I took apart my Gateway Solo 2500 to fix the backlight. The palmrest has to be removed to access the inverter, and it's sandwiched between the base chassis and the hinges. To remove the palmrest, there are screws under the battery, floppy drive, and touchpad. Then you need to remove a metal cover over the inverter/LCD board (inverter, status LCD, main display connector, and power button), which is wedged under the CPU heatsink and card slot daughterboard. Next, you need to remove the entire board to remove the plastic insulator over the inverter, and the two screws near the middle are slightly too large for the holes in the PCB. After all of that, you need to reinstall all of the boards that you just removed (without the metal panels) to be able to power it up and start probing.

After all of that, it started working again. I could see that there was no switching activity on any of the transistors (PQ1 was the easiest to probe), and then I suddenly noticed that the LCD was lit up. Now I'm tracing the enable circuit, which appears to simply switch Vcc to the TL494 through some transistors (a PEMD12 dual PNP/NPN transistor at PU1) driven by an AND gate (next to the K7 marking, but it's probably U5 - the U is obscured by a via). One input to the AND gate goes to the video/card board, so I might also find out why the Xaos demo turns the backlight off until I reboot. The other input goes to a trace on an inner layer, but I eventually traced it to the second cable that goes to the motherboard. I can see that the input going to the video board is slightly below the high-level input voltage for the AND gate, but that's it.

I didn't find a schematic online (it appears to be a Quanta SQ2), and it would probably only be for the motherboard anyway. I would need schematics for three boards.

Kahenraz wrote on 2024-11-16, 19:41:

Split the output into a TV to see if this dropping out at the output (VCR) or at the capture device. There's no way to know for sure without testing it.

Sorry, I should have mentioned it in the first post, that it's definitely the VCR, because the same behavior was observed on the TV.

Yes, if I get my hands on another VCR, I will try it for sure, but they don't seem to be common in these parts anymore.

Worked on trying to repair this GeForce FX 5200 PCI video card.

The card had a couple bad electrolytic caps, so I opted to recap the entire card.

Unfortunately the card is producing visual artifacts. Not sure what this could be, possibly memory, a problem with the GPU chip, or some other component.

Shponglefan wrote on 2024-11-16, 20:36:

Worked on trying to repair this GeForce FX 5200 PCI video card.

The card had a couple bad electrolytic caps, so I opted to recap the entire card.

Unfortunately the card is producing visual artifacts. Not sure what this could be, possibly memory, a problem with the GPU chip, or some other component.

Crud ! I have a similar looking FX 6200 with bad caps and was thinking replacing but maybe these low end cards burn up things when the caps go bad ?

Although unlikely the cause of the artifacting, a green cap typically indicates that it's low-ESR. I don't know what happens when the wrong kind of cap is used in this scenario.

Shponglefan wrote on 2024-11-16, 20:36:

Worked on trying to repair this GeForce FX 5200 PCI video card. […]
Show full quote

Worked on trying to repair this GeForce FX 5200 PCI video card.

The card had a couple bad electrolytic caps, so I opted to recap the entire card.

Unfortunately the card is producing visual artifacts. Not sure what this could be, possibly memory, a problem with the GPU chip, or some other component.

The attachment GeForce FX 5200 before.jpg is no longer available

The attachment GeForce FX 5200 cap removal.jpg is no longer available

The attachment GeForce FX 5200 after recap.jpg is no longer available

The attachment GeForce FX 5200 DOS artifacts.jpg is no longer available

The attachment GeForce FX 5200 VGA artifacts.jpg is no longer available

I saw that recently when I started 3Dbench with the system in low speed turbo setting. 486 board, Trident 8900D ISA. So ensure it's not just a 3Dbench problem if you have some unusual clock setting, cache disabled or something.

Kahenraz wrote on 2024-11-16, 22:23:

Although unlikely the cause of the artifacting, a green cap typically indicates that it's low-ESR. I don't know what happens when the wrong kind of cap is used in this scenario.

Aside from the two bad caps, the other four green caps (all 1000uF) measured 0.04. Whereas the caps I was using to replace them measured 0.08.

So the ESR of the replacement caps is higher, but is it too high?

BitWrangler wrote on 2024-11-16, 22:49:

I saw that recently when I started 3Dbench with the system in low speed turbo setting. 486 board, Trident 8900D ISA. So ensure it's not just a 3Dbench problem if you have some unusual clock setting, cache disabled or something.

It's not a 3D Bench issue. The card is producing artifacts in everything.

I go the Gateway Solo 2500 reassembled enough to boot DOS. I thought that both Xaos demos turned the backlight off, but it's actually only Xaos 2. It's actually the enable signal from the motherboard that goes low. I also found out that the NeoMagic graphics chip has a backlight enable output (pin 108 on this MN2160), but it isn't connected to that AND gate. I don't know what drives that signal, but I'll have to wait until I lose the backlight again to do any more troubleshooting. There is a thread on Badcaps from 10 years ago with the same problem, and it showed the AND gate having a low output with both inputs high (one at 5V and the other at 3.3V). Maybe replacing it with a 74AHCT1G08 will help. Quanta has a habit of bad practices, so I'm not surprised that they're driving 5V CMOS logic from 3.3V logic. They love driving the gates of 2N7002s from 3.3V logic, which is below the worst-case maximum gate threshold voltage.

Kahenraz wrote on 2024-11-16, 22:23:

Although unlikely the cause of the artifacting, a green cap typically indicates that it's low-ESR. I don't know what happens when the wrong kind of cap is used in this scenario.

