Victim selection time...

Tridents??

There's actually some okay ones in there I am meaning to test. I'm eyeballing that Aamazing one though, if I ain't stealing it's RAM I wanna test RAM in it....

But miracle of miracles I actually found the weird 256kb SIMMs with DIP RAM on.... gotta look up some numbers..

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space to play with all my old and broken computers.

First up is troubleshooting an instability problem on my P2B-DS system. I built this one before I got serious about things, and I never checked its caps, so that was an easy place to start. One had a little bit of poopy on it, and after pulling it, it tested pretty dead. The rest of them looked ok visibly, but I pulled a couple others of the same value and they tested marginal, so I decided to change all (29!) of them.

Only other caps on the board are Rubycon branded and the one I pulled tested fine, so I left them. (They're also really tall - 20mm - and the tallest 1000uf caps I have are 16mm. I'm not smart enough to know if those would be ok. It might solve a slight clearance issue with the CPU1 fan though... 🤔)

So far so good - it posts, so I didn't break the thing. 🙂

Stability improvements yet to be seen. I was mucking about with a multi-boot setup, but I think I'm just going to all in with Win2K, in the name of period correctness. 😉

ubiq wrote on 2024-10-20, 23:37:

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space […]
Show full quote

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space to play with all my old and broken computers.

First up is troubleshooting an instability problem on my P2B-DS system. I built this one before I got serious about things, and I never checked its caps, so that was an easy place to start. One had a little bit of poopy on it, and after pulling it, it tested pretty dead. The rest of them looked ok visibly, but I pulled a couple others of the same value and they tested marginal, so I decided to change all (29!) of them.

Only other caps on the board are Rubycon branded and the one I pulled tested fine, so I left them. (They're also really tall - 20mm - and the tallest 1000uf caps I have are 16mm. I'm not smart enough to know if those would be ok. It might solve a slight clearance issue with the CPU1 fan though... 🤔)

The attachment IMG_1877.jpeg is no longer available

So far so good - it posts, so I didn't break the thing. 🙂

Stability improvements yet to be seen. I was mucking about with a multi-boot setup, but I think I'm just going to all in with Win2K, in the name of period correctness. 😉

Love that setup! What slotkets are you using, and are they modified to run dual cpu?

POD83 cannot work stably at 100Mhz under the default voltage, and the Octek HIPPO15 motherboard I used did not have the required 4V voltage, so I removed the LM912 on the CPU and installed a DC-DC module with 12V input and output set to 4.1V, which worked very stably at 100Mhz.

H3nrik V! wrote on 2024-10-21, 04:10:

ubiq wrote on 2024-10-20, 23:37:

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space […]
Show full quote

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space to play with all my old and broken computers.

First up is troubleshooting an instability problem on my P2B-DS system. I built this one before I got serious about things, and I never checked its caps, so that was an easy place to start. One had a little bit of poopy on it, and after pulling it, it tested pretty dead. The rest of them looked ok visibly, but I pulled a couple others of the same value and they tested marginal, so I decided to change all (29!) of them.

Only other caps on the board are Rubycon branded and the one I pulled tested fine, so I left them. (They're also really tall - 20mm - and the tallest 1000uf caps I have are 16mm. I'm not smart enough to know if those would be ok. It might solve a slight clearance issue with the CPU1 fan though... 🤔)

The attachment IMG_1877.jpeg is no longer available

So far so good - it posts, so I didn't break the thing. 🙂

Stability improvements yet to be seen. I was mucking about with a multi-boot setup, but I think I'm just going to all in with Win2K, in the name of period correctness. 😉

Love that setup! What slotkets are you using, and are they modified to run dual cpu?

Asked and answered last time I was posting about this build back in March! 😉 But yeah, they're coppermine-supporting slockets with one pin bodged on each to get the dual CPU working.

