VOGONS


Bought these (retro) hardware today

Topic actions

Reply 59240 of 59257, by Shader_BiH

User metadata
Rank Newbie
Rank
Newbie
Dan386DX wrote on 2026-06-11, 04:08:
zuldan wrote on 2026-06-10, 19:56:
Shader_BiH wrote on 2026-06-10, 19:35:
Meanwhile I ordered a pair of those ATX to AT PSU connector converters... I've read somewhere that they are perfectly fine to us […]
Show full quote

Meanwhile I ordered a pair of those ATX to AT PSU connector converters... I've read somewhere that they are perfectly fine to use as long as you have strong 3.3 and 5v rails... I guess I'll find out 😁

s-l1600-13.webp

Those converters are great. I use them in all my 386 and 486’s. Connect those 2 leads directly to the 4 pin power switch of your AT case. Also much safer than having mains (120v/240v) going to the power switch.

I agree, I swear by them; I like to keep the hardware as period correct as possible but it's hard to trust a 35 year old PSU if it's not been recapped/refurbished lately.

The only problem with these, is that those four metal terminals behind the power switch can vary in size, I've had to modify the connectors on the adaptor to make them fit before.

Yes those can be tricky... but then again, that's a much easier DIY correction than hack-jobbing entire ATX to AT connection 🤣 I'm just happy people make these and you can get them brand new. Concerning old PSUs... every time I use them it feels like gambling on a slot machine, but regardless, PSUs from 95-2000 proved to be much higher quality than those from 2001-2005, at least in my experience. I had so many MS and Codegen PSUs blow up... and those like "brands" from early 2000s era, they must have got those bad capacitors as well. (and I'm not even considering inherent bad design here...) Recently I got a pentium mmx rig with very old Seasonic PSU and it works great... checked inside, no bad caps, at least visually.

Reply 59241 of 59257, by Fazeshift

User metadata
Rank Newbie
Rank
Newbie
Shader_BiH wrote on 2026-06-12, 06:20:

Yes those can be tricky... but then again, that's a much easier DIY correction than hack-jobbing entire ATX to AT connection 🤣 I'm just happy people make these and you can get them brand new. Concerning old PSUs... every time I use them it feels like gambling on a slot machine, but regardless, PSUs from 95-2000 proved to be much higher quality than those from 2001-2005, at least in my experience. I had so many MS and Codegen PSUs blow up... and those like "brands" from early 2000s era, they must have got those bad capacitors as well. (and I'm not even considering inherent bad design here...) Recently I got a pentium mmx rig with very old Seasonic PSU and it works great... checked inside, no bad caps, at least visually.

Anecdotally, I agree. The worst power supply cost-cutting and false specs seems to have happened after the 90's.

I opened some of the power supplies I have, and the some of the mid-00's are pretty bad in build quality and garbage caps that have obvious bulging and electrolyte leakage, even the ones that were supposedly a "premium" brand. Then I looked at a used Astec 200W from 1997 with Japanese Nichicon caps (good, visually), good overall build quality, and tests in-spec with minimal ripple.

Seasonic was top-tier, at least back then.

It has also been interesting reading this PS discussion:
Any sense buying modern PSU for old hardware?

I don't have anything meaningful to contribute there, but it does illustrate the challenge of replacement power supplies for the most 5V-hungry hardware from the pre-12V ATX era. Apparently the DC-DC converters can overheat with light 12V load but heavy 5V load. I'm looking at new PSU's in the 650-750W range, with 5V ratings of 18-20A. That 200W Astec has 18A rating on the 5V, and that isn't shared with the 3.3V rail. I'm considering doing a full re-cap of that (I have a few of them) for a retro build (probably PII 400MHz) instead of buying a modern PSU.

Reply 59242 of 59257, by BloodyCactus

User metadata
Rank Oldbie
Rank
Oldbie

Picked up this, A 3COM 3c982 rev b, basically a dual nic 3c905c /3c920 on a single card using an intel PCI-to-PCI bridge.