The new caps appear to be Panasonic FR series, and the original caps are probably Sanyo WG or WF series. Sanyo caps seem to be failing a lot now (the WF series failed faster than the rest). With the wrong caps, it will probably work fine. I don't think new caps would have a high enough ESR to cause stability problems, but they won't last as long (unless the old caps were total garbage like GSC or Licon/Fujicon - then even cheap 85°C audio caps will last longer than the originals).

Yup, I can confirm the replacement 1000 uF caps are Panasonic FR series.

Shponglefan wrote on 2024-11-16, 22:53:

BitWrangler wrote on 2024-11-16, 22:49:

I saw that recently when I started 3Dbench with the system in low speed turbo setting. 486 board, Trident 8900D ISA. So ensure it's not just a 3Dbench problem if you have some unusual clock setting, cache disabled or something.

It's not a 3D Bench issue. The card is producing artifacts in everything.

I'd check the regulators. That brown stuff on them isn't normal.

I don't see any damage to any of the tiny ceramic capacitors on the top of the card. Have you inspected the bottom carefully? It's not uncommon for these to get damaged if the card was previously stored in a bin with other cards.

Unfortunately the card now appears to be DOA.

I went to do some more testing and it's not even being detected. I tried on a couple different motherboards, same result. I even tried applying some gentle heat via a hot air gun just in case the issue was thermal contraction/expansion related. But no dice, the card is not booting at all.

I'm guessing this card was on its last legs to begin with...

PcBytes wrote on 2024-11-16, 23:41:

I'd check the regulators. That brown stuff on them isn't normal.

It looks like the pads had oxidized. At any rate, it's a bit moot at this point.

Friday and yesterday my retro activities were pretty much recapping low-esr capacitors of Epox 7kxa Slot A board. Couple of them were already bulged next to the CPU and they were Tayeh LE crap, another alias for Evercon/GSC/Sacon. Oddly, I measured all the caps I removed and most of those where pretty much 100% in spec, except one of the smaller 100uf LE series caps somewhere on the board, which was completely dead but still looked fine. I was almost amazed how reliably they held on this board not including the three dead ones.

Today I flashed the bios to the newest version with a programmer, threw the board with 750MHz Pluto and V3 3000 on a bench and tested if this puppy still runs. And yes it does. I haven't attached drives and such, I just let it sit some time in BIOS menu and see that it at least works fine there and doesn't start behaving oddly when it warms up and plan is to attach drives and start installing stuff later today. So far everything looks good. According to bios health monitor 3.3V is a tad low at 3.16V, but I'm not that worried about it as it is still within spec and those measurements aren't reliable. PSU delivers 32A on 3.3V rail, so it is definitely adequate.

New 1000uf caps were couple of mms taller than the original, so pretty much Zero clearance with V3 3000 heatsink. I knew it won't be a problem as new caps are pretty much exactly the height of the AGP slot, but this is certainly a thing that needs to be taken into account when choosing the caps.

gerry wrote on 2024-11-14, 11:48:

kinetix wrote on 2024-11-14, 02:17:

a couple of days ago I finished 90% of the restoration of a very damaged 386 board I salvaged from destruction a right a year ag […]
Show full quote

a couple of days ago I finished 90% of the restoration of a very damaged 386 board I salvaged from destruction a right a year ago.
some help to repair/restore a leaking battery ravaged motherboard
today I use it to test one more 30 pin memory module I found. It was OK.
A couple of days ago, in the process of testing some modules, I tough I had broke the MB again, but thanks Zeus it was not, but a bad module mixed with almost identical good ones.

that must be satisfying, to bring it back to life over time like that 😀

yes, it is

Took my Soyo 5EHM v1.2 (1MB cache version) out for a spin with the classic combo of TNT2+Voodoo 2.

I might try out S3 Savage 4 up next.

Oh man, make sure the straps on your gaming chair are tight... don't want to fall off laughing. 🤣

Shponglefan wrote on 2024-11-17, 00:16:

Unfortunately the card now appears to be DOA.

I went to do some more testing and it's not even being detected. I tried on a couple different motherboards, same result. I even tried applying some gentle heat via a hot air gun just in case the issue was thermal contraction/expansion related. But no dice, the card is not booting at all.

I'm guessing this card was on its last legs to begin with...

That looks like the HP OEM Asus P2B board. More than once I've thought a PCI video card was dead because I was testing it on a mainboard that didn't output 3.3v on the PCI slots, both ATI & Nvidia cards and I'm pretty sure the P2B in this picture doesn't output 3.3v to the PCI slots since I had the same issue with my Aopen AX6BC.
That FX5200 card does have the notches for a 3.3v pci card...

Try it in a Pentium 4 system and you might get a different result, then hopefully you could run MATS on it and find out which memory banks / bits are bad, but like others have said it can often be little capacitors on the back that get bashed off and break things. I've fixed more than a couple FX5200 cards just putting caps back in place where they got bashed off from being stored in a drawer.

Thermalwrong wrote on 2024-11-17, 20:38:

That looks like the HP OEM Asus P2B board. More than once I've thought a PCI video card was dead because I was testing it on a mainboard that didn't output 3.3v on the PCI slots, both ATI & Nvidia cards and I'm pretty sure the P2B in this picture doesn't output 3.3v to the PCI slots since I had the same issue with my Aopen AX6BC.
That FX5200 card does have the notches for a 3.3v pci card...

Try it in a Pentium 4 system and you might get a different result, then hopefully you could run MATS on it and find out which memory banks / bits are bad, but like others have said it can often be little capacitors on the back that get bashed off and break things. I've fixed more than a couple FX5200 cards just putting caps back in place where they got bashed off from being stored in a drawer.

It had previously worked in this board. I also tested it in a couple other boards and it doesn't work in any of them. I further tried another working GeForce FX 5200 PCI in the same systems (including the above Asus P2B), and that card did work.

Unfortunately this particular card appears to be dead.