As for getting Win2K up and going... well it's stable, but it's also running dog sh*t slow for some reason. Not sure at this point if it's hardware or drivers - CMOS got wiped, so maybe a setting there... Gonna swap out the Voodoo3 as a first easy troubleshooting step before diving deep into bios settings.

Fixed the CF card to the HDD bracket and replaced the amber LED with a red one. The old LED was too dim in combination with the CF adapter.

Tested this MP-586ACT motherboard. RetroWeb lists it as a variant of the Soltek SL-54P2, though this has a different layout and switches than are listed on RetroWeb.

Took a few tries to get it to boot up, but seems to working and stable now. Caps seem visibly okay, but it might benefit from a recap all the same.

It also has a couple spaces for additional cache chips. I wonder if this board could be upgraded to 1MB of L2 cache and if there would be any benefit?

Thermalwrong wrote on 2024-10-20, 21:49:

I might make a thread on it, there are quite a few variations of 3com card that I think would benefit from documenting and possi […]
Show full quote

BitWrangler wrote on 2024-10-20, 19:07:

Thermalwrong wrote on 2024-10-20, 18:33:

Today following the advice of this old usenet post: https://groups.google.com/g/uk.comp.sys.lapto … /m/6K4Edf-N78QJ and some kno […]
Show full quote

Today following the advice of this old usenet post: https://groups.google.com/g/uk.comp.sys.lapto … /m/6K4Edf-N78QJ and some knowledge gained doing the same thing on a Dlink DE650 card, I've just put a permanently attached RJ45 lead onto a 3COM 3C562D. The end result was pretty nice, just trimmed away the plastic bit between the two dongle jacks and ran the cable through there. The 4 un-used wires are insulated and run on the back of the card for better mechanical stability, then the wires are carefully routed so the case can close.

The attachment re-dongle-3com-3c562d-1 (Custom).JPG is no longer available

The attachment re-dongle-3com-3c562d-2 (Custom).JPG is no longer available

It worked 1st try and got an IP with an old router as a DHCP server and can ping just fine 😀
Once that was confirmed then the cable was hot glued into place and the case re-closed with superglue.

Opening up these old PCMCIA cards, some of them were actually made for repair - if you can see a seam on the plastic at the corners then the case can be pried into two parts with alignment pegs etc so they can be re-glued later. The 3COM cards seem to be pretty good in that regard, I just opened up an even older 3C589C card and that's got essentially the top cover clipped onto the frame / lower cover piece, but there are places designed for jamming a pin into to pry open the metal clips without damaging anything. In fact this card appears to have been opened in the past and has some evidence of re-work on the PCMCIA connector.

Some of them are horrendous to repair though, the D-Link DE-650 PCMCIA card was welded shut and I had to grind all the welds off with the dremel. The dongle jack on that one was broken so it was completely removed and the cable runs through that spot now. Then it was spot-welded closed again as best as I could manage with my little powerbank spot-welder I got for rebuilding lithium packs.

Oh that's brilliant, I've got 3 or 4 hanging round with no dongles. I didn't really imagine they could be got into without destruction or damage that made them not go back in a PCMCIA hole ever again. I guess they are super trivial compared to todays phone repair, but seemed like they were made by elves or something in the 90s 🤣 Thanks very much for posting about it.

I might make a thread on it, there are quite a few variations of 3com card that I think would benefit from documenting and possibly I could go into more detail on how to open these things up.
Those colour code positions on the LAN magnetic do seem to be correct for the 3C562, 3C589C & 3C589E and they actually look about the same inside with the same pinout / shape transformer.