Ye9ANHr.jpeg

--/\-[ Stu : Bloody Cactus :: [ https://bloodycactus.com :: http://kråketær.com ]-/\--

Reply 59243 of 59257, by pan069

User metadata
Rank Oldbie
Rank
Oldbie

I have a couple of ET6000s already but grabbed this Hercules version a while back. I have some ram upgrade chips. Should I get sockets or blast them directly onto the pcb?

The attachment PXL_20260613_023447826.jpg is no longer available

Reply 59244 of 59257, by Shader_BiH

User metadata
Rank Newbie
Rank
Newbie
pan069 wrote on 2026-06-13, 05:37:

I have a couple of ET6000s already but grabbed this Hercules version a while back. I have some ram upgrade chips. Should I get sockets or blast them directly onto the pcb?

The attachment PXL_20260613_023447826.jpg is no longer available

I love these Tseng cards... I would go for the sockets if you have some, it will look much better in my opinion.

Reply 59245 of 59257, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
pan069 wrote on 2026-06-13, 05:37:

I have a couple of ET6000s already but grabbed this Hercules version a while back. I have some ram upgrade chips. Should I get sockets or blast them directly onto the pcb?

The attachment PXL_20260613_023447826.jpg is no longer available

This is probably just due to my lack of experience, but I find soldering surface mount sockets reliably much more difficult than just soldering chips. If you have no intention of removing them in the future, I'd just solder the chips and be done with it.

Now for some blitting from the back buffer.

Reply 59246 of 59257, by rasz_pl

User metadata
Rank l33t
Rank
l33t
Ozzuneoj wrote on 2026-06-13, 16:46:

This is probably just due to my lack of experience, but I find soldering surface mount sockets reliably much more difficult than just soldering chips. If you have no intention of removing them in the future, I'd just solder the chips and be done with it.

+ this

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 59247 of 59257, by pan069

User metadata
Rank Oldbie
Rank
Oldbie
Ozzuneoj wrote on 2026-06-13, 16:46:

This is probably just due to my lack of experience, but I find soldering surface mount sockets reliably much more difficult than just soldering chips. If you have no intention of removing them in the future, I'd just solder the chips and be done with it.

Interesting. What makes it in your opinion more difficult? I am not a super experienced solder-dude but I image that solder these types sockets, you approach from the top whereas soldering directly you approach from the side.

If I was to go for sockets, should I cut out the bottom plate or leave it? I have seen people cutting it out as it makes it easier to solder the sockets as the nearby plastic tends to melt.

I have a couple of these PLCC68P sockets:

The attachment s-l960.webp is no longer available

Technically I have two ET6000s that can use a memory upgrade (hence the four chips). I could try both techniques... 🤔

Reply 59248 of 59257, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
pan069 wrote on 2026-06-13, 21:33:
Interesting. What makes it in your opinion more difficult? I am not a super experienced solder-dude but I image that solder thes […]
Show full quote
Ozzuneoj wrote on 2026-06-13, 16:46:

This is probably just due to my lack of experience, but I find soldering surface mount sockets reliably much more difficult than just soldering chips. If you have no intention of removing them in the future, I'd just solder the chips and be done with it.

Interesting. What makes it in your opinion more difficult? I am not a super experienced solder-dude but I image that solder these types sockets, you approach from the top whereas soldering directly you approach from the side.

If I was to go for sockets, should I cut out the bottom plate or leave it? I have seen people cutting it out as it makes it easier to solder the sockets as the nearby plastic tends to melt.

I have a couple of these PLCC68P sockets:

The attachment s-l960.webp is no longer available

Technically I have two ET6000s that can use a memory upgrade (hence the four chips). I could try both techniques... 🤔

PLCC and SOJ chips are fairly simple to solder with a heat gun and solder paste (or some flux if you have to desolder). I haven't done a lot of them, but the ones I have done have been kind of spur-of-the-moment "I'm going to dive into this and see what happens" projects and they end up working without much fuss, and with no damage.

