Over the years, and the help of some you, I developed a sequence of check before powering up "unknown state" boards/circuits.

I bought a PCChips 741LMRT board as is "untested / unknown working order" for close to nothing.
1. shorts or suspicious low resistance
2. visual inspection
3. power on free of cards, cpu
4. power on
5. power on with cpu and diag card
6. add stuff
-----
The board came with a Celeron 500.

Board was ok until power on with cpu: the -- -- was on 😀
I tried to swap the cpu, add ram + buzzer: board would issue a cycle of BIOS codes starting from 00 to 9F (the last possible value, maybe this one is wrong)

Suspects: cpu voltage, BIOS with bit rot.
And copious amounts of dirt. Cleaned thoroughly.

Replaced the BIOS with one from another board, C1 appeared. Reflashed old BIOS rom and tested again, POST.
C-500 wasn't working. Tested on another board: POST.
Why??? 🤣

After letting the 741 warm it'll POST with the C-500, caps need a replacement session.

I tested 2 BIOS roms, AMI and Award. Can't tell which is better.

I got a Windows 98 VM running under Linux using QEMU and virt-manager. Does that count as modern or retro?

What wasted the most of my time was trying to figure out why the audio and network adapters weren't detected and the DMA option didn't exist for the hard drive. I eventually figured out that Windows 98 doesn't like the way QEMU handles Plug and Play, so it didn't install the base chipset drivers. After I manually installed the "PCI bus" driver through the Add New Hardware Wizard, everything worked.

Then I tried installing SoftGPU, and I learned that it breaks the Windows hardware detection routines. If you run the Add New Hardware wizard, refresh Device Manager, or run certain driver installers, it will crash by switching to text mode with a blinking cursor in the corner and alternating black and green vertical bars. The "solution" is to make SoftGPU the last driver you install.

I spent a little bit of time looking for why Windows setup crashed (Patcher9x fixed that) and trying different audio devices (including options that QEMU supports, but virt-manager doesn't present in its drop-down menu). The SB16 option got Windows to install a driver, but it didn't make any sound. The ES1370 option might have worked, but I had already installed SoftGPU at that point, so Windows crashed during driver installation (which is already a better experience than I've ever had with a real AudioPCI card). What finally worked was AC97 using some random Realtek VxD driver (manually installed because the installer detects that it is running on Windows 98 and refuses to install, and the WDM driver crashes).

Treated my VIA Apollo Pro 133T/Tul1.4 system with proper SSD(with IDE->SATA adapter). I have few 120GB Kingston SSD laying around - perfect size for Win98SE
I have been using for years SD adapter. No complaints actually, however they have controller limit around 30MB/s. So performance difference is clear. Also Win seems snappier or maybe I'm imaging it.
Used Norton Ghost 5.1d in DOS for cloning. It was painfully slow, but hassle free.

Today I got my 1999 Guillemot Digital gamepad running in W98 by installing the Guillemot Dual Analog gamepad drivers. Same thing minus two thumb sticks.

This gamepad is NOT recognized by W98 using default gamepad drivers.

vutt wrote on 2025-06-01, 19:34:

Treated my VIA Apollo Pro 133T/Tul1.4 system with proper SSD(with IDE->SATA adapter). I have few 120GB Kingston SSD laying around - perfect size for Win98SE
I have been using for years SD adapter. No complaints actually, however they have controller limit around 30MB/s. So performance difference is clear. Also Win seems snappier or maybe I'm imaging it.
Used Norton Ghost 5.1d in DOS for cloning. It was painfully slow, but hassle free.

Very nice. I expect that window would feel faster if you can double your transfer rate while reducing latency.

The speed limit on the SD-IDE device you are running into is the SDHC “high speed bus mode”, which is the fastest SD access mode available to the FC1307 bridge.

https://en.m.wikipedia.org/wiki/SD_card

The Secure Digital High Capacity (SDHC) format, announced in January 2006 and defined in version 2.0 of the SD specification … introduces a high-speed bus mode for both SDSC and SDHC cards, which doubles the original Standard Speed clock to produce 25 MB/s.

Since the proprietary clamps/clips were broken, I tried to tie-wrap the heatsinkfan to the heatspreader. Looks easy, wasn't.

Yeah, you have to use 8.
Basically you poke 4 through like"bolts" and use the other 4 as "nuts"
When satisfied, trim.

PD2JK wrote on 2025-06-02, 11:33:

The attachment DSC_4417.JPG is no longer available

Since the proprietary clamps/clips were broken, I tried to tie-wrap the heatsinkfan to the heatspreader. Looks easy, wasn't.

No holes on the HSF to use case screws? That's how I basically did on older Pentium 2s and at least a 750MHz T-Bird I sent to Robert B.

Indeed no holes in the heatsink. I could drill some of course. A downside could be that the plastic tie-wraps loose theire flexibility and will snap over a decade. We'll see.