The attachment IMG_4713 (Custom).JPG is no longer available

Using the un-used strands held in place with hot-glue on the reverse side to anchor the cable in place is working really well, unless the hot glue melts in use of course 😁

The attachment re-dongle-cable-anchoring.JPG is no longer available

Figuring out how to wire up the first one I did was tricky, I spent ages trying to reverse engineer the pinout of this D-Link DE650 card which has options for both RJ-45 and BNC and wasn't getting anywhere, tracing it all out in paint.net. Then I looked up the part code on the LAN magnetic / transformer and it listed the exact pinout to use. Once I hooked the RX+ / RX- / TX+ / TX- to the pins specified on the Pulse PE-68030 datasheet it just worked.
If not for that it would've been simple trial and error as demonstrated here: http://labs.cexx.org/dongle.htm
And to give credit where it's due, that page gave me most of the initial info on this such as opening the card up and that soldering to the magnetic is easier than soldering to the pins. The RJ-45 sockets themselves came from some aliexpress specials, which had 2x rj45 sockets going to one rj45 plug, kind of a stupid cable since that's not going to work with switched ethernet, but it was cheap and the wires are actual copper.
These cheapo rj-45 sockets have an unexpected 'feature' too in that when you push a cable in, it can't lock into place. But that's actually kind of a good thing since it's like a magsafe cable and won't rip the PCMCIA card to pieces if the cable gets pulled 😀

Great info thanks. I guess if the sockets are too high a quality one could tape or elastic band down the latch on the hookup cable so it doesn't engage well. I think I might have two dongles with sockets on in the stash somewhere that don't belong to any of my cards, just ones I saw bagged up on the electrical cords wall in the thrift and I bought them crossing my fingers and they didn't fit anything I had. I believe I did make an effort to ID them some decade back because I know ppl want them, but no markings or distinctive features so might as well cannibalise them for the sockets.

Shponglefan wrote on 2024-10-21, 15:48:

Tested this MP-586ACT motherboard. RetroWeb lists it as a variant of the Soltek SL-54P2, though this has a different layout and […]
Show full quote

Tested this MP-586ACT motherboard. RetroWeb lists it as a variant of the Soltek SL-54P2, though this has a different layout and switches than are listed on RetroWeb.

Took a few tries to get it to boot up, but seems to working and stable now. Caps seem visibly okay, but it might benefit from a recap all the same.

It also has a couple spaces for additional cache chips. I wonder if this board could be upgraded to 1MB of L2 cache and if there would be any benefit?

The attachment MP-586ATC Testing 1.jpg is no longer available

The attachment MP-586ATC Testing 2.jpg is no longer available

That design looks closer to Gigabyte, with all that text going on between the SIMM slots and PCI. Closest of GB's boards would be the 586TX3.

ubiq wrote on 2024-10-21, 14:49:

H3nrik V! wrote on 2024-10-21, 04:10:

ubiq wrote on 2024-10-20, 23:37:

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space […]
Show full quote

Been away from retro stuffs for a bit due to a cross-country move. Finally got my life sorted enough to start setting up a space to play with all my old and broken computers.

First up is troubleshooting an instability problem on my P2B-DS system. I built this one before I got serious about things, and I never checked its caps, so that was an easy place to start. One had a little bit of poopy on it, and after pulling it, it tested pretty dead. The rest of them looked ok visibly, but I pulled a couple others of the same value and they tested marginal, so I decided to change all (29!) of them.

Only other caps on the board are Rubycon branded and the one I pulled tested fine, so I left them. (They're also really tall - 20mm - and the tallest 1000uf caps I have are 16mm. I'm not smart enough to know if those would be ok. It might solve a slight clearance issue with the CPU1 fan though... 🤔)

The attachment IMG_1877.jpeg is no longer available

So far so good - it posts, so I didn't break the thing. 🙂

Stability improvements yet to be seen. I was mucking about with a multi-boot setup, but I think I'm just going to all in with Win2K, in the name of period correctness. 😉

Love that setup! What slotkets are you using, and are they modified to run dual cpu?

Asked and answered last time I was posting about this build back in March! 😉 But yeah, they're coppermine-supporting slockets with one pin bodged on each to get the dual CPU working.

As for getting Win2K up and going... well it's stable, but it's also running dog sh*t slow for some reason. Not sure at this point if it's hardware or drivers - CMOS got wiped, so maybe a setting there... Gonna swap out the Voodoo3 as a first easy troubleshooting step before diving deep into bios settings.

