7K+ CRT. Gotta love how well it runs.

Building a "new" 6809 CUBIX lab workstation.

I originally developed CUBIX in 1983 as the operating system for my homebuilt
6809 portable. I later built a workstation for my lab, and built a number of
test tools on that platform. (you can see these on "Daves Old Computers")

My D6809 portable died last year, and I gave it to another guy who was very
interested in reviving it. I decided the workstation was too big (not using it
nearly as much these days), so I disassembled it.

Lately I've had need of it again, so I decided to "build another smaller one".
The original CUBIX 6809 system had:
0000-1FFF = I/O space
2000-DFFF = RAM
E000-FFFF = 2764 (8K) ROM

I happened to have some boards from a NewBridge 1032 Data PBX that I designed
for them years ago, which came close:
6809 CPU
0000-1FFF = I/O space
0000-BFFF = RAM
C000-FFFF = 27128 (16k) ROM

I piggybacked another 8K RAM on one of the existing ones, and brought the
now extra address line on the 27128 out to a front panel jumper.
I put CUBIX OS (8k) into half the 27129 and my own MON09 (good stand-alone
6809 debug monitor) in the other half - easily switched between.

I had previously made some 765 based FDC boards that fit the same slots as
were on the 1032 - and a 1032 I/O card gave me 8 6551 uarts.

All this assembled in a much smaller homebuilt rack!
And so far, everything seems to be working!

Dave ::: https://dunfield.themindfactory.com ::: "Daves Old Computers"->Personal

Do you have any photos of this hardware ? It sounds very interesting ~

Tiido wrote on 2024-11-01, 10:24:

Do you have any photos of this hardware ? It sounds very interesting ~

You can see both original systems on "Dave Old Computers" under "HomeBuilts".

The "New smaller" workstation is basically the same hardware as the original, I did cut-down a backplane by removing the part with the
power connector and soldering wires directly to the main part. This system is only the height of the cards, has a much smaller power supply
on the end, and uses external drives (a bit more work to set up when I need it, but much easier to store).

----

If you want, you can even try/experience using it!

A few years back I wrote an emulator for it.
"my site" -> Downloads -> "Older downloads from previous site"
Look for "Simulators and Emulators" ; "D6809"

It is a 16 bit DOS program, but it runs under DosBox very well!

This will let you see my CUBIX OS in action. Back when I designed it, I was working on
VAX mainframes, so the interface does look a little vax-like 😀

As it was one of the earliest OSs that I created.
It doesn't have a full tree directory structure but "fakes" a single level of
directories fairly well. It has a "directory prefix" in the filename which
gets auto-added if you don't include "[directory]" in the filename ... you can
of course set that "default" directory.

A sometimes useful advantage of this approach is that you can use wildcards in
the directory prefix: dir [MC*]*.*
shows all files in "directories" beginning with "MC" (typically relating to my
Micro-C compiler on my systems: [MCSOURCE], [MCBIN], [MCLIB] etc.

Obviously this approach wouldn't work efficiently on large hard drives, but for a
typical D6809 system with 360k floppies it worked quite well!

Dave ::: https://dunfield.themindfactory.com ::: "Daves Old Computers"->Personal

Very fun ! Good use of wood as construction material too ~

I'll skip on the experience though, there's no interesting video or sound hardware which is the main appeal to me (I'm slowly working on something fun on both of those aspects, using PLDs)

PcBytes wrote on 2024-10-31, 20:30:

Started some work on my fresh EPoX EP-51MVP3E-M.

- added a K6-II+ 500MHz, 2.1V Vcore as it seems the board doesn't go lower (@Sphere478, you happen to know a lower voltage for this board?)

What is typical is that all the jumpers off give 2.0 or maybe 1.8 some boards... if they're already off, then it's down to looking up the data sheet of the voltage regulator and seeing what you have to change the bias resistors to to take it lower.

Hey guys
I have already asked this question in a separate thread, but since the range is greater here, I wanted to find out from all GA-586HX users whether these memory modules, each of which is 64 MB in size, work with my Rev 1.55.
They could be ECC buffered fpm modules.

ECC needs CPU support methinks, these must be for P Pro or early slot 1.