When dealing with sockets that accept PLCC or SOJ chips, I have found that they cannot handle the same amount of heat that I have used when working with the chips themselves. The plastic tends to get shiny if you hit them with a little too much heat, and will deform fairly easily if you keep going. Also, if you want to use a soldering iron you will need an extremely steady hand to avoid melting the plastic or soldering multiple pins together.

I'm sure if I had more experience I would develop a technique that works well for sockets, but using my current tools and skills they are things that I generally avoid desoldering\soldering unless absolutely necessary.

If you don't have a heat gun and don't have a ton of experience with a soldering iron, I wouldn't even attempt this. Just get a heat gun and some solder paste and practice on some other devices first. It's kind of amazing the first time you watch a part pull itself into place while the solder paste heats up. When I first started soldering tiny SMD caps under a microscope this way it made me laugh... it looks like little nano robots are doing the job for you. Sure beats stabbing the poor board with a hot iron that is 3 times the width of the component being soldered.

Just make sure to try to separate the solder paste on each pad before hitting it with heat and check very thoroughly for shorts between pins afterward.

Last edited by Ozzuneoj on 2026-06-14, 04:11. Edited 1 time in total.

Now for some blitting from the back buffer.

Reply 59249 of 59257, by rasz_pl

User metadata
Rank l33t
Rank
l33t
pan069 wrote on 2026-06-13, 21:33:

should I cut out the bottom plate or leave it? I have seen people cutting it out as it makes it easier to solder the sockets as the nearby plastic tends to melt.

Why do you ask whats wrong with sockets when you already know in the next sentence? 😀 Cutting internal frame makes socket non rigid and prone to bending out and losing contact. Its a lot of additional hassle in exchange for non deterministic outcome.

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 59250 of 59257, by dominusprog

User metadata
Rank Oldbie
Rank
Oldbie

Bought a new old stock case for the P!!! build. The build quality is not the best, and it didn't come with a fan holder for the front, but it's new, and I bought it for a very reasonable price.

The attachment IMG_20260615_202245.jpg is no longer available
The attachment IMG_20260615_202316.jpg is no longer available

Duke_2600.png
A-Trend ATC-1020 V1.1 ❇ Cyrix 6x86 150+ @ 120MHz ❇ 32MiB EDO RAM (8MiBx4) ❇ A-Trend S3 Trio64V2 2MiB
OPTi MAD16 / Creative Vibra16XV ❇ 8.4GiB Quantum Fireball ❇ Win95 OSR2 Plus!

Reply 59251 of 59257, by Shader_BiH

User metadata
Rank Newbie
Rank
Newbie
dominusprog wrote on Yesterday, 17:17:

Bought a new old stock case for the P!!! build. The build quality is not the best, and it didn't come with a fan holder for the front, but it's new, and I bought it for a very reasonable price.

The attachment IMG_20260615_202245.jpg is no longer available
The attachment IMG_20260615_202316.jpg is no longer available

Cool find... feels so weird to see these new 😀 What hardware did you have in mind for the build?

Reply 59252 of 59257, by dominusprog

User metadata
Rank Oldbie
Rank
Oldbie
Shader_BiH wrote on Yesterday, 17:36:
dominusprog wrote on Yesterday, 17:17:

Bought a new old stock case for the P!!! build. The build quality is not the best, and it didn't come with a fan holder for the front, but it's new, and I bought it for a very reasonable price.

The attachment IMG_20260615_202245.jpg is no longer available
The attachment IMG_20260615_202316.jpg is no longer available

Cool find... feels so weird to see these new 😀 What hardware did you have in mind for the build?

Thanks. I'll put a Gigabyte GA-BX2000+ with a Coppermine 800MHz (133*6), Geforce 2 MX-400 32MiB and 448 or 512MiB of RAM.

https://theretroweb.com/motherboards/s/gigabyte-ga-bx2000-2

Duke_2600.png
A-Trend ATC-1020 V1.1 ❇ Cyrix 6x86 150+ @ 120MHz ❇ 32MiB EDO RAM (8MiBx4) ❇ A-Trend S3 Trio64V2 2MiB
OPTi MAD16 / Creative Vibra16XV ❇ 8.4GiB Quantum Fireball ❇ Win95 OSR2 Plus!