Damn I need to get more sleep ... Or maybe I'm just old 🤣

PcBytes wrote on 2024-10-21, 17:11:

That design looks closer to Gigabyte, with all that text going on between the SIMM slots and PCI. Closest of GB's boards would be the 586TX3.

It does look kinda similar to Gigabyte boards. I'm half-tempted to peal off the model number stickers to see what might be underneath. 😅

Monday chores seemed to expand to 3x the usual, so didn't get long this evening. Spent my retro time assembling sundry support "stuff" that I might be needing in tackling the parity error on the PB800 board. If you've got any clues as to narrowing it down, I'm accepting your 2c here ... NEAT chipset memory problem on PB800/900 rev D 286 board....

First up, just grabbing a fresh set of picks for poking, prying and scraping, some turned pin socket headers because I haven't found any sockets yet, a 74LS280 I found in the logic bin, parity generator doohickey, I haven't found a discrete one onboard yet, but if there's a chance teh parity circuits are pooched, then I miiiight be able to jury rig something with that deadbugged to it. To the right, SIMMs I am gonna be grabbing the chips off, they are a match I think, though I notice they have had the backside nubs shaved so might not have as much pin on them as is usual, but they're standing off the board a bit so can't be too bad. The PSU hah, yeah, that one lost a fight with a can of tar based sealant, it was stored on the same surface, the can leaked, it got covered... That was about 15 years ago, wiped the worst off then, but it took a decade for it to dry to merely a bit tacky in the thicker parts. I guess I could have dunked it in kerosene or something, but at least it's more rustproof 🤣 Then of course we need a proper motherboard tray for the test bench, so there's a slab of wood to nail it to.

Boards, the 286 at the back is the Concorde boarde, mainly I got it out to threaten the PB800 board that it could be replaced and make it behave, and to maybe borrow it's 286 if required. Then a FIC 486-VC the big one, for main tester of SIMM and maybe other boards as required, maybe will have a trident testing individual, well pairs, of RAMs. Maybe a Trident will end up with the bad one, if it's a single pixel above 768k I'll see it basically never, okay yah, I'll pick one of my slow tridents for a high res windows card, sure. Anyhoo also got out a random ALI chipset board to make the FIC behave... or backup... Might need an ISA machine to run the EPROM programmer before we are done here too so one of those boards will serve.

The FIC fired up first try, just to POST screen, was about "done" for tonight. Oddly it complained Dallas battery was "low" but also seemed to be complaining about changed settings from when I last had it powered a decade back, 🤣 I guess that Dallas from 1993 still has a wee bit of juice in it, prolly not enough to keep time.

edit: ah okay, couldn't find that other board under Acer because it was an Asus ... https://theretroweb.com/motherboards/s/acer-i … gxi-model-i433a

So I have been playing around with my recent acquisition after full recap - MSI MS 6163 Master Pro v2
First I did put officially supported Slot1 Coppermine 933 PIII in and run it trough my regular stress tests - no problems whatsoever. Rock solid at 133FSB speed.

Then I decided to be more bold and pair 1100A Tueleron (SL5VQ ) with Soltek SL02-A++. First rom.by BIOS mod and then usual PIN AN3 AJ3 AK4 isolation + manually set jumpers for 1.5V VCORE on Slotket. Got hint from this page: https://geocities.ws/_lunchbox/tualeron_success_table.html
First boot nothing. Then set also FSB speed manually and it booted...
Run trough all my DOS tests, but with windows I couldn't get it stable. Might be something to do with my 5 years old used and countless time cloned Win98 installation. Couldn't boot to windows with AGP card. Then replaced with PCI Matrox G450. It is fine in desktop mode, but with Sisoft Sandra stress test it keeps rebooting after few iterations.

Will try to run memtest tomorrow. Maybe do Win98 clean install...