BitWrangler wrote on 2024-11-01, 15:32:

ECC needs CPU support methinks, these must be for P Pro or early slot 1.

So absolutely no chance to use it in the GA-586HX board?

Chipset is supposed to support but seeing nothing on Gigabyte's side to say they fully implemented it. Think what I said before was just for those lines of CPU, when Intel started blocking things from working unless you bought the spendy crap instead of just letting it if the silicon was there anyway.

BitWrangler wrote on 2024-11-01, 15:58:

Chipset is supposed to support but seeing nothing on Gigabyte's side to say they fully implemented it. Think what I said before was just for those lines of CPU, when Intel started blocking things from working unless you bought the spendy crap instead of just letting it if the silicon was there anyway.

In that case I have no choice but to try it out. Are there even such large modules without ECC buffers? After all, the board can support up to 512 MB of RAM with a second TAG module

For 128mb modules? I've only ever seen registered simms. I'm sure 'normal' simms exist for 128mb, but I haven't seen any.

***generally speaking*** most ecc ram will work in non-ecc boards. The chipset is just going to ignore the extra ram chip. Registered ram typically only works in server chipsets for which it's designed.

Also, in my experience, intel chipsets are much more tolerant in this department than others. At least in this era of hardware.

Thank you, that helps me a lot to know that.

I just came across a small box of stuff tucked behind things for years that is super welcome... it's full of all kinds of unexciting parts, drivebay covers, slot covers, 3.5 to 5.25 rails and adapters, drive rails, 2xCDROM audio cable, a couple of drive cables, all the stupid stuff you end up short of to finish off systems, the stuff that nickel and dimes you, or more likely these days $5 and $10s you. Anyway in that I found my precious... my original IBM XT speaker, the sole part remaining of my first XT I had to abandon in long distance move.. up until early 2000s I had managed to keep an unbroken line of upgrades, replace everything but the speaker... so I was still using my "Ship of Theseus" original IBM XT, but I had misplaced it, until now... So I gotta jump it through some quick upgrades, Tbred to X2 to X4 to FX8350 to bring it up to date...

So tested it on mods, on same setup I am using that speaker I liberated from a modem. I dunno if it's just that this one has been through hell and back, but it's bad, worse than the smaller 2". Seems to have no response below 500 and above 7000, whereas I guesstimate the modem speaker covers about 50 to 10,000, neither of them at all linear, response at each end is poor, but talking about the cliff where it falls off to nothing. Anyway, IDK if that in particular is why the original IBM speaker is maligned, or whether this one just kinda shagged.

Weird BIOS ROM off to the left in pic there only says H. T. M11003 on it... I'm guessing it might be our favorite penny pinchers Hsing Tech, and remember a bad ROM on a PCChips board way back that this might actually be, but any ideas of it's equivalent to try burning it real good not just on a PCChips minimum viable EEPROM interface?

Also shown there, some 5.25 floppy connector to 3.5" drive connector convertors, useful in the case where you have onboard I/O on OEM formfactors where the cable has it's own end or is soldered down. Or when you need a really long cable to get around the other drive cages for a right mount sidewinder floppy, and the 3.5" is 3" down the cable from the 5.25 on the end... and otherwise strange situations... I would say I need to put them somewhere safe in case I need them for that, but you know how that goes, better just leave them with the rest of this "random shit that's super important when you haven't got it." stuff.

Edit: also happy because I "discerned the presence of" a dot matrix printer, not sure which one it is yet, was beginning to think they all got discarded/destroyed.... it's a kind of peek in the box coz there's not clearance to get it out without a couple hours heavy lifting kinda thing.

Hi everyone...

Today was cleaning and assemble "pentium 2 project", replace vga card from "pentium 1 project"

Small case
Pentium 233mmx, 64mb ram, Diamond stealth 3d 2000, diamond monster 3D (3dfx), SoundBlaster32 PnP (28mb ram), 3com 3c905 10/100, Philips CDD3610 (first cdrw drive 2x2x6x), Toshiba 24X , Win98 SE

Big case
Pentium 2 450, 128Mb ram, Matrox G200, SoundBlaster AWE64 Gold, 3Com 905 10/100, HP cdrw 9500i, maybe install creative infra 1800, Win98 SE