Reply 59254 of 59257, by Ahrle

User metadata
Rank Newbie
Rank
Newbie
dominusprog wrote on 2026-06-11, 05:48:
Ahrle wrote on 2026-06-09, 20:49:
Project 357575 case arrived last Friday: […]
Show full quote

Project 357575 case arrived last Friday:

Quite the troubleshooting lesson!
Defective turbo LED panel (gave up, now solid 66), keylock on (no key), and nearly 20 hours spent on power issues (turned out as wrong facing IDE cable).
Finally finished yesterday.

ET4000 ISA was unfortunately sent up in smoke after direct contact with NE2000 during test process and network troubleshooting.

The attachment IMG_20260609_030945269.jpg is no longer available

Screwholes of m/b plate are misaligned, never seen this before.
Board has to be slightly slanted.
This makes the bottom ISA inaccessible, due to no room for a backplate.

The attachment IMG_20260609_020314367.jpg is no longer available

CT3900 and ET4000/W32 arrived as well.
~7mm between CT and PC speaker housing 😁

The attachment IMG_20260609_180355075.jpg is no longer available
The attachment IMG_20260609_182536491.jpg is no longer available
The attachment IMG_20260609_223557648.jpg is no longer available

The screw on the sound card is a nice touch. But why not use a VLB HDD controller too?

Thanks, will probably help staying upright a little longer 😁

For the controller:
- already had that one and already spent some $2300 on this project (ebay nowadays is no joke haha)
- ISA card will allow high mounting, minimizing harness
- mainly, I read somewhere VLB will clock down 10 MHz for each VLB card present (ie 50-40-33 MHz due to bus sharing)?

Project 357575: ECS U4914-G | i486DX2-66 | ET4000/W32 | CT3900
Main: IBM PC300PL | PIII 750 | Viper AGP + V2 SLi | CT4500

Reply 59255 of 59257, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++
Ahrle wrote on Today, 01:46:
Thanks, will probably help staying upright a little longer :D […]
Show full quote
dominusprog wrote on 2026-06-11, 05:48:
Ahrle wrote on 2026-06-09, 20:49:
Project 357575 case arrived last Friday: […]
Show full quote

Project 357575 case arrived last Friday:

Quite the troubleshooting lesson!
Defective turbo LED panel (gave up, now solid 66), keylock on (no key), and nearly 20 hours spent on power issues (turned out as wrong facing IDE cable).
Finally finished yesterday.

ET4000 ISA was unfortunately sent up in smoke after direct contact with NE2000 during test process and network troubleshooting.

The attachment IMG_20260609_030945269.jpg is no longer available

Screwholes of m/b plate are misaligned, never seen this before.
Board has to be slightly slanted.
This makes the bottom ISA inaccessible, due to no room for a backplate.

The attachment IMG_20260609_020314367.jpg is no longer available

CT3900 and ET4000/W32 arrived as well.
~7mm between CT and PC speaker housing 😁

The attachment IMG_20260609_180355075.jpg is no longer available
The attachment IMG_20260609_182536491.jpg is no longer available
The attachment IMG_20260609_223557648.jpg is no longer available

The screw on the sound card is a nice touch. But why not use a VLB HDD controller too?

Thanks, will probably help staying upright a little longer 😁

For the controller:
- already had that one and already spent some $2300 on this project (ebay nowadays is no joke haha)
- ISA card will allow high mounting, minimizing harness
- mainly, I read somewhere VLB will clock down 10 MHz for each VLB card present (ie 50-40-33 MHz due to bus sharing)?

Wow, that's a lot to spend without having a 133 Mhz Cyrix 5x86 upgrade a Mach 64 2MB VLB and a Voodoo 1 in it. (That's just an "easy" $1000 worth, not a recommendation) You must have had very specific hardware in mind.