For everyday retro stuff, I hook my beloved Iiyama Vision Master Pro to my rigs, although it suffers from flyback issues. Besides another 15" Highscreen monitor, there are two tubes more in storage. Today it was time to check if there was still any sign of life in it.

First up, a Philips Pro 7BM749 monochrome monitor with a d-sub 15 connector. I'm not stating it has a VGA connector, because there are only about half the pins in the connector. Haven't count them.
This monitor's image is really pleasant to see, almost like an e-reader, that calm to ones eyes.
This unit is small, lightweight and the surface of the screen is almost flat.

Next up, a Volleman, probably some rebranding is going on here, for the Dutch or Belgian market I guess. Is it EGA ? It has a 9 pin d-sub connector. And the switch / filter at the back of this thing is intriguing. Vogon user Waterbeesje messaged me about the different modes. Green, amber, whitepaper and normal.
It's heavy, large but still sharp after 36 years.

Sorry about the amateur collage thing.

Just the other day my retro activity, was booting up my Windows 98 machine again, and not realising it had lost its Bios settings.

it was boot looping, and despite all my efforts I could not get it to boot. ended up formatting and reinstalling windows, still the same issue.

Turns out I had to set the HDD setting back to work with a SATA and disable the primary IDE. Gives me an opportunity to try some other patches and such that I hadnt before (including 98 2 ME) 😁

Seems the FIC 486-VC is stupid and I hate it... Not quite so much life left in that Dallas, it is acting super flaky on all forms of floppy activity though and have tried several drives and controllers in it. Either it all comes nice again when the dallas is properly modded or it needs as much work as the PB800, but in different areas. I don't remember it being like this, I had linux going on it this century. Strangely, memtest86 ran great and reported no errors 😒 .... anyway the whole deal was seeming like it was gonna rely on a stack of workarounds as shaky as a game of Jenga about 20 moves in... so for this purpose I am done with that board.... all this newfangled AT crap giving me the runaround, makes me wanna stick the XT back on the bench and play stupid CGA stuff on it for a week. Tomorrow I start over with the Asus.

Good news though, maybe the EGA monitor is not quite as tired internally as I was fearing, the front brightness control was way dirtier/glitchier than I realised, and the drastic loss of brightness few times may have been attributable to it. Anyway, seemed to improve with some exercising of it, maybe the wiper was a bit oxidised. Still will be under observation, yah alright smartasses, monitors are typically under observation, I mean a degree or two more than that.

Note to self though, try FIC with a different CPU next time, that one (DX33) was getting close to ouchy hot just in CMOS setup.

Had a look at why my K7S5A locked up during gameplay. Didn't fix the issue at 133/133, but it's stable at 100/100 FSB/CPU base clock, and that's probably good enough for now (and seems stable for prolonged periods of gameplay). As a side effect of playing around with different settings and trying to see if it's a RAM issue, I read somewhere online that the board doesn't POST at the latest BIOS with DDR RAM, so found and flashed one version earlier than the latest one, and now rocking DDR RAM on that board, so that's good.

ABIT AB-PB4 recovery day)))

Nice, another ABIT taken care of.
Speaking of them... added a fan on my KG7-RAID's heatsink. Gotta keep that 761 chipset in tip-top shape!

PD2JK wrote on 2024-10-22, 21:18:

Next up, a Volleman, probably some rebranding is going on here, for the Dutch or Belgian market I guess. Is it EGA ? It has a 9 […]
Show full quote

Next up, a Volleman, probably some rebranding is going on here, for the Dutch or Belgian market I guess. Is it EGA ? It has a 9 pin d-sub connector. And the switch / filter at the back of this thing is intriguing. Vogon user Waterbeesje messaged me about the different modes. Green, amber, whitepaper and normal.
It's heavy, large but still sharp after 36 years.

The attachment 20241022_231225-COLLAGE.jpg is no longer available

Sorry about the amateur collage thing.

That's neat that it has all those different modes. Kinda wish more monitors back then had features like that.