Nexxen wrote on 2025-06-01, 11:01:

Over the years, and the help of some you, I developed a sequence of check before powering up "unknown state" boards/circuits. […]
Show full quote

Over the years, and the help of some you, I developed a sequence of check before powering up "unknown state" boards/circuits.

I bought a PCChips 741LMRT board as is "untested / unknown working order" for close to nothing.
1. shorts or suspicious low resistance
2. visual inspection
3. power on free of cards, cpu
4. power on
5. power on with cpu and diag card
6. add stuff
-----
The board came with a Celeron 500.

Board was ok until power on with cpu: the -- -- was on 😀
I tried to swap the cpu, add ram + buzzer: board would issue a cycle of BIOS codes starting from 00 to 9F (the last possible value, maybe this one is wrong)

Suspects: cpu voltage, BIOS with bit rot.
And copious amounts of dirt. Cleaned thoroughly.

Replaced the BIOS with one from another board, C1 appeared. Reflashed old BIOS rom and tested again, POST.
C-500 wasn't working. Tested on another board: POST.
Why??? 🤣

After letting the 741 warm it'll POST with the C-500, caps need a replacement session.

I tested 2 BIOS roms, AMI and Award. Can't tell which is better.

Nice, I was initially confused as to whether the board was working but that's a quick repair 😀 So the board won't start from cold with the Celeron 500 fitted, how bad are those capacitors?

I bought another broken laptop! This one was an Olivetti Echos P120E which I have its cheaper relative the P100E and I'm quite fond of it - this one was shown as broken in the listing and has a completely destroyed hinge plastic on one side:

Quite a while back I bought an Acer laptop with a similarly busted hinge and 3d scanned it and made a big faff of it with too much complexity for a small plastic part: Re: What retro activity did you get up to today?
This time since it's a basic shape of a cylinder with some screw mounts and other minor features, this time I made a design in fusion 360 using calipers and a radius measuring tool:

The new part is printed in clear PETG, the original screw threads were liberated from the broken plastic and melted into this new replacement. This could only really work with the right hinge which doesn't have too much cabling in it - the 3d print has much more material in the empty space where there's room to give it some strength, it's got to accomodate the backlight cable so there's a cable trench modeled in. Overall I'd say this design took a couple of hours to design and an hour to print.
Something to note about this 3d print is that it's printed left-to-right so that the layer lines are in the rotation axis of the hinge, this is so that layer separation can't be a cause of plastic failure. If it was printed in the other orientation i.e. from the bottom to the top then potentially the layers of plastic around the screws could become a weakspot, but because it was printed vertically with supports, the stress and forces from the hinge are spread across all the layers.
The print is also set up to use enough perimeters that there is no infill in this part, it's almost completely solid and is much stronger because of that.

Then cheap epoxy was spread over the contact area of the two parts and they were squeezed together, somehow this white epoxy cures very quickly and gets very hot. But it works well and works better than the clear epoxy I was using before.

Now it's sanded back a bit, sadly the epoxy cured too quickly to be able to smooth the area properly but it's super strong. With a little sanding and some fresh paint on top it looks pretty good:

Once I could actually use the laptop, I then discovered that the mainboard seems to be broken - it can start up but like my other olivetti is just showing a blinking cursor and never does anything after. I think the DS12885S RTC chip is faulty since it's got power and the data signals are good, but it's not running the 32.768khz clock crystal. Now to get a replacement for that to see if this can be got up and running after I spent all this time fixing the hinges
(the other hinge was also just about to break but this was fixed with epoxy reinforcement on the inside and melted-in staples on the outside).

A question.
If I cut the ISA slots all together from a motherboard (destroyed, completely unrecoverable), I file the edges to avoid contacts and insulate them, and check that there are no shorts or cross-connects of any kind.
The idea is, can I use this as a some sort of "backplane," adding what's necessary for the power supply? Let's say I make a card I can insert in one slot to deliver the necessary voltages . Or a separated circuit , connected to the right pins.
The same for a raiser card.

Thermalwrong wrote on 2025-06-03, 01:14:

Nice, I was initially confused as to whether the board was working but that's a quick repair 😀 So the board won't start from cold with the Celeron 500 fitted, how bad are those capacitors?

Wasn't my best english either 😀
The caps look pretty, it's going to be a full recap of the cpu voltage first. I'm pretty sure it's those, but as I'm at it I'll replace them all.
-----------

Thermalwrong wrote on 2025-06-03, 01:14:

I bought another broken laptop!

Nice job, really nice job! Patience goes a long way.
Congrats!

Spending hours looking for the right PSU for my pending retro build with an XP 3000+, on an MSI KT6V, with GF4 TI 4200/4400 up to a 6800 Ultra(if it still lives inside my old Dell P4 Northbridge system) but for now my old PCs EVGA 750 G2 is getting it running(24A on 3.3v and 5v and its current Sempron 2400+ and an AGP + molex powered 7300 GT isn't too thirsty for amps as far as I am aware).