VLB... two random cards have a chance that they will only play nice at 33. Random single cards, you might find only like 33 by themselves, where if they'll do 40 they are glitchy even with the wait states set. Back in the later 90s, when VLB were $1 a piece out of computer fair junk boxes, I was able to determine that by mixing and matching and cherrypicking, that you could indeed match VGA and I/O that would do 50 and even 60 together. It's not such a viable option now when they get to $150 a piece for tested/working cards. The most viable candidates though will do 40 by themselves with no wait states, tightest timings. Find a VGA and i/o that do that and there's a good chance they will pair at 50 with wait states set. In general the most willing VGAs I found for higher speeds had GD5428 or GD5429 VGA chipsets and 60ns SOJ RAM on. The i/o cards were typically UMC based. I think one pair I had that went well were a PT-429 VGA and a PT-426 i/o both from Pine, could still be a coin flip though. Also I am poor at model number recall for my ultimate 60mhz pair, might have been one of the Pine cards and one similar. Card models may be less relevant than "silicon lottery" of the chipsets, only tested a dozen or so, not the thousand you need to figure that out. 66Mhz may even have been possible with my golden pair, I could boot at 66, but being at the raggedy edge with the motherboard, CPU, RAM, cache etc does not give you much clue about what particular component is the weak link, and it would crash on almost everything in a short time. Motherboard also may be a factor, either in model or silicon lottery, so a perfect pair of VLB might not do 50, or even 40 on a "bad" motherboard.

Board to case fit. One of the standoff holes, think it was one of the ones in the middle, near ISA slot length distance from back, would have a somewhat random variance in position. Some cases might have a long slot for the hole, some might only support one position, but it was always a maybe it will, maybe it won't thing. Thus frequently necessary to use a standoff with the bottom nub cut off, so it just sat freely on the backplate. One of the corner screw to standoff holes was also not entirely standardised, many boards have an ovalled or double hole there. If a board doesn't and doesn't screw down straight, might be necessary to drill and tap a new standoff hole to suit. The two "most standard" holes seem to be the one on the back right, and the middle back, get those two aligned and then work on fudging, relocating, working around any that don't align after you've got those in. Be aware that sometimes the card slot frame is screwed on and can have minor variance in position, which may require loosening screw and tapping into place.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 59256 of 59257, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Found a previous personal "unicorn" yesterday, and let it run free...

It was a Genius (The mouse people, not the ones that own a big chunk of Florida, the ones that made computer mice) hand scanner with a parallel port connection. The parallel port connection is the big deal here. They were advertised but hard to find in the day, or when I saw one I didn't have the cash, and the type that requires an interface card is at least 10 times more common. I had wanted one from about 1993 to probably 2003. The parallel port made it machine agnostic, you could use it on different machines, and particularly, laptops, so you could go to the library, get the SAMs or whatever repair manual out from the non-loanable section and scan whatever schematic or tech info you needed. Also applicable to other pursuits of that type. Since 2003,
there were USB options, and since 2013, good enough cell phone cameras. (And earlier and alternate options, but usually more expensive or less common even.)

Thus I gazed upon it's magnificence for some minutes and decided to leave it there. I could have been more tempted if it was complete in box with it's software and manuals, but it was just a loose piece. Bit weird to see something you wanted for so long, but no longer have a use for. Possibly I may have bought it if I didn't already have one complete interface type hand scanner in box and some cards and another scanner loose, that all just "turned up" with other stuff through the years. So if I get the itch to hand scan just for novelty rather than practicality I can install that somewhere and have at it. ... and not care that I can't use it on different machines or take it to the library.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 59257 of 59257, by tehsiggi

User metadata
Rank Oldbie
Rank
Oldbie

Another Targa Visionary with an XP-M 2400 - this time the notebook is partly defective (cooler not working and shutting down due to over-temperature) - but I first and foremost need the CPU.

And an Athlon XP 2500 with date before the superlocking of CPUs.

Two new contestants for the Socket A fun.

AGP Card Real Power Consumption
AGP Power monitor - diagnostic hardware tool
Graphics card repair collection