Sadler2010 wrote on 2025-06-03, 06:27:

Spending hours looking for the right PSU for my pending retro build with an XP 3000+, on an MSI KT6V, with GF4 TI 4200/4400 up to a 6800 Ultra(if it still lives inside my old Dell P4 Northbridge system) but for now my old PCs EVGA 750 G2 is getting it running(24A on 3.3v and 5v and its current Sempron 2400+ and an AGP + molex powered 7300 GT isn't too thirsty for amps as far as I am aware).

That board has the supplemental 12v P4 power connector on the motherboard along with a 12v VRM so it doesn't need a high AMP 5v rail PSU unlike earlier 462 boards that don't have either. So any good modern ATX PSU that has the P4 4 pin power connector would work just fine for it.

Earlier 462 boards got stuck with a 5v VRM and no supplemental power and needed the extra AMPs to power everything else along with the motherboard, later boards 462 like yours are no different than modern ATX boards that run off the 12v rail.

With my two 486 systems in hardware-complete form and with full OS installs, I put off the rest of the software configuration to restore the PCChips M590 I used to build my parents a computer in the late 1990s. Including figuring out the PS/2 and USB wiring since I don't have the ATX form card. One hiccup was that I was planning to put in a PCI TNT card instead of the onboard SIS video, but neither card would POST properly despite being fine in my Athlon so for the time being I'll just set it up for 2D DOS/Win98 gaming. It also can't recognize the 64GB SATA drives I have, but 32GB SD is more than enough for what it can run well.

Fortunately I was able to find the thread here explaining that the cards I have both require 3.3V on the PCI bus despite being keyed as universal, and the board doesn't deliver it. The posted fix looks pretty easy once I get the parts, so I'll add it to future modding plans.

Did some benchmarking with an Asus K7M mainboard, something is wrong.

SuperPi 128k
===================================
Athlon 700 Pluto: 94 seconds
Athlon 700 Thunderbird: 104 seconds
Athlon 800 Thunderbird (goldfinger): 100 seconds

Temperatures are okay.

The hinges on my Compaq LTE Lite 25E were starting to crack in a few places and it was time to fix them properly, because if left then it's just going to get worse and maybe break further up. The hinge design on the LTE Lite laptops isn't great, I think it just puts a lot of stress in too small of an area.

They were broken kinda when I got this laptop and I just used a bit of superglue and some low-temp mouldable plastic to remake the screw mounts on the right side but that didn't have enough rigidity and was flexing the plastic to breaking point each time I opened the screen up.

Cleaned that mouldable plastic out, used the soldering iron to melt a bunch of keying features into the good plastic in the area around those right hinge screw mounts. Marked the area out with blutack and then started putting in epoxy resin, this time it's very quick curing white stuff which seems to be strong enough for this job.
Since it takes some time to get the full hardness, I could clean up the edges with a knife to get it looking pretty good:

Melted in some keying features for the left one as well, where the lower screw mount just broke apart like it was made of chalk, so epoxy is doing all the screw thread holding instead:

This is the same laptop that I custom wired a direct drive floppy into: Re: What retro activity did you get up to today?
I got the PCBs made and it works so much better than the enamel-wire prototype, but I found that after I put a 720KB disk in, then put a 1.44MB disk in that the drive was broken. Somehow I missed that in my other testing, looks like just disconnecting the HD_MEDIA signal allows things to work and not break. It's a shame I got something like 20 of those PCBs made and they mostly fit and work but they need traces cut and the design could be better.
But it works really well and software installation using the floppy drive is now nice and reliable on my Compaq LTE Lite laptops.

I've now also got a Compaq LTE Lite 4/25C with the glorious TFT colour screen where somehow the mainboard has only superficial battery damage. I think the previous owner looked after this one but it had a lot of hours of usage and the display's capacitors gave out and started leaking, then it was given to a recycler where they removed the hard drive (and the hdd cable, ergh) and sold it to me. Thankfully I had a spare cable from another parts machine and the BIOS which is famously picky with its hard drives, was willing to work with a 200MB Seagate I had spare.

The screen was the bad part on this one, needed recapping fully and electrolyte cleaned up:

And on the other side there was electrolyte which had got under chips and possibly under the chip on film. Lots of original flux on the PCB as well but that wasn't doing any harm. Amazingly this all cleaned up nicely with some flux and fresh solder:

I don't have all the correct capacitors for this LQ9D011 unfortunately but I wanted to test if it's working first. The 100uF cap is replacing a pair of 16v 47uF capacitors that were connected together and that should be okay unless it really needs the lower ESR, I think they're bulk capacitance for the main chip or the TFT in general.
The other two 16v 47uF caps are replaced with 16v 56uF because that's what I've got and it's kind of in tolerance, I think they're important for the LCD voltage generation rather than just bulk capacitance.
But the worst ones are the 25v 10uF that I replaced with 35v 22uF, they're too big but they work. Might try harder to find 10uF capacitors since I'm pretty sure I have them, this is not the first LQ9D011 that I've had to recap with the other being in my Toshiba T4400SXC

All that time fixing and the LCD has a big horizontal line across the screen where two rows of pixels are bad, 640 x 478 is still